47242 ECONOMIC APING ESH GEOGRAPH Y R in east asia Edited by Yukon Huang and Alessandro Magnoli Bocchi Reshaping Economic Geography in East Asia Reshaping Economic Geography in East Asia Edited by Yukon Huang and Alessandro Magnoli Bocchi THE WORLD BANK Washington, DC © 2009 The International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org E-mail: feedback@worldbank.org All rights reserved 1 2 3 4 5 12 11 10 09 This volume is a product of the staff of the International Bank for Reconstruction and Devel- opment / The World Bank. The findings, interpretations, and conclusions expressed in this volume do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgement on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development / The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; telephone: 978-750-8400; fax: 978-750-4470; Internet: www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2422; e-mail: pubrights@worldbank.org. ISBN-13: 978-0-8213-7641-6 eISBN-13: 978-0-8213-7642-3 DOI: 10.1596/978-0-8213-7641-6 Cover design: Zefyr Design, info@zefyr.co.uk Typesetting: Precision Graphics Library of Congress Cataloging-in-Publication Data has been applied for. Contents Foreword xvii Preface xix Acknowledgments xxv Contributors xxvii Abbreviations xxxi Section I Context and concepts: density, distance, and division 1 1 Regional integration, agglomeration, and income distribution in East Asia 1 Nobuaki Hamaguchi Production networks in East Asia 4 Regional income inequality 11 Discussion 14 Notes 15 References 16 Annex: Data for analysis of the China-Japan border effect 18 2 Geography of cluster-based industrial development 19 Keijiro Otsuka and Tetsushi Sonobe A theory of cluster-based industrial development 20 Evidence from case studies 22 Evidence of changing industrial locations 25 Policy implications 30 Notes 30 References 31 v vi Contents 3 Rural clustering at incipient stages of economic development: hand-weaving clusters in Lao PDR 33 Akihiko Ohno Market segregation at an incipient stage of development 34 Hypotheses on the emergence of traders 35 Profile of the Lao PDR hand-loom industry 36 Marketing and contractual arrangements 37 Weaving clusters and traders 40 Decline of trust 44 Retailers as urban-based traders 45 Conclusions 45 Notes 46 References 46 4 Spatial networks, incentives, and the dynamics of village economies: evidence from Indonesia 48 Futoshi Yamauchi, Megumi Muto, Reno Dewina, and Sony Sumaryanto Data 49 Descriptive analyses 50 Empirical framework 58 Empirical results 59 Policy discussion 63 Conclusions 63 Notes 64 References 64 5 The Iskandar Development Region and Singapore 66 Manu Bhaskaran Background 66 Key factors driving the relationship between the two regions 67 Potential synergies between Singapore and the IDR 71 The way forward 75 Conclusion: can these obstacles be overcome? 78 Notes 78 References 78 6 Spatial integration and human transformations in the Greater Mekong subregion 79 Jonathan Rigg and Chusak Wittayapak The GMS: an idea becomes a subregion 79 Openness, progress, and inequality in the GMS 81 Contents vii Scales and sites: the empirics of spatial transformations in the GMS 83 Agents, agency, and impacts of spatial transformation in the GMS 86 Policies and politics of spatial transformation 92 Notes 95 References 96 Section II Southeast Asia: Vietnam, the Philippines, Indonesia, and Thailand 100 7 Rural development and issues in Vietnam: spatial disparities and some recommendations 100 Dang Kim Son Current spatial disparities and policy issues 100 Disparities between rural and urban areas 101 Regional disparities 103 Causes of disparities 109 Recommendations 112 Notes 113 References 113 8 Economic geography of Indonesia: location, connectivity, and resources 115 Hall Hill, Budy P. Resosudarmo, and Yogi Vidyattama Economic geography 116 Regional economic growth and change 121 Regional inequality and convergence 125 Conclusions 132 Notes 133 References 134 9 Spatial considerations on decentralization and economies of concentration in Indonesia 135 Francisco Javier Arze del Granado Decentralization of expenditures, urban-to-rural migration, and urban density in congested regions 136 Agglomeration economies 143 Room for further decentralization and policy implications 146 Conclusions 148 Notes 149 References 151 Annex: Definition of the variables and estimation results 154 viii Contents 10 Spatial agglomeration, firm productivity, and government policies in Indonesia: concentration and deconcentration in the manufacturing sector 156 Ari Kuncoro Evolution of the manufacturing industry's spatial configuration in Java 156 Choice of firm location 158 Empirical methodology: externalities and firm decentralization 162 Results 164 Conclusions 166 Notes 167 References 168 11 Spatial disparities and development policy in the Philippines 169 Arsenio M. Balisacan, Hal Hill, and Sharon Faye Piza Regional development patterns 171 Determinants of local growth and poverty reduction 177 Conclusions 179 Notes 180 References 181 Annex: Determinants of growth and poverty reduction in the Philippines: descriptive statistics 183 12 Spatial disparities in Thailand: does government policy aggravate or alleviate the problem? 184 Nitinant Wisaweisuan Growth and spatial disparities 185 Explaining spatial disparities 186 Attempts to alleviate the problem 190 Conclusions 194 Notes 194 References 194 Section III Northeast Asia: China and the Republic of Korea 196 13 Reshaping economic geography in China 196 Yukon Huang and Xubei Luo China's historic and geographic legacy 196 Accelerating growth: coordinating structural, incentive, and fiscal policies 198 Contents ix Spatial factors and government policies: growth and equity implications 199 Looking to the future 213 Notes 216 References 216 14 The political economy of government policies toward regional inequality in China 218 Yang Yao Regional disparities in China 218 The path to the uneven development model 223 Preferential government policies and economic geography 226 Recent government initiatives to reduce regional disparities 230 Alternative ways to address regional disparities 235 Conclusions 238 Notes 238 References 238 15 Is China sacrificing growth when balancing interregional and urban-rural development? 241 Zhao Chen and Ming Lu Industrial agglomeration and city development 242 China's urban-rural and interregional development: is there a tradeoff between efficiency and equality? 247 Interregional and urban-rural economic development: policy adjustment and fiscal transfer 250 Conclusions and policy implications 253 Notes 255 References 255 16 Industrial agglomeration and economic performance in transitional China 258 Canfei He Theoretical understanding of industrial agglomeration in China 258 Industrial agglomeration and industrial specialization in China 261 Industrial agglomeration and labor productivity in China 273 Industrial clusters in China: a county-level analysis 277 Conclusions 279 Note 280 References 280 x Contents 17 Capital allocation, regional specialization, and spillover effects in China 282 Chong-En Bai and Xu Lin Returns to capital across provinces 283 Regional specialization 285 Spatial factors behind productivity growth 289 Conclusions 291 Notes 292 References 292 18 Coastal China's urban-rural spatial restructuring under globalization 294 Yue-man Yueng and Jianfa Shen Guangdong and the Pearl River delta 296 Shanghai and the Yangtze River delta 304 The Bohai Bay region 310 Conclusions 315 Notes 317 References 317 19 A history of the Republic of Korea's industrial structural transformation and spatial development 320 Sam Ock Park Industrial policies, structural changes of industry, and spatial transformation 321 Spatial development and change 327 Policy implications 334 Note 336 References 336 Section IV Conclusion: lessons from experience 20 Lessons from experience: reshaping economic geography in East Asia 338 Yukon Huang and Alessandro Magnoli Bocchi Context: the rise of regionalism and the role of production-sharing networks 338 Accompanied by increasing disparities 339 The new economic geography 339 Vietnam and Lao PDR: emerging spatial patterns at low income levels 341 Contents xi Indonesia, the Philippines, and Thailand: diverse settings and varied outcomes in three middle-income ASEAN countries 342 China: agglomeration, rapid growth, and major spatial consequences 346 Korea: from developing to developed status and eventual equalization in living standards 349 Lessons learned 350 Notes 356 References 357 Index 359 Boxes 5.1 Density, distance, and division: Singapore and Johor 67 18.1 Tianjin Binhai New District: the third pole in 6.1 Development through concentration? The Lao PDR China 313 government's focal site strategy 90 19.1 The case of Sunchang 335 Figures 1.1 East Asia's share of intraregional trade, 1980­2004 4 4.4 Change in nonagricultural income share in select villages of 1.2 Composition of intraregional trade in East Asia, by category Indonesia 54 of use, 1980­2004 5 4.5 Impact of change in the proportion of asphalt roads at the 1.3 Regional production shares of information subdistrict level on per capita income growth and change in technology­related goods, 2005 6 nonagricultural income in select villages of Indonesia 55 1.4 Triangular trade in East Asian electronics industry 7 4.6 Impact of years of schooling of household head on per capita income growth and change in nonagricultural 1.5 Framework for fragmentation 8 income in select villages of Indonesia 56 1.6 Modified framework for fragmentation 9 4.7 Self-employment activities, by province-level road density 1.7 Structure of interregional trade between China and in select villages of Indonesia 57 Japan 11 4.8 Self-employment activities, by speed to district center in 1.8 Share of East Asia's population living in agglomerations select villages of Indonesia 58 greater than 750,000 inhabitants, 1950­2000 12 6.1 The Greater Mekong subregion 80 1.9 Rank-size rule of large agglomerations (population over 6.2 International migrant flows in the GMS 91 750,000) in East Asia, 1950 and 2005 13 7.1 Employment in Vietnam, by sector, 1988­2005 102 1.10 Nominal income per capita in select Asian economies, 1990 and 2005 14 7.2 Urban and rural expenditure in Vietnam, 2002­06 102 1.11 Regional income inequality measured using the coefficient of variation for select East Asian countries, 1990­2004 15 7.3 Labor productivity in Vietnam, 1995­2006 104 2.1 Development of the motorcycle industry in Japan, 7.4 Cereal production per capita in Vietnam, by region, 1945­65 23 2006 105 3.1 The marketing process 38 7.5 Distribution of malnourished children under five years of age in Vietnam 105 3.2 Contractual arrangements along a "Make or Buy" spectrum 39 7.6 Source of migrants to Hanoi and Ho Chi Minh City, 3.3 Map of survey sites in Lao PDR 41 1999 107 4.1 Location of surveyed villages in Indonesia 50 7.7 Poverty rate and density in Vietnam 108 4.2 Change in average intervillage road quality (asphalt 7.8 Inequality in Vietnam 109 roads as a proportion of all roads) in select provinces of 7.9 Relation between GDP per capita and agricultural value Indonesia 52 added in GDP in Vietnam 111 4.3 Per capita income growth in select villages of 7.10 Total investment in the economy and in agriculture in Indonesia 54 Vietnam, 1995­2005 112 xii Contents 8.1 Structural change and growth in Indonesia, 13.8 GDP per capita and ratio of central transfers to 1975­2004 125 subnational government expenditure at the provincial 8.2 Initial regional GDP with and without mining and level 2004 205 household consumption expenditure per capita vs. growth 13.9 Subnational government expenditures per capita on in Indonesia, by province, 1975­2004 128 education and health care in coastal and inland provinces of 8.3 Provincial income inequality in Indonesia, China, 1999­2006 205 1975­2003 129 13.10 Relationship between growth in provincial GDP per capita 9.1 Net migration of population in Indonesia, by districts, and adjusted distance from the major coastal commercial 2005 138 areas in China, 1979­2003 206 9.2 Simulation: Urban gravity at alternative distances and levels 13.11 Density of highways in China, 1978, 1999, and 2003 207 of urbanization 139 13.12 Density of railways in China, 1978, 1999, and 2003 208 9.3 Out-migration from large metropolitan districts to the 13.13 Interprovincial migration flows to the coast in China, fringes in Indonesia 140 1995­2000 209 10.1 Java island and its vicinity 158 13.14 Income inequality in China, 1978­2004 210 11.1 Regions of the Philippines, 2007 170 13.15 Income growth in rural and urban areas of China, 1989­2004 212 11.2 Regional growth versus initial regional income 172 13.16 Projected household income inequality in China, 11.3 Variation in provincial prices, 1985­2003 176 1981­2021 216 11.4 Income growth and poverty reduction, Philippine 14.1 Divergence of income among Chinese provinces, provinces, 1985­2003 177 1952­98 219 12.1 GDP per capita in Thailand, 1980­2005 185 14.2 Divergence of income among Chinese provinces, 12.2 Absolute poverty and income distribution in Thailand, 1999­2006 219 1990­2006 185 14.3 Divergence of growth rates among Chinese provinces, 12.3 Regional share of GDP in Thailand, 1990­2005 186 1952­78 220 12.4 Proportion of the poor people in Thailand, by region, 14.4 Divergence of growth rates among Chinese provinces, 1990­2004 186 1978­98 220 12.5 GDP per capita in Thailand, by region, 14.5 Divergence of growth rates among Chinese provinces, 1990­2005 187 1999­2006 220 12.6 Composition of regional GDP in Thailand, by region and 14.6 Per capita GDP and urban-rural divide in China, economic activity, 1981­2005 188 1999 and 2006 222 12.7 Gini coefficient and economic growth in Thailand, 14.7 Interprovincial inequality in China, 1981­2005 189 1952­2004 223 12.8 Allocation of the government budget in Thailand, by region, 14.8 Contribution of urban-rural divide and 2003­06 192 regional divide to interprovincial inequality, 12.9 Allocation of the government budget in Thailand, by 1952­2003 223 activity and region (excluding Bangkok and its vicinities), 15.1 Industrial share in China, by province, 1987, 1995, and 2001­06 193 2005 243 13.1 Regions and open cities in China 197 15.2 Globalization and industrial growth in China, 13.2 GDP growth in China, by region, 1981­2006 198 1987­2005 244 13.3 Ratio of total government revenue and expenditure to GDP 15.3 Urbanization and industrial growth in China, in China, 1980­2005 201 1987­2005 245 13.4 Ratio of central government revenue and expenditure 15.4 Per capita GDP and industrial shares of cities in China, to total government revenue and expenditure in China, 1991­2005 246 1980­2005 202 15.5 Population density and per capita GDP of cities in China, 13.5 Share of total investment in fixed assets in coastal and 1991­2005 247 inland areas of China, 1982­2007 202 15.6 Urban-rural and interregional income disparities in China, 13.6 Ratio of total provincial government expenditure to 1978­2005 248 regional GDP in coastal and inland regions of China, 15.7 Urban-rural income disparities and infringement cases in 1999­2006 203 China, 1981­2004 249 13.7 Central transfers and GDP per capita in China at the 15.8 Share of net fiscal transfers in China from the central provincial level, 2004 204 government, by geographic region, 1998­2004 253 Contents xiii 15.9 Level of economic development and central fiscal transfers, 16.16 Temporal changes of industrial specialization in select 1998­2004 253 western provinces of China, 1980­2004 275 15.10 Changes in the share of agricultural expenditures, 16.17 Temporal changes of industrial specialization in central 1998­2004 254 provinces of China, 1980­2004 275 15.11 Central fiscal transfers and economic 16.18 Spatial distribution of manufacturing employment in growth 254 China, by county, 2004 277 16.1 Geographic concentration of manufacturing industries in 16.19 Spatial distribution of employment in telecommunications China, 1980­2004 262 equipment, computers, and other electronic equipment in 16.2 Geographic agglomeration of industrial output of China, by county, 2004 278 globalized industries in China, 1980­2004 264 17.1 Returns to capital in China, by province, 1978­2005 284 16.3 Provincial distribution of industrial output of highly 17.2 Standard deviation of returns to capital across provinces in globalized industries in China 265 China, 1978­2005 284 16.4 Geographic agglomeration of domestic market­oriented 17.3 Average of Hoover coefficients across regions in China, industries in China, 1980­2004 266 1999­2003 287 16.5 Provincial distribution of industrial output of domestic 17.4 Average (across time) Hoover coefficient in China, by market­oriented industries in China in 2004 267 region 287 16.6 Geographic agglomeration of favored and protected 18.1 Three coastal regions in China 296 industries by local governments in China, 18.2 The Pearl River delta region and its constituent cities 299 1980­2004 268 18.3 Urban cluster-coordinated development plan of the Pearl 16.7 Geographic agglomeration of less-protected and -favored River delta, 2020 301 industries in China, 1980­2004 268 19.1 Administrative divisions of Korea: provinces, mega cities, 16.8 Provincial distribution of favored and protected industries and provincial cities 321 by local governments in China 269 19.2 Four regions, nine provinces, and major cities in 16.9 Provincial distribution of least-favored or -protected Korea 322 industries in China 270 19.3 A brief history of Korean industrial policies 323 16.10 Relations between exports (panel A) and foreign capital 19.4 Inward and outward FDI in Korea, 1981­2006 326 (panel B) and industrial agglomerations in China in 19.5 Share of population in Korea, by region, 1970­2004 328 2004 271 19.6 Entropy index 329 16.11 Relations between ratio of income tax and value-added tax to sales revenue (panel A) and ratio of total profits in 19.7 Annual growth rate of GDP in Korea, 1960­2005 330 sales revenue (panel B) and Gini coefficient of three-digit 19.8 Regional GDP in Korea, by region, 1985­2003 330 industries in 2004 272 19.9 Per capita regional GDP in Korea, by region, 16.12 Relation between ratio of state capital to total capital and 1985­2005 331 Gini coefficient of three-digit industries in 2004 273 20.1 Provincial disparities: human development indexes 16.13 Relationship between industrial agglomeration (weighted Gini in East Asia 353 coefficients across industries) and industrial specialization 20.2 Share of subnational government expenditure in (weighted Gini coefficients across provinces) 273 total government expenditure in East Asia during 16.14 Temporal changes of industrial specialization in centrally the 1990s 355 administered municipalities of China, 1980­2004 274 20.3 Coefficient of variation in provincial per capita 16.15 Temporal changes of industrial specialization in select revenues before and after transfers in select coastal provinces of China, 1980­2004 274 East Asian countries 355 Tables 1.1 Intraregional trade of intermediate goods in East Asia, 1995, 2.2 Major features of the machine tool enterprises in Taichung, 2000, and 2005 5 Taiwan, China 23 1.2 Transactions of semiconductors and integrated circuits in 2.3 Transition to quality improvement: Average enterprise size East Asia, 2000 6 and marketing channels in Wenzhou, China, 1999, 1995, 1.3 Gravity model estimates of China-Japan intraregional and 2000 24 trade 10 2.4 Share of manufacturing employment and annual 2.1 An endogenous model of cluster-based industrial growth rates of employment in Taiwan, China, development 20 by industry 26 xiv Contents 2.5 Changes in employment shares in Taiwan, China, by area 5.5 Competitiveness indicators for Malaysia and and industry, 1976, 1986, and 1996 27 Singapore 75 2.6 Average number of enterprises and workers and 6.1 Openness, progress, and inequality in the GMS, real value of production per sample township in 1990­2006 82 Jiangsu and Zhejiang provinces, China, by location, 1990, 6.2 Incidence of poverty in Lao PDR, by ethnolinguistic family, 1996, and 2002 29 2001 84 2.7 Share of production and employment of the most 6.3 Rural and urban distribution of poverty based on national important industry in the township in Jiangsu and poverty lines, various years 84 Zhejiang provinces, China, by area, 1990, 1996, and 6.4 Annualized growth rates of per capita expenditure and 2002 29 income, by country and quintile 85 3.1 Trust of villagers in various economic agents in Lao 6.5 Expanded trade flows in the Greater Mekong subregion PDR 35 two-way trade 85 3.2 Source of income of rural households in Lao PDR 36 6.6 Effects of improved roads and transport 88 3.3 Post-contractual problems under the different contracting systems 39 6.7 Decline in the availability of nontimber forest products in Ban Nong Hin, Champassak province, Lao PDR, 3.4 Surveyed clusters in Lao PDR 40 1989­99 89 3.5 Social characteristics of the large-scale master weavers 6.8 Distilling the GMS "success factors" 93 surveyed in Lao PDR 42 7.1 GDP growth rate and GDP per capita in Vietnam, by 4.1 Proportion of asphalt roads in intervillage roads in ecological zone 101 Indonesia, 1996 and 2006 51 7.2 Average living expenditure in urban and rural areas of 4.2 Changes in intervillage road quality (asphalt, concrete, or Vietnam, 2006 103 cone block or not) in Indonesia, by province, 1996­2006 52 7.3 Energy in the Vietnamese diet in rural and urban areas, 2000 103 4.3 Distance to economic centers in select villages of Indonesia 53 7.4 Income and health indicators in the plains and highlands of Vietnam, 2004 103 4.4 Descriptive statistics: household income, nonagricultural income share, landholding, and education in select villages 7.5 Living standards in the highlands and plains of Vietnam, of Indonesia 54 2004 104 4.5 Provincial averages: household income, nonagricultural 7.6 Labor and migration situation in Vietnam, 2004 106 income share, landholding, and education in 7.7 Economic growth rate in Vietnam, by province, Indonesia 55 2006 107 4.6 Nonagricultural income share and share of households with 7.8 Poverty in Vietnam, by region, 2004 109 self-employment activity in select villages of Indonesia, by 7.9 Investment in Vietnam, by region 110 distance from economic centers 56 7.10 Accessibility of infrastructure in Vietnam, by region, 4.7 Type of self-employment activities in select villages of 2004 110 Indonesia, by distance to economic centers 57 7.11 Structure of GDP and economic development in Vietnam, 4.8 Change in average road quality and per capita income by region, 2006 111 growth in select villages of Indonesia 60 7.12 Effective rate of protection in Vietnam, by sector, 4.9 Change in average road quality and nonagricultural income 1997­2003 111 share in select villages of Indonesia 61 8.1 Shares of regional GDP with and without mining and 4.10 Change in average road quality and nonfarm self- household consumption expenditure in Indonesia, by employment income share in select villages of province, various years, 1975­2004 118 Indonesia 62 8.2 Regional GDP with and without mining and household 4.11 Summary of parameter signs in select villages of consumption expenditure per capita in Indonesia, by Indonesia 62 province, various years, 1975­2004 119 5.1 Share of Singapore foreign direct investment in Asia, by 8.3 Annual growth rates of regional GDP with and without country, select years, 1996­2005 67 mining and household consumption expenditure per capita 5.2 Comparison of Singapore and the IDR 68 in Indonesia, by province, 1976­2004 122 5.3 Economic growth before and after the Asian financial crisis, 8.4 Social and demographic indicators in Indonesia, by by country, 1991­2006 71 province, 1971 and 2000 126 5.4 Benefits of relocating and undertaking complementary 8.5 Indicators of social vulnerability in Indonesia, by province, activities 73 various years, 1971­2004 127 Contents xv 8.6 Absolute convergence 129 11.4 Population and intraregional migration in the Philippines, 9.1 Urban and rural population and population growth rate in by region, 2000 174 Indonesia, 1960­2005 137 11.5 Infrastructure indicators in the Philippines, by region, 1988 9.2 Determinants of districts' net migration 139 and 2004 or 2005 175 9.3 Population dynamics: public expenditures and net 11.6 Determinants of local growth and poverty reduction in the migration in Indonesia 141 Philippines 178 9.4 Population convergence 142 12.1 Economic indicators in Thailand, 1980­2005 185 9.5 Distribution of population in Indonesia, by district, 12.2 Urban-rural income gap in Thailand measured by per capita province, and island, 1983 and 2005 142 income, 1994­2004 186 9.6 Number of people relocated under Indonesia's 12.3 Per capita regional GDP in Thailand, 1990­2005 187 transmigration program 142 12.4 Gini coefficient of consumption spending in Thailand, by 9.7 Population and annual population growth in Indonesia, region, 1988­2006 190 2000 and 2005 143 12.5 Openness and income distribution in Thailand, 9.8 Composition and concentration of employment, by 1995­2005 190 sector 144 12.6 Net inflows of foreign direct investment in Thailand, 10.1 Comparison of high-income per capita and low-income per 1970­2006 191 capita regions in Java, select years, 1986­2003 157 12.7 Major emphasis of national economic and development 10.2 Ratio of local GDP per capita to Jakarta's GDP in Java, select plans in Thailand 191 years, 1986­2003 158 12.8 Ratio of Q5 to Q1 in Thailand, by region, 10.3 Spatial concentration of manufacturing GDP in Java, select 1994­2004 192 years, 1986­2003 158 12.9 Availability of health care resources in Thailand, by 10.4 Fraction of manufacturing value added in local GDP in region 193 Java, select years, 1986­2003 159 12.10 Proportion of the poor in Thailand, by education of the 10.5 Fraction of villages with paved roads in Java, select years, head of household, 1996­2004 193 1986­2000 159 13.1 GDP growth rates of central and western regions as a 10.6 Concentration of manufacturing firms in Java, select years, percentage of the coastal region in China, 1980­2003 160 1980­2006 212 10.7 Concentration of manufacturing employment in Java, select 14.1 Convergence of growth rates among the three regions in years, 1980­2003 161 China, 1978­2006 221 10.8 District industrial concentration index in Java, select years, 14.2 Comparison of the three richest and the three poorest 1990­2003 161 provinces in China, 2006 222 10.9 Change in the district industrial concentration index in 14.3 Urban-rural and regional divides in China, 2005 222 Java, 1990­2003 162 14.4 Central government capital investment, 10.10 Annual growth of stock of firms and labor employment 1953­2005 227 in large and medium manufacturing enterprises in Java, 14.5 Net transfers received from the central government, 1990­2003 162 1953­2005 228 10.11 Externality and productivity in Java: textiles, garments, 14.6 Preferential policies offered to different zones 229 leather, and footwear, 1990­2003 164 14.7 Policy versus geography: Regression results 230 10.12 Externality and productivity in Java: chemicals, 1990­2003 165 14.8 Comparison of the western region and the country, 2000­05 232 10.13 Externality and productivity in Java: nonmetallic minerals, 1990­2003 165 14.9 Composite tax rates of inland and coastal regions 237 10.14 Externality and productivity in Java: machinery, 14.10 Marginal contribution of public investments to regional 1990­2003 165 income inequality 237 10.15 Test of dynamic externalities in Java, 1990­95 166 15.1 Historical development of opening areas 251 11.1 Regional growth and structure in the Philippines, by region, 16.1 Gini coefficient of manufacturing industries in China in 1975­2005 171 select years, 1980­2004 263 11.2 Key economic indicators in the Philippines, by region, 1988 16.2 Regression analysis of the relationship between productivity and 2003 172 and industrial agglomeration for two-digit industries in 11.3 Social indicators in the Philippines, by region, 1988 and China, 1980­93 276 2003 173 17.1 Variable mean and rank 288 xvi Contents 17.2 Pair-wise correlations between main variables 288 19.1 Structure of production in Korea, by sector, 17.3 Estimation results with dependent variable: 1960­2005 323 LogitHoover 288 19.2 Share of manufacturing industry in Korea, by type of firm, 17.4 Estimation results with dependent variable ln(TFP) 291 1981­2005 324 19.3 Share of manufacturing employment in Korea, by region, 18.1 Demographic and GDP indicators of Guangzhou, 1963­2005 325 Shenzhen, Shanghai, Beijing, and China, 1980­2005 297 19.4 Share of population in Korea, by region, 18.2 Share of secondary and tertiary industries in GDP, exports, 1970­2005 329 and realized foreign capital in Guangzhou, Shenzhen, Shanghai, Beijing, and China, 1980­2005 298 19.5 Share of headquarters and plants of top 100 firms in Korea, by region, 2002 and 2007 332 18.3 Distribution of population in Shanghai, 2006 307 19.6 Share of industrial patents in Korea, by region, 1995, 2000, 18.4 Industrial distribution in Shanghai, 2006 308 and 2005 332 18.5 Population and GDP density in the Yangtze River delta 19.7 Regional GDP per capita in the capital region of Korea, region, 2005 309 1985­2005 333 18.6 Comparison of the three coastal regions in China, 19.8 Share of dot-kr domains in Korea, by region, 2005 311 2001­07 334 18.7 Major indicators of the JJJ city-region, 2005 312 20.1 Key indicators for growth, urbanization, and income 18.8 Employment in three cities of China, 2005 314 distribution in East Asia, by country, various years 339 Foreword I am pleased to be associated with this collection of studies, a companion volume to the World Development Report 2009 (WDR 2009). The WDR 2009 provides a comprehensive overview, from a global perspective, of the importance of economic geography. In turn, this volume focuses on East Asia and the role of economic geography in shaping its development. As a member of the advisory panel for the WDR 2009, and having spent much of my career pio- neering the principles that underpin the "new economic geography" or spatial economics, it is gratifying to see how these principles are now being used to deepen our understanding of the most dynamic region of the world and its development process. An earlier volume of studies on East Asia that I edited focused on how spatial factors influ- enced both the process of regional integration and the location of production across coun- tries (Fujita 2007). These outcomes reflect the tenets of the new economic geography: given first-nature factors and natural conditions, agglomeration forces lead to "clustering" (that is, the concentration of economic activity among firms) and give rise to scale economies. Once clustering takes off, proximity to markets reduces transport costs, and "agglomeration econo- mies" encourage more and more firms to congregate. In this process, economies of scale and transport costs are important factors in shaping the size and nature of both cities and produc- tion centers. These principles help to explain the so-called "flying geese model" of regional development that characterized East Asia from the 1970s to the early 1990s. In that model, Japan played the role of "lead goose"; over time, it shed, with a cascading effect, industrial activities in which it no longer had a comparative advantage to lower-income countries--first to the newly industrializing economies, then to several Association of South East Asian Nations (ASEAN) countries, and finally to China. This process was supported by the emergence of agglomeration economies in several "core" metropolitan areas of Japan, while complementary production centers in Asia's mega cities represented the "periphery." However, over the past decade and a half, East Asia has moved from a mono-polar sys- tem, dominated by Japan, to a multi-core system, which is the outgrowth of greater regional integration, driven by specialization and scale economies, interindustry relocation, and trade-related production-sharing networks and by more diverse patterns of industrialization within the same country. Integral to this phenomenon have been the growing importance of innovation in driving economic growth and the emergence of China as the major assembly plant for the region (Gill and Kharas 2007). Also, despite the resounding success of East Asia, within countries there are signs of great stress, such as rising internal disparities, urban congestion, and environmental degradation. The "flying geese model" does not fit well with this recent experience. The new economic geography, however, allows for such variation, and while it does not explicitly address issues of income distribution, there is no doubt that the existence or lack of scale economies--and the manner in which they are being exploited--greatly affect the creation and distribution of wealth. For example, the density of economic activity and the greater compensation being xvii xviii Foreword given to those with technology-related skills have strong distributional impacts. Hence, the rural and urban divide is not the only dimension to create income disparities. This collection of studies emphasizes the process of domestic integration: in East Asia, the continuing pressures for sustaining growth are bringing about a growing divergence in incomes between lagging and leading regions and urban and rural areas. In particular, these studies provide two perspectives. First, they describe the broad spatial transformations, as measured by trends in income, industrial output, population movements, and social indica- tors. Second, they assess how government policy, at the local, regional, and national levels, has affected the pattern of spatial development, especially its impact on growth and equity objectives. For low-income countries such as Lao PDR and Vietnam, the concentration of economic activity is just beginning; there, economic growth is shaped primarily by how well the main agglomerations of population are connected to the major commercial centers. In larger, more diverse, and urbanized countries such as Indonesia, the Philippines, and Thailand, the concentration of economic activity around large metropolitan centers is well entrenched; however, there continue to be concerns about the extent of the agglomeration benefits and whether policies can deal adequately with long-standing spatial disparities. China is perhaps the most notable example of how a large and spatially diverse country has managed to reshape its economic geography to achieve rapid growth, but challenges remain because of the sharply rising spatial disparities. In the Republic of Korea, however, the move from a developing to a developed country led to a gradual convergence in living standards across regions and eventually to broad equality in per capita incomes. I have characterized East Asia before as the region of ultimate diversity: in incomes, lan- guage, culture, and inherited conditions. This is exemplified in the wide range of outcomes of these studies, which reflect each country's stage of development, historic precedents, and differing policy approaches. Still, these studies illustrate how well the concepts underpin- ning the new economic geography can explain what is happening in East Asia. Despite a few exceptions, urbanization and related agglomeration benefits are part of the region's success story, along with the related pressures on policy makers to deal with increasing spatial dis- parities. The challenge is how best to sustain efficient growth processes, while ensuring over time that even though incomes may diverge in the early stages of development, with good policies, living standards can--and will--converge. Masahisa Fujita References Fujita, Masahisa, ed. 2007. Regional Integration in East Asia: From the Viewpoint of Spatial Economics. London: Palgrave Macmillan. Gill, Indermit, and Homi Kharas. 2007. An East Asian Renaissance: Ideas for Economic Growth. Wash- ington, DC: World Bank. Preface Reshaping Economic Geography in East Asia illustrates how extensively spatial factors have influenced and informed by growth and development in the region. This study was conceived as a companion volume to and informed by the World Development Report 2009: Reshaping Economic Geography (WDR 2009). By providing case studies and illustrative examples and by deepening our understanding of the forces of economic geography in the East Asia region, this work helped to substantiate some of the key concepts in the WDR 2009. There is full con- sistency in terms of the analytical framework used and broad agreement on how economic geography has influenced growth trends across a diverse range of countries. As in the WDR 2009, the underlying framework for this volume originates with the body of thought encompassed in the "new economic geography." The new economic geography explains how spatial factors affect production and, in turn, trade and growth patterns (Gill and Kharas 2007). It all starts with the choices that firms make about the location of their activity--that is, the desire to concentrate production in one location so as to enjoy plant-level economies of scale and to be near customers and suppliers to reduce transport costs. Once a market has reached a certain scale, other firms are encouraged to locate there to take advan- tage of market size, thereby giving rise to "agglomeration economies"--in other words, the advantages of coalescing geographically. But history also matters. Normally, the existence of a large manufacturing sector gives suppliers an incentive to locate in major commercial centers to take advantage of the larger market and greater potential access, reinforcing the original advantages. Factors of produc- tion, however, especially labor, are not mobile between countries in the same way that they are mobile within countries; thus cost structures may drive firms from larger, higher-wage areas to smaller, lower-wage areas both within and across countries. The lower the transport costs that firms face, the less likely firms are to congregate in one major center or in one country. Changing spatial and production patterns can adversely affect income distribution. Econo- mies of scale are likely to exist in manufacturing and in urban areas but are largely absent in agriculture and in rural areas. If this is true, then it provides an explanation for the persistence of urban-rural wage differences as well as the widening gaps between leading and lagging areas. Indeed, economies of scale may also result in a premium for skilled relative to unskilled workers and thus explain the widening income gaps in relatively open and rapidly growing economies, where innovation is likely to be an important factor shaping growth. This is likely to be the case in the middle-income economies of East Asia today. As discussed in the WDR 2009, as a country or a region develops, both people and firms tend to concentrate in cities and prosperous areas. The resulting disparities in living standards across space underlie the unevenness of the growth process. While location matters at all stages of development, it matters less in a rich country than in a poor one. While WDR 2009 looks at this issue at three spatial levels--at the international level between countries, at the country level, and at the local level--this volume is largely about country experiences and, within xix xx Preface countries, about lagging versus leading regions and cities versus rural areas. This collection of studies focuses on how market forces and government policies influence the concentration of economic activities and the convergence of living standards across locations. In this volume, the three dimensions of development highlighted in the WDR 2009--density, distance, and division--provide a conceptual framework for the geographic transformations. · Density refers to the geographic "compactness" of economic activity and is measured by the level of output produced, or income generated, per unit of land area. It is thus highly correlated with both employment and population density. This is the defining charac- teristic of urbanization, and historically there is a very strong correlation between rising concentration of activity and level of development. · Distance refers to the ease or difficulty of moving goods and factors of production between two locations. In this sense, distance is a physical as well as an economic con- cept. It is not just related to the physical distance between two locations but is also influ- enced by factors such as the quality and mode of transport and regulatory barriers. The main point is that locations closer to markets have an advantage. While the easiest way to reduce distance is through labor migration, improved connectivity through better trans- port and communications is also essential. · Division refers mainly to the negative consequences of political borders separating coun- tries and thus restricting the flow of goods, people, services, and ideas. However, division can also mean boundaries and regulatory barriers within countries, which impede the internal movement of factors of production. In our studies, the focus is largely on inter- nal divisions within countries, although two studies discuss how cross-border issues can affect economic and social developments across countries. For policy makers, the challenge is getting density right: fostering the appropriate con- centration of economic activities to realize the potential benefits of agglomeration econo- mies. If done well, convergence of living standards between lagging and leading areas and urban and rural areas will occur over time. But the distance between concentrated activities is also an important factor, which can be addressed both by favoring the mobility of labor and by reducing transport costs with infrastructure investment. In the process, any artifi- cial or politically driven division--due to jurisdictional boundaries, ethnicity, language, or religion--can divide people, hamper economic activities, impede growth, and exacerbate social development. Reshaping Economic Geography in East Asia also highlights the dramatic urbanization process under way in the region, evidenced by the number of globally recognized "mega cities" (those with populations over 10 million). Seven of the world's 21 mega cities are in East Asia. Cities in East Asia generate about three-quarters of annual output and between a half and two-thirds of exports. Often, much of this is concentrated in one major city: Bangkok. Bangkok accounts for 40 percent of Thailand's gross domestic product (GDP) and Manila for 30 percent. Other major centers such as Guangzhou, Jakarta, Seoul, Shanghai, and Tokyo are seen as driving their economies. East Asian cities have been able to deliver the agglomeration benefits required for growth and are now exceptionally well connected to the global economy. The region, excluding Japan, is home to 16 of the largest 25 seaports in the world and 14 of the largest 25 container ports. Without this improved connectivity, the region's rapid expansion in trade volumes would not have been possible. This collection of studies is organized in four sections. The first section comprises chapters dealing with the "context and concepts" for this volume. The second deals with Southeast Asia, specifically, the Association of South East Asian Nations (ASEAN) countries, and the third deals with Northeast Asia: China and the Republic of Korea. The fourth section provides an overview of lessons learned. The time perspective for most of the studies spans several decades; in many cases, outcomes and policies can be traced back half a century or more. Preface xxi Leading off section 1, Nobuaki Hamaguchi provides the regional context in his description of the evolution of East Asia's growth and trade patterns over the past several decades. He sur- veys the origins of East Asia's success story as depicted first by the "flying geese model," then by "the East Asian miracle," and, more recently, by the "East Asian renaissance," which highlights the rise of regional trade and financial networks and the emergence of China as an important economic player (Gill and Kharas 2007). A common theme is the role of agglomeration eco- nomies, the pattern of regional development, and the implications for income distribution at the regional and national levels. The other papers in section 1 illustrate particular aspects of the three dimensions of the spatial transformations for economic development: density, distance, and division. In chapter 2, Keijiro Otsuka and Tetsushi Sonobe explore an important facet of density, the agglomeration phenomenon: how do industrial clusters form, and what factors influence the degree and success of the processes involved? Drawing on several country case studies, they examine the pattern of cluster-based industrial development to identify key determi- nants, notably the role played by innovation. In chapter 3, Akihiko Ohno's study of Lao PDR illustrates some of the complexities that arise when both density and distance affect small, landlocked, low-income countries, where the challenge is connecting isolated clusters of rural artisans (producers) to national and global markets. His solution is the emergence of a special type of trader--the culture broker--who can gain the trust of local artisans but also under- stands the demands of the external market and thus can overcome the distance factor that hampers development. In chapter 4, Futoshi Yamauchi, Megumi Muto, Reno Dewina, and Sony Sumaryanto deal with the "distance" factor, explaining how spatial connectivity through improvements in the quality of roads relative to distance to commercial centers can affect incomes in rural vil- lages in Indonesia. Aside from investment in transport infrastructure, the chapter argues that improved education is also important. In chapter 5, Manu Bhaskaran draws attention to the negative consequences of divisions, highlighting how the largely artificial borders that separate contiguous nations can discour- age factor mobility and distort investment choices. In this case, a more coordinated and free flow of labor and investment between Singapore and the southern tip of Malaysia--known as the Iskandar Development Region--could, in theory, convey substantial benefits to both countries. But, not unexpected, overcoming political sensitivities may be the real challenge. In chapter 6, Jonathan Rigg and Chusak Wittayapak illustrate another aspect of the "divi- sion" problem. The Greater Mekong subregion straddles six countries, with areas inhabited largely by relatively poor, ethnic minorities and logistically difficult to access but potentially rich in resources. As documented in their chapter, stronger regional cooperation supported by major infrastructure investments is vital to overcoming international barriers, but the poten- tial for social disruptions and the need to compensate losers in the process also need to be considered. Section 2 deals with the larger countries in Southeast Asia: Indonesia, the Philippines, Thai- land, and Vietnam. Section 3 deals with Northeast Asia: China and Korea. These country case studies provide a broad historic overview of how economic geography has been shaping devel- opment. The outcome is reflected in the spatial patterns of production, growth, population settlements, and, in many cases, spatially differentiated measures relating to income and access to social services. In all countries, governments have sought to influence the process, and, as proposed in the framework of the WDR 2009, these policies reflect the nature of institutions in providing basic services and influencing market forces, the significance of connective infra- structure, and the impact of spatially targeted incentives. Section 2 begins with chapter 7, by Dang Kim Son, which documents how rapid growth in a largely agrarian, low-income economy such as Vietnam has created regional disparities. Although all regions have benefited from the country's recent successes, given Vietnam's stage of development, high priority has been--and should continue to be--given to policies designed to improve connectivity across regions and raise rural productivity. xxii Preface The case of Indonesia's spatial development does not lend itself to easy caricature. In chapter 8, Hall Hill, Budy P. Resosudarmo, and Yogi Vidyattama provide a comprehensive view of the relationship between location and the patterns of growth, population movement, and dis- parities over the past several decades. Although Jakarta has benefited from being the capital and hence has better connectivity globally, some resource-rich areas have also flourished over time, but more episodically. Indeed, there is no clear natural resource advantage, and Jakarta continues to perform relatively better than other areas. There is a great diversity in social and economic outcomes, and large differences in living standards exist across the major island groups, but progress has been remarkably even overall. Conflict, as expected, is a negative factor. Not explicitly addressed in this or the other Indonesian chapters is whether a large population spread across many islands puts pressures on maintaining more diverse and self- sufficient productive capacities than would be the case otherwise. In chapter 9, Francisco Javier Arze del Granado takes a more in-depth view of how recent fiscal decentralization efforts have affected the spatial pattern of development in Indonesia. It is perhaps too early to reach definitive conclusions, but the initial results suggest that decentralization may be an effective means to improve local service delivery and discourage migration, but it has not made much of a difference in terms of reshaping the location of industrial activity. In chapter 10, Ari Kuncoro analyzes the characteristics of Indonesian industries that the government has encouraged to relocate to less congested areas, to see if agglomeration benefits are derived from localization factors that would favor smaller urban centers--or from urbanization factors that require larger and more diverse urban settings. He finds that, depending on the nature of the indus- trial subsector, relocating industries to smaller cities can be consistent with market solutions, especially if accompanied by investments that improve transport connectivity. The cases of the Philippines and Thailand illustrate the importance of economic geography in shaping developmental outcomes, especially the regional dimension of poverty. In chapter 11, Arsenio M. Balisacan, Hal Hill, and Sharon Faye Pizasurvey the relatively disappointing growth of the Philippines over the past decades. Whether these outcomes reflect a spatial misallocation of public expenditures or a failure to unify national markets and encourage more factor mobil- ity is not yet clear. However, the glaring regional disparities and concentrated pockets of poverty suggest that a sharp increase in infrastructure investments to improve connectivity between the lagging and leading areas is long over due. In chapter 12, Nitinant Wisaweisuan studies Thailand's experience, indicating that a quarter century of rapid growth has resulted in a sharp decline in the number of poor people. Nevertheless, significant regional disparities continue, although some regions appear to be catching up with the more densely settled Greater Bangkok area, in part because of industrialization dynamics linked to globalization. Section 3 on China and the Republic of Korea begins with chapter 13, and an overview by Yukon Huang and Xubei Luo, of how China has reshaped its economic geography over a quar- ter of a century to create a more competitive productive structure and, in the process, has expe- rienced major regional disparities. Drawing on the other China studies, the authors analyze the spatial impact of incentive and fiscal policies on investment patterns, industrial agglom- eration, and regional growth trends. These policies have had a major distributional impact, as reflected in regional and rural-urban indicators of income and access to social services. In chapter 14, Yang Yao provides a broad sweep of how government policies over the past half century have shifted from decentralization to recentralization in response to historic events. He then relates these shifts to growth and distributional outcomes from a spatial perspective. His analysis provides revealing insights on the factors that have shaped government policies. In chapter 15, Zhao Chen and Ming Lu discuss how agglomeration economies have affected provincial growth rates and, in turn, given rise to rural-urban disparities. They note that more balanced interregional and urban-rural development is needed and that such policies will not necessarily compromise efficiency considerations. In chapter 16, Canfei He provides an in-depth analysis of the impact of agglomeration economies on the geographic clustering of industries. Using a comprehensive database, he shows how the location of industrial activity has shifted over time, with more agglomerated industries locating along the coast and more Preface xxiii localized, resource-based, or domestic market­oriented industries expanding in the interior provinces. His findings are reinforced in chapter 17, in which Chong-En Bai and Xu Lin con- clude that the spatial pattern of firm-level specialization and competitive pressures are leading to a more efficient industrial structure in China. In chapter 18, Yue-man Yueng and Jianfa Shen show how the three major growth centers of Guangzhou, Shanghai, and Beijing and their related subregions (Pearl River delta, Yangtze River delta, and the Bohai Bay area) have catalyzed China's modernization. In the process, this has triggered rapid urban restructuring and generated major spillover benefits to the surrounding secondary cities and rural areas. The last chapter in this section (chapter 19), by Sam Ock Park, covers the experience of the Republic of Korea. With a per capita income far exceeding that of other countries (exclud- ing Japan and Singapore), Korea illustrates the other end of the spectrum: how a successful country can transition from developing to developed status and from a low to a high degree of urbanization and, in the process, move from spatial disparity and bipolar concentration of industries to a new spatial division of labor between the core (Seoul) and the rest of the country. In the process, the spatial disparity of incomes and access to social services has been considerably reduced over time. No region is as diverse as East Asia, endowed with a wide range of incomes and levels of development and with differing cultural and historical precedents. Section 4 contains a con- cluding chapter by Yukon Huang and Alessandro Magnoli Bocchi, which summarizes the diversity of country experiences, highlighting how the principles underpinning the new eco- nomic geography explain remarkably well the developmental experience of most East Asian economies. The "take-home" message is the following: independent of their particular stage in the development process, countries should work with--rather than try to counter--the forces of geography in shaping economic growth and development. Even under ideal circum- stances, spatial disparities can be evidence of success rather than failure. But these processes take decades, if not generations, to work through. Policies that focus on making the growth dynamics more inclusive rather than more "balanced" will ultimately lead to outcomes that result not only in higher levels of income but also in more equitable living standards. Yukon Huang and Alessandro Magnoli Bocchi References Gill, Indermit, and Homi Kharas. 2007. An East Asian Renaissance: Ideas for Economic Growth. Wash- ington, DC: World Bank. World Bank. 1993. The East Asian Miracle: Economic Growth and Public Policy. New York: Oxford Uni- versity Press. Acknowledgments This volume is the product of a strong collaborative effort between the World Bank and many scholars and institutions in East Asia. We are especially indebted to Professor Masahisa Fujita for his advice and guidance. We also wish to express our deep appreciation for the personal support we received from Indermit Gill, director of the World Development Report 2009, for conceiving and initiating work on this volume. His intellectual involvement has been critical to the success of this undertaking. Our work benefited greatly from the assis- tance of Chorching Goh, Doris Chung, Patricia Katayama, Rick Ludwick, Maria Hazel Macadangdang, and Ofelia Valladolid in the World Bank's Washington office and that of Mika Iwasaki and Yoshiko Maruyama, who, together with others, provided support for two workshops held in the World Bank's Tokyo office. Many staff in the various World Bank country offices throughout East Asia also provided assistance. Elizabeth Forsyth provided excellent editorial services. Finally, we wish to express our thanks to the Japanese Govern- ment's Ministry of Finance, whose Policy and Human Resources Development grant made this volume a reality. xxv Contributors Javier Arze del Granado is with the International Monetary Fund (IMF). He graduated from Universidad Mayor de San Simón in Bolivia and has a doctorate from Georgia State University. Prior to joining the IMF in 2008, he was with the World Bank's office in Jarkarta. His research interests include fiscal policies, public expenditures, and decentralization. Chong-en Bai is the Mansfield Freeman Chair Professor and chairman of the Economics Department at Tsinghua University in China. He did his graduate work at Harvard in eco- nomics and at the University of California at San Diego in mathematics. He is on the editorial or advisory boards of several journals and professional associations. His research interests are in the fields of economic organizations and incentives, corporate governance and public economics, economics of transition and development, industrial economics, and the Chinese economy. Arsenio M. Balisacan is director of the Southeast Asian Regional Centre for Graduate Study and Research in Agriculture (SEARCA) of the Southeast Asian Ministers of Education Orga- nization (SEAMEO). He is concurrently a professor of economics at the University of the Philippines, Diliman. He has held leadership positions in many professional associations, both national and international. Prior to his appointment at SEARCA in 2003, he served as agricul- ture undersecretary (deputy minister) in the Philippine government. His research interests are poverty alleviation, inequality, food security, agricultural and rural development, globaliza- tion, and political economy of policy reforms. Manu Bhaskaranis adjunct senior research fellow, Institute of Policy Studies, Singapore. He is a partner and member of the board of the Centennial Group, a strategic advisory firm based in Washington, DC. He coordinates the Asian business of the Centennial Group and heads the group's economic research practice, which provides in-depth analysis of macro trends in Asia for investment institutions, government agencies, and companies with interests in Asia. He is a chartered financial analyst and holds degrees from Cambridge University and the Kennedy School of Government. Zhao Chen is professor at the China Center for Economic Studies, Fudan Univesity, China, and director of the Fudan Institute for Industrial Development. He has bachelor's and doctoral degrees from Fudan University. His research interests include the Chinese economy, industrial economics, and information and incentive theory. Reno Dewina is a research analyst at the International Food Policy Research Institute (IFPRI), Washington, DC. She has a master's degree in economic policy management from Columbia University, New York, and a bachelor's degree in sociology from the University of Indonesia. Her research interests include economic development, poverty dynamics, small- holders' participation in high-value agriculture, and agricultural marketing. Masahisa Fujita is currently a professor of economics at Konan University and president of the Research Institute of Economy, Trade, and Industry. He has been a major contributor to spatial economic theory during his 20-year tenure at the University of Pennsylvania and xxvii xxviii Contributors subsequently at Kyoto University, where he served as director of the Institute of Economic Research. He has also served as president of the Institute of Developing Economies, Japan External Trade Organization (JETRO). His scholarship relates to regional science, location theory, economic geography, urban economics, and spatial economics. He is the author or co-author of several authoritative books on economic geography and urban economics. The Spatial Economy (co-authored with Paul Krugman and Anthony J. Venables) defines the field of new economic geography and economics of agglomeration. He is the recipient of the 1983 Tord Palander Prize, the 1998 Walter Isard Award in regional science, the 2000 Nikkei Eco- nomic Book Prize, and the 2002 William Alonso Prize in regional science. Nobuaki Hamaguchi is an associate professor, in the Research Institute for Economics and Business Administration, Kobe University, Japan. He graduated from the Osaka University of Foreign Studies and has a doctorate from the University of Pennsylvania. His research interests include the economic development of Latin America and East Asia, regional economics, and international trade and investment. Canfei He is an associate professor in the Department of Urban and Regional Planning, Peking University, China. He graduated from Beijing Normal University and has a doctorate from Arizona State University. His research interests include industrial clusters and agglomer- ation, foreign direct investment and trade, locational analysis of industries, urban and regional economic development, and human migration. Hal Hill is the H. W. Arndt Professor of Southeast Asian Economies in the Research School of Pacific and Asian Studies, Australian National University. He is on the editorial board of a number of academic journals and has been associated with a variety of institutions in the course of his career. His general research interests are the Southeast Asian economies, espe- cially Indonesia and the Philippines, industrialization and foreign investment in East Asia, and Australia's economic relations with the Asia-Pacific region. Yukon Huangwas the World Bank's country director for China from 1997 to 2004 and, prior to that, the Bank's director for Russia and Central Asia. He graduated from Yale University and has a doctorate from Princeton University. Prior to joining the World Bank, he was a professor at the UniversityofVirginiaandalsotaughtattheUniversityofMalaysiaandUniversityofDaresSalaam. His expertise is in the area of development strategies, rural development, and fiscal policies. Ari Kuncoro is a research associate at the Institute of Economic and Social Research, Uni- versity of Indonesia. He graduated from the University of Indonesia and has a doctorate from Brown University. His research interests include industrial economics, competition and pro- ductivity, governance, and the Indonesian economy. Xu Lin is an assistant professor in the Department of Economics, Tsinghua University, China. She graduated from Renmin University and has a doctorate from Ohio State University. Her research interests include econometrics, health, and labor economics. Ming Lu is a professor in the Department of Economics and the Employment and Social Security Research Center, and director of the Center for China Development and Policy Studies, Fudan University, China. His bachelor's and doctoral degrees are from Fudan Uni- versity. His research interests are in the areas of labor economics, development, and applied microeconomics. Xubei Luo is an economist in the East Asia and Pacific Region of the World Bank. She entered the World Bank after graduating from Sun Yat-sen University in China; she received her doctorate from the University of Auvergne in France. Her research interests cover a variety of topics in economic development, including poverty and inequality, spatial factors affecting development, and economic integration. Alessandro Magnoli Bocchi is an economist at the World Bank. Previously, he was a research associate at Harvard University and an economist at the Inter-American Devel- opment Bank (IADB). His main interests include economic growth, fiscal policy, public finance, and welfare theory. He has published a number of articles and books. In 2003 he authored What Do You Mean?, a book on efficiency and equity in social services delivery Contributors xxix published by the IADB and Johns Hopkins University Press. In 2007 he co-authored The World Bank, a book published in Italian by Il Mulino. Megumi Mutois a senior economist and director at the Japan Bank for International Coop- eration Institute (JBICI). She received her master's degree in public affairs from the Woodrow Wilson School, Princeton University, and is currently a doctoral candidate at the National Graduate Institute for Policy Studies in Japan. Before joining JBICI in 2005, Ms. Muto's tasks included evaluation of JBIC's medium-term strategy and assessment of JBIC's contribution to the Millennium Development Goals. Her research interests include the effects of physi- cal infrastructure on poverty dynamics, market integration, human resource outcomes, and agglomeration of industries. Akihiko Ohnois a professor at the School of International Politics, Economics, and Commu- nication, Aoyama Gakuin University, Japan. He graduated from the Hitotsubashi University Graduate School of Economics. His research interests include development issues relating to rural development, technology, and market integration. Keijiro Otuskais a professorial fellow at the Foundation for Advanced Studies International Development (FASID) and director of the Graduate Research Institute for Policy Studies (GRIPS) and FASID Joint Graduate Program, Japan. He has degrees from Hokkaido University and the Tokyo Metropolitan University and a doctorate from the University of Chicago. He is on the editorial or advisory boards of several journals and associations. His research interests are in the areas of industrial and agriculture issues in developing countries, including technol- ogy, productivity, incentives, and resource use. Sam Ock Park has served as chairman of the Department of Geography and dean of the College of Social Sciences, Seoul National University, the Republic of Korea. He has headed or been a member of many academic associations and on the editorial panels of academic jour- nals. He has published extensively on issues relating to economic geography, innovation and technology, globalization, and economic development of Korea. Sharon Faye Piza is a senior research associate at the Asia-Pacific Policy Center, the Philip- pines. Her education has been at the University of the Philippines. Her research interests are regional development, poverty, migration, and agriculture. Budy P. Resosudarmo is a fellow at the Research School of Pacific and Asian Studies, Aus- tralian National University. He graduated from Bandung Institute of Technology in Indonesia and has a doctorate from Cornell University. His main research interests are in environmental and resource issues as they relate to economic development, decentralization, governance, and the Indonesia economy. Jonathan Rigg is a professor and deputy head of the Department of Geography, University of Durham in the United Kingdom. His graduate education was at the London School of Oriental and African Studies. His research interests are broad and include problems, tensions, and potentialities of development in the Southeast Asia region. Some of the topics addressed in recent publications include agriculture development, ethnicity, environmental implications, and rural transformation. Jianfa Shen is a professor in the Department of Geography and Resource Management in the Chinese University of Hong Kong and co-director of Urban and Regional Development in the Pacific Asia Program, Hong Kong Institute of Asia-Pacific Studies. He is a member of several editorial boards. He obtained his doctoral degree in geography from the London School of Economics in 1994. His research interests include urbanization, migration, and urban and regional development in China. Dang Kim Son is general director of the Institute of Policy and Strategy for Agriculture and Rural Development in Vietnam. He graduated from Hanoi Agricultural University and has a master's degree from the Food Research Institute at Stanford University and a doctoral degree from Vietnam Agricultural Science Institute. He has been a member of or led many expert groups dealing with economic issues for the Vietnamese government. His research interests are in agricultural development and poverty reduction in Vietnam. xxx Contributors Tetsushi Sonobe is a professorial fellow with the Foundation for Advanced Studies on Inter- national Development (FASID) and the National Graduate Research Institute for Policy Stud- ies (GRIPS) Joint Graduate Program in Tokyo. He has a bachelor's degree from the University of Tokyo and a doctoral degree from Yale University. His research interests include industrial economics, economic geography, and development. Sony Sumaryanto is a researcher at the Indonesian Center for Agriculture Socio-Economic Policy Studies (ICASEPS), the Indonesian Agency for Agricultural Research and Development, Bogor. He has a doctoral degree in agricultural economics from Bogor Agricultural University. His research interests include agricultural and rural dynamics, irrigation management, and poverty dynamics. Yogi Vidyattama is a senior research officer at the National Centre for Social and Economic Modelling (NATSEM), the a research center at the University of Canberra. He graduated from University of Indonesia and is completing his doctoral thesis at the Australian National Uni- versity on Indonesia's provincial economic growth. Other research interests include develop- ment and regional economics, migration, urban planning, and public finance. Nitinant Wisaweisuan is an assistant professor and vice dean for international affairs and external relations, Faculty of Economics, Thammasat University in Thailand. Her research interests are urbanization, land and housing policies, economic integration, and trade issues in Thailand. Chusak Wittayapak is an assistant professor, Department of Geography, Faculty of Social Sciences, Chiang Mai University, Thailand. He graduated from Chiang Mai University and has a doctoral degree from the University of Victoria in Canada. His research interests focus on the history and socioeconomic geography of Thailand. Futoshi Yamauchi is a research fellow at the International Food Policy Research Institute (IFPRI), Washington, DC. He graduated from Hitotsubashi University, Japan, and received a doctorate in economics from the University of Pennsylvania. His fields of research are devel- opment economics, labor economics, the economics of population and health, and applied microeconomics, with recent interests in the dynamic formation of human capital, poverty dynamics, social protection, and HIV/AIDS. His expertise include the countries Indonesia, the Philippines, Thailand, Kenya, Malawi, and South Africa. Yang Yaois a professor of economics and deputy director of the China Center for Economic Research, Beijing University, China. He graduated from Beijing University and has a doctoral degree from the University of Wisconsin, Madison. He is a member of several editorial and advisory boards for academic journals and associations. His research interests focus on eco- nomic development and institutional change in China. Yue-man Yueng is emeritus professor of geography, University of Hong Kong, China. During his career, he has been director of several institutes at the University of Hong Kong and a member of numerous advisory and editorial boards for academic journals and asso- ciations. He holds degrees from the University of Hong Kong and Western Ontario and a doctorate from the University of Chicago. His publications include a wide range of books and articles dealing with globalization, housing, economic geography, urbanization, and China's development. Abbreviations ADB Asian Development Bank IMDG Infrastructure on Millennium Development AFTA ASEAN Free Trade Area Goals, Indonesia APEC Asia-Pacific Economic Cooperation IMF International Monetary Fund ARMM Autonomous Region of Muslim Mindanao, INPRES social capital program, Indonesia the Philippines IT information technology ASEAN Association of South East Asian Nations JABOTABEK BOTABEK plus Jakarta, Indonesia BOTABEK Bogor, Tangerang, and Bekasi, Indonesia JBIC Japan Bank for International Cooperation BOTADEBEK Bogor, Tangeran, Depok, and Bekasi, JBICI Japan Bank for International Cooperation Indonesia Institute B2B business to business JETRO Japan External Trade Organization B2C business to consumer JJJ Jing-Jin-Ji, China CAR Cordillera Administrative Region, the LADB Inter-American Development Bank Philippines MAR Marshall-Arrow-Romer CARP Comprehensive Agrarian Reform Program, NAFTA North American Free Trade Agreement the Philippines NATSEM National Centre for Social and Economic CE household consumption expenditure Modelling CRS constant return to scale NCR National Capital Region, Philippines CV coefficient of variation NDRC National Development and Reform Com- ETDZ economic and technological development mission, China zone, China NEP New Economic Policy, Malaysia EU European Union NIE newly industrializing economy FASID Foundation for Advanced Studies on Inter- PATANAS National Farmer's Panel, Indonesia national Development PCE household consumption expenditure per FDI foreign direct investment capita or personal consumption per capita GDP gross domestic product PIL polynomial inverse lag GE general entropy PODES village-level census conducted by Indonesia's GMS Greater Mekong subregion Central Bureau of Statistics GNI gross national income PPP purchasing power parity GNP gross national product PTP Port of Tanjung Pelepas, Singapore GRIPS Graduate Research Institute for Policy R&D research and development Studies SARS severe acute respiratory syndrome GSO General Statistical Office SEAMEO Southeast Asian Ministers of Education HDI human development index Organization IADB Inter-American Development Bank SEARCA Southeast Asian Regional Centre for Graduate ICASEPS Indonesian Center for Agriculture Socio- study and Research in Agriculture Economics and Policy Studies SEZ special economic zone ICL income-contingency loan SME small and medium enterprise ICT information and communications technology TBND Tianjin Binhai New District IDE Institute of Developing Economies TFP total factor productivity IDR Iskandar Development Region TVE township-and-village enterprise, China IFPRI International Food Policy Research Institute WTO World Trade Organization xxxi SECTION I Regional integration, agglomeration, and income Context and concepts: density,distribution in East Asia distance, and division Nobuaki Hamaguchi 1 East Asia has achieved economic growth technologically advanced industries, while of greater than 8 percent annually over the successively shedding industries in which last two decades.1 Influenced by this high it no longer held a comparative advantage; growth, the economic geography of the these industries, in turn, were relocated to region has been transformed by the oppos- nearby less-developed countries (the Asian ing forces of dispersion and concentration. newly industrializing economies, or NIEs). On the one hand, industrialization has Over time, the "following geese" upgraded c h a p t e r spread across the region as the intraregional their industrial structures, following the trade of manufactured goods has grown "lead goose" trajectory, while shedding out- substantially. On the other hand, productive dated industries to their own neighboring activities have become geographically con- less-developed countries (members of the centrated in each country, reinforcing the Association of South East Asian Nations leading role of industrial agglomerations in [ASEAN] and China). When less-developed the development process. countries had prepared their basic situations To explain the distribution of economic correctly, freer international trade provided activities and associated patterns of trade, them with opportunities to integrate them- traditional international trade theory based selves into the regional production network, on comparative advantage emphasizes the enabling the sequential takeoff of these diversity of natural endowments, which economies. On the whole, the flying-geese creates differences in relative factor prices or analogy aptly describes the catch-up indus- technological advantages among countries. trialization in East Asia. Given the assumption of production tech- In this context, the World Bank's report nology with constant returns to scale and The East Asian Miracle (World Bank 1993) immobility of production factors, where presents extensive analyses of those basics goods are traded under perfect competition that qualify a country as a"following goose." without transportation costs, the compara- The report stresses that, unlike other devel- tive advantage perspective predicts that free oping economies, East Asia achieved rapid trade will promote the efficient allocation of growth with equity through the use of two economic activities consistent with the first sets of measures: fundamental policies and nature of each location across space. selective intervention. Fundamental poli- The manifestation of this argument in the cies include macroeconomic stability, large East Asian context is known as the "flying investments in human capital, stable and geese" pattern of catch-up industrialization secure financial systems, limited price dis- (Kojima 2000). Up to a certain moment, tortions, and openness to foreign technol- Japan was the lead economy in this pattern. ogy. Selective intervention includes mild Intraregional division of labor developed financial regulation, directed credit, selective as Japan became increasingly specialized in industrial promotion, and trade policies that 1 2 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA push nontraditional exports (World Bank and higher productivity based on the large 1993: 10­11). pool of educated workers and intermedi- The report contends that the two types ate goods, as well as the availability of good of policy tools are mutually complementary infrastructure. Urbanization also stimu- and cannot be addressed separately. For lates the interaction of people, encouraging example, macroeconomic stability is funda- technological innovation and new kinds of mental to cultivate high savings as well as economic activities. to achieve exchange rate stability, required Agglomeration economies have enabled for economic opening, which, in turn, Japanese industries to play the role of the engender feedback to growth and stability cutting-edge "lead goose." According to through large investments and exports. The Fujita and others (2004), Japan accounted accumulation of savings and achievement of for 72 percent of total gross domestic highly productive exports were sustained by product (GDP) in East Asia in 1990; within wealth-sharing public policies on education, Japan, core economic regions represented 40 landholdings, and small and medium-size percent of the national total, which implies enterprises. Because of their crucial com- that Japan's core regions, with a mere 0.18 plementarities, developing economies can percent of the total area and 2.5 percent miss development opportunities if they fail of the total population of East Asia, repre- to coordinate these measures. Consequently, sented 29 percent of the total regional GDP, the World Bank (1993) concludes that the displaying remarkable geographic concen- government's commitment to social coor- tration. Recently,Asian NIEs caught up with dination through consistent and unbiased Japan in many technological areas such as policies is the key to growth with equity. semiconductors and information and tele- Although providing the basic conditions communications equipment manufactur- for sound economic development remains ing (for example, notebook computers and necessary, a somewhat new scenario has mobile telephones). Leading technological unfolded in East Asia since the 1990s. Above firms in Asian NIEs compete intensely with all, the remarkable growth of the Chinese Japan's firms in the global market. In these economy subsequent to market-oriented countries, the geographic concentration of reforms has been a decisive feature. Based high-technology firms is growing in places on China's abundance of labor and the such as Daejeon, Hsinchu, and Seoul. Indus- explosive growth of its market of middle- trialization in China clearly is concentrated class consumers, scale economies in China in coastal regions; agglomeration has inten- have become the dominant factor attract- sified in many parts of the region, and the ing investment. Middle-income countries East Asian economy has been transformed were rapidly surpassed by China's leapfrog- from the traditional one-dimensional flying ging growth. Countries in a lower stage of geese pattern to a pattern encompassing development cannot take it for granted that multiple technological centers. merely having the right basics will put them According to the framework of the new on track to catch up given the existence of economic geography (Fujita, Krugman, and exceedingly strong agglomeration econo- Venables 1999), agglomeration is a self- mies in China. It is now impossible for any organizing process that results from the East Asian country to consider its develop- balance of concentration and dispersion ment strategy without seriously considering forces.As explained by Fujita (2007a),at least the impact of China. three types of concentration forces (forward In addition, in contrast to the simplified linkages) are identified: the wide variety of version of the flying geese analogy, which consumption enhances consumers' real stipulates a vertical division of labor, the income, the wide variety of intermediate manufacturing of high-technology products inputs increases firms' productivity, and the has spread to countries at heterogeneous wide variety of talented people stimulates levels of development. These industries tend the creation of knowledge. These attract to agglomerate in large urban areas in each consumers, final goods producers, and country to benefit from access to consumers innovative research, respectively. Scale Regional integration, agglomeration, and income distribution in East Asia 3 economies, in turn, exert a pull on an even Inequality cannot be overlooked because greater variety of consumer products, inter- the concentration of wealth and power can mediate inputs, and talented people (back- foment discontent in the bypassed regions ward linkages). Because the new economic and threaten social stability. Government geography models are built on this positive programs for income transfer from urban feedback mechanism, no a priori assump- to rural areas are usually implemented in tion of the difference in the first nature, as this context. However, if farmers residing required in the models based on compara- in disadvantageous locations were to con- tive advantage, is necessary to explain the tinue producing only generic goods under formation of uneven economic geography. perfect competition, intensifying pressure Because of the nature of increasing returns from global trade liberalization would to scale, agglomeration enhances long-run require subsidies, which are not sustainable economic growth (Romer 1986). in the long run. In Japan, for example, the Yet, as a counterpoised dispersion force, dwindling prospects for traditional farm- high transportation costs necessitate that ing have encouraged farmers to migrate to production be dispersed in proportion to cities, thereby accelerating the aging of soci- the size of each local market without real- ety in rural areas and exacerbating related izing agglomeration economies. Therefore, problems such as the difficulty of providing actual formation of agglomeration requires essential public services in such areas. In sufficiently low transportation costs (Fujita, many developing countries, large cities tend Krugman, and Venables 1999). Consistent to be overcrowded, leaving huge populations with this result, agglomeration in East Asia living in makeshift conditions. has developed together with the deepening Innovative ideas are needed to establish of regional economic integration through nontraditional agricultural production and increasingly numerous free trade agree- make the periphery lively and livable with- ments as well as unilateral and bilateral deals out depending heavily on income transfers that reduce the costs of trading. from the core region. In this context, Fujita The foregoing discussion emphasizes (2007b) argues that the introduction of that not only factor price differences but highly differentiated branded agriculture is also scale economies play an important a viable strategy. Branded agriculture makes role in reshaping the economic geography full use of cheap land and labor, which are of East Asia. A recent report published by abundant resources in the periphery, while the World Bank, An East Asian Renaissance overcoming the disadvantages of unfavor- (Gill and Kharas 2007), addresses this point. able market access because consumers will Compared to the focus on coordination fail- buy differentiated products even at higher ure of the earlier World Bank report (World prices.For instance,Japan imported 359 tons Bank 1993), the later report contends that of roses from Kenya in 2006, corresponding to sustain economic growth, especially for to roughly 8 percent of the quantity and middle-income economies, product differ- 20 percent of the value of total imports of entiation, knowledge creation, and agglom- that product. As the data suggest, the unit eration based in cities are expected to play price of Kenyan roses is very high not only key roles. If scale economies prevail, further because of the distance but also because improvement in market integration (both roses are transported by air via the cold international and domestic) must foster the storage facilities of Dubai airport. Still, sales advantages of agglomeration, while mitigat- are growing thanks to high product qual- ing the negative effects, such as road conges- ity. Being in the highland more than 1,000 tion, pollution, and inflated housing prices, meters above sea level and right on the equa- through the appropriate provision of urban tor, Kenya offers ideal natural conditions for infrastructure and regulation of land use. such horticulture: constant daylight hours Although we expect agglomeration all year long and a large temperature gap to enhance growth, this strategy inevita- between day and night, lowering the risk of bly exacerbates regional income dispari- insect infestations. This example suggests ties, especially in the rural-urban context.2 that remote rural areas can be connected to 4 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA a large market if they produce sufficiently outside the region (mainly from the EU and differentiated products, take advantage of East Asia). the local natural conditions, and establish As described in the preceding section, innovative market access.Product differenti- reasons for the steady growth of intrare- ation of branded agricultural products must gional trade include the increasing trade be understood in a broader sense, which of intermediate goods. As shown in figure involves the whole value chain--including 1.2, trade in intermediate goods dominates quality control and logistics management-- intraregional trade, corresponding to nearly rather than innovations in the product itself. 60 percent of total intraregional trade in In contrast to the general perception of the 2005. It is multidirectional: as presented periphery as a static supplier of generic in table 1.1, Japan, the NIEs (Hong Kong foods, innovation is needed in the periphery and Taiwan, China; Republic of Korea; and as much as in large cities. Singapore), ASEAN-4 (Indonesia, Malaysia, the Philippines, and Thailand), and China Production networks in export and import intermediate goods East Asia within the region. Although Japan is still the Intraregional trade accounted for 57.3 per- net supplier of intermediate goods in the cent of all imports and 54.5 percent of all region, its imports of intermediate goods exports of East Asian countries in 2005. from the remainder of East Asia are growing These shares increased over the past quar- rapidly. Although the NIEs show a deficit of ter century, as shown in figure 1.1, except interregional trade in intermediate goods, for the temporary setback during the eco- the collective exports from these economies nomic crises of the 1980s and the 1990s, are already greater than those of Japan. The which increased the share of exports out of export of intermediate goods from China the region to compensate for the precipi- and ASEAN-4 has also grown substantially. tous drop in regional demand. The current According to METI (2007), in the electric share of intraregional trade approaches that machinery industry, which accounts for of the European Union (EU); the pattern more than one-third of all intraregional contrasts with that of the North American exports of intermediate goods, the share Free Trade Agreement (NAFTA), where of Japan's exports dropped from 42 to 22 the gap between the intraregional share percent between 1995 and 2005, although of exports and imports in trade is widen- the share of ASEAN-4 rose from 25 to 31 ing because of the growth of imports from percent and the share of China rose from 5 to 17 percent. Localization of intermediate goods firms following the expansion of for- eign direct investment for the assembly of Figure 1.1 East Asia's share of intraregional trade, 1980­2004 final goods in ASEAN-4 and China and the reduction in tariffs on intermediate goods 60 traded among ASEAN-4 countries have 55 contributed to the dispersion of intermedi- ate goods production. 50 Using data from the Institute of Devel- oping Economies (IDE), we have compiled 45 a trade matrix including transactions of percent semiconductors and integrated circuits to 40 present an example of intraregional pro- 35 duction linkage of electronic parts (IDE 2006; see table 1.2). The table presents the 30 total value of output in the second column; 1980 1983 1986 1989 1980 1992 1995 1998 2001 2004 the third column details the destination of shipments consisting of (1) same-country Import (share) Export (share) intermediate inputs, (2) intermediate inputs Source: METI (2007: fig. 2-1-14). for electronic parts produced in other East Regional integration, agglomeration, and income distribution in East Asia 5 Asian countries, (3) intermediate inputs for mation technology­related products in electronics and electronic final products in East Asia. Of world production, more than other East Asian countries, (4) intermedi- 73 percent of VCRs and DVD players and ate inputs for other types of industries in 80 percent of personal computers are made other East Asian countries, and (5) other in China. On the other hand, about 70 per- goods (for final consumption in East Asia cent of hard disk drives and 43 percent of and shipments to outside East Asia); and the DVD-ROM drives are produced in other fourth column gives the share of each des- Asia,which includes 62 percent of the former tination. Among the listed countries, Japan and 38 percent of the latter in ASEAN coun- boasts the highest value of regional ship- tries. These products are used for assem- ments of these products (destinations 2, 3, bling personal computers; for that reason, and 4). Especially, US$9.7 billion worth of the production linkage between ASEAN and semiconductors and integrated circuits are China is readily explainable. Japan still has exported to other East Asian countries for large market shares in some products, such use as inputs for electronic parts production as 25 percent of flat-panel televisions and there. They are then used for local assembly 39 percent of digital cameras, whereas Korea of final goods or are exported. This suggests has 26 percent of mobile phone production. upstream characteristics of the Japanese Technological advantages in these prod- semiconductors and integrated circuits for ucts sustain the competitiveness of firms in other East Asian countries, partly because of high product differentiation and partly because of intrafirm trade between mother Figure 1.2 Composition of intraregional trade in East Asia, by category of use, 1980­2004 factories in Japan and affiliated plants in 1600 other East Asian countries. In contrast, 70 1400 percent of Chinese semiconductor and 1200 integrated circuits are consumed locally, 1000 and intraregional exports are few. Exports dollars 800 to final goods assemblers within the region 600 (destination 3) are quantitatively similar billion 400 for exports to Japan; Republic of Korea; 200 Malaysia; and Taiwan, China. Sales outside 0 of East Asia (destination 5) show higher shares for Korea, Malaysia, Singapore, and 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 the Philippines. Consumption goods Intermediate goods Raw material Next, figure 1.3 portrays the remarkable concentration of the production of infor- Source: METI (2007: fig. 2-1-15). Table 1.1 Intraregional trade of intermediate goods in East Asia, 1995, 2000, and 2005 US$ billion Indicator and year Japan NIEs ASEAN-4 China Intraregional exports 1995 137.2 129.8 62.1 40.0 2000 143.9 174.3 108.3 63.8 2005 216.5 309.5 175.4 171.4 Intraregional imports 1995 -51.1 -201.9 -83.8 -40.9 2000 -72.1 -252.5 ­93.9 -78.5 2005 -110.4 -386.1 ­148.5 -228.7 Balance 1999 86.1 -72.1 ­21.6 -0.9 2000 71.8 -78.2 14.4 -14.7 2005 106.1 -76.5 26.8 -57.4 Source: METI (2007: fig. 2-2-8). 6 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 1.2 Transactions of semiconductors and integrated circuits in East Asia, 2000 US$ billion Shipment to Shares (percent) Total Economy output (1) (2) (3) (4) (5) (1) (2) (3) (4) (5) Japan 58.1 23.5 9.7 7.4 0.5 16.9 40.5 16.8 12.7 0.9 29.1 China 42.2 29.6 0.8 0.9 0.1 10.8 70.2 1.8 2.1 0.1 25.7 Korea, Rep. of 30.4 5.3 3.5 6.8 0.5 14.2 17.6 11.6 22.3 1.8 46.7 Malaysia 22.3 2.5 3.0 5.1 0.4 11.3 11.4 13.7 22.8 1.6 50.5 Taiwan, China 20.1 0.8 3.1 7.8 0.6 7.8 4.0 15.4 38.9 3.0 38.7 Singapore 19.6 1.6 3.5 3.5 0.3 10.6 8.2 18.0 18.0 1.7 54.0 Philippines 15.4 0.2 1.8 3.0 0.3 10.2 1.0 11.6 19.4 1.7 66.3 Source: IDE (2006). Note: (1) used as intermediate inputs in own country, (2) used as intermediate inputs in electronic parts production in other East Asian countries, (3) used as intermediate inputs for electronics and electronic final products in other East Asian countries, (4) used as intermediate inputs for other types of industries in other East Asian countries, and (5) other goods (for final consumption in East Asia and shipments outside East Asia). Figure 1.3 Regional production shares of information technology­related goods, 2005 DVD-ROM drives Hard disk drives Personal computersa Mobile phones Digital cameras Video cassette and DVD players Flat-panel TVs Color TVs (CRT) 0 20 40 60 80 100 percent of world production China Other Asia Europe Japan North America South America Source: JEITA (2006). a. Notebook and desktop computers. these industries despite higher costs. Not- ics sector. Major trade flows are represented withstanding, their advantages might not as solid arrows; broken arrows represent last long because of "commoditization," or minor flows. This figure is fundamentally rapid deterioration of prices resulting from identical to the triangular trade scheme short product cycles, which are expected described by Fujita (2007a). to force firms to seek cost savings through off-shoring, as has already occurred for the Fragmentation production of laptop personal computers in The growth of intraregional trade with the Taiwan, China. relevant share of intermediate goods and Summarizing these observations, figure the triangular trade scheme is related closely 1.4 portrays the value chain of the electron- to the development of the intraregional Regional integration, agglomeration, and income distribution in East Asia 7 Figure 1.4 Triangular trade in East Asian electronics industry Japan and NIEs Electronic parts (high quality) China Cutting-edge products Parts and components Final goods ASEAN Components (hard disk drives) Semiconductors, integrated circuits United States and the rest of the world Final goods Source: Author's calculations based on IDE (2006). production network.Ando (2006) identifies establishing extra production plants and a the explosive increase of vertical intrain- "service-link cost" for using transportation dustry trade in the machinery industry in and communication services to link the two East Asia in the 1990s. According to Kimura operations. The firm's choice of whether or and Ando (2005), reductions in the costs not to split up production depends on the of transportation and communication have balance between the marginal cost saving enabled firms to cut production processes and the additional costs. into pieces of tasks and to allocate each Borrowed from Jones and Kierzkowski to the most suitable location given factor (2001), figure 1.5 depicts the decision mak- price differences. ing of the firm. The horizontal axis shows This phenomenon is dubbed "fragmen- the quantity of production; the vertical axis tation" in the literature on international shows the total costs of production. The trade theory, which includes the important sum of the setup costs and the service-link report of Jones and Kierzkowski (2001). costs of fragmentation is considered as a Fragmentation refers to the splitting up of fixed cost represented by F. The line for the a previously integrated production process fragmentation case is drawn flatter, imply- into two or more stages. For example, con- ing the marginal cost savings. Expanding sider a production process consisting of a output further than Q1 entails a switch to capital-intensive stage and a labor-intensive fragmented production if a firm chooses the stage. Fragmentation allows the firm to lower-cost production modality. locate the former in a country endowed This illustration is not complete because with more capital and the latter in a coun- the decision depends only on the scale of try endowed with abundant labor. The firm output. The diagram also does not address can thereby reduce marginal costs by tak- the interaction between transport costs ing advantage of factor price differences and scale economies. To add geographic in contrast to locating the two production perspectives, we turn to a straightforward components together in either of the two extension by modifying the graph to pro- countries. However, such cost savings are duce figure 1.6. Let F now represent the made possible by incurring a"setup cost"of setup costs only. Total cost is given as TC 8 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 1.5 Framework for fragmentation mentation) turns out to be profitable if the firm chooses to locate its affiliate in the TC TC (integration) location with a lower service-link cost such as TR**. The setup cost is compensated by sufficiently large scale economies. However, if the service link cost is too high, point C is still not profitable. TC Figure 1.6 enables us to examine the (fragmentation) interaction between scale economies and A transportation involved in the fragmenta- tion. Clearly, if each firm's output is given at Q1, no individual firm will choose fragmen- F tation. It is interesting that, with Q2, frag- mentation might be an outcome under the same service-link cost TR** and the same production technology TC (fragmenta- tion), suggesting the possibility of multiple 0 equilibriums. Q1 Q What kind of a reality does this result describe? We can infer the following effect of externalities. Imagine that, initially, all (integration) if a firm decides to integrate firms integrate production in an industrial the production process at one location and country. Technological developments in produce under higher marginal cost with- transportation and communication open out the fixed cost. Under free entry and exit, the possibility of fragmentation, but each the total cost is equal to the total revenue firm alone will find it unattractive to do so (TR). Presuming that the service-link cost if the output size is insufficiently large for imparts a cost of transportation and com- the given service-link cost. Presuming some munication per unit of fragmented produc- sort of coordination that induces all firms tion, which is discounted from the revenue, to opt for fragmentation, the move creates the total revenue for the fragmented firm industrial jobs and raises the income in the denoted as TR* or TR** is lower than TR less-industrial country, increasing the total by the magnitude of the cost of linking the demand to Q2 and enabling firms to operate fragmented operations across the distance. profitably at point C under the service-link Here, TR* is depicted as the total revenue of cost TR**. Therefore, the big push­like con- a firm that locates the affiliate in a distant certed shift (Murphy, Shleifer, and Vishny location. As a result, the service-link cost is 1989) toward fragmentation is important higher than in the case of TR**. Imagine when the service-link cost is reduced to a that each firm produces Q1. Consequently, moderate level. This shift does not neces- the integrated firm produces at point A sarily require government coordination, but with zero profit. Given the output level, the rush for Japanese investment in China in the firm can set up an affiliate abroad and the 1990s might have been a self-organizing conduct a multiplant operation. In figure shift from point A to C. 1.6, such a decision is represented by the If the service-link cost is sufficiently move to point B. Although the firm can high, as in the case of TR*, fragmenta- reduce its total cost by fragmentation, the tion with output Q2 is still not profitable. strategy is not profitable because B passes This observation implies that fragmenta- above TR**; that is, the cost is greater than tion is more likely to occur with a lower revenue because of the service-link cost. service-link cost, suggesting the case of For that reason, at the individual firm level, regional integration among the neighbor- fragmentation will not occur. However, if ing countries. Using firm-level micro data the firm is able to sell Q2 instead of Q1, the of Japanese multinational firms, Kimura move from point A to point C on TC (frag- and Ando (2005) find that Japanese firms Regional integration, agglomeration, and income distribution in East Asia 9 investing in East Asia are more likely to de- Figure 1.6 Modified framework for fragmentation internalize their production processes flex- TC TR TC (integration)=TR ibly and to conduct outsourcing activities than those investing in other regions such as Europe and North America. Fragmenta- tion might be more sensitive to distance than the case of ordinary trade because the TR** service-link cost entails frequent travel of TR* people in need of technical assistance and TC just-in-time delivery of intermediate prod- A C (fragmentation) ucts across countries. East Asian countries have lowered their international transac- B tion costs through trade policies facilitating imports of intermediate goods, favorable F treatment of foreign direct investment, and development of infrastructure. Kimura and Ando (2005) also suggest the existence of scale economies in infrastructure, strength- 0 Q1 Q2 Q ening the benefits of more intensely used service links. Receiving the spin-off labor-intensive factories of the fragmentation process facil- economic partnership agreements with six itates the industrialization of developing ASEAN member countries (Brunei, Indo- countries. Governments in the region com- nesia, Malaysia, the Philippines, Singapore, petitively offer unilateral and bilateral pro- and Thailand), which are going to be visions to reduce the setup costs of offshore extended with an economic partnership factories and operational costs of linking agreement between Japan and ASEAN as with factories in other countries. Neverthe- a group (including Cambodia, Lao PDR, less, Baldwin (2006) asserts that East Asian Myanmar, and Vietnam). Aside from Thai- integration is still fragile because each coun- land, because of disagreements related to try's preferential trade deals are neither dis- opening of the rice market, Korea and 9 ciplined by the World Trade Organization of 10 ASEAN members reached an agree- (WTO) rules nor supported by a supra- ment in 2007 to form a free trade area, and national regional-level management body negotiations are being held to include trade such as the European Union; consequently, in services. In fact, ASEAN's aggressiveness countries in the region are expected to in AFTA diplomacy is partly a response strengthen such de jure features. to the rise of China, which ASEAN mem- In this respect, it is notable that trade bers fear will bring a hollowing out of policy in East Asia has shifted from mere investment.3 The substantial progress of export promotion, which is a fundamen- the AFTA is expected to contribute to tal policy tool analyzed by the World Bank consolidation of the fundamentals for pro- (1993),to regional integration and free trade duction networks. agreements. Initially, ASEAN was launched to form the ASEAN Free Trade Area (AFTA) for the 15 years since 1993. Subsequently, Border effect between China ASEAN incorporated Vietnam, Myanmar, and Japan Lao PDR, and Cambodia in the late 1990s; Our next task is to evaluate the magnitude leaders of China, Japan, and Korea have of transportation costs, emphasizing the been invited to the annual ASEAN sum- cost of crossing national borders (that is, mit since 1997 in an effort to establish a the border effect). For this, we have con- political framework for ASEAN+3. China structed a simple version of the McCallum agreed to establish a free trade area with (1995) type of gravity model to analyze bor- ASEAN by 2010. Japan signed bilateral der effects involved in interregional trade 10 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA between China and Japan. The estimated equation 1, which includes observations of model is given as follows: domestic regional trade in both China and Japan. These coefficients are estimated as LN(xij) = constant + a11LN(yi) + a2 LN(yj) larger in equation 2, which includes cross- + a3 LN(dij) + b DUMMIES border interregional trade. However, elas- + e ticity with respect to distance is of similar ij (1), magnitude in both equations. In equation 1, where x ij denotes the shipment between the coefficient of the dummy variable China region i and j, and where yi and yj, respec- implies that the domestic trade among tively, represent the GDP of regions i and j. regions in China tends to be twice as large Furthermore, dij is the distance between i as that of Japan (e1.14­ 1 = 2.13). In fact, and j, and eij is an error term (for a detailed as figure 1.7 shows, the domestic trade in explanation of the data, see the annex to this China is heavily biased toward intraregional chapter). As DUMMIES, we included the trade. The share of trade within each region following dummy variables: in China is distributed between 62.5 and 85.8 percent for intermediate transactions · China: intraregional and interregional and between 81.1 and 92.8 percent for final trade in China; demand, compared to that of regions in · Border_China: cross-border interregional Japan, which is distributed between 48.3 and trade from a Japanese region to a Chinese 62.5 percent for intermediate transactions region; and between 69.6 and 83.3 percent for final · Border_Japan: cross-border interregional demand. Given the large factor price differ- trade from a Chinese region to a Japanese ences within the country, if China reduces region; and transport costs internally and its provinces become better linked, industrial special- · Inland: cross-border interregional trade ization within the country is expected to from a Japanese region to a Chinese inland develop, and interregional trade is expected region. to grow. Without such development, pro- The data set includes intraregional trade, duction is expected to concentrate heavily that is, i = j. in the coastal regions: low-cost production The estimated results are portrayed in there will require a large inflow of migrant table 1.3.It is apparent that this simple model workers from inland regions. has reasonably good explanatory power,with Our primary interest is the magnitude adjusted R2 greater than 0.8. The elasticities of the border effect. Our results show that, of trade with respect to GDP of the region all things being equal, Chinese regions trade of origin and of the region of destination with the Japanese regions about 9 times are, respectively, 0.86 and 0.72 according to less than they do with Chinese regions Table 1.3 Gravity model estimates of China-Japan intraregional trade 1 2 Variable Coefficient Standard error Coefficient Standard error Constant -8.28 2.20 -21.84 2.37 LN(yi) 0.86 0.08 1.21 0.08 LN(yj) 0.72 0.07 1.00 0.07 LN(dij) -1.22 0.05 -1.24 0.08 China 1.14 0.14 1.48 0.18 Border_China -2.38 0.19 Border_Japan -2.31 0.18 Inland -2.76 0.15 Number of observations 226 450 Standard error 0.79 1.11 Adjusted R2 0.82 0.90 Source: Author's calculations. Note: All estimated coefficients are statistically significant at the 1 percent level. Equation 1, domestic regional trade, China and Japan; equation 2, both domestic and cross-border interregional trade. Regional integration, agglomeration, and income distribution in East Asia 11 Figure 1.7 Structure of interregional trade between China and Japan 2.4 100 9.2 10.6 13.8 80 34.0 Interregional final consumption 60 Interregional intermediate 43.7 Intraregional final consumption percent 40 Intraregional intermediate 54.4 20 31.9 0 China Japan Source: Compiled based on IDE (2007). (e2.31 ­ 1 = 9.07), although the Japanese as from the Kanto region (1,771 kilometers). regions'cross-border trade with the Chinese The Kanto region's regional GDP is 12.7 regions is 10 times less than their domestic times larger than that of the Huanan region, regional trade (e2.38­1=9.80). These mag- suggesting the border effect. However, if nitudes are half of McCallum's estimate: we examine transactions for intermediate that the cross-border provincial trade from inputs only, Huadong region exports almost Canada to the U.S. states is 22 times less twice as much in intermediate goods to the than Canada's interprovincial trade.4 The Kanto region as to Huanan region. For that border effect from China to Japan must also reason, we can infer that the border effect on be emphasized; that from Japan to China intermediate goods is smaller than the bor- has almost equal magnitude, which sug- der effect on final goods (for a description of gests that the border handling of China is each region, see the annex). as efficient as that of Japan. Therefore, for Japanese multinational firms operating Regional income inequality fragmented production operations between According to the United Nations (2006), Japan and China, the service-link cost has in 2005, 21.0 percent of East Asia's people been substantially lowered. Nevertheless, we resided in 182 urban agglomerations with find that if Japanese cross-border regional populations greater than 750,000 (see figure trade is with inland regions of China, the 1.8). This ratio has increased steadily, from trade flow (both exports and imports) is 8.4 percent in 1955, when only 35 such about 15 times (e2.76­1=14.80) less than agglomerations existed. Rapid urbanization when it is with coastal regions. Therefore, is a spectacular feature of East Asia. we can conclude that the low border effect Next, we construct the ranking of East in China is restricted to the coastal regions, Asian cities by population size for 1950 which implies that a fragmentation type of and 2005. In 1950 there were 35 cities with multinational setup in inland China is best populations greater than 750,000; in 2005 discouraged. there were 182. Figure 1.9 depicts the rela- It is also worth pointing out that the tionship between cities' population size Huadong region (including Shanghai) (log-transformed) and their rank numbers exports 2.3 times more to the Huanan region (log-transformed). They are placed on a (including Guangdong province) than to remarkably straight line of almost identical Japan's Kanto region (including Tokyo), slope ( 0.75) for the two years, illustrating although the Huadong region is almost as far the rank size rule.5 Although this regular- from the Huanan region (1,650 kilometers) ity is known to pertain in the context of the 12 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA hierarchical urban system of a particular of the hierarchy of cities in East Asia as a country, it is striking to see that the random whole has not changed, the concentra- growth of East Asian large cities has evolved tion of higher-ranking cities has tended according to the same kind of regularity. The to intensify in each country because of the degree of primacy represented by the slope agglomeration process. of lines has not changed in East Asia. Although most of the 20 largest cities in Income distribution 2005 have remained in the ranking since Some other characteristics of the interna- 1950, Shenzhen and Dongguan of Guang- tional economic catch-up in East Asia are dong province, which is located next to interesting. Figure 1.10 depicts the relative Hong Kong, China, were not even included size of nominal per capita GDP converted in the list in 1950, but in 2005 they were into U.S. dollars, taking Japan as the refer- ranked ninth and eighteenth, respectively. ence (Japan = 100). Because these figures are Including the two cities, 119 Chinese cit- not PPP (purchasing power parity)-based ies newly entered the list in 2005. It is also data, they do not represent the purchasing noteworthy that Jakarta, Manila, and Seoul power of the people in each country. Rather, rose in rank, respectively, from twelfth to because the location decision of the foreign third, ninth to sixth, and nineteenth to direct investment (FDI) generally is made seventh, thereby transforming the top according to the nominal wage, the nomi- 10 largest agglomerations in East Asia. nal figures are more appropriate. During Although Tokyo and Osaka-Kobe remain 1990­2005, each economy in East Asia made first and fourth, other Japanese cities such progress toward catching up with Japan. A as Kyoto, Nagoya, Fukuoka-Kitakyushu, remarkable catch-up achievement was made and Sapporo have lowered their position. by NIEs, but among NIEs, the importance An increasing number of people in East of Singapore and Korea increased relative Asia are living in large urban areas. In to Hong Kong, China, and Taiwan, China. China the number of such agglomerations The Chinese position also advanced, from is increasing rapidly. These new entrants just 1.4 percent of Japan to 4.9 percent, to the city ranking thicken the lower tail surpassing Indonesia and the Philippines. of the rank size rule distribution, whereas Among the least-developed countries, in other countries, population growth is Vietnam experienced leapfrogging growth. concentrated in fewer cities, shifting the The disparity among ASEAN countries is line upward. shrinking. The difference between Malaysia Therefore, figures 1.8 and 1.9 reveal and Cambodia dropped from one-sixteenth that cities in East Asia have grown both to one-eleventh, although the relative in numbers and in size above the thresh- importance of Thailand and the Philippines old population of 750,000 between 1990 declined slightly. For that reason, in East and 2005. Although the degree of primacy Asia in the last 15 years, although each country narrowed the gap with the lead- Figure 1.8 Share of East Asia's population living in agglomerations greater than 750,000 ing economies, some countries made great inhabitants, 1950­2000 strides, changing the order of the income 25 level among countries. The East Asian regional economy has 20 been transformed from a one-dimensional 15 structure led by Japan into an internation- ally diverse and balanced one after the percent 10 emergence of industrial agglomeration in 5 various countries. Meantime, the problem of income disparity has become more seri- 0 1950 1960 1970 1980 1990 2000 ous within each country because the core- year periphery structure has been clarified.6 Source: Author's calculations based on United Nations data (2006, 2007). Figure 1.11 presents the trend of regional Regional integration, agglomeration, and income distribution in East Asia 13 Figure 1.9 Rank-size rule of large agglomerations (population over 750,000) in East Asia, 1950 and 2005 11.0 10.5 10.0 9.5 size) 9.0 8.5 (population 8.0 log 7.5 7.0 6.5 6.0 0 1 2 3 4 5 6 log (rank) 1950 2005 Source: Author's calculations based on United Nations (2006, 2007). income inequality measured using the the initial stage of economic opening in coefficient of variation (standard error/ 1985­94; industrial production showed mean) of regional GDP per capita. The strong agglomeration toward the coastal intensification of regional inequality is areas, although a trend toward convergence more pronounced in dynamically growing was apparent within the coastal provinces. economies such as China and Thailand. Higher growth was related to production Inequality in Korea is rising slightly but agglomeration, prompted by exposure to steadily, whereas Japan's recent economic globalization (exports and foreign direct recovery is being led by agglomeration in investment) and economic liberalization the Tokyo metropolitan area, whose cen- (reduction in the state enterprise share). See tral business districts are witnessing a rush the other studies on China in this volume to build new buildings. Consequently, we for more detailed analysis. can infer that, although the income dispar- The case of Thailand also portrays a clear ity between regionally integrated countries tendency toward strengthening of the core- is shrinking, the regional disparity within periphery structure.7 In this case, the core each country is rising as these economies includes provinces in Bangkok and its vicini- grow. Because of agglomeration econo- ties,the central region,and the eastern region. mies, some small areas of each country are Many provinces with per capita regional driving national economic growth, among GDP higher than the national average in which income gaps are growing. These cit- 1981 are in noncore regions in the north- ies correspond to the increasing primacy in east. However, in 2003, most provinces with the upper tail of the rank size distribution income higher than the national average were given in figure 1.9. in the core regions. Moreover, the number of For China, Fujita and Hu (2001) show such provinces decreased from 36 in 1981 to that income disparities between the coastal 14 in 2003, leaving the remaining provinces areas and the interior increased during below the average. It is also noteworthy that 14 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 1.10 Nominal income per capita in select Asian economies, 1990 and 2005 although the income gap within the core can be narrowed because of the sprawl of A. 1990 100 agglomeration economies. 100 90 80 Discussion 70 The East Asian economic geography has 60 53.7 49.7 50 been transformed by the opposing forces 40 32.5 of dispersion and agglomeration. Disper- 30 23.9 sion is related to factor price differences 20 9.9 10 6.4 2.9 2.5 based on comparative advantage. Through 1.4 0.9 0.6 0.4 0 such transformations, sequential catch-up Japan China China Rep. Korea industrialization, often described using Thailand Vietnam Singapore Malaysia Cambodia Kong, PhilippinesIndonesiaMainlandDem. the metaphor of flying geese, has occurred. Taiwan, China, Hong Regional integration has lowered the cost of Peoples Lao linking services and broadened the oppor- tunities to divide labor by tasks in different B. 2005 100 locations. Intraregional trade in interme- 100 diate goods is rapidly growing within the 90 80 75.8 regionally extensive production network. 70.8 70 The international spread of industries 60 has contributed to more rapid growth of 50 46.5 42.9 40 low-income countries and to a narrowing 30 of the income gap between the rich and 20 14.3 7.9 poor countries. Regional integration, on 10 4.9 3.6 3.3 1.7 1.4 1.3 0 the other hand, increases the relevance of and scale economies, which in turn stimulate Japan China Korea China Rep. Singapore Malaysia ThailandMainl VietnamDem. Cambodia agglomeration. High economic growth Kong, IndonesiaPhilippines Taiwan, is accompanied by urbanization. For this China, Hong Peoples reason, economic development tends to Lao concentrate geographically in each coun- Source: IMF, International Financial Statistics (various years). try. Because of increasing returns to scale, agglomeration enhances productivity and the income gap between the poorer prov- innovation, providing sources of long-run inces and the national average widened. This growth. These benefits of regional integra- core-periphery structure, which is more tion contributed to East Asia's dominant accentuated than in the Chinese case, might position in the production of many types of be related to the higher mobility of labor in industrial products, especially in the elec- Thailand, which strengthens the agglomera- tronics industry. tion effect through backward and forward Two main concerns might arise in rela- linkages of the core region. tion to the agglomeration-based develop- It follows that deeper economic integra- ment strategy. First, excessively high density tion and the related structural changes in in certain agglomerations might diminish economic geography can generate a mix the advantages that they provide because of of convergence and divergence of income diseconomies from congestion and higher inequality at different levels. First, within prices of immobile resources such as land East Asia, some countries that have attracted and unskilled labor. Cities might grow industry have tended to grow faster, beyond their optimal size, but industries although others have not taken advantage might have difficulty relocating to a remote of such trends and remain in the economic periphery because such areas frequently periphery. Second, within each country, have poor access to markets and interme- industrial agglomeration occurs in a lim- diate goods. Therefore, local governments ited spatial range, sharpening the regional must implement appropriate urban poli- contrast between the core and the periphery, cies to mitigate diseconomies by providing Regional integration, agglomeration, and income distribution in East Asia 15 Figure 1.11 Regional income inequality measured using the coefficient of variation for select East Asian countries, 1990­2004 0.50 A. Japan 0.50 B. Korea 0.45 0.45 0.40 0.40 0.35 0.35 variation 0.30 variation 0.30 of of 0.25 0.25 0.20 0.20 0.15 0.15 coefficient coefficient 0.10 0.10 0.05 0.05 0 0 1990 1992 1994 1996 1998 2000 2002 2004 1990 1992 1994 1996 1998 2000 2002 2004 year year 1.00 C. China 1.20 D. Thailand 0.95 1.15 0.90 1.10 0.85 1.05 variation 0.80 variation 1.00 of 0.75 of 0.95 0.70 0.90 0.65 0.85 coefficient0.60 coefficient 0.80 0.55 0.75 0.50 0.70 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 1990 1992 1994 1996 1998 2000 2002 2004 year year Sources: For Japan, Statistics Bureau (http://www.esri.cao.go.jp/jp/sna/toukei.html#kenmin [in Japanese]); for Korea, National Statistics Office (http://www.kosis.kr/eng/index.html); for China, National Bureau of Statistics of China, China Statistical Yearbook; for Thailand, National Statistical Office (http://web.nso.go.th/eng/index.htm). infrastructure and regulating land use, while 2. An interesting contrast can be made: the encouraging specialization in knowledge- World Bank (1998) report highlights coordina- intensive activities. tion to mitigate income inequality as an essential Second, emphasizing the role of agglom- policy component for sustained growth, whereas Gill and Kharas (2007) predict that the strategy eration inevitably widens regional income to sustain growth will exacerbate inequality. gaps.It is necessary to improve transportation 3. On the other hand, there has been little connections with the periphery, which would progress in de jure integration among the three enable urban industries to move activities that major economies in East Asia. In fact, free trade no longer are competitive to the periphery. area talks between China and Korea remain at Another possibility is to introduce product a preliminary stage, whereas the negotiations differentiation (in a broad sense), thereby regarding an economic partnership agreement taking advantage of the diversity of the natu- between Japan and Korea have been interrupted ral conditions of the remote periphery. for several years. 4. McCallum (1995) explains that the inten- sive use of transportation by air and land is Notes partly responsible for the high magnitude of the Nobuaki Hamaguchi is a professor in the border effect in North American regional trade, Research Institute for Economics and Business whereas most international trade in East Asia is Administration, Kobe University. transported by water. 1. The total GDP of Northeast and Southeast 5. The rank size rule is widely studied in Asia, excluding Japan, grew 8.9 percent in 1990­ urban economics. We generally expect Zipf's law 2004 (World Bank, World Development Indica- to hold, showing the gradient of -1. In this case, tors, 2006). however, the slope is flatter. 16 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA 6. Baldwin and Wyplosz (2003) emphasize Henderson and Jacques-Franēois Thisse, pp. this problem in relation to European integration. 2911­77. Amsterdam: Elsevier. 7. Detailed figures can be found in Fujita and Gill, Indermit, and Homi Kharas. 2007. An Hamaguchi (2008). East Asian Renaissance. Washington, DC: World Bank. References IDE (Institute of Developing Economies). 2006. Ando, Mitsuyo. 2006."Fragmentation and Verti- Asian International Input-Output Table 2000. cal Intra-industry Trade in East Asia." North Chiba: Japan External Trade Organization. American Journal of Economics and Finance 17 ------. 2007. Transnational Interregional (3): 257­81. Input-Output Table between China and Baldwin, Richard. 2006. Managing the Noodle Japan 2000. China: Japan External Trade Bowl: The Fragility of East Asian Regional- Organization. ism. CEPR Discussion Paper 5561. London: IMF (International Monetary Fund). Various Centre for Economic Policy Research, March years. International Financial Statistics. (http://hei.unige.ch/~baldwin/PapersBooks/ Washington, DC: IMF. Manage_CEPR_DP.pdf). JEITA (Japan Electronics and Information Baldwin, Richard, and Charles Wyplosz. 2003. Technology Industries Association). 2006. The Economics of European Integration. World-wide Production of Major Electronics London: McGraw-Hill. 2004­2006. Tokyo: JEITA. Fujita, Masahisa. 2007a."Globalization, Jones, Ronald, and Henryk Kierzkowski. 2001. Regional Integration, and Spatial Economics: "A Framework for Fragmentation." In Frag- An Introduction." In Regional Integration in mentation: New Production Patterns in the East Asia: From the Viewpoint of Spatial World Economy, ed. Sven Arndt and Henryk Economics, ed. Masahisa Fujita, pp. 1­20. New Kierzkowski, pp. 17­34. Oxford: Oxford Uni- York: Palgrave Macmillan. versity Press. ------. 2007b."Spurring Economic Develop- Kimura, Fukunari, and Mitsuyo Ando. 2005. ment by Capitalizing on Brand Agriculture: "Two-Dimensional Fragmentation in East Turning Development Strategy on Its Head." Asia: Conceptual Framework and Empir- In Annual World Bank Conference on Devel- ics." International Review of Economics and opment Economics 2007, Global: Rethinking Finance 14 (3): 317­48. Infrastructure for Development, ed. Franēois Kojima, Kiyoshi. 2000."The Flying-Geese Model Bourguignon and Boris Pleskovic. Washing- of Asian Economic Development: Origin, ton, DC: World Bank. Theoretical Extensions, and Regional Policy Fujita, Masahisa, and Nobuaki Hamaguchi. Implications." Journal of Asian Economics 11 2008."Regional Integration in East Asia: (4): 375­401. Perspectives of Spatial and Neoclassical McCallum, John. 1995."National Borders Mat- Economics." In Economic Integration in East ter: Canada-U.S. Regional Trade Patterns." Asia: Perspectives from Spatial and Neoclassical American Economic Review 85 (3): 615­23. Economics, ed. Masahisa Fujita, Satoru Kumagai, and Koji Nishikimi, pp. 13­42. METI (Ministry of Economy, Trade, and Indus- Cheltenham: Edward Elgar. try). 2007. White Paper on International Econ- omy and Trade. Tokyo: METI. Fujita, Masahisa, and Dapeng Hu. 2001."Regional Disparity in China, 1985­1994: The Effect of Murphy, Kevin, Andrei Shleifer, and Robert W. Globalization and Economic Liberalization." Vishny. 1989."Industrialization and the Big Annals of Regional Science 35 (1): 3­37. Push." Journal of Political Economy 97 (5): 1003­26. Fujita, Masahisa, Paul Krugman, and Anthony Venables. 1999. The Spatial Economy: Cities, National Bureau of Statistics of China. Vari- Regions, and International Trade. Cambridge, ous years. China Statistical Yearbook. Beijing: MA: MIT Press. National Bureau of Statistics of China. Fujita, Masahisa, Tomoya Mori, J. Vernon Hen- Romer, Paul. 1986."Increasing Returns and derson, and Yoshitsugu Kanemoto. 2004. Long-Run Growth." Journal of Political Econ- "Spatial Distribution of Economic Activities omy 94 (5): 1002­37. in Japan and China." In Handbook of Regional United Nations. 2006. World Urbanization Pros- and Urban Economics, vol. 4, ed. J. Vernon pects: The 2005 Revision. New York: United Regional integration, agglomeration, and income distribution in East Asia 17 Nations, Population Division, Department Economic and Social Affairs (http://esa. of Economic and Social Affairs (http://esa. un.org/unpp). un.org/unpp). World Bank. 1993. The East Asian Miracle. New ------. 2007. World Population Prospects: The York: Oxford University Press. 2006 Revision. New York: United Nations, ------. Various years. World Development Indi- Population Division, Department of cators. Washington, DC: World Bank. Annex Data for analysis of the China-Japan border effect T he estimation uses interregional transnational interregional flows of trade us to track the input-output relationship in transaction data from IDE's Trans- directed for intermediate inputs as well as detail. We designate a representative city for national Interregional Input-Output for final demand. The following table gives each region: Shenyang, Beijing, Shanghai, Table between Japan and China 2000 (IDE the provinces included in each region. Guangzhou, Wuhan, Xian, and Chengdu for 2007). This data set comprises informa- The data classify the transaction data into theChineseregionsinthesameorderasabove; tion from seven Chinese regions (Dongbei, 10 sectors (agriculture, livestock, forestry, and and Sapporo, Sendai, Tokyo, Nagoya, Osaka, Huabei, Huadong, Huanan, Huazhong, fishery; mining and quarrying; household Hiroshima, Takamatsu, and Fukuoka for the Xibei, Xinan) and eight Japanese regions consumption products; basic industrial mate- Japanese regions. Road distances between (Hokkaido, Tohoku, Kanto, Chubu, Kinki, rials; processing and assembly; electricity, gas, cities in China are from the Web site http:// Chugoku, Shikoku, Kyushu); it reports and water supply; construction; trade; trans- www.yusen.co.jp/china/english/distance/ intraregional and both intranational and portation; services), which does not allow index.html. Using software Eki-spert of Val Laboratory Corporation, we obtain railroad Country and region Provinces included distances between Japanese cities. The aver- age distance within each region is defined as China Dongbei Liaoning, Jilin, Heilongjiang dii = Si ×(Rurali), where Si is the land Huabei Beijing, Tianjin, Hebei, Shandong area, p is the circular constant (3.14), and Huadong Shanghai, Jiangsu, Zhejiang Rurali is the ratio of rural to urban popula- Huanan Fujian, Guangdong, Hainan tion in region i. Namely, we assume that the Huazhong Shanxi, Anhui, Jiangxi, Henan, Hubei, Hunan region is a circle with the same land area and Xibei Inner Mongolia, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang calculate its radius. The region with a higher Xinan Guangxi, Chongqing, Sichuan, Guizhou, Yunnan, Tibet ratio of rural to urban population has higher costs of intraregional trade because the popu- Japan lation is more scattered. Therefore, the radius Hokkaido Hokkaido is multiplied by Rurali. To ascertain the dis- Tohoku Aomori, Iwate, Miyagi, Akita, Yamagata, Fukushima tances between regions of China and Japan, Ibaraki, Tochigi, Gunma, Saitama, Chiba, Tokyo, Kanagawa, Niigata, Kanto Yamanashi, Nagano, Shizuoka we measure the great circle distance between Chubu Toyama, Ishikawa, Gifu, Aichi, Mie representative cities using Google Earth. Kinki Fukui, Shiga, Kyoto, Osaka, Hyogo, Nara, Wakayama Market size variables yi and yj, respectively, Chugoku Tottori, Shimane, Okayama, Hiroshima, Yamaguchi represent the total output and input of each Shikoku Tokushima, Kagawa, Ehime, Kochi region, except for the case of trade flows for Kyushu and Okinawa Fukuoka, Saga, Nagasaki, Kumamoto, Oita, Miyazaki, Kagoshima, Okinawa final demand, for which yi denotes aggregate Source: http://www.ide.go.jp/Japanese/Publish/Books/Tokei/xls/TIIO(00).xls. demand in the recipient region. Geography of cluster-based industrial development Keijiro Otsuka and Tetsushi Sonobe 2 Many successfully growing industries are development is common among different based in clusters, particularly in develop- industrial clusters in different countries, ing countries where small and medium whether the common pattern of devel- enterprises (SMEs) dominate. There is no opment, if any, is induced by internally question that industrial clusters stimulate generated common economic forces, and industrial development by providing enter- what successful clusters have and unsuccess- prises and their industries with advantages ful ones lack. These questions are critically c h a p t e r conducive to growth, such as informa- important, because if we can identify the tion spillovers and low transaction costs common causes of successful cluster-based (Sonobe and Otsuka 2006). Yet industrial industrial development, we can prescribe clusters have developed disproportionately policies that support or accelerate such pro- in advanced countries. Although there are cesses effectively, while anticipating their industrial clusters in developing countries, future development. Another question that very few are growing and thriving. Nonethe- arises is whether and how it is possible to less, governments in developing countries as rectify the geographic imbalance of indus- well as international organizations seldom trialization accompanied by the develop- pursue deliberate policies to promote the ment of industrial clusters in selected areas development of industrial clusters. without sacrificing industrial production The absence of industrial development and growth. To put it differently, what is the policy would not be much of a problem if appropriate strategy to create new industrial there were no market failures. According to clusters in less-developed regions? the theoretical literature on economic geog- We have conducted 12 case studies in East raphy, however, the geographic imbalance Asia (China; Japan; the Philippines; Taiwan, in industrial development arises principally China; and Vietnam), 3 case studies in South from market failures (Fujita, Krugman, and Asia (Bangladesh, India, and Sri Lanka), and Venables 1999). Thus, a critical task boils 5 case studies in Sub-Saharan Africa (Ethio- down to identifying the market failures pia, Ghana, and Kenya).1 The purpose of that have prevented poor countries and this study is to address the issues identified poor regions in developing countries from above based on our own case studies in Asia establishing industrial clusters. To realize and Africa. Our main conclusions are that a geographically balanced economic devel- common pattern of industrial development opment without sacrificing economic effi- is observed among successful cases and that ciency, empirical research must be carried the main difference between successful and out to identify and characterize the nature unsuccessful cases lies in the lack of "inno- of market failures. vations" in the latter rather than in funda- In this chapter, we examine the extent to mental differences in the nature of develop- which a pattern of cluster-based industrial ment processes. A policy implication is that 19 20 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA it is possible to stimulate the development cheap labor is used intensively and market- of stagnant industrial clusters by inject- ing is the key to success, such as garment and ing missing knowledge conducive to the footwear industries, tends to be led by mer- innovations. chants and to be located in the suburbs or in This study is organized as follows.It begins the vicinity of large cities, where agriculture by presenting a stage theory of cluster-based does not have a strong comparative advan- industrial development, followed by evi- tage.3 If transport and communication costs dence for our theory from our own case are very high, however, merchant-led indus- studies. It then examines the changing loca- trialization may take place in large cities tions of industries in the process of eco- because the advantages of close proximity to nomic development in China and Taiwan, large markets outweigh those of cheap labor China. A final section discusses policy impli- in less urbanized areas. Markets for interme- cations of the study. diate inputs specific to the product do not yet exist, and because of the small scale of A theory of cluster-based production, there is not yet any advantage in industrial development the division of labor among manufacturing An industrial cluster is defined as a "con- enterprises and between such enterprises centration of enterprises producing similar and merchants. Thus a pioneering enterprise or closely related products in a small area" must procure all of the required materials (Sonobe and Otsuka 2006). Based on our and parts directly and sell final products case studies, we have conceptualized the directly to consumers and end users. process of cluster-based industrial develop- ment by stages: (1) initiation, (2) quantity The quantity expansion stage expansion, (3) quality improvement, and The initiation of a new industry requires (4) "eruption."2 The distinction among establishing appropriate production meth- these stages is important because the type of ods, marketing channels, and procure- market failure is different in different stages. ment sources, suitable to the given business The major characteristics of the develop- environment. Compared with the cost of ment stages to be discussed are summarized initiation, the cost of imitating the estab- in table 2.1. lished production methods and marketing is miniscule. Once a pioneering enterprise The initiation stage succeeds in making a sizable profit, the In general, there are good reasons for a new massive entry of new enterprises follows. industry to locate in a certain place, typi- Typically, a swarm of new entrants who are cally in an urban or suburban area. A new former workers of incumbents, or spin- industry in which production is technically offs, appear at this stage, contributing to complicated, such as machinery industries, the formation of industrial clusters. In this tends to be established initially in large process, the industry grows in terms of the cities because it tends to be led by engineers, volume of production but not in terms of who are more readily available in large cit- the quality of products and productivity, ies. Another type of new industry, in which because only imitation takes place. Hence Table 2.1 An endogenous model of cluster-based industrial development Prior experience Innovation, imitation, and Stages of managers Education productivity growth Institutions Initiation Merchants, engineers Low Imitate foreign technology directly Internal production of parts, components, and or indirectly final products Quantity expansion Spin-offs and entry from Mixed Imitate imitated technology; stagnant Interenterprise transactions and formation of various fields productivity; declining profitability industrial cluster Quality improvement Second generation of Very high Multifaceted innovations; exit of many Reputation and brand names; direct sales; founders and newcomers enterprises; increasing productivity subcontracts or vertical integration; with new ideas emergence of large enterprises Eruption The same as in quality Very high Sustained multifaceted innovations and Relocation of leading enterprises from improvement productivity growth congested cluster to less-developed areas Geography of cluster-based industrial development 21 this stage is termed the quantity expansion advantage of established brand names and stage. Industrial clusters facilitate transac- possibly embark on exports. Such"multifac- tions between assemblers and parts suppli- eted innovations" are carried out by highly ers and between producers and merchants educated managers who are able to learn because the geographic proximity among from the experience of more-developed enterprises in industrial clusters reduces not countries. If such managers are not avail- only transportation costs but also opportu- able, the multifaceted innovations usually do nistic behavior and other types of transac- not take place. As a result, industrial clusters tion costs. In a sense, the industrial cluster is either do not grow or eventually are over- an artificially created"community"intended whelmed by cheap imports. In our view, the to reduce transaction costs to make markets critical difference in industrial development work. It is important to emphasize that dif- between Asia and Africa lies in the pres- ferent industries in different countries share ence and absence of innovations. A num- striking similarities in the pattern of cluster- ber of industrial clusters have succeeded in based industrial development, at least up to the transition to the quality improvement this quantity expansion stage. stage, not only in East Asia but also in South Toward the end of this stage, the Asia,4 whereas few have succeeded in doing industrial cluster sets the stage for qual- so in Africa (Akoten and Otsuka 2007; Ako- ity improvement. First, a variety of useful ten, Sawada, and Otsuka 2006; Iddrisu and human resources, such as engineers, design- Sonobe 2007). ers, merchants, parts suppliers, and skilled Ideas embodied in imported high-quality workers, have been attracted to the clus- parts and components are often useful for ter. Second, because of the active entry of the "imitative innovations." In this sense, a followers, the price of products falls and, liberal trade system facilitates the transition consequently, the profitability of producing to the quality improvement phase. Also use- low-quality products declines. Under such ful are ideas brought about by global buyers conditions, the profitability of producing and foreign direct investments. However, high-quality products by employing useful learning from abroad is not a panacea: the human resources is high. knowledge acquired from the experience gained in the quantity expansion stage is The quality improvement stage often necessary to assimilate and adapt If the entrepreneurs in the sense of Joseph new knowledge to the local production Schumpeter (1912) are available in the environment. cluster, they tend to innovate at this stage As aptly pointed out by Marshall (1920), by carrying out "new-combination," which the advantages of industrial clusters are is another name for innovation. Then the (a) information spillovers or imitation, industrial cluster graduates from the quan- (b) division of labor among enterprises, and tity expansion stage and enters the quality (c) development of skilled labor markets. improvement stage (Sonobe and Otsuka Sonobe and Otsuka (2006) argue that the 2006). The innovative enterprises improve availability of useful human resources for the quality of products by employing design- innovation is another advantage of indus- ers, engineers, and skilled workers. Such trial clusters. This is critically important, enterprises establish a reputation and brand because the first advantage of the industrial name and develop their own distribution cluster cited by Marshall--that is, imita- network to solve the problem of the "mar- tion--cannot be of any significance without ket for lemons" that arises from asymmet- innovation. Two negative externalities, how- ric information as to whether the product ever, are generated in the developed cluster: is really an improvement. The enterprises one is congestion, and another is rampant offer long-term subcontracts to dependable imitation of innovative ideas, which reduces parts suppliers to procure specially designed the private returns to innovation below the parts or undertake a vertically integrated social returns. Thus the quality improve- production system. The innovative enter- ment stage may be followed by the eruption prises expand the scale of operation to take stage. 22 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA The eruption stage abroad. Furthermore, the original cluster If we draw the map of an industrial clus- may remain as a technology and marketing ter at the end of the quality improvement center, even though the scale of produc- stage, while showing production "isoquant" tion declines. Another key factor inducing contours, what emerges would look like a the relocation of industries to remote areas volcano in which the amount of produc- where the industry has not developed is tion is the largest in the center of the cluster the policies of local governments to attract and declines as the distance from the cen- new industries in underdeveloped areas. In ter increases. The pioneering enterprises, contrast, the eruption is "small" or "short which initiated the industry, are naturally distance" if the innovations are minor, the located in the center. They tend to be more subcontracting with suppliers or the trans- innovative than others because their found- actions with local merchants are important, ers had foresight and their children tend to or local government provides new and spa- be highly educated. Indeed, there are many cious industrial zones nearby. At this stage, cases in which educated sons of the found- industrial policies in the less developed ing entrepreneurs are the first to attempt region may be able to attract innovative quality improvement (Sonobe and Otsuka enterprises so as to reduce the geographic 2006). Their successful quality improve- imbalance of industrialization. ment boosts the scale of their production, Evidence from case studies thereby increasing the "altitude of the vol- cano." Such innovations give rise to further To substantiate our arguments, we attempt congestion in the central area of the cluster. to provide supportive evidence taken from Moreover, innovative enterprises expect to our case studies of the motorcycle industry suffer losses from imitation by local enter- in Japan (Yamamura, Sonobe, and Otsuka prises that are not innovative. Thus, partly 2005), the machine tool industry in Taiwan, because of congestion and partly because China (Sonobe, Kawakami, and Otsuka of the fear of imitation by rivals, innovative 2003), and the electric machinery industry enterprises tend to relocate their produc- in China (Sonobe, Hu, and Otsuka 2004). tion base outside the existing cluster. This Figure 2.1 (panel A) shows the change relocation may be termed an eruption, in in the number of motorcycle enterprises in which lava (or innovative enterprises) flows Japan from 1946 to 1964. Clearly the num- out from the crater (or the center of the ber of enterprises increased sharply up to cluster) and lands at the foot of the moun- the early 1950s, because of the sizable entry tain or farther away. of new enterprises. Roughly speaking, the The eruption is likely to be "big" or "long annual growth rate of the total number distance" to the extent that new innova- of motorcycles produced was as high as tions are major, because motivations to 100 percent in the early 1950s. According expand the scale of production in remote to figure 2.1 (panel B), the engine qual- areas where congestion is absent and to ity index, computed by Taylor's formula, escape from the imitation are large. Behind did not rise until 1953, indicating that this such eruption would be superior new tech- period corresponded to the quantity expan- nological ideas and superior management sion stage. After the mid-1950s, however, the and marketing abilities of enterprise man- quality index rose steadily, and a number of agers. The eruption is also likely to be big enterprises that used lower-quality engines if the main purpose of the relocation is to exited. While the growth rate of produc- seek cheap labor abroad. In our view, the tion decreased to less than 50 percent a year, relocation of industry from developed to the average size of the surviving enterprises developing countries envisaged by Vernon's grew approximately 10 times in the six-year (1966) product cycle theory is an example period in the late 1950s. A key role in this of a big eruption. Although a big eruption quality improvement stage was played by may eliminate industrial clustering in the Soichiro Honda, the founder of the Honda original location, it may create a new one Motor Company. Indeed, Hamamatsu city, Geography of cluster-based industrial development 23 Figure 2.1 Development of the motorcycle industry in Japan, 1945­65 A. Changes in the number of motorcycle enterprises B. Improvement in the quality index of motorcycle engines 140 0.45 0.40 120 0.35 100 0.30 firms 80 0.25 of 60 0.20 number 0.15 40 0.10 20 0.05 0 0 1945 1950 1955 1960 1965 1945 1950 1955 1960 1965 year year New entrants Exiting firms Total of existing firms Exiting Surviving Source: Yamamura, Sonobe, and Otsuka (2005). Table 2.2 Major features of the machine tool enterprises in Taichung, Taiwan, China Characteristic Pioneers Early imitators Innovators New imitators Number of samples 7 24 2 10 Year of establishment 1957 1979 1979 1994 Year of numerically controlled production 1,980 1,990 1,980 1,995 Years before machine tool production 12.6 0.7 0 0 Prior job of founders (percent) Machine tool enterprises 0 59 100 90 Other machinery enterprises 88 13 0 10 Schooling of founders (percent) Primary 71 13 0 0 Secondary 0 12 0 0 High or vocational schools 29 54 0 60 University or graduate school 0 21 100 40 Total 100 100 100 100 Number of parts suppliers 19 30 41 39 Source: Authors' calculations. where Honda was originally located, became processes in the 1950s and 1960s. Six of the the leading cluster of this industry, dominat- 7 pioneers used to work at other machin- ing the clusters in Tokyo and Nagoya. But ery-producing enterprises, and 5 had only a Honda "erupted" to Suzuka city, far from primary education. It took them 12.6 years, Hamamatsu, to begin vertically integrated on average, to commence the production of mass production of high-quality products machine tools after founding their enter- in huge factories. prises. By contrast, most enterprises estab- Table 2.2 shows the characteristics of lished in the 1970s and the early 1980s were managers of machine-tool enterprises in spin-offs and began producing machine Taiwan, China, by type of enterprise (that tools immediately after their establish- is, innovator or imitator), which roughly ment. During this period, a large number correspond to the establishment peri- of specialized parts suppliers emerged, and ods. Among the 43 sample enterprises, 7 both final products and parts were highly are identified as pioneers of the industry, standardized. To keep profitability from which attempted to produce machine tools declining, the founders of the industry for the first time through trial-and-error began attempting to produce numerically 24 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA controlled machines, again through trial- on the basis of quality in the second half and-error processes. of the 1980s, when a few enterprises began The major innovation, however, was using machines to inspect product quality carried out in the early 1980s by two new and attempted to establish brand names. In enterprises, which table 2.2 refers to as inno- 1990 the importance of engineers was com- vators. Before founding their enterprises, paratively low, a long-term subcontracting the managers completed graduate studies system did not exist, and final products were and worked for the pioneering enterprises. commonly marketed through anonymous They were innovative not only because they marketplaces and local merchants. These were among the first to produce numerically are precisely the characteristics present in controlled machines in Taiwan, China, but the quantity expansion stage. Indeed, the also because they introduced a new busi- number of enterprises increased in the early ness model of outsourcing all of the parts to 1990s.5 Throughout the decade, particularly suppliers through long-term subcontracting in the late 1990s, the size of enterprises arrangements, which drastically reduced the expanded in terms of real value added and costs of producing numerically controlled number of employees. Also, the ratio of machines. As soon as they began mass pro- engineers to employees and the number duction and instituted drastic price cuts, the of subcontractors increased dramatically. innovators overwhelmed the pioneers and Moreover, direct marketing by means of the early imitators. A number of spin-offs sales agencies and own retail outlets became from the innovators imitated the produc- much more important than indirect market- tion methods, but they were not as success- ing through wholesalers and other interme- ful as the innovators. The pioneers turned diate merchants. Because large enterprises out to be efficient imitators, while the worst that established brand names merged with imitators in the quality improvement phase unsuccessful enterprises, the number of were the early imitators. independent enterprises decreased 25 per- In the machine tool industry in Taichung, cent from 1995 to 2000. The average value only a short-distance eruption to suburban added shown in table 2.3 excludes that of areas took place, partly because access to a subsidiaries, and, if that were included, the large number of subcontractors in the cluster average value added of independent enter- was critically important and partly because prises in 2000 would be 1.6 times higher city governments in the Taichung area set than the number shown in this table. In this up new industrial zones to attract innovative growth process,successful enterprises moved enterprises. to nearby industrial zones constructed by The last example comes from the electric local governments. machinery industry in Wenzhou (see table While the figure and tables shown in this 2.3). According to our interviews with expe- section are merely suggestive, the three cases rienced managers, firms began to compete discussed are consistent with the predictions of the endogenous model of cluster-based Table 2.3 Transition to quality improvement: Average enterprise size and marketing channels industrial development formulated in the in Wenzhou, China, 1999, 1995, and 2000 previous section. The other five cases in East Characteristic 1990 1995 2000 Asia that are not discussed here are equally Number of sample enterprises 66 102 112 supportive of our arguments (Sonobe and Number of independent enterprises 66 96 73 Otsuka 2006). Real value added (Y 10,000 in 2000 prices) 123.7 375.8 3,671.4 It is no exaggeration to say that microen- Number of workers 46.7 104.1 338.3 Ratio of engineers (percent) 1.5 2.7 4.2 terprises and SMEs in developing countries Number of long-term subcontractors 0 2.8 34.8 in general and Sub-Sahara Africa in par- Marketing channels (percent) ticular are all located in industrial clusters. Market places 23.5 20.4 3.6 In our view, this is because transportation Local Wenzhou traders 26.5 23.8 5.7 costs and transaction costs are too high for Agents 22.0 30.7 50.6 Own retail shops 9.5 12.6 27.1 microenterprises and SMEs outside indus- Others 18.5 12.5 13.0 trial clusters. The vast majority of these Source: Authors' interviews. firms are microenterprises, which have just Geography of cluster-based industrial development 25 a few workers and are operated by artisans available in many countries. This is why who received apprentice training but no for- many studies on this subject rely on episode mal training. The scope and depth of their or case studies. Our review is based on case understanding of management, marketing, studies in the western side of Taiwan, China, and technology tend to be highly limited. and the Yangtze River region in China con- Thus industrial clusters have been generally ducted by the authors in collaboration with stagnant and remain in the quantity expan- local experts (Sonobe and Otsuka 2006).6 sion stage. Taiwan, China, is known for the impor- A notable exception is the leather-shoe tance of industrial clusters and the preva- industry in Addis Ababa, Ethiopia (Sonobe, lence of subcontracting systems, but it is Akoten, and Otsuka 2007). This industry also known for its geographically dispersed began about 70 years ago with a factory pattern of industrial development (Otsuka established by an Armenian merchant. The 2007). This section examines how the geo- workers of this factory started their own graphic concentration and dispersion of workshops, which in turn produced a num- industries took place. In China in the 1980s, ber of new entrepreneurs. The repetition of one of the most successful models of indus- spin-offs created a large cluster consisting of trial development was the "Sunan model,"7 more than 1,000 shoe-making workshops in which collective township-and-village in the city. The long history of the industry enterprises supported by the township-and- also produced another type of important village governments grew rapidly based on human resource: the highly educated, young cooperation with state-owned enterprises entrepreneurs, who are the sons and grand- and learned from them through the recruit- sons of the owners of the long-established ment of engineers and managers (Otsuka, shoe enterprises. These new entrepreneurs Liu, and Murakami 1998). This section have recently been carrying out multifaceted advances and examines the broad hypoth- innovations. esis that, since the 1990s, the formation of Multifaceted innovations are also being industrial clusters has been one of the criti- attempted by relatively highly educated cal factors underlying the sustained, rapid entrepreneurs in a few other industrial clus- growth of the Chinese economy. ters in Sub-Saharan Africa, including a small cluster of metal-processing enterprises in Development of suburban clusters Nairobi, Kenya, after a period in the quan- in Taiwan, China tity expansion stage, during which the prof- We now examine how industrial locations itability of producing low-quality products and the division of labor evolved in five declined. These firms are adopting new mar- manufacturing industries in Taiwan, China, keting and new procurement strategies and from the mid-1970s to the mid-1990s. trying to improve their products. Some of For this study, we obtained data on the them have already erupted to more spacious production and employment of manufac- industrial areas to expand their production turing establishments, aggregated up to the and take full advantage of the favorable effect township level, by industry, for 1976, 1986, of their attempts. These cases from Africa and 1996, from the Director-General of lend support to the endogenous model of Budget, Accounting, and Statistics of the cluster-based industrial development. Executive Yuan, Taiwan, China.8 A township is an administrative unit below the county Evidence of changing industrial level (and county-level city) and above the locations village level. The areas under study are lim- It is difficult to trace the birth, formation, and ited to the western part of Taiwan, China, subsequentdevelopmentof anindustrialclus- which consists of the northern, central, and ter, as well as its eventual relocation. Because southern regions. The mountainous eastern the industrial cluster is usually located in a region is excluded from our analysis because geographically small area, township-level it accounts for only 5 percent of total man- data with fine classification of industrial ufacturing employment in Taiwan, China. subsectors are needed, which are rarely From 1976 to 1996, there were several 26 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA subdivisions and mergers of townships.After As table 2.4 shows, these five industries adjusting for these changes, we obtained had sharply contrasting growth records over the consistent data of 275 "townships" for the last few decades.The apparel industry is in this study. decline: it accounted for nearly 12 percent of Until the mid-1980s, the manufacturing total manufacturing employment in 1976,but sector of the Taiwanese economy enjoyed less than 4 percent in 1996. Its extremely large relatively favorable growth led by expand- negative growth rate of employment in the ing export markets. The wages of unskilled period from 1986 to 1996 is explained mainly workers, however, rose sharply in the late by the relocation of production from Taiwan, 1980s,and the currency appreciated abruptly China, to mainland China (Tu 2000), which against the U.S. dollar in 1986 and 1987. The is a big eruption, according to our theory. contribution of net exports to the economic The other industries, except the machinery growth rate declined from 50 percent in the and computer industries, were also increas- mid-1980s to 20 percent in the mid-1990s, ingly affected by competition from low- and exported products became increasingly wage countries. The machinery industry lost intensive in the use of skilled labor, accord- employment share slightly in the first period, ing to Chan, Chen, and Hu (1999). but it regained employment share in the sec- We chose five major manufacturing ond period, presumably because it succeeded industries in Taiwan, China: apparel, plas- in shifting from standardized and conven- tic products, machines, electric appliances, tional machines to high-quality numerically and computers. The production processes controlled machines, as discussed earlier. The of these industries can be divided into a highest growth rate in the second period was number of subprocesses, and the extent of recorded by the computer industry. vertical disintegration varies across enter- To examine from where and to where prises, areas, and industries as well as over the spatial dispersion and concentration time. The five industries correspond roughly took place, table 2.5 exhibits changes in the to the two-digit classification adopted in share of employment in urban, suburban, Taiwan, China's, census data. To represent and rural areas in total employment of each a new and growing industry, however, we industry. The classification of areas is based selected the computer industry from the on the administrative classification as fol- two-digit electric machinery industry. lows: (a) urban areas consist of the 44 wards Moreover, the electric appliances industry of Taipei and Kaohsiung municipalities and does not include the subsector producing Keelung, Taichung, and Tainan cities; (b) audio and visual equipment and parts, as suburban areas consist of 104 townships data for this subsector are not available for that are either adjacent to the urban areas 1976. Likewise, the computer industry in defined above or designated as (township- 1976 is not included in our analysis because level) cities; and (c) rural areas consist of the of the lack of data. remaining 127 townships. Table 2.4 Share of manufacturing employment and annual growth rates of employment in Taiwan, China, by industry percent Plastic Electric Indicator and year Apparel products Machinery appliances Computer Share of manufacturing employment 1976 11.7 9.1 4.8 1.3 -- 1986 9.1 11.6 4.5 1.4 1.2 1996 3.9 7.5 7.9 1.4 4.2 Annual growth rate of employment 1976­86 1.1 6.1 2.9 3.9 -- 1986­96 -9.2 -5.1 4.9 -0.1 11.3 Source: Authors' calculations based on the employment data for the areas under study. -- Not available. Geography of cluster-based industrial development 27 Table 2.5 Changes in employment shares in Taiwan, China, by area and industry, 1976, 1986, and 1996 percent Location and year Apparel Plastic products Machinery Electric appliances Computer Urban areas 1976 30.9 24.8 42.4 30.3 -- 1986 27.7 16.6 26.7 20.9 25.8 1996 29.0 15.9 20.9 22.0 25.2 Suburban areas 1976 53.8 54.3 51.5 61.1 -- 1986 56.3 56.9 63.0 68.8 62.2 1996 57.6 65.1 65.1 70.7 69.4 Rural areas 1976 15.3 21.0 6.1 8.5 -- 1986 16.0 26.6 10.3 10.2 11.9 1996 13.4 19.0 14.0 7.2 5.4 Source: Authors' calculations. For each year, urban + suburban + rural = 100 percent. -- Not available. From this table,it is clear that the employ- been established in these suburban areas by ment share of the suburban areas increased the late 1970s (Chang 1992). Furthermore, steadily throughout the two periods in every the Industrial Technology Research Institute, industry under study. Furthermore, the which facilitated international technology employment share of suburban areas tended transfer to Taiwanese enterprises, is located to increase when the industry was growing. near Taoyuan (Hong and Gee 1993). Thus This seems to support the hypothesis of suburban areas had a high share of employ- eruption, which argues that the industrial ment in the computer industry from its cluster tends to move to suburban areas inception, and this share continued to rise, when the industry grows. In contrast, the which is consistent with our contention that share of the urban areas decreased in the the center of manufacturing is often estab- first period in every industry, but it increased lished in suburban areas. slightly in the second period in the apparel and electric appliances industries. The share Development of clusters in the of the rural areas increased in every indus- southern Yangtze River region try in the first period but decreased in the Southern China, such as Guangdong prov- second period in every industry except the ince, grew most rapidly in the 1980s, pro- machinery industry. These observations ducing relatively low-quality products indicate that, due to the eruption, the cen- using unskilled young migrant workers. ter of manufacturing sectors moved from As the Chinese economy began the quest urban areas to suburban areas, but not to for quality improvement and started shift- rural areas, most likely because of the disad- ing from unskilled labor­intensive to more vantages associated with remoteness. skilled labor­intensive industries in the New industries generally are born in urban early 1990s, the center of economic devel- environmentsthathavewell-developedtrans- opment shifted from the south to the lower portation and communication facilities and Yangtze River region, such as southern favorable access to new information, a vari- Jiangsu. In Jiangsu, industrial clusters did ety of intermediate inputs, and skilled labor not develop in the 1980s, as the location (see, for example, Henderson, Kuncoro, and of collective enterprises was determined Turner 1995).Although the computer indus- largely by township-and-village govern- try was new in Taiwan, China, in the mid- ments. Although it is not recognized in the 1980s, its location was concentrated in the literature on the Chinese economy, once the suburban areas between Taipei and Taoyuan collective township-and-village enterprises rather than in urban areas, because the base were privatized and the development of pri- of the electric and electronics industries had vate enterprises was promoted in Jiangsu 28 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA province, enabling the establishment and the latter caught up with the former in the location of enterprises to be determined by 1990s by taking advantage of the dominance profit incentives, clustering began in this of private industries and the development of province.9 This seems to provide the prima industrial clusters. facie evidence that industrial clusters have Table 2.6 exhibits the number of sample clear advantages over stand-alone enter- townships from which we obtained data, prises in production and management effi- the average number of enterprises, and the ciency. Here we compare the experience of average number of workers and real value Jiangsu with that of Zhejiang, where indus- of production by location (that is, north- trialization began at poor farmers' houses ern Jiangsu, southern Jiangsu, and Zheji- and was supported by the development of ang) and year. We do not include the data dense SME-based industrial clusters in a for state-owned enterprises or urban col- countless number of locations. lective enterprises. As shown in the table, We attempted to collect unpublished data rapid industrialization clearly took place in at the township level for 1990, 1996, and all three regions. Compared with the other 2002. Typically, townships in the areas under two regions, Southern Jiangsu had a sub- study have a population of about 50,000. The stantially smaller number of enterprises per following sampling scheme was adopted. We township, although they were larger in size randomly selected 50 counties each in Zheji- or had slightly less employment and pro- ang and Jiangsu provinces and then chose 2 duction value, which indicates that large- townships from each county. Because there scale enterprises originally supported by are approximately 40 townships in each township-and-village governments were county, applying a random selection tech- prevalent in southern Jiangsu. In this region, nique might have yielded many townships the real value of production increased six- where major industrialization did not take fold over the 12-year period, whereas the place. Therefore, we asked the county gov- number of workers increased only 50 per- ernments to choose the 2 most important cent, which implies that labor productiv- industries in the county and then chose 2 ity increased explosively. The fact that the townships with the most prosperous indus- number of workers almost doubled from tries. The collection of appropriate data 1990 to 2002 in northern Jiangsu indicates was difficult primarily because large efforts that the enterprises were catching up with were required to prepare the detailed data those in southern Jiangsu. The number of requested and also because past data are not workers and the value of production were well kept in the government offices. Thus we higher in Zhejiang than in southern Jiangsu received data from 135 townships, out of 200, partly because we do not include state- for 2002 and only 97 for 1990. The response owned enterprises and urban collectives, rate was much lower from Zhejiang province, which are much more common in the latter where the distinction between collective and region and partly because we failed to obtain private enterprises, which we asked them to data from the southern regions of Zhejiang, report, might be difficult to draw because of which were less developed. the so-called "red-cap" enterprises.10 Almost all of the enterprises were private According to official statistics, gross in Zhejiang from the beginning, and half domestic product (GDP) per capita in Zhe- of them were private in northern Jiangsu jiang and Jiangsu provinces is about twice in 1990. Despite the prohibition of private as high as the average in China, and it is enterprises, the share of the private sector about 15 percent higher in Zhejiang than in was comparatively high in northern Jiangsu, Jiangsu. The income figure in Jiangsu prov- because small family-based enterprises ince, however, includes poor areas north of operated by farmers and ex-farmers with the Yangtze River, which account for two- seven workers or fewer, which were con- thirds of the area in this province. According sidered legal, were included in the number to our interviews with government officials of private enterprises. In southern Jiangsu, at various levels, southern Jiangsu was more privatization began in the early 1990s and prosperous than Zhejiang in the 1980s, but was almost completed by the early 2000s. Geography of cluster-based industrial development 29 Table 2.6 Average number of enterprises and workers and real value of production per sample township in Jiangsu and Zhejiang provinces, China, by location, 1990, 1996, and 2002 Location and indicator 1990 1996 2002 Northern Jiangsu Number of sample townships 48 59 66 Average number of enterprises 237 366 558 Average number of workers 3,807 5,680 8,495 Real value of productiona 123.5 424.6 1,136.5 Southern Jiangsu Number of sample townships 16 21 23 Average number of enterprises 83 147 262 Average number of workers 6,160 6,254 8,995 Real value of productiona 268.4 762.6 1,649.8 Zhejiang province Number of sample townships 33 39 46 Average number of enterprises 320 663 950 Average number of workers 10,894 16,489 23,350 Real value of productiona 663.3 1,256.5 2,739.7 Source: Authors' calculations. a. Million yuan in 2002 prices. The factory shipment price index reported by the State Statistical Bureau (2003) was used as deflator. Table 2.7 Share of production and employment of the most important industry in the township in Jiangsu and Zhejiang provinces, China, by area, 1990, 1996, and 2002 percent Location and indicator 1990 1996 2002 Northern Jiangsu Employment share 25.8 30.5 32.2 Production share 31.9 34.6 37.3 Southern Jiangsu Employment share 27.3 28.9 33.4 Production share 35.8 33.7 44.8 Zhejiang province Employment share 33.9 35.8 36.4 Production share 31.5 30.0 36.8 Source: Authors' calculations based on township data. To explore the role of industrial clusters difficulty in interpreting the specialization in regional development,we obtained data on rate arises from the production of interme- the most important industry of each township diate inputs. When parts and components in accordance with the two-digit industrial are produced internally by enterprises in the classification,which consists of approximately most important industry, they are counted 200 subsectors. For example, the subsector as the production of this industry. They are of the transportation equipment industry not counted, however, if they are produced includes the automobile, motorcycle, and by specialized parts suppliers. Because the ship-building industries. We computed the collective enterprises tend to produce parts production and employment shares of the internally, the specialization rate tends to be most important industry in total production higher where collective enterprises domi- and employment in the township, which are nate. This may explain why the specialization intended to capture the extent of specializa- rate is not necessarily higher in Zhejiang tion.These shares,called"specialization rates," province, where collective enterprises are are shown in table 2.7 by location and year. relatively few and private enterprises rarely While the regional average of the special- adopt a vertically disintegrated production ization rates varies from 30 to 40 percent, the system. Yet the specialization rate in terms figure at the township level ranges from less of employment share was higher in Zhejiang than 5 percent to more than 90 percent. One province than in Jiangsu province, which 30 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA strongly suggests that small labor-intensive know-how, and marketing strategies are industries tend to form dense industrial clus- severely limited. ters in this province. Another important policy implication of The specialization rate increased over this study is the desirability of promoting time particularly in the two Jiangsu regions, a "big eruption" by setting up production which indicates that industrial clusters were environments conducive to the operation established and strengthened over time. This of large factories in less-developed areas, by is likely to reflect the impacts of privatization constructing industrial zones,and by provid- because it allows industries a free choice of ing training for the future cadre of the indus- location. In our own observation, spin-offs try. It may also be possible to establish new are common as privatization progresses. A industrial clusters by constructing model major effect of privatization may well be to plants and technology transfer centers. Cur- enhance the efficiency of industrial produc- rently, however, our knowledge is far from tion by promoting the formation of indus- adequate to select appropriate industries in trial clusters in specialized areas (Sonobe appropriate locations while using appropri- and Otsuka 2003). ate technologies. In other words, we have to Judging from the fact that the forma- seek development strategies that ensure that tion of industrial clusters began less than a the benefits of correcting "market failures" decade ago, it is likely that a greater num- and"geographic imbalance"exceed the costs ber of industrial clusters will be formed, of "government failures." strengthened, and refined in the southern Yangtze River region. It is also notewor- Notes thy that private enterprises in this region Keijiro Otsuka is a professorial fellow with the seldom move to inland areas where wages Foundation for Advanced Studies on Interna- are lower.11 tional Development (FASID) and program direc- tor of the Graduate Research Institute for Policy Policy implications Studies (GRIPS) in Tokyo. Tetsushi Sonobe is a professor at GRIPS and deputy program director It is a mistake to assume that geographic dis- of FASID. parity is always undesirable from the social 1. Although we have completed the data col- point of view, because industrial clusters lection in these studies, the analysis is currently have the advantages of reducing transaction under way in several cases. costs and providing expanded opportunities 2.Since we are interested in development,we do for innovation. Because of lower transaction not elaborate on the declining stage of industries. costs, markets work in industrial clusters, 3. For example, the Bingo Working Clothes and ample job opportunities can thereby Cluster in Hiroshima prefecture in Japan used to be a major production area of traditional be created for the poor. Because innovation Japanese casual wear, called mompe. When the possibilities are enlarged, sustainable growth demand for mompe declined, traders selling of industries becomes possible. Because in mompe throughout Japan brought in the idea of the early stage of development, industrial producing new types of garments (Yamamura, clusters develop primarily in urban and Sonobe, and Otsuka 2003). suburban areas, these social benefits can be 4. See case studies of the garment and motor- obtained by sacrificing geographic balance. cycle clusters in China (Sonobe, Hu, and Otsuka Since a number of industrial clusters already 2002, 2006) and the garment cluster in Bangla- exist in large cities in developing countries desh (Mottaleb and Sonobe 2007). and since they are generally stagnant, we 5. However, some enterprises might have strongly advocate policies to stimulate mul- exited in this period. 6. Sonobe and Otsuka (2006) also report the tifaceted innovations in these clusters by case of Kanto, in the vicinity of Tokyo, during the providing appropriate training programs for postwar period, where the geographic dispersion managers of enterprises. This is particularly far outweighed the geographic concentration. relevant in Sub-Saharan Africa, where both 7. Sunan literally means southern Jiangsu. the supply of entrepreneurs and the access 8. In Taiwan, China, manufacturing firms to advanced technologies, management with multiple establishments are the exception. Geography of cluster-based industrial development 31 The number of establishments per firm was less Hong, Chi-Ming, and San Gee. 1993. "National than 1.05 during the periods under study. Systems Supporting Technical Advance in 9. These changes in industrial location have Industry: The Case of Taiwan." In National been discussed in the recent empirical literature Innovation Systems: A Comparative Analy- on economic geography. See, for example, Fujita, sis, ed. Richard R. Nelson. Oxford: Oxford Krugman, and Venables (1999). University Press. 10. Fake collectives or red-cap enterprises Iddrisu, Alhassan, and Tetsushi Sonobe. 2007. were essentially private but disguised themselves "Human Capital and Industrial Develop- as township-and-village enterprises (Oi 1999). ment: A Case Study of a Car Repair and Met- Their formation of industrial clusters in Wen- alworking Cluster in Ghana." Unpublished zhou in the early 1980s was facilitated by the mss. Foundation for Advanced Studies on establishment of specialized local markets for International Development, Tokyo. products and materials by local governments, and when these clusters made inroads into Marshall, Alfred. 1920. Principles of Economics. national and international markets later, impor- London: Macmillan. tant roles were played by traders who migrated Mottaleb, K. A., and Tetsushi Sonobe. 2007. "An out from Wenzhou (Xu and Tan 2001). Inquiry into the Rapid Growth of the Knit- 11. In our observations, the wage gap between wear Industry in Bangladesh." Unpublished the coastal and inland regions in China is not so mss. Foundation for Advanced Studies on large as a result of the active migration of work- International Development, Tokyo. ers, despite the restrictions placed on the migra- Oi, J. C. 1999. Rural China Takes off: Institutional tion to urban areas. Foundations of Economic Reform. Berkeley: University of California Press. References Otsuka, Keijiro. 2007. "Rural Industrialization Akoten, John, and Keijiro Otsuka. 2007. "From in East Asia: What Influences Its Nature and Tailors to Mini-Manufacturers: The Role of Development?" In Transforming the Rural Traders in the Transformation of Garment Nonfarm Economy, ed. Steve Haggblade, Peter Enterprises in Kenya." Journal of African Hazell, and Thomas Reardon. Baltimore, MD: Economies 16 (4): 564­95. Johns Hopkins University Press. Akoten, John, Yasuyuki Sawada, and Keijiro Otsuka, Keijiro, Deqiang Liu, and Naoki Otsuka. 2006. "The Determinants of Credit Murakami. 1998. Industrial Reform in China: Access and Its Impacts on Micro and Small Past Performance and Future Prospects. Enterprises: The Case of Garment Produc- Oxford: Clarendon Press. ers in Kenya." Economic Development and Schumpeter, Joseph A. 1912. The Theory of Cultural Change 54 (4): 927­44. Economic Development. New York: Oxford Chan, Vei-Lin, Lii-Tarn Chen, and Sheng-Cheng University Press. Hu. 1999. "Implications of Technology and Sonobe, Tetsushi, John Akoten, and Keijiro Education for Wage Dispersion: Evidence Otsuka. 2007. "An Exploration into the from Taiwan." In The Political Economy of Successful Development of the Leather Shoe Development into the 21st Century, ed. Gustav Industry: Ethiopia." Unpublished mss. Foun- Ranis, Sheng-Cheng Hu, and Yun-Peng Chu. dation for Advanced Studies on International London: Edward Elgar. Development, Tokyo. Chang, Chung-Chau. 1992. "The Development Sonobe, Tetsushi, Dinghuan Hu, and Keijiro of Taiwan's Personal Computer Industry." In Otsuka. 2002."Process of Cluster Formation Taiwan's Enterprises in Global Perspective, ed. in China: A Case Study of a Garment Town." N. T. Wang. New York: M. E. Sharpe. Journal of Development Studies 39 (1): 118­39. Fujita, Masahisa, Paul Krugman, and Anthony ------. 2004. "From Inferior to Superior Prod- Venables. 1999. The Spatial Economy: Cities, ucts: An Inquiry into the Wenzhou Model of Regions, and International Trade. Cambridge, Industrial Development in China." Journal of MA: MIT Press. Comparative Economics 32 (3): 542­63. Henderson, J. Vernon, Ari Kuncoro, and Matt ------. 2006. "Industrial Development in the Turner. 1995. "Industrial Development in Inland Region of China: A Case Study of the Cities." Journal of Political Economy 103 (5): Motorcycle Industry." Journal of Comparative 1067­90. Economics 34 (4): 818­38. 32 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Sonobe, Tetsushi, Momoko Kawakami, and Taiwan." Unpublished mss. Chung-Hua Insti- Keijiro Otsuka. 2003. "Changing Role of tution for Economic Research, Taipei, Taiwan. Innovation and Imitation in Development: Vernon, Raymond. 1966. "International Invest- The Case of the Machine Tool Industry in ment and International Trade in the Product Taiwan." Economic Development and Cultural Cycle." Quarterly Journal of Economics 80 (2): Change 52 (1): 103­28. 197­207. Sonobe, Tetsushi, and Keijiro Otsuka. 2003. Xu, Wei, and K. C. Tan. 2001. "Reform and the "Productivity Effects of TVE Privatization: Process of Economic Restructuring in Rural The Case Study of Garment and Metal Cast- China: A Case Study of Yuhang, Zhejiang." ing Enterprises in the Greater Yangtze River Journal of Rural Studies 17 (2): 165­81. Region." In Governance, Regulation, and Privatization, ed. Takatoshi Ito and Anne O. Yamamura, Eiji, Tetsushi Sonobe, and Kei- Krueger. Chicago: University of Chicago Press. jiro Otsuka. 2003. "Human Capital, Cluster Formation, and International Relocation: The ------. 2006. Cluster-Based Industrial Develop- Case Study of the Garment Industry in Japan, ment: An East Asian Model. Hampshire: 1968­98." Journal of Economic Geography 3 Palgrave Macmillan. (1): 37­56. State Statistical Bureau. 2003. China Statistical ------. 2005. "Time Paths in Innovation, Yearbook. Beijing: State Statistical Bureau. Imitation, and Growth: The Case of the Tu, Chaw-Hsia. 2000. "Industrial Clusters and Motorcycle Industry in Postwar Japan." Network Restructuring for Competitiveness: Journal of Evolutionary Economics 15 (2): The Example of Textiles and Clothing in 169­86. Rural clustering at incipient stages of economic development: hand-weaving clusters in Lao PDR Akihiko Ohno 3 Spatial agglomeration of economic activity is produced by modern technologies penetrate an ineluctable corollary of economic devel- the rural market. Several papers, however, opment, and this presents a major policy reveal that rural nonfarm household indus- concern regarding rural-urban disparity. tries based on indigenous technologies Conventional economic theory predicts (hereafter, rural indigenous industries) that higher factor prices in the agglomerated intensify in tandem with economic develop- area create a dispersion effect. In develop- ment, especially at its incipient stages. ing economies, however, the urban-to-rural Added to this, a large number of studies c h a p t e r trickledown effect has not delivered sat- reveal that rural industrialization takes the isfactory results. For this reason, building form of clustering (Itoh and Tanimoto 1998; prosperous rural industries constitutes an McCormic 1999; Ohno 2001; Sandee and effective step toward achieving more equi- Rietveld 2001; Weijland 1999). The domi- table economic growth (Otsuka, Estudillo, nant form of industrial cluster in developed and Sawada forthcoming). economies is the large company that con- Rural nonfarm industries can be classi- trols networks of small suppliers, while the fied into two types: one based on exogenous dominant form of rural cluster in develop- manufacturing technologies and the other ing economies often is a mere geographic based on indigenous technologies. The first concentration of independent household appears as a result of the spillover of urban producers. As Weijland (1999) points out, industrial activities in accordance with dif- as long as rural households produce small ferentiated factor prices. This type of rural quantities of goods, attaining sufficient industrialization occurs only at the periph- economies of scale to attract traders is a ery of cities as a byproduct of urban indus- major advantage of rural clustering. trialization and only after industrialization In Lao PDR numerous weaving clusters has generated a rural-urban disparity in emerged after economic liberalization. They factor prices. produce cloth intended for urban and over- The second type consists of industries that seas markets. These clusters are village based were originally artisan manufacturing indus- in most, if not all, cases. In the neighbor- tries producing low-quality commodities ing villages of a weaving cluster, only a few intended for self-consumption or, at most, weavers earn enough to support a minimum local markets. They are sluggish by nature standard of living. The difference between and unable to raise their competitive edge villages in a cluster and neighboring villages is sufficiently to embark on a commercial drive the presence of master weavers. These master into wider markets. These features of rural weavers take the role of traders for the clus- industrialization give rise to the argument by ter and work under a putting-out contract. Hymer and Resnick (1969) that rural indus- Individual weavers in neighboring villages, in try is bound to wither as the commodities contrast, must sell their own products in the 33 34 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA marketplace. Traders thus play an important Lao PDR opened its doors under the rubric role in the emergence of rural clusters. of the New Economic Mechanism. This chapter investigates the weaving Being a mountainous country, Lao PDR clusters of Lao PDR with reference to the role has high internal transportation costs, which of traders. It is based on unique case stud- have hindered the formation of a unified ies of weavers surveyed from 1995 to 2006. national market. The main road--Route Hand weaving is a typical rural indigenous 13--that links northern, central, and south- industry, and it opens the way for support- ern Lao PDR was unpaved and poorly main- ing industries, such as sericulture, cotton tained until the turn of this century. This growing, silk reeling, cotton spinning, dye- constricted regional commerce, especially ing industries, and the like. Promoting this during the rainy season. This still holds true type of industry is a more effective measure for the most remote areas, many of which for alleviating poverty in wider rural areas are cut off from the rest of the country dur- than promoting an industry based on exog- ing the rainy season. Commerce with rural enous technologies. This chapter argues that economies thus incurs high transporta- the role of traders in rural weaving clusters tion costs. Reducing transportation costs is more than what the Weijland thesis envis- would contribute a great deal to market ages and that the multimodal characteristics integration. of traders lead to various growth trajectories of weaving clusters. In addition, this chap- Transaction costs ter emphasizes peculiar conditions for rural Asymmetric information and the resulting clustering in a society at an incipient stage opportunistic behavior in market exchanges of development. impede efficient market exchanges by rais- ing the level of transaction costs. The lack Market segregation at an of effective sanction mechanisms aggravates incipient stage of development the problem (Humphrey and Schmitz 1998; Rural and urban markets are often segregated, Mead 1984; Schmitz and Nadvi 1999). especially at early stages of development. In a society at an incipient stage of devel- Transportation costs and transaction costs opment, as is the case for Lao PDR, formal are the two major elements of segregation. sanction mechanisms such as the justice Lao PDR is a landlocked, resource-scarce system and the police are either absent alto- country with the lowest per capita income in gether or, at best, available only in urban Asia. Market development--the integration areas and, even then, to a limited extent. of rural markets into urban markets or even However, recent studies reveal that, even into overseas markets--remains at an initial without formal sanction mechanisms, the stage. Lao PDR has not yet reached high lev- existence of informal trust-based mecha- els of urbanization. However, as a result of nisms checks agents' opportunism. the ASEAN (Association of South East Asian Nations) Free Trade Area (AFTA) regime, Trust the Lao economy has been integrated into To bolster trust, sanction mechanisms are the Thai economy at a rapid pace. In addi- required. The mechanisms range from vil- tion, there exists little difference between lage ostracism, bad reputation, or humilia- Lao and Thai languages. Agglomeration in tion to the simple termination of business the Lao context thus needs to be considered relationships. It is useful to distinguish within a wider regional perspective. between two types of trust: process-based trust and community-based trust. Transportation costs Process-based trust arises through long- After the communist takeover of 1975, Lao term recurrent transactions, as the tit-for-tat PDR was sealed off from the outside world. strategy dictates (Axelrod 1984). This trust However, rapid social transformation of emerges in a dyadic relationship. Commu- the countryside did not take place, and the nity-based trust, on the other hand, rests on rural economy remained largely the natural the norms of obligations and cooperation economy (Evans 1995). In November 1986 that are rooted in social similarity, such as Rural clustering at incipient stages of economic development: Hand-Weaving clusters in Lao PDR 35 ethnicity, being native to the same place, and Table 3.1 Trust of villagers in various economic agents in Lao PDR the like. Community-based trust is reliable percent because it involves social sanction mecha- Economic agent Trustworthy Cannot say Not trustworthy Total nisms as a collective punishment. Thus Insiders once any pair of traders recognizes that they Relatives 81.6 14.6 3.8 100.0 share social similarities, community-based Village people 54.2 32.3 13.5 100.0 Village retailers 41.1 30.3 28.6 100.0 trust comes into play without the need for Village traders 36.4 42.4 31.2 100.0 frequent interactions. Outsiders The two types of trust provide differ- Retailers in towns 10.2 28.3 61.5 100.0 ent sanction mechanisms. Processed-based Traders coming from towns 9.2 31.5 59.3 100.0 trust is attached solely to a relevant trading Source: Author's survey. Note: The number of respondents is 826. Arranged by level of trust. pair, so the cheated will sanction the cheater by terminating their business relationship. Community-based trust entails sanctions arbitrage trader to sell their products, while by other members of the community. In a they wait near the market. The weavers pay community, once sanctioned, the mem- the arbitrage trader a mediation fee of 1 per- ber loses future opportunities to do busi- cent of the sales price. The arbitrage trad- ness with other community members. Fear ers are migrants to Vientiane from the same of such censure works as a self-enforcing area as the weavers: the weavers consider norm that ensures compliance with infor- them to be insiders. mal contracts. Urban and rural markets in developing economies are assumed to be regulated by Severance of trust a different nexus of institutional arrange- The discussion so far suggests that urban ments. Urban agglomeration can be viewed and rural societies maintain incompatible as a systematic integration of various com- sanction mechanisms.1 This incompatibil- mercial customary laws as well as formal ity is assumed to disturb the transactions legal regulations. This accumulation of con- between urban and rural societies. suetudinary law in urban society curtails Table 3.1 shows the level of villagers' trust transaction costs and accelerates further for different groups of economic agents. agglomeration. In rural societies, commu- There exists a distinct severance of trust nity sanction mechanisms govern transac- between insiders and outsiders. Higher trust tions within a community. of insiders suggests the existence of informal Business dealings are, in principle, much sanction mechanisms, while higher distrust easier among people who share the same of outsiders predicts difficulties in transac- institutional arrangements, because trading tions between villagers and outsiders due to parties "speak the same language" (Hicks the lack of effective sanction mechanisms. 1969). In contrast, dealing with outsiders For this reason, commerce beyond the vil- is hindered by the severance of trust, which lage boundary has to cope with transac- works as a force of segregation. tion costs associated with distrust between trading parties. In urban markets, urban Hypotheses on the emergence traders have a marked advantage over vil- lagers in the sphere of market information of traders and negotiation skills. Traders have been evicted from the main- The following anecdote tells how the sev- stream textbooks of economics. Recently, erance of trust hampers the ability of villag- however, there is growing recognition ers to transact with outsiders. In the largest that traders play a critical role in market public market in Vientiane City, Talaat Sao, development (Fafchamps and Minten there are approximately 100 cloth shops.2 1999, 2001). Research on the emergence Many weavers come to the city to sell their of traders in peasant societies classifies cloth to these shops. Some weavers from traders according to three hypotheses: remote areas, however, hesitate to sell their the insider, the outsider, and the culture products themselves. Instead, they use an broker hypotheses. This classification is 36 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA deeply associated with the distinct sanc- come transactional difficulties. Investigating tion mechanisms mentioned above. how these traders engage in their business throws light on the emergence of various · The insider hypothesis (Hayami and types of rural clusters and their different Kawagoe 1999; Landa 1994) argues that growth paths. Taking the severance of trust the difficulty of performing business into account, it is possible to hypothesize transactions under imperfect information that culture brokers play the most crucial can be mitigated when trading partners role in integrating rural markets into urban are fellow villagers, because information and overseas markets. among villagers is relatively symmetrical and informal sanction mechanisms ensure Profile of the Lao PDR the enforcement of contracts. Thus insid- hand-loom industry ers have a better chance of trading suc- This section profiles the Lao hand-loom cessfully with villagers than outsiders do. industry. It discusses income from nonfarm · The outsider hypothesis suggests that peas- activities, offers a brief history of Lao tex- ants in traditional societies are too passive tiles, and discusses techniques for weaving to be entrepreneurial traders and they are Lao cloth. disadvantaged when it comes to having access to urban markets. Thus innovators Income from nonfarm activities emerge from the outside world, such as Table 3.2 shows annual sources of cash the overseas Chinese, the Parsees of India, income for rural households in Lao PDR and the like (Geertz 1963; Weber 1930). (N = 826,surveyed in 2003). A major portion The notion of the trader's dilemma also of cash income comes from self-employed endorses the outsider hypothesis (Evers businesses, while agriculture yields only 18 and Schrader 1994). The dilemma argues percent of cash income. This indicates the that insiders are often required to share importance of self-employed businesses income with their fellow villagers accord- for poverty reduction. These businesses are ing to the social norms of the peasant made up of a variety of rural industries, such community. Insiders are thus subject to as hand weaving, basket making, brewing of social constraints on the maximization of indigenous drinks, bamboo mat making, private profits. In contrast, remaining an and so forth. outsider allows traders to operate a busi- ness without being constrained by the Brief history of Lao textiles community norms of peasant societies. Lao cloth is broadly classified into figured · Culture brokers facilitate and mediate cloth woven in the northern and central relationships among different cultures. part of Lao PDR and ikat (mad mii) woven The culture broker hypothesis stresses mainly in the southern part. This chapter that in developing economies neither deals solely with the former, which is here- urban traders nor village producers are after referred to simply as "cloth." familiar with the rules and norms of the other world. Thus interpreters familiar Table 3.2 Source of income of rural households with both worlds play a pivotal role in in Lao PDR the commercial integration of rural Source of income Amount (1,000 kip) Percent and urban economies. Such a marginal Agriculture 1,854 18.00 person is referred to as a culture broker Agricultural wage 388 3.77 (Neale 1984). Self-employment 4,149 40.55 Salary 2,851 27.67 These hypotheses are not mutually exclu- Remittance sive. A field study on which this chapter is Domestic 83 0.81 based found that a wide variety of traders Overseas 558 5.41 Other income 418 4.06 endowed with different socioeconomic char- Total income 10,305 100.00 acteristics are actively conducting business Source: Author's survey. by using their respective advantages to over- Note: US$1 = 10,500 kip. Rural clustering at incipient stages of economic development: Hand-Weaving clusters in Lao PDR 37 Lao cloth is patterned with ample designs mushrooming of weaving clusters since Lao using supplementary wefts: nonstructural PDR opened its doors? As shown in this threads added to the basic weave. The pat- chapter, traders did much for the prosperity terns of Lao cloth reflect the cultural, reli- of the weaving clusters by coping with the gious, and ritual symbols of ethnic groups, constraints. regions, and even villages. When cloth is woven for home consumption or traded Weaving techniques of Lao cloth within a local community, villagers only The organization of production and struc- need to weave cloth with indigenous patterns ture of marketing vary considerably with the specific to their ethnic group or local com- type of commodity and thus involve distinct munity. In reality, weavers know only a few transactional difficulties (Siamwalla 1978). ancestral patterns. Lao cloth is woven on a hand loom with the For centuries Lao weavers spun cotton help of a vertical heddle.3 A vertical heddle and reeled silk, dyed fabrics with plants, and is the lacy mesh with pattern rods above the wove mainly for personal use rather than warp, which reflects the number of changes for commercial ends (Connors 1996). Hand required to make figured cloth with complex weaving declined rapidly in the 1950s as a patterns. While weavers can produce a verti- result of stiff competition from imported cal heddle of a few traditional patterns, pat- machine-woven cloth. After the communist tern designers can produce various patterns takeover, this situation changed drastically, to be produced on a vertical heddle. as the isolationist policy was conducive to A vertical heddle helps less experienced revival of the hand-weaving industry. weavers to produce cloth with a twill and The recent history of Lao PDR has been satin weave design. In addition, it makes turbulent. Greater contact between ethnic patterns transferable. Weavers can produce and regional groups during the turbulence a vertical heddle by referring to a sample blurred the historical and cultural signifi- cloth, although doing so is arduous work. cance of traditional patterns, making them They copy the information on the pattern merely fashionable designs. of sample cloth by counting the number In November 1986 Lao PDR opened its of warp yarns that appear on the surface doors to the international market economy. for one weft yarn at a time and transfer the This affected the hand-weaving industry in information onto a vertical heddle through two significant ways. First, new markets were the reverse action of the weaving process. brought to the industry by the overseas Lao With this method, weavers can infringe on who left the country at the time of the Civil patterns, causing post-contractual prob- War, the Thais who share a similar culture lems. As the quality of cloth can be visually regarding cloth making, and foreign tour- inspected with ease, adverse selection is not ists. Second, imported factory-spun yarns a serious problem. Thus moral hazard is the enabled weavers to produce high-quality major contractual problem inherent in cloth cloth. These two events led to a weaving transactions. renaissance in Lao PDR. Although the emergence of the new mar- Marketing and contractual kets presented income-generating oppor- arrangements tunities for the villagers, it also imposed new constraints on them, such as the need This section describes marketing channels, to obtain information about patterns the taxonomy of contractual arrangements, demanded in the market and to secure funds and contractual choice. to purchase high-priced imported yarns. Rural weavers seldom have formal training Marketing channels or education, and they face severely con- The majority of weavers are women living strained working capital. Thus rural artisan in villages, because hand weaving is a side clusters are deemed to be passive by nature business for farm households. Master weav- and unable to break into wider markets by ers either employ weavers at their workshop themselves. If this is true, what explains the (hereafter referred to as workshop weavers) 38 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA or deal with weavers working at home (here- the continuum of transactions from the spot after referred to as home weavers), or they market to the factory system, the putting- use a combination of the two (see figure out system is closer to the factory system, 3.1). Master weavers, who are generally pat- while the advance-order system is most akin tern designers, furnish market information to spot-market transactions. on patterns in the form of vertical heddles. The putting-out system often appears Retailers, basically town-based traders, as an interim stop between the craft shop, have their shop in an urban marketplace. or the guild system, and the factory system Other traders observed in our research are (Wardell 1992). Landes (1969) finds the rea- itinerant traders and commission traders. sons for the transition in"the usual difficulty The former reside in a village and deal with of compelling performance by cottage work- fellow villagers, while the latter cover several ers" and "the poor quality of hand work," villages. Independent weavers deal directly which are collectively referred to as "the with retailers without relying on interme- frictions inherent in putting-out." However, diaries. They reside in the vicinity of con- hand weaving is a side business for women, sumption centers. They purchase yarns at mostly married, in rural households. They their own expense and weave cloth of tra- weave when they find time outside of house- ditional patterns they have inherited. The hold chores. Little needs to be said about the cloth is generally of low quality. difficulty of guiding such weavers into a cen- tralized unit. Hence, relational contracting is Taxonomy of contractual the rule in the cloth market. arrangements The relational contract is subject to the Contractual arrangements observed in this agents' opportunism, as Landes points out. field study are found along a spectrum Three major problems are related to Lao that has "make or buy" as polar cases (see cloth: (1) infringement of patterns that prin- figure 3.2). Various relational contract sys- cipals develop; (2) sale of the covenanted tems exist in between. They can be classi- cloth to other traders who offer a higher price fied into three contractual arrangements: than the agreed-upon one; and (3) embezzle- the putting-out system, the yarn-on-credit ment of yarns provided by a principal. system, and the advance-order system. On This section begins by describing the contrast between the putting-out and the yarn-on-credit system. Under the for- Figure 3.1 The marketing process mer, a principal provides his or her agents with yarns for processing and later collects woven cloth for a piece-rate payment. The Domestic provision of yarns is not merely for the con- Overseas market market venience of the agents, who face severe con- straints on working capital; rather, it is also advantageous to the principal, as it binds his or her agents to deliver the cloth with the patterns to which the principal can claim Retailer title. Furthermore, the fact that the prin- Itinerant Commission cipal can claim ownership of the provided trader trader yarns also makes the exclusive delivery binding on the agents. Although the agents' opportunity to sell the covenanted cloth is Master weaver Weaver deterred, embezzlement of yarns remains a serious problem, as the agents (weavers Independent in this case) are tempted to embezzle the Workshop weaver Home weaver weaver provided weft yarn by reducing the number of picks or by saying that not enough weft yarn was provided to weave the covenanted Note: The thickness of the lines corresponds to the volume of transactions. The two-headed arrow denotes that market information on fashionable patterns flows from principal to agents in a dyadic relation. amount of cloth. Rural clustering at incipient stages of economic development: Hand-Weaving clusters in Lao PDR 39 Figure 3.2 Contractual arrangements along a "Make or Buy" spectrum [Make] [Buy] Factory system Yarn-on-credit system Spot transaction Putting-out system Advance-order system [Cloth of high quality] [Cloth of low quality] Under the yarn-on-credit system, in constraints, and they find it difficult to contrast, the principals sell yarns to the obtain market information on fashionable agents on credit in return for stipulating the patterns in urban and overseas markets. purchase price of cloth. When the cloth is Therefore, it is hard for weavers to produce delivered, the agents are paid the remain- good-quality cloth by themselves, even ing balance of an agreed-upon price after though it yields higher profits. Therefore, deducting the price of yarns advanced. The to obtain high-quality cloth, traders need to principals check the quality of the delivered advance appropriate yarn and furnish weav- cloth and may reduce the price if the qual- ers with fashionable patterns. However, this ity is found to be unsatisfactory. The pos- triggers post-contractual problems, such as sibility of punishment for embezzlement of infringement of patterns, embezzlement of yarns is built into the contract in the form yarns, and sale of the covenanted cloth to a of reduced payments after inspection of third party. Traders have to cope with these the delivered cloth. However, this contract problems to run a successful business. attenuates the moral obligation of the agents These contracting systems have distinct to deliver cloth exclusively to the principal, features with respect to post-contractual because the agents can claim ownership of problems (see table 3.3). This chapter con- the yarns. Therefore, the infringement of siders community-based trust and the abil- intellectual property rights on cloth patterns ity to place a hold on patterns to be the or the sale of covenanted cloth to a third sanction mechanisms that constitute the party is more likely to be serious. principal's ascendancy over an agent. Under the advance-order system, the A hold on patterns depends on the pri- principals only guarantee the purchase of vate information of the principals. Master the products at a stipulated price. As yarns weavers, when they are principals, can pre- are not advanced, the agents have to pur- serve their hold on patterns because they chase them at their own expense. Therefore, are the designers as well. Urban retailers as the agents cannot help but use low-quality principals can preserve a hold on patterns yarns due to their weak financial position. as a result of day-to-day contacts with con- Needless to say, yarn embezzlement does not sumers. As cloth with an attractive pattern take place in this system. However, agents enjoys stable orders and higher prices, the are vulnerable to the temptation to sell the agents contracting with principals who have products to traders offering higher prices. a strong hold on patterns can secure constant Therefore, under this system the principals orders. Hence, the threat of termination of refrain from furnishing their agents with the business relationship deters the agents' market information on patterns, and the opportunism. Process-based trust is not con- cloth remains of low quality. sidered here, because relational contracting Besides these relational contracts, spot- generally involves recurrent transactions. market transactions are common. Although no post-contractual problems take place in spot transactions, the cloth transacted is of Table 3.3. Post-contractual problems under the different contracting systems very poor quality. Sale to others and Contracting system pattern infringement Yarn embezzlement Contract choice Putting-out system Weak Large Yarn-on-credit system Medium Medium Rural weavers cannot afford to purchase Advance-order system Large None high-quality yarns due to working capital Source: Author's survey. 40 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA In the interest of brevity, the discussion chosen even though only weaker benefits is limited to the choice of contract between can be obtained by linking rural producers weavers and master weavers. A similar pic- to urban markets. Spot-market transactions ture can be applied to the choice of contract prevail when the principals are unable to in other transactions. restrain the agents' opportunism, and the For rural weavers, the putting-out system quality of cloth deteriorates accordingly. is the first choice, because the principals, who are master weavers as a general rule, advance Weaving clusters and traders good-quality yarns and furnish market infor- Among numerous weaving clusters in Lao mation on fashionable patterns. In addition, PDR, this study selected 11 clusters that the principals guarantee the purchase of the show distinct growth trajectories (see table processed products. For the principals, this 3.4). It is assumed that the dynamics at work effectively protects their intellectual prop- can be explained by the ability of traders to erty rights on patterns, as this system assures use different sanction mechanisms. exclusive delivery of the products. Thus the putting-out system provides the best condi- Why does space matter? tions for both agents and principals. We start with two cases in which distance The agents' opportunism is, however, an determines the features of clusters. Xam inherent problem in the putting-out system. Nua and Xam Tai of Houa Phan province, It can be assumed that the principals who the most remote region of Lao PDR, are have strong ascendancy over agents are more renowned for their hand weaving (see fig- likely to choose the putting-out system. As ure 3.3). Xam Nua is three days by bus from a natural corollary, traders who adopt the Vientiane, while Xam Tai is four. The conse- putting-out system rather than conduct- quence is that the margin between the whole- ing business through spot transactions can sale price and the purchase price is about 5 establish prosperous weaving clusters. percent for the Xam Nua trader and 47 per- When the principals can rely on only one cent for the Xam Tai trader. This difference or the other sanction mechanism, meaning reflects the distance of "an extra day." that they have weak ascendancy over agents, The distance of "an extra day" does not post-contractual problems arise.The problem denote a mere increment of transporta- of yarn embezzlement surfaces first, because tion cost, because the bus fare is too small a sense of guilt is weaker for yarn embezzle- to explain the difference in profit margin. ment than for the sale of cloth to a third Traders in Xam Nua, mostly master weavers, party. The principals are obliged to choose take orders from urban retailers in a recur- the yarn-on-credit system to cope with yarn rent manner, receiving product information embezzlement,even though the agents'moral on fashionable patterns and colors under the obligation of exclusive delivery is weak. advance-order system.4 Some master weav- When the agents' opportunism is more ers have constant orders from Thai retailers hazardous, the advance-order system is who specify patterns and colors in detail. In Xam Tai, master weavers do not exist. Table 3.4 Surveyed clusters in Lao PDR Xam Tai weavers consign their cloth to com- mission traders, setting the price limit. The Cluster Province traders sell the products to urban retailers Distance matters? in Luang Prabang and Vientiane. Although 1 Xam Tai in Houa Phan province 2 Xam Nua in Houa Phan province Xam Tai weavers are known to be highly Traders having weak governance skilled, they weave cloth of low quality. 3 Itinerant trader in Vientiane province When asked why they do not weave qual- 4 Commission trader in Vientiane ity cloth, they responded that they fear the Master weavers in different settings risk of frozen stock. It takes several months 5­8 See table 5 for details Three contrasting clusters in Luang Prabang to weave a piece of high-quality cloth. This 9 Village X: Several insiders exporting to Thailand makes them risk averse. The same holds true 10 Village Y: No effective traders for the traders. Distance of "an extra day" 11 Village Z: A public marketplace for cloth does not impose prohibitive transportation Rural clustering at incipient stages of economic development: Hand-Weaving clusters in Lao PDR 41 Figure 3.3 Map of survey sites in Lao PDR Hoei Xam Xai Nua Luang Phrabang Xam Tai Vientiane Capital Savannakhet Survey Sites costs, but it affects the attitudes of the weav- the traders do not share a similar social back- ers as well as the traders in association with ground with the weavers. Therefore, itinerant risk. Thus distance hampers the emergence traders cannot resort to community-based of master weavers. trust for resolving conflicts. As Ms. P is from a village renowned for hand weaving, she Clusters producing low-quality possesses market information on fashionable cloth: itinerant and commission patterns. However, she hesitates to furnish traders the information to her weavers, because she An itinerant trader, Ms. P deals with approx- cannot count on them to deliver cloth exclu- imately 300 weavers under the yarn-on- sively to her: she has no power over them.5 credit system, covering several villages in Thus Ms. P trades cloth of low quality. Vientiane province about 80 to 90 kilome- Ms. Q, a commission trader in a village ters north of Vientiane City. As the weavers about 40 kilometers fromVientiane,contracts know only a few indigenous patterns, there with 56 village weavers under the yarn-on- is weak demand for that line of cloth. credit system. She also runs a small general Itinerant traders facilitate the market store. As there is neither a pattern designer activities of weavers in remote areas. They nor a master weaver in the village, the vil- cover wider areas than commission traders lagers weave cloth with a few indigenous or master weavers do. However, wider cover- patterns that fetch low prices. The products age itself inevitably entails a drawback in that go mainly to local markets. Ms. Q not only 42 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA advances yarn on credit but also sells daily for overseas markets (see table 3.5). Two of necessities to the weavers on credit to secure them reside in Vientiane, and one resides the delivery of cloth from them. In addition, in Savannakhet, the second largest town she lends money to her weavers without in Lao PDR. Each master weaver contracts charging interest and accepts repayment in exclusively with her own weaving cluster on cloth. The credit-tying contract is a device a large scale. As the table shows, the master that allows the commission trader to secure weavers are not insiders: they live in urban the delivery of cloth from the weavers. areas and are medical doctors or graduates Although Ms. Q shares a similar social of foreign universities. They have acquired background with the weavers, she suffers the rules and norms of urban society. At the from the weavers' opportunism. Four to five same time, they contract with migrant weav- weavers per month sell their cloth directly ers from their native area, with the exception to retailers in the market without delivering of Ms. C. it to Ms. Q. As she knows the cloth pattern Ms. A migrated to Vientiane from Houa of each weaver, such opportunism is eas- Phan province during the Vietnam War. She ily detected when she visits the market in quit a government post (as a medical doc- Vientiane. However, Ms. Q thinks there is tor) due to low wages and started a weav- no way to prevent this from taking place. ing business in 1986 when Lao PDR opened Because she is not a pattern designer, her its doors. She deals exclusively with more ascendancy over the weavers is weak. than 200 weavers from two villages under a Itinerant and commission traders have putting-out contract. The villagers migrated weak ascendancy over their agents in that from Houa Phan province for the same rea- the former cannot resort to community- son as their principal. Ms. A is known as an based trust and the latter have a weak hold eminent pattern designer in Vientiane. She on patterns. Thus they are subject to the provides her weavers with vertical heddles agents' opportunism. Even though they con- embodying patterns that she has designed tribute to the formation of weaving clusters, and silk yarns that she has dyed. During the the clusters produce low-quality cloth. So dry season, Ms. A accepts the delivery of an long as these traders do not furnish market average of 500 to 600 pieces of cloth from information on patterns, they cannot be a her weavers. The number drops to about substitute for the master weavers. half during the rainy season when farm work reaches its peak. Her weaving business Clusters producing products brings monthly earnings of 8.19 million kip intended for overseas markets (US$8,712 in August 1996) in the dry sea- This section reviews three of several doc- son. Her success is due to community-based umented cases of master weavers who trust and her strong hold on patterns. contributed to the formation of weaving Neighboring villages are different from vil- clusters in which the products are intended lages with a weaving cluster, even though they Table 3.5 Social characteristics of the large-scale master weavers surveyed in Lao PDR Social similarity Number of weavers Name and birthplace Description with weavers under control Ms. A. (Hua Phan) A medical doctor and Yes 219 an eminent pattern designer Ms. B. (Saravan) Has a long-term contract Yes More than 200 with a large cloth shop in Japan; lives in Savannakhet and runs a weaving workshop there Ms. C. (Savannakhet) University graduate No 57 (France) and can communicate in French and English Source: Author's survey. Rural clustering at incipient stages of economic development: Hand-Weaving clusters in Lao PDR 43 have similar historical backgrounds. While In conclusion, the master weavers who almost all of the households in the cluster are involved in rural weaving clusters on engage in weaving, only some households a large scale are those who can exploit an in the neighboring villages weave. This is ambivalent position as outsiders as well as because the weavers there do not have a con- insiders. They are socially differentiated tract with master weavers. They do not have from their agent weavers in that they are access to the fashionable patterns desired in highly educated and have urban experience. urban markets. In addition, they lack work- They are culture brokers. As they reside in ing capital and can only afford cotton yarns. or in the vicinity of large towns far from Cotton cloth has a narrow profit margin their agent weavers, their businesses are less compared to silk cloth. Master weavers who likely to be constrained by the community are also traders thus have an important role norms that demand income redistribution. to play in the formation of weaving clusters. Although direct monitoring is difficult, Ms. B was born in a village in Saravan community-based trust and a strong hold on province and moved to Savannakhet (the patterns deter the opportunistic behavior of capital of Savannakhet province) after her agent weavers. This allows the master weavers marriage. In 1998 she started a weaving to conduct a large-scale weaving business. business with support from a Japanese dyer The discussion so far reveals the two faces who recommended indigo dyeing. Since of community norms.While the community 1999 she has contracted with a medium-size norms often facilitate efficient transactions cloth shop in Japan that furnishes designs by producing community-based trust, the for the cloth. What is noteworthy is that Ms. community may impede capital accumula- B contracts with the weavers of her native tion in the hands of inside traders, creating village about 70 kilometers away from her a trader's dilemma. It is the culture broker residence. She adopts the putting-out sys- who can maneuver through these two faces tem, even though monitoring is difficult. of the community most effectively. The cloth woven in this case is plain, unlike the figured cloth of other master weav- Three clusters taking distinct ers. Thus Ms. B does not provide vertical trajectories in Luang Prabang heddles to her weavers. Her strong hold on A final set of weaving clusters are from product specifications is derived from her Luang Prabang, the former royal capital and long-term business relations with the for- a World Heritage site that lies on the upper eign retailer. This relation and community- Mekong River. On the fringe of the town, based trust enhance her ascendancy over the there are three weaving clusters: clusters X, agent weavers. Y, and Z. The residents of clusters X and Y Community-based trust helps to curtail were on the side of the ousted regime. After monitoring costs. Ms. C, a master weaver the communists came to power in 1975, from Vientiane, illustrates the role of com- they were forced to settle on the Mekong munity-based trust. Ms. C graduated from riverbank, where little farmland is available. university in France and exports her prod- A typical picture of proto-industrialization ucts to France as well. She is a migrant from emerged there. the southern province of Savannakhet where In 1994 a weaver of cluster X (Ms. E) was weaving is not prosperous. In Vientiane and approached by a Lao trader, who was a bro- its environs, she is unable to find weavers ker for Thai traders from Hoei Xai, a trading from her native village, which has forced post on the opposite shore of Chiang Kong her to deal with migrant weavers from Houa in Thailand. The trader travels by boat along Phan province. Due to the lack of commu- the Mekong River. The weaver entered into a nity-based trust with the weavers, Ms. C has contract with the broker and embarked on a to employ three local people to monitor the weaving business as a master weaver. Several weavers' opportunistic behavior. She is the weavers in cluster X followed the same path only master weaver discussed here to rely on as Ms. E. such supervisors; the others rely instead on At the end of the 1990s,a road to Nan (Nan community-based trust. province of Thailand) through Xayabury 44 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA province was opened to traffic. Master weav- the second half of the 1990s. The kip depre- ers in cluster X began to transact directly with ciated sharply against the U.S. dollar, with cloth retailers in Nan. Ms. F had worked as a wide fluctuations, from US$1 equivalent to broker of master weaver Ms. E. She brought 940 kip in August 1996, 2,150 kip in Decem- the cloth to a retailer in Nan and conveyed ber 1997, and more than 10,000 kip in 2000. to Ms. E the information on pattern and Consequently, the price of imported yarns as color specified by the retailer. In the course well as the wholesale price of cloth rose dras- of her work, Ms. F became a trader and spun tically. This is particularly true for quality silk off, taking some of the weavers away from yarns for warps imported from China. Ms. E. In the meantime, several other master The devaluation of the kip over a pro- weavers emerged in cluster X. longed period made agents think that the In cluster Y, no master weavers emerged. profit from one-shot opportunism would This is possibly because there are more outweigh the future benefits obtained from weavers in cluster X, which consists of recurrent transactions. In addition, several several villages located on the riverbank. traders started to offer higher prices for This scale-economy effect may explain why cloth, encouraging weavers to breach their Hoei Xai traders first approached cluster X. contracts. In the meantime, the weaving activities of This section considers the example of cluster Y were absorbed by the traders of master weaver Ms. A to illustrate the decline other clusters. of trust and consequent institutional Cluster Z followed a different pathway. In changes. Although Ms. A is an eminent pat- 1978 the provincial government constructed tern designer who contracts with weavers a market hall for selling cloth to tourists with whom she shares a similar social back- on the condition that the villages sell their ground, she could not avoid the turmoil. products only at the hall. The hall was reno- During a subsequent interview in 2000, after vated in 1991 and has 73 seats at present. the turmoil had subsided, she said that she Some villagers collect cloth from other areas, had changed from using putting-out con- including cluster Y, through spot transac- tracts to using the yarn-on-credit system. tions. As the majority of customers are for- She introduced contract letters that have the eign tourists who buy indigenous cloth as following four provisory clauses: (a) a per- a souvenir, the weavers do not have to take son who receives advanced yarns is required into account the taste of wider markets. This to weave the stipulated sheets of cloth; cluster does not foster entrepreneurship. (b) the cloth delivered needs to be inspected, While the weaving industry of cluster Z has and substandard quality leads to a reduction stagnated, several shops opened in cluster X in the agreed-upon price; (c) breach of this to cater to the growing tourist traffic. This is contract incurs penalty charges worth twice because the weavers of cluster X are capable as much as the stipulated price of cloth; and of taking in market information on patterns (d) when the pattern provided by the prin- and colors, especially those of Thais. They cipal is transferred to a third person, 50,000 were quick to seize the opportunity to weave kip has to be paid as a fine. It is a good high-quality cloth. guess that Ms. A has suffered from agents' The above contrast emphasizes the opportunism. importance of traders in facilitating rural Process-based trust can be obtained after clusters. In addition, any policy intervention engaging in time-consuming recurrent that blocks the transmission of market infor- transactions, as the Nash equilibrium in the mation will not promote rural clusters. repeated prisoner's dilemma game (Axelrod 1984). This equilibrium can be maintained Decline of trust on condition that the payoff matrix remains As trust in trade is calculating by nature, stable. The currency turmoil affected the it may decline as the level of punishment payoff over a long period of time, increas- decreases or the level of profit from a breach ing the gains from one-shot opportunism.As of contract increases. Decline of trust the case of Ms.A indicates, the upheaval even occurred during the economic turmoil in destroyed community-based trust. After the Rural clustering at incipient stages of economic development: Hand-Weaving clusters in Lao PDR 45 currency crisis, the trust-based contracts retailers who cannot rely on community- were in some cases replaced by spot transac- based trust, the relational contract is vul- tions. Policy makers should take note: a sta- nerable to agents' opportunism. Currency ble macroeconomy is critically important for turmoil added momentum to this shift. preserving informal sanction mechanisms, Urban retailers cannot be active market which support market development. integrators because they do not share a simi- lar social background with weavers and they Retailers as urban-based traders do not design patterns. When the retailers This section shifts the focus to retailers, adopted the relational contracting system, examining whether they contribute to rural they gained a distinct selection of goods. clustering in a positive manner. The fol- Since the currency turmoil, however, things lowing is based on interviews with 13 cloth have turned around drastically. The retailers retailers in Talaat Sao. in Talaat Sao now sell a similar range of cloth, Most retailers in Talaat Sao neither have because they began to buy cloth on a spot- a hold on patterns nor are able to rely on market basis. A retailer said that they now community-based trust. A typical example have to compete not on quality but on price. is Ms. K of Vientiane origin. She has been purchasing cloth from independent weavers Conclusions residing on the periphery of Vientiane on a This chapter has shown that the emergence spot basis for more than 20 years. Over time, of traders is indispensable for the devel- she has selected "skilled and reliable weav- opment of rural clusters; traders who can ers" (in her words) one by one and entered maneuver incompatible sanction mecha- into yarn-on-credit contracts with them. nisms between rural and urban societies play In this case, recurrent transactions a particularly important role. This notion is engendered process-based trust. Clien- critically vital for the transaction of com- tilism, however, does not always assure modities, which entails post-contractual effective enforcement of contracts. An oft- problems. The weaving clusters observed heard complaint from retailers, including have followed distinct growth trajectories. Ms. K, is that some weavers do not deliver This is largely explained by the multimodal the promised products even if the retailers characteristics of traders, who have different advance yarns on credit. Embezzlement sets of sanction mechanisms. is also commonplace. According to Ms. K Insiders such as commission and itiner- (interviewed in December 1997, just after ant traders contribute to the establishment the currency turmoil broke out), 7 agent of rural clusters. However, they engage in weavers among 50 did not deliver the prod- cloth business on a small scale, dealing with ucts. She visited them to urge delivery, but cloth of poor quality intended for domestic they pleaded illness or the like. She suspects markets. They are unable to access informa- them of selling off the advanced yarns for tion regarding the taste of overseas consum- rice. To reduce such damage, she tries to ers or are unable to resort to community contract with weavers of different vil- sanction mechanisms. In addition, they are lages as much as possible. This is because, subject to the trader's dilemma. according to her, when such a problem Culture brokers deal with quality cloth occurs with one weaver, it spreads to others for overseas high-end consumers on a large in the same village. This is a case of conta- scale. They not only have access to informa- gious equilibrium. tion on the taste of wider markets but also Ms. K decided to replace the yarn-on- can resort to community sanction mecha- credit system with the advance-order system. nisms. Thus they can furnish rural produc- Some retailers in Talaat Sao also switched ers with information on fashionable patterns from the yarn-on-credit system or even the while evading post-contractual problems. putting-out system to the advance-order Hicks (1969) and North and Thomas system. The main reason for this shift was (1973) argue that the evolution from per- the rise in agents' opportunism, as the num- sonalized exchange to impersonal exchange, ber of traders in the market increased. For supported by legal sanction mechanisms, is 46 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA central to the process of growth and devel- 2. Vientiane City (hereafter Vientiane) is the opment. However, this chapter has discussed capital of Lao PDR, while Vientiane province the evolution before formal sanction mech- extends to the north of Vientiane. anisms have been implemented effectively. 3. Heddle is the name of the small cords through which the warp is passed in a loom The findings suggest that policies to fos- after going through the reed and by means of ter traders need to specify a target. Fostering which the warp yarns are separated into two sets culture brokers is not an easy task. Educat- to allow passage of the shuttle bearing the weft. ing able villagers from target areas in urban 4. Urban businesses do not advance yarns to settings serves as the initial step. Educa- the weavers of Xam Nua because yarns are avail- tion not only enhances the human capital able in local markets. In the case of bulk orders, of villagers, but also helps them to acquire the urban principals advance money to the mas- the property of outsiders, including urban ter weavers. experiences when education is obtained in 5. In spite of the weavers' opportunism, an urban setting. Thus education facilitates Ms. P has to advance yarns to secure the the emergence of culture brokers. collection of cloth because several itinerant trad- ers are doing business in the area. A possible measure is to hold trade fairs to assist local producers in establish- References ing connections with urban and overseas businesses. Take the Japanese experience in Axelrod, Robert. 1984. The Evolution of Coopera- the early Meiji period, for example, when tion. New York: Basic Books. numerous trade fairs were held all over the Connors, Mary F. 1996. Lao Textiles and Tradi- country. This disseminated information tions. Kuala Lumpur: Oxford University Press. on rural commodities to urban businesses. Evans, Grant. 1995. Lao Peasants under Socialism Ohno and Jirapatpimol (1998) point out and Post-Socialism. Bangkok: Silkworm Books. that some master weavers in northern Thai- Evers, Hans, and Heiko Schrader, eds. 1994. The land gained a link with urban and overseas Moral Economy of Trade: Ethnicity and Devel- markets through such trade fairs. oping Markets. London: Routledge. Needless to say, informal sanction mech- Fafchamps, Marcel, and Bart Minten. 1999. anisms need to be superseded by formal "Relationships and Traders in Madagascar." sanction mechanisms in the course of eco- Journal of Development Studies 35 (6): 1­35. nomic development. The implementation ------. 2001. "Property Rights in a Flea Market of formal sanction mechanisms needs to Economy." Economic Development and Cul- prioritize areas related to the transactions tural Change 49 (2): 229­67. between foreign outsiders and domestic Geertz, Clifford. 1963. Peddlers and Princes. traders as long as the commodity possesses Chicago: The University of Chicago Press. an international competitive edge. Another important policy implication Hayami, Yujiro, and Toshihiko Kawagoe. 1999. of this chapter is that economic uncer- "Middlemen in a Peasant Community: Vegitable Marketing in Indonesia." tainty jeopardizes trust by inducing one- In Communities and Markets in Economic shot opportunistic behavior and leading to Development, eds. Masahiko Aoki and Yujiro a payoff matrix and a prisoner's dilemma. Hayami. New York: Oxford University Press. A stable macroeconomy is thus an impor- Hicks, John. 1969. A Theory of Economic History. tant ingredient of both process-based and Oxford: Oxford University Press. community-based trust for economic development. Humphrey, John, and Hubert Schmitz. 1998. "Trust and Inter-firm Relations in Developing Notes and Transition Economies." Journal of Devel- opment Studies 34 (4): 32­61. Akihiko Ohno is a professor at Aoyama Gakuin University. Hymer, Stephen, and Stephen Resnick. 1969. "A 1. Transactions relying on community-based Model of an Agrarian Economy." American trust segregate rural communities, impeding the Economic Review 59 (4): 493­506. spillover effects of clustered activities in rural Itoh, Motoshige, and Masayuki Tanimoto. 1998. areas. "Rural Entrepreneurs in the Cotton-Weaving Rural clustering at incipient stages of economic development: Hand-Weaving clusters in Lao PDR 47 Industry of Japan." In Toward the Rural-Based Ohno, Akihiko. 2001. "Market Integrators for Development of Commerce and Industry, ed. Rural-Based Industrialization: The Case Yujiro Hayami. Washington, DC: World Bank, of the Hand-Weaving Industry in Laos." Economic Development Institute. In Communities and Markets in Economic Landa, Janet T. 1994. Trust, Ethnicity, and Iden- Development, ed. Masahiko Aoki and Yujiro tity: Beyond the New Institutional Economics Hayami. New York: Oxford University Press. of Ethnic Trading Networks, Contract Law, Otsuka, Keijiro, Jonna P. Estudillo, and Yasuyuki and Gift Exchange. Ann Arbor: University of Sawada, eds. Forthcoming. Rural Poverty and Michigan Press. Income Dynamics in Asia and Africa. London: Landes, David S. 1969. Unbound Prometheus: Routledge. Technological Change and Industrial Develop- Sandee, Henry, and Piet Rietveld. 2001. ment in Western Europe from 1750 to the Pres- "Upgrading Traditional Technologies in ent. Cambridge, MA: Cambridge University Small-Scale Industry Clusters: Collaboration Press. and Innovation Adoption in Indonesia." Jour- McCormick, Dorothy. 1999. "African Enterprise nal of Development Studies 37 (4): 150­72. Clusters and Industrialization: Theory and Schmitz, Hubert, and Khalid Nadvi. 1999. Reality." World Development 27 (9): 1531­51. "Clustering and Industrialization: Introduc- Mead, D. C. 1984. "Of Contracts and Subcon- tion." World Development 27 (9): 1503­14. tracts: Small Firms in Vertically Disintegrated Siamwalla, Ammar. 1978. "Farmers and Middle- Production/Distribution Systems in LDCs." men: Aspects of Agricultural Marketing in World Development 12 (11­12): 1095­106. Thailand." Economic Bulletin for Asia and the Neale, Walter C. 1984. "The Role of the Broker Pacific 29 (1): 38­50. in Rural India." In Rural South Asia: Linkages, Wardell, Mark. 1992. "Changing Organizational Change, and Development, ed. Peter Robb. Forms: From the Bottom up." In Rethinking London: Curzon Press. Organization: New Directions in Organiza- North, Douglass C., and Robert P. Thomas. tion Theory and Analysis, ed. Mike Reed and 1973. The Rise of the Western World. Michael Hughes. London: Sage. Cambridge: Cambridge University Press. Weber, Max. 1930. The Protestant Ethic and Ohno, Akihiko, and Benja Jirapatpimol. 1998. the Spirit of Capitalism. London: Allen and "The Rural Garment and Weaving Industries Unwin. in Northern Thailand." In Toward the Rural- Weijland, Hermine. 1999. "Microenterprise Based Development of Commerce and Industry, Clusters in Rural Indonesia: Industrial Seed- ed. Yujiro Hayami. Washington, DC: World bed and Policy Target." World Development 27 Bank, Economic Development Institute. (9): 1515­30. Spatial networks, incentives, and the dynamics of village economies: evidence from Indonesia Futoshi Yamauchi, Megumi Muto, Reno Dewina, and Sony Sumaryanto 4 Economic growth often accompanies spa- that are available in larger economic centers tial inequality. Spatial connection to high- farther than the local town (without migrat- growth centers offers a pathway out of ing). In this case, road access to the larger poverty in local economies, by improving economic center is more important. economic returns to investment and reduc- Similarly, if the local town has efficient ing costs in transportation and the search for markets for agricultural products, landed both human and physical resources, which farmers will benefit from the new local c h a p t e r alters the allocation of household resources. road, but landless households will not, In general, an improvement in spatial con- because farmers have surplus products to nectivity is expected to increase allocative sell, whereas landless households do not. efficiency in the local economy, because the Increasing demand for food from larger mobility of resources becomes faster and less economic centers may induce some landed costly and price disparity becomes smaller farmers to invest in agroprocessing, increas- (for example, Minten and Kyle 1999). ing nonfarm income. In this case, the effects However, it is not clear how better spa- could be heterogeneous across different tial connectivity--among neighborhood locations and across households with dif- local areas or between local areas and dis- ferent endowments. tant economic centers--changes poverty The recent literature provides some and income distribution. In other words, studies suggesting that returns to human it is not clear who gains from better spatial and physical capital in rural areas critically connectivity. Improved spatial connectivity depend on spatial connectivity, which affects in the local economy may have heteroge- the allocation of household resources,such as neous impacts on households with different labor (see Fafchamps and Shilpi 2003, 2005; endowments. To investigate this issue, we Fafchamps and Wahba 2006). Fafchamps use recently available household panel data and Shilpi (2003) show that the distance to from Indonesia. cities is crucial for determining wage oppor- In rural contexts, once a village is con- tunities and employment structure in Nepal, nected by a new road to a nearby town and thus nonfarm employment (either where jobs are available, the household wage or self-employment) is concentrated allocation of labor is expected to change so in and around cities. Since road construc- that they gain from earning opportunities tion improves the access to (nonagricul- in the town's labor market. If entry to the tural) labor markets or urban consumers, it labor market is easier for educated agents, increases wages and employment choices for the allocation of labor changes among rural residents. Cerain types of employment households with educated members. More become available with improved spatial link- educated agents may try to capture better ages. For example, Fujita and Muto (2007) employment or urban market opportunities show that the effect of spatial linkages on Spatial networks, incentives and the dynamics of village economies 49 brand agriculture depends on the differen- Potential gain in allocative efficiency is also tiation of products. affected by the distance to economic centers The connectivity to urban centers can at different levels, as these economic centers benefit laborer households more than farm offer different economic opportunities. (landed) households by improving the access Previous studies on the spatial connec- to nonagricultural employment opportuni- tivity of rural households were limited in ties. Foster and Rosenzweig (2001) present the sense that they perceived connectivity recent evidence from India that the landless only as access to local towns or as distance prefer public investment in local road con- from growth centers and were unable to dis- struction because it improves their access cuss the combination of both. But in actual to labor markets, while the landed prefer policy choices, public investment planners investment in irrigation because it augments face decisions regarding the allocation of returns to land. resources among trunk roads (which lead The improvement in spatial connectivity to economic centers) and local roads. They also has implications for product markets, also face policy choices regarding the bal- reducing transportation margins. Minten ance between fiscal spending on education and Kyle (1999) show that price variations and on roads. Therefore, this paper can are largely due to transportation costs in bridge the gap between academic studies the Democratic Republic of Congo. An and infrastructure planning. interesting finding is that traders benefit Empirical results show that improve- from bad road conditions, which lower the ments in the quality of local roads in the purchase price of products (thus increasing local area (which are positively correlated their profit). Therefore, spatial connectivity with speed of transportation) have an can potentially increase farmers' income by impact on income growth and the transition reducing traders' profit margin. to nonagricultural activities and that the Numerous studies have estimated impact depends on the distance to economic the returns to infrastructure investment centers and household education. Education such as road construction under various significantly increases the benefit from an assumptions, mostly at the aggregate level improvement in spatial connectivity, which (Binswanger, Khandker, and Rosenzweig is augmented by distance from the provin- 1993; Fan, Zhang, and Zhang 2004). To cial center. Education and local road qual- analyze the dynamic effects of infrastruc- ity are complementary, increasing income ture investment on income growth at the growth. Therefore, whether the improve- household level, it is necessary to combine, ment in local connectivity (measured by by household and village locations, both average road quality) is pro-poor or not household and spatial panel data over a long depends on village location and the initial span of time with sufficiently large changes household-level endowment of education. in infrastructure. In this paper, we endeavor to capture Data the improvement in spatial connectivity by We use data from two sources. First, the main constructing a measure that captures inter- data come from village- and household-level village road quality in a region (from the surveys that we conducted in 2007 for 98 vil- Indonesian village census). We combine this lages in seven provinces (Lumpong, Central measure and distance to economic centers: Java, East Java, West Nusa Tenggara, South subdistrict, district, and provincial capi- Sulawesi, North Sulawesi, and South Kali- tals (from the village survey we conducted mantan) under the Japan Bank for Inter- in 2007). Our main idea is that intervil- national Cooperation (JBIC's) Study of the lage road quality determines the means of Effects of Infrastructure on Millennium transportation used in the local economy Development Goals in Indonesia (IMDG). and therefore the average speed of resource The 2007 village survey captured the physical mobility (including human), which affects distance and time to various points of eco- allocative efficiency in the local economy. nomic activity such as markets, stations, and 50 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA capital towns. Figure 4.1 shows the location using both the 2007 household and 1994­95 of surveyed villages. PATANAS surveys.1 The survey was designed to overlap with Second, 1996 and 2006 PODES data were villages in the 1994­95 National Farmer's used to construct road quality data. PODES Panel (PATANAS) survey conducted by is a village census conducted by the Republic Indonesian Center for Agriculture Socio- of Indonesia's Central Bureau of Statistics. Economics and Policy Studies (ICASEPS) to build household panel data. The 1994­95 Descriptive analyses PATANAS survey focused on agricultural This section describes the data dealing with production activities in 48 villages cho- spatial connectivity (specifically intervil- sen from different agroclimatic zones in lage road improvement and distance to seven provinces. In 2007 we revisited those economic centers) and household income villages to expand the scope of research (specifically income dynamics and nonfarm through a general household survey con- self-employment). ducted under the IMDG survey. In the 2007 round, therefore, we added 51 new villages Spatial connectivity in the seven provinces. In this section we describe village census In the revisited villages, we resampled data (PODES) with a focus on transporta- 20 households per village from the 1994­95 tion and road quality variables and charac- sample and followed the split households. terize changes in local road quality in the In the new villages, we sampled 24 house- period of 1996 to 2006. The data cover all holds from two main hamlets in each village. villages in the census years. For our research, Because one of the 48 villages in the 1994­95 we use the 1996 and 2006 rounds of PODES, PATANAS (in West Nusa Tenggara prov- as our household panel data were collected ince) was not accessible for safety reasons in in 1995 and 2007. In the panel analysis, we 2007, 98 villages were available for various take the difference between 1996 and 2007 research objectives. In our panel analysis, we to represent changes in the average quality constructed household income panel data of roads in local economies. from 34 villages in six provinces (Lumpong, The PODES data have the information Central Java, East Java, West Nusa Teng- on major intervillage traffic. If the major gara, South Sulawesi, and North Sulawesi) traffic is on land, the survey asks about the Figure 4.1 Location of surveyed villages in Indonesia MALAYSIA Manado KALIMANTAN Tanjungpinang TIMUR SULAWESI RIAU GORONTALO KEPULAUAN Pontianak Gorontalo UTARA KALIMANTAN BARAT Samarinda Palu KALIMANTAN SULAWESI TENGAH SULAWESI TENGAH MALUKU Pangkalpinang Palangkaraya BARAT Mamuju UTARA BANGKA-BELITUNG Palembang KALIMANTAN SUMATERA Bandjarmasin Kendari SELATAN SULAWESI SELATAN SULAWESI TENGGARA LAMPUNG SELATAN Bandar D.K.I. JAKARTA Makassar Lampung JAKARTA Serang JAWA TENGAH 12 Bandung Semarang JAWA TIMUR BANTEN JAWA BARAT Surabaya NUSA TENGGARA BARAT NUSA TENGGARA TIMUR Yogyakarta BALI D.I. YOGYAKARTA Survey Sites TIMOR-LESTE Denpasar Mataram Source: National Coordinating Agency for Surveys and Mapping for boundaries as of 1990; GPS coordinates collected during IMDG 2007 for the location of surveyed villages. Spatial networks, incentives and the dynamics of village economies 51 type of widest road for this purpose: asphalt, Table 4.1 Proportion of asphalt roads in concrete, or cone block; hardened; soil; and intervillage roads in Indonesia, 1996 and 2006 (provincial average) others. Another question identifies whether four-wheel or larger vehicles are able to use Province 1996 2006 the road all year long. From this information, 11 0.45562672 0.39410377 12 0.48859242 0.52783693 it is possible to construct indicator variables 13 0.69230769 0.92619926 for (a) major intervillage traffic = land or not, 14 0.39776952 0.48143236 (b) type of widest road = asphalt, concrete, or 15 0.61111111 0.73608903 16 0.63424867 0.68574200 cone block or not, (c) type of widest road = 17 0.74492498 0.72736521 hardened or not,(d) type of widest road = soil 18 0.52244898 0.47041636 or not,(e) type of widest road = others or not, 31 0.98850575 1.00000000 32 0.68730866 0.65761397 and (f) four-wheel or larger vehicles can use 33 0.64077898 0.74067070 the road all year long = yes or no. 34 0.80593607 0.79156909 We use the measure of the type of widest 35 0.55911418 0.67632006 51 0.98452012 0.98798799 road to capture the speed of transportation 52 0.81891026 0.78364566 in the local economy. The average is taken at 53 0.44480171 0.40334378 the subdistrict, district, and provincial levels 61 0.41470588 0.46736842 62 0.36184211 0.43560606 in each round. 63 0.63270504 0.66544923 64 0.32412791 0.49311295 71 0.75829726 0.75510204 ztm 72 0.57568627 0.63330300 zt (j) mN (j) (4.1) , 73 0.49590893 0.60324617 #N(j) 74 0.52157830 0.55233853 81 0.56921488 0.64210526 m 82 0.24639671 0.44170404 where zt is the indicator variable, which takes Source: Authors' calculations using PODES 1996, 2006. the value of 1 if major intervillage traffic is on Note: The unit of observation is the village. land and the road is constructed of asphalt, concrete, or cone block (good quality) and 0 for deteriorating road quality is not obvious otherwise (bad quality),N(j) is a set of villages from the data, but it may be related to inad- within village j's neighborhood, and #N(j) is equate maintenance or the construction of the number of villages in N(j). Therefore, poor-quality new roads. zt(j) is the probability of having good-quality Next, taking the difference between the transportation, which is assumed to be posi- two rounds, we can see improvement and tively correlated with the average transporta- deterioration in the quality of roads in local tion speed in the local economy. economies: Table 4.1 shows the provincial averages of asphalt road indicators in 1996 and 2006. z(j) = z1(j) - z0(j). (4.2) To have comparability between the two years, we use 1996 provinces for villages In all regions, the changes are symmetri- that changed province or district from 1996 cally distributed, with either improvement or to 2006. First, in both years, we observe deterioration, although the majority shows interprovincial disparities in average road relatively small changes around 0 (see figure quality. Second, the average proportion of 4.2). At the subdistrict level, improvement intervillage roads that are made of asphalt and deterioration coexist over the 10 years has improved in many provinces. in Indonesia, which allows us to examine the Table 4.2 shows tabulations of villages impact of intervillage changes in road quality matched between 1996 and 2007 based on on household income dynamics.Comparison changes in intervillage road quality (asphalt of the change in road quality (at the subdis- or not). In many provinces, many villages trict level) between Java and non-Java regions have seen an improvement in intervillage shows that areas in Java experienced a faster road quality, although a large number of improvement than areas outside Java. villages have seen no change in quality and Regarding distance to economic centers, a non-negligible number of villages have we assume that the physical distance has seen deterioration in quality. The reason been constant throughout the period, so it is 52 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 4.2 Changes in intervillage road quality (asphalt, concrete, or cone block or not) in Indonesia, by province, 1996­2006 Number of villages Percent of villages in each province No change No change Difference: improved Remain Remain Remain Remain minus Province good bad Deteriorated Improved Total good bad Deteriorated Improved deteriorated Jawa Barat 516 546 230 128 1,420 36.3 38.5 16.2 9.0 -7.2 Lampung 373 60 53 35 521 71.6 11.5 10.2 6.7 -3.5 Maluku 249 349 91 70 759 32.8 46.0 12.0 9.2 -2.8 Jambi 586 154 101 77 918 63.8 16.8 11.0 8.4 -2.6 South Kalimantan 303 47 42 35 427 71.0 11.0 9.8 8.2 -1.6 East Java 1,067 438 279 250 2,034 52.5 21.5 13.7 12.3 -1.4 Aceh 989 1,907 689 649 4,234 23.4 45.0 16.3 15.3 -0.9 Kalimantan Timur 602 3 8 10 623 96.6 0.5 1.3 1.6 0.3 Bali 1,277 1,277 385 424 3,363 38.0 38.0 11.4 12.6 1.2 Sulawesi Tengah 349 125 71 82 627 55.7 19.9 11.3 13.1 1.8 Central Java 258 0 0 7 265 97.4 0.0 0.0 2.6 2.6 Riau 860 599 139 189 1,787 48.1 33.5 7.8 10.6 2.8 West Nusa Tenggara 188 378 56 78 700 26.9 54.0 8.0 11.1 3.1 Sumatra Barat 261 207 56 78 602 43.4 34.4 9.3 13.0 3.7 Sumatra Selatan 190 357 12 36 595 31.9 60.0 2.0 6.1 4.0 Irian Jaya 1,162 646 157 261 2,226 52.2 29.0 7.1 11.7 4.7 Nusa Tenggara Timur 101 759 25 81 966 10.5 78.6 2.6 8.4 5.8 North Sulawesi 968 695 179 314 2,156 44.9 32.2 8.3 14.6 6.3 Sumatera Utra 152 251 17 49 469 32.4 53.5 3.6 10.4 6.8 Bengkulu 215 37 8 28 288 74.7 12.8 2.8 9.7 6.9 Sulawesi Tenggara 561 423 73 159 1,216 46.1 34.8 6.0 13.1 7.1 South Sulawesi 139 502 18 73 732 19.0 68.6 2.5 10.0 7.5 DKI Jakarta 378 137 64 123 702 53.8 19.5 9.1 17.5 8.4 Kalimantan Barat 4,379 1,361 684 1,441 7,865 55.7 17.3 8.7 18.3 9.6 DI Yogyakarta 268 536 61 171 1,036 25.9 51.7 5.9 16.5 10.6 Kalimantan Tengah 3,653 1,756 807 1,746 7,962 45.9 22.1 10.1 21.9 11.8 Total 20,044 13,550 4,305 6,594 44,493 45.0 30.5 9.7 14.8 5.1 Source: Authors' calculations using PODES 1996, 2006. Figure 4.2 Change in average intervillage road quality (asphalt roads as a proportion of all roads) in select provinces of Indonesia Group 1: Sumatra Group 2: Java (excluding Group 3: Kalimantan Jakarta) .274286 0 ­1 1 frequency Group 4: Sulawesi Group 5: Others (excluding Bali) .274286 0 ­1 1 ­1 1 change in proportion Source: Authors' calculations using PODES 1996, 2006. taken as predetermined. This information to the economic centers in all 98 villages, is important because we hypothesize that using data from the 2007 village survey. the development of spatial connectivity has an uneven impact on village economies, Household income depending on the distance to major points of In the analysis of household income dynam- economic activity. Table 4.3 shows distances ics, we use household panel data from two Spatial networks, incentives and the dynamics of village economies 53 Table 4.3 Distance to economic centers in select villages of Indonesia kilometers Province and Province and village Subdistrict District Province village Subdistrict District Province Lampung South 1 9 37 53 Kalimantan 2 13 56 120 1 0.5 4 102 3 5 14 75 2 4 12 124 4 7 7 67 3 3.5 37 40 5 3 15 125 4 3 10 180 6 3.5 42 145 5 0.1 22 170 7 12 85 55 6 4 22 90 8 38 104 12 7 18 18 61 9 7 85 37 8 17 20 67 10 37 95 14 9 0.1 29 79 11 35 95 14 10 0.05 17 86 12 1 10 45 11 15 32 45 13 5 5 50 12 1.5 16 81 14 4 45 82 13 3.5 10 93 15 20 80 120 14 21 45 60 16 15 60 150 15 50 40 50 Central Java 16 50 20 50 1 3 13 110 North 2 3 15 50 Sulawesi 3 3 30 93 1 0.3 27 54 4 10 60 120 2 0.7 18 100 5 0.05 30 250 3 1 5 25 6 2 60 225 4 4 6 27 7 0.1 8 114 5 4 40 335 8 4 14 90 6 6 5 5 9 6 5 93 7 0.5 18 60 10 6 15 60 8 6 25 105 11 7 15 270 9 3.5 16 97 12 5 8 250 10 1 30 60 East Java 11 4 23 59 1 3 15 190 12 13 20 50 2 5 20 137 South 3 5 14 35 Sulawesi 4 4 20 38 1 3 60 600 5 0.7 27 90 2 5 42 279 6 5 14 115 3 2 7 258 7 6 20 218 4 3 48 126 8 4 17 80 5 9 33 352 9 2 25 93 6 0.5 28 114 10 1 8 145 7 1 30 140 11 2 27 145 8 3 17 189 West Nusa 9 3 16 186 Tenggara 10 3.5 13 183 1 5 5 50 11 8 45 282 3 5 25 60 12 16 51 280 4 0.1 62 300 13 2 16 185 5 6 25 500 14 1 60 600 6 2.5 44 640 15 2 60 530 7 2 19 57 16 7 70 570 8 5 12 50 17 7 17 197 9 8 54 250 18 7 24 250 10 3 4 22 11 0.3 44 45 Mean 6.9 32.7 141.1 12 0.1 30 500 13 7 49 650 14 12 13 39 Source: Authors' calculations using IMDG 2007 Version 1. 54 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 4.3 Per capita income growth in select villages of Indonesia rounds conducted in 1995 and 2007 in six provinces, as mentioned. In both surveys, we .125203 collected detailed information on income- generating activities. From each activity, we aggregated incomes to construct a house- hold-level income measure. To merge the income data for 2007 with the data for 1995, we aggregated incomes frequency from original and split households using the 1995 household units. Some households split from the 1995 households (called original 0 households), but it is important to aggre- ­7.14378 9.09736 gate incomes from both original and split growth rate ST T a am households in 2007 to be comparable with Source: Authors' calculations using PATANAS 1994/95 and IMDG 2007 Version 1. the original households in 1995. The results are quite similar, which implies that attrition (split) bias in our panel analysis is not large Figure 4.4 Change in nonagricultural income share in select villages of Indonesia (see figures 4.3 and 4.4 on per capita income .318991 growth and change in nonagricultural income share). Table 4.4 shows descriptive statistics of key variables: number of household mem- bers ages 15­64, household income, growth of household income, share of nonagricul- tural income and nonfarm self-employment income in total income, landholding size, frequency and household head's education in 1995. First, the share of both nonagricultural and nonfarm self-employment income increased in the period. Second, about 23.6 percent of the sample households were landless. Third, 0 about 10 percent of the household heads ­1 1 had completed high school or above. Lastly, ST T a am change in share nominal household income grew about 1.8 Source: Authors' calculations using PATANAS 1994/95 and IMDG 2007 Version 1. percent. However, regression analysis always includes location averages (dummies), Table 4.4 Descriptive statistics: household income, nonagricultural income share, landholding, and education in select villages of Indonesia Standard Variable Number Mean deviation Minimum Maximum Number of household members ages 15­64, 1995 673 3.6 1.9 0 11 Number of household members ages 15­64, 2007 673 3.3 1.6 0 11 Household income, 2007 (100,000 rupiah) 674 478 4,330 -3,600 101,000 Household income, 1995 (100,000 rupiah) 674 22.5 39.9 -16.6 712.0 Income growth (percent) 616 1.8 1.9 -6.7 8.6 Nonagricultural income share, 2007 (percent) 674 0.5 0.4 0 1 Nonagricultural income share, 1995 (percent) 674 0.3 0.4 0 1 Nonfarm self-employment income share, 2007 (percent) 674 0.2 0.4 0 1 Nonfarm self-employment income share, 1995 (percent) 674 0.1 0.3 0 1 Landholding size, 1995 (hectares) 674 0.7 1.1 0 10.3 Landless indicator, 1995 674 0.2 0.4 0 1 Head's years of schooling, 1995 658 5.3 3.8 0 17 Head completed at least primary school, 1995 (0 = not completed) 658 0.5 0.5 0 1 Head completed high school or above 1995 (0 = not completed) 658 0.1 0.3 0 1 Source: Authors' calculations using PATANAS 1994/95 and IMDG 2007 Version 1. Spatial networks, incentives and the dynamics of village economies 55 which control price changes specific to each variables with household and village-level location (village). variables such as education and distance to Provincial averages are compared in the district center. table 4.5. First, the shares of nonagricultural Figure 4.5 (panel A) shows the relation- income in 2007 are higher in Java provinces ship between a change in the proportion than outside Java. Second, this does not of asphalt roads (at the subdistrict level) necessarily imply higher income (or growth) and per capita income growth in our sam- in Java provinces. Third, landholding size is ple. Since price change and province-level smaller in Java provinces than outside Java. aggregate factors affect income growth (as It is easy to link the diminishing role of land well as the change in road quality), we con- with the increase in nonagricultural activi- trol province effects to obtain the residuals. ties in rural areas, but this does not mean Therefore, the figure shows the effect of a higher income or higher income growth in change in local road quality on the residuals. our sample. Changes in local road quality and income To merge the household panel data with growth are positively related, which sup- spatial data on road quality constructed from ports our hypothesis. PODES (1996­2006), we use the informa- Second, figure 4.5 (panel B) depicts the tion on subdistrict, district, and provincial relationship between changes in the pro- identification. In the analysis, we interact portion of asphalt roads and the share of subdistrict- and district-level road quality nonagricultural income. It clearly shows a Table 4.5 Provincial averages: household income, nonagricultural income share, landholding, and education in Indonesia West Nusa North South Variable Lampong Central Java East Java Tenggara Sulawesi Sulawesi Number of household members ages 15­64, 1995 3.7 3.4 3.5 3.3 2.9 4.2 Number of household members ages 15­64, 2007 3.2 3.2 3.3 3.6 3.1 3.7 Household income, 2007 (100,000 rupiah) 223.0 593.0 208.0 527.0 414.0 977.0 Household income, 1995 (100,000 rupiah) 17.3 21.2 34.4 10.3 25.3 21.0 Income growth (percent) 2.2 1.9 1.6 1.9 2.1 1.3 Nonagricultural income share, 2007 (percent) 0.2 0.7 0.6 0.4 0.7 0.4 Nonagricultural income share, 1995 (percent) 0.1 0.5 0.3 0.4 0.5 0.2 Nonfarm self-employment income share, 2007 (percent) 0.1 0.3 0.3 0.2 0.3 0.3 Nonfarm self-employment income share, 1995 (percent) 0.0 0.2 0.1 0.2 0.2 0.1 Landholding size, 1995 (hectares) 1.2 0.2 0.3 1.0 1.0 1.0 Landless indicator, 1995 0.1 0.5 0.4 0.2 0.1 0.0 Head's years of schooling, 1995 5.0 5.2 4.1 4.6 7.0 6.3 Head completed at least primary school, 1995 0.5 0.4 0.5 0.5 0.6 0.6 (0 = not completed) Head completed high school or above, 1995 0.1 0.1 0.0 0.1 0.2 0.2 (0 = not completed) Source: Authors' calculations using PATANAS 1994/95 and IMDG 2007 Version 1. Figure 4.5 Impact of change in the proportion of asphalt roads at the subdistrict level on per capita income growth and change in nonagricultural income in select villages of Indonesia A. Per capita income growth B. Change in nonagricultural (controlling province fixed effects) income share 0.22 0.05 share 0.21 0 income 0.2 ­0.05 residuals 0.19 ­0.1 0.18 ­0.15 non-agriculture ­0.1 ­0.5 0 0.5 1 ­0.1 ­0.5 0 0.5 1 change in the asphalt road proportion at subdistrict level change in the asphalt road proportion at subdistrict level kernel = epanechnikov, degree = 0, bandwidth = .3 kernel = epanechnikov, degree = 0, bandwidth = .28 Source: Authors' calculations using PODES 1996, 2006; PATANAS 1994/95; and IMDG 2007 Version 1. 56 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA positive association between the two changes. seems to be a threshold in level of schooling Although we face some identification issues beyond which a change in local road quality in the estimation we conduct below, these and education jointly increase the impact on relationships back up our hypothesis. income growth. Next we investigate the relationship In contrast, figure 4.6 (panel B) shows a between the household head's years of clear negative (monotonic) effect on change schooling and income growth or change in nonagricultural income share. Less- in nonagricultural income share. In this educated households (measured by the exercise, we use observations (villages) that household head's schooling) are likely to have experienced a positive change in road qual- a higher share of nonagricultural income ity in their subdistrict. Figure 4.6 (panels (activity) when road quality improves in A and B) shows per capita income growth their neighborhood. and nonagricultural income share, respec- We describe nonagricultural income tively. By controlling village effects we get opportunities in rural Indonesia, using the residuals to, observe intravillage varia- the 2007 household survey data for 98 tions. An interesting finding is that, as villages, with the focus on nonfarm self- the household head's years of schooling employment and its linkage to the spatial increase, income growth stays intact up to connectivity of villages to economic centers. around completion of junior high school, As shown in table 4.6, the mean share of but it increases substantially from comple- nonagricultural income in total house- tion of senior high school or higher. There hold income is about 44 percent, and the Figure 4.6 Impact of years of schooling of household head on per capita income growth and change in nonagricultural income in select villages of Indonesia A. Per capita income growth B. Change in nonagricultural (controlling village fixed effects) income share 0.6 0.22 share 0.4 0.20 income 0.2 residuals 0.18 0 ­0.2 non-agriculture 0.16 0 5 10 15 0 5 10 15 household head's years of schooling household head's years of schooling kernel = epanechnikov, degree = 0, bandwidth = 2.13 kernel = epanechnikov, degree = 0, bandwidth = 2.45 Source: Authors' calculations using PATANAS 1994/95 and IMDG 2007 Version 1. Note: Both figures use observations with change in the subdistrict-level intervillage proportion of asphalt roads greater than zero. Table 4.6 Nonagricultural income share and share of households with self-employment activity in select villages of Indonesia, by distance from economic centers share in percentages; distance in kilometers Distance to district center Distance to provincial center Indicator Total 0­15 16­30 30 or more 0­60 61­120 121­400 400 or more Mean share in household income by Nonagricultural sector Self-employment (nonfarm) 22 21 24 21 26 22 20 19 Nonagriculture employment 22 30 21 17 27 22 19 16 Agricultural sector Farm activities 36 32 38 39 28 38 43 42 Agriculture employment 19 18 17 23 20 19 18 23 Share of households with At least one self-employment activity 37 32 40 37 41 37 33 33 At least one manufacturing activity 13 11 12 16 15 10 12 18 Distribution of all households by distance 100 30 37 33 33 28 31 9 Source: Authors' calculations using IMDG 2007 Version 1. Spatial networks, incentives and the dynamics of village economies 57 share of nonagricultural income declines Manufacturing activities account for nearly as the distance from either the district or half of self-employment activities in the sur- provincial center increases. Within nonag- vey. The main products include processed ricultural income, which is composed of food, such as dried fish and crackers, wood nonagricultural labor income and nonfarm (or bamboo) products, and garments (see self-employment income in this analysis, table 4.7). the share of nonagricultural labor income Next, we illustrate the density (frequency) declines with distance from economic cen- of households with self-employment activi- ters. However, the mean share of nonfarm ties by distance from district centers, using self-employment income in household figures based on kernel density estimates.2 income does not necessarily decline. Our goal is to understand how nonfarm In fact, the distance from the district self-employment activities are linked with center does not necessarily reduce the share spatial connectivity of villages to economic of households that engage in nonfarm self- centers and what type of self-employment employment activity. In particular, this is activities are made possible by spatial link- the case for self-employment activities that ages. In figures 4.7 and 4.8, we compare the involve manufacturing or processing activi- density pattern of households having at least ties. For example, the share of households one self-employment manufacturing activ- with at least one nonfarm self-employment ity with the density pattern of households manufacturing activity is 11 percent among having other self-employment activities. households living within 15 kilometers of The density pattern of all households (either the district center, while the share is 16 per- with or without self-employment) is also cent among those farther than 30 kilometers. presented as a reference (dotted lines). We Table 4.7 Type of self-employment activities in select villages of Indonesia, by distance to economic centers share in percentages; distance in kilometers Distance to district center Distance to provincial center Activity Total 0­15 16­30 30 or more 0­60 61­120 121­400 400 or more Manufacturing Processed food 26.9 25.1 26.3 29.1 26.3 25.9 28.8 26.3 Wood, bamboo products 8.2 4.2 3.8 16.7 10.5 5.5 3.4 25.0 Cloth, textiles 6.3 8.7 6.9 3.5 5.9 4.8 8.5 5.3 Building materials 0.5 0.7 0.5 0.3 0.3 0.7 0.3 1.3 Others 6.6 7.7 5.7 6.6 6.7 10.0 3.4 5.3 Nonmanufacturing 51.6 53.7 56.7 43.8 50.4 53.1 55.6 36.9 Total 100 100 100 100 100 100 100 100 Source: Authors' calculations using IMDG 2007 Version 1. Figure 4.7 Self-employment activities, by province-level road density in select villages of Indonesia A. High road density provinces B. Low road density provinces 0.05 0.05 0.04 0.04 Others 0.03 0.03 Others density 0.02 density 0.02 Manufacturing Manufacturing 0.01 0.01 0 0 0 20 40 60 80 100 0 20 40 60 80 100 distance to district center distance to district center All households Manufacturing Others All households Manufacturing Others Source: Authors' calculations using IMDG 2007 Version 1. 58 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 4.8 Self-employment activities, by speed to district center in select villages of Indonesia A. High speed to district center B. Low speed to district center 0.06 0.06 0.04 0.04 Others density density 0.02 Manufacturing 0.02 0 0 0 20 40 60 80 100 0 20 40 60 80 100 distance to district center distance to district center All households Manufacturing Others All households Manufacturing Others Source: Authors' calculations using IMDG 2007 Version 1. also separate households into two groups-- that the possibility for distant households to one with better spatial connectivity to the engage in nonfarm self-employment activi- economic centers and one without--and ties may not be changed by an improvement compare the results.3 in district-level roads via faster speed (reduced Figure 4.7 (panels A and B) shows that time) of transportation. self-employment manufacturing activities exist more at distant (but not very distant) Empirical framework places from district centers. This relation- In the analysis we estimate the following ship is particularly evident among house- equations on income growth and change holds in provinces where the density of in nonagricultural income share, both first national and provincial roads is relatively differenced between 1995 and 2007 to elim- high than among those where it is relatively inate fixed effects. The equations for both low. This implies that improving road net- income growth and nonagricultural income works beyond the district level may enable share are written as: manufacturing self-employment activities to 0 emerge at distant places from district centers yij = + 11z(j) + 12xij z(j) + 21 djz(j) (but not very distant places). Although we + 22xijdj z(j) + ij , 0 (4.3) need further investigation about why manu- facturing activities are generally high in vil- where y ji is income growth (or change in lages located between 40 and 55 kilometers nonagricultural income share) for household from the district center, this may be related i in village j, z(j) is change in the average to better access to local resources (for exam- road quality in the neighborhood of village ple, woods)4 and the reasonable range of j, dj is the distance to a center (discussed transportation time or cost (for example, below), xij is household i's landholdings 0 within two hours) needed to reach more and level of education in the initial period, consumers, including those in urban pro- and ij is an error term. As mentioned, fixed vincial centers. effects are differenced out. Figure 4.8 (panels A and B) shows similar We assume that distance to the economic density patterns to the previous ones. That activity center is predetermined and so is is, self-employment manufacturing activities taken as exogenous. The economic activity exist at distant (but not very distant) places point can be the subdistrict, district, or pro- from the district center. However, there is vincial center. The interaction of z(j)and dj no clear difference in this pattern between captures how the benefit from an improve- households in villages where speed of access ment in spatial connectivity varies with vil- to the nearest district center is relatively high lage location and distance from economic or where speed is relatively low.5 This implies activity points. Spatial networks, incentives and the dynamics of village economies 59 In the above specification,we also attempt because the development of localized spatial to capture heterogeneous effects of the spa- connectivity is important to opening access tial development by the initial-stage holding to wider economic activities (such as are of assets and endowment of education at the available at district and provincial centers). household level. We use the information on To capture potential heterogeneous effects landholding size and household head's edu- of improvement in the subdistrict average cation in 1995. road quality on income growth, we introduce The error term potentially consists of some heterogeneity in the analysis: household aggregate and household-specific shocks: head's education level and landownership in ij = v j +i . To control province-specific 1995 at the household level and distance to shocks, we could include province dummies. subdistrict, district, and provincial centers at However, village-specific shocks are cor- the village level.6 related with local economic development, The main analytical point is to investigate which is again correlated with dynamic the role of postprimary education and initial change in average road quality. Thus landholding in income growth when spatial E vjz(j) 0. In the estimation below, connectivity is improving in the local neigh- therefore, we control village-level dynamic borhood and then to investigate the relation- shocks in the first-differenced specification. ship between this and the connectivity with more distant economic centers. We include yij = + 12xij z(j) + 22xijdj z(j) 0 0 village dummies to control village-specific (4.4) + villagedummies + ij . shocks containing price changes specific to the village economy. This specification enables us to see intra- In table 4.8, columns 1 and 2 use years of village variations in the response to the schooling completed, interacted with the dis- development of spatial connectivity (as the tance to subdistrict, district, and provincial village average is controlled).Village-specific centers. The results confirm that the school- income shocks (affecting growth) are con- ing effect is significantly positive (in the spec- trolled by village dummies. We assume that ification with the squared term). Interactions the correlation between household-specific with distances are not significant. Column shocks and area-wide spatial development is 3 uses the indicator that takes the value of not important. 1 if the household head has completed high We use income aggregated from both orig- school or higher and 0 otherwise. Consistent inal and split households in 2007. Therefore, with figure 6 (panel A), the effect is signifi- our results are robust to attrition bias poten- cantly positive. tially arising from endogenous household split Theeffectincreasesasthedistancefromthe dynamics.In the analysis,however,the migra- provincial center increases, and it decreases as tion process of individuals is taken as exoge- the distance from the district center increases. nous, which may bias our estimates given that Returns to schooling decrease if the village is the migration process defines the denomina- far from the district center, but distance from tor used to calculate per capita income. the provincial center significantly augments the returns.Thus if the village is near the local Empirical results center (district center) but the local economy Inthissectionwesummarizeourmainresults is located far from the provincial center, the from the household analysis. Specifically, we benefit from the improvement in spatial con- examine household income growth, changes nectivity is larger among relatively educated in the share of nonagricultural income, villagers. and changes in the share of nonfarm self- These results suggest that being a local employmentincome.Inpreliminaryanalyses, center in a remote area is key. The marginal we found that the subdistrict-level measure benefit from an improvement in local road of road quality explains these changes bet- quality is large in remote areas, probably ter than district-level and province-level because capital accumulation is at a low measures of road quality, probably because level. However, our results show that the it has enough variations in the sample and district center is always important in the 60 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 4.8 Change in average road quality and per capita income growth in select villages of Indonesia Independent variable (1) (2) (3) (4) (5) Educ 0.0903 0.2191 1.7370 0.2184 1.7614 (1.16) (1.86) (1.80) (1.79) (1.93) Educ squared -0.0121 -0.0112 (1.08) (1.00) Educ_Distance subdistrict -0.0027 -0.0030 -0.0154 -0.0021 -0.0083 (1.56) (1.62) (0.49) (1.00) (0.28) Educ_Distance district -0.0044 -0.0033 -0.1312 -0.0044 -0.1356 (1.03) (0.66) (2.25) (0.84) (2.43) Educ_Distance province 0.0004 0.0003 0.0129 0.0004 0.0132 (1.17) (0.93) (2.70) (1.04) (2.89) Land size 0.0122 -0.0143 (0.03) (0.04) Land_Distance subdistrict -0.0125 -0.0099 (1.13) (0.97) Land_Distance district 0.0230 0.0232 (1.42) (1.57) Land_Distance province -0.0026 -0.0027 (1.86) (2.07) Village dummies Yes Yes Yes Yes Yes R2 0.1069 0.1080 0.1093 0.1105 0.1119 Number of observations 589 589 589 589 589 Source: Authors' calculations using PATANAS 1994/95, PODES 1996, 2006, and IMDG 2007 Version 1. Note: Numbers in parentheses are absolute t-values, using robust standard errors with village-level clusters. Education variable for columns 1, 2, and 4: years of household head's schooling completed. Education variable for columns 3 and 5: 1 = household head com- pleted high school or higher; 0 = otherwise. The dependent variable is per capita income growth; the independent variables are interacted with change in average road quality. local economy, given localized economic transition, which does not exclude its static interactions at the district level. There seem contribution to agricultural production. to be two important dimensions in eco- Next we examine a change in the share of nomic connectivity: links to the local econ- nonagricultural income (see table 4.9). Col- omy (district capital) and a larger economic umns 1 and 2 examine the effects of school- demand center (provincial capital). In the ing on the share of nonagricultural income. former, proximity to the center is always Consistent with figure 4.6 (panel B), we beneficial for the educated, but areas far find that schooling decreases the change from the center (that is, districts far from in nonagricultural income share. That is, the provincial capital) are more likely to the (positive) effect of road quality is larger benefit from an improvement in local road among uneducated households. Distance quality. Regardless of the interaction with from the subdistrict center diminishes this distance, education always increases the effect. marginal benefits from an improvement in Columns 3 and 4 include the size of land- local road quality. holding interacted with the distance to eco- Columns 4 and 5 include the effects of nomic centers. Landholding does not matter landholding size. Although landholding in the transition to nonagricultural activi- does not show significant effects on income ties. Education effects remain robust with growth, the exercise proves the robustness of landholding size. our previous findings on schooling. There are three possible reasons for the Land is an important conventional input negative effect of schooling on the change in in agricultural production. But because the share of nonagricultural income. First, the land is already in use in 1995, its conversion educated are more likely to have nonagri- to nonagricultural or financial resources cultural income opportunities than the less always incurs opportunity costs. In our find- educated at the initial stage, and therefore ings on income dynamics,land does not mat- the improvement in local road quality has a ter in income growth or in nonagricultural smaller marginal effect on the transition to Spatial networks, incentives and the dynamics of village economies 61 Table 4.9 Change in average road quality and nonagricultural income share in select villages of Indonesia Independent variable (1) (2) (3) (4) Educ -0.0652 -0.5900 -0.0692 -0.6453 (2.33) (2.2) (2.44) (2.27) Educ_Distance subdistrict 0.0013 0.0549 0.0009 0.0612 (0.78) (5.45) (0.59) (5.66) Educ_Distance district 0.0019 0.0045 0.0022 0.0061 (0.77) (0.20) (0.90) (0.26) Educ_Distance province -0.0001 -0.0001 -0.0001 -0.0002 (0.47) (0.03) (0.65) (0.10) Land size 0.2200 0.2060 (1.25) (1.16) Land_Distance subdistrict 0.0052 -0.0036 (1.22) (0.67) Land_Distance district -0.0061 -0.0044 (1.22) (0.93) Land_Distance province -0.0001 -0.0001 (0.05) (0.30) Village dummies Yes Yes Yes Yes R2 0.1049 0.1004 0.1049 0.1073 Number of observations 644 644 644 644 Source: Authors' calculations using PATANAS 1994/95, PODES 1996, 2006, and IMDG 2007 Version 1. Note: Numbers in parentheses are absolute t-values, using robust standard errors with village-level clusters. Education variable for columns 1 and 3: years of household head's schooling completed. Education variable for columns 2 and 4: 1 = household head completed high school or higher; 0 = otherwise. The dependent variable is the change in nonagricultural income share; the independent variables are interacted with change in average road quality. the nonagricultural sector among the edu- hold head, tend to move out of their house- cated. Second, the more educated house- holds (Dewina and Yamauchi 2008). If out- holds also have more assets for agricultural migrants work in nonagricultural sectors,the production, and thus the improvement in share of labor supplied in non-agricultural road quality increases the productivity of sectors in the extended family (including their farm activities. out-migrants) increases as the local spatial Third, individual-level selectivity may connectivity improves and the household cause this result. At the individual level, head's education is greater, and the distance the educated are more likely to move out of from provincial capital augments the posi- the household over time to pursue higher- tive complementary effect (Yamauchi and income opportunities in nonagricultural others 2008). sectors. The comparison of completed Table 4.10 shows results for a change schooling between current members and in nonfarm self-employment as a share nonmembers shows higher average school- of income. We use the same specifications ing among nonmembers. In the household adopted in the previous analyses. In column with an educated head, other members are 1, we find that the schooling effect critically also likely to be educated. Therefore, if the depends on village location. Distance from migration selection is important in the the provincial center reduces the schooling period of 1995­2007, an inverse correla- effect, while distance from the subdistrict tion between schooling (at the household center increases the effect. This probably level) and the transition to nonagricultural means that nonfarm business activities tend work is feasible. This is because educated to pay off in areas close to economic centers agents migrate, and relatively less educated with large (heterogeneous) demand such as agents stay. the provincial center. Large demand enables Yamauchi et al. (2008) and Dewina and households to cover a relatively large setup Yamauchi (2008) recently showed evidence cost. The distance effects are all negative in supporting the above third possibility. The the landholding effect, which is also consis- young and educated, relative to the house- tent with our finding. 62 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 4.10 Change in average road quality and nonfarm self-employment income share in select villages of Indonesia Independent variable (1) (2) Educ 0.0676 -0.0016 (0.33) (0.01) Educ_Distance subdistrict 0.0077 0.0183 (1.95) (3.26) Educ_Distance district 0.0081 0.0099 (1.13) (1.20) Educ_Distance province -0.0021 -0.0022 (4.61) (4.24) Land size 0.1860 (1.48) Land_Distance subdistrict -0.0065 (2.01) Land_Distance district -0.0017 (0.52) Land_Distance province -0.0004 (1.63) Village dummies Yes Yes R2 0.0708 0.0747 Number of observations 644 644 Source: Authors' calculations using PATANAS 1994/95; PODES 1996, 2006; and IMDG 2007 Version 1. Note: Numbers in parentheses are absolute t-values, using robust standard errors with village-level clusters. Education variable: 1 = house- hold head completed high school or higher; 0 = otherwise. The dependent variable is the change in nonfarm self-employment income share; the independent variables are interacted with change in average road quality. Table 4.11 Summary of parameter signs in select villages of Indonesia Nonfarm self-employment Per capita income growth Nonagricultural income share income share Independent variable (1) (2) (3) (4) (5) (6) Educ + + - - - - Educ_ Distance subdistrict + + + + + + + + + + Educ_Distance district - - - - - Educ_Distance province + + + + + + - - - - - - Land size n.a. n.a. n.a. Land_Distance subdistrict n.a. n.a. n.a. - - Land_Distance district n.a. n.a. n.a. Land_Distance province n.a. - - n.a. n.a. Village dummies Yes Yes Yes Yes Yes Yes R2 0.1093 0.1119 0.1004 0.1073 0.0708 0.0747 Number of observations 589 589 644 644 644 644 Source: Authors' calculations using PATANAS 1994/95, PODES 1996, 2006, and IMDG 2007 Version 1. Note: Three signs are significant at 1 percent level. Two are significant at 5 percent level. One is significant at 10 percent level. Minus (­) means coefficient is minus and plus (+) means coefficient is plus. Blank cells: Variables are not statistically significant. n.a. Not applicable (not included in the specifications). Note: Education variable: 1 = household head completed high school or higher; 0 = otherwise. The independent variables are interacted with change in average road quality. Table 4.11 summarizes the signs of role of distance to different economic cen- our parameter estimates. First, interest- ters. In the transition to nonagricultural ing results are concentrated in education activities among educated households, the effects. In general, land does not matter marginal impact of an improvement in in the dynamics of household income and local road quality is large in locations dis- nonagricultural transition. Second, while tant from local economic centers (subdis- education augments the impact of road trict capital), but in income dynamics, the quality improvement (spatial connectivity) impact is large in villages far from the pro- on per capita income growth, it decreases vincial capital. the impact on nonagricultural transition. In our definition, nonagricultural activi- Third, a similar contrast is observed in the ties only cover activities done by current Spatial networks, incentives and the dynamics of village economies 63 household members. This excludes non- crackers) and marketing of staple foods. In members who work in locations distant this case, proximity to the economic center is from their village (not able to commute a key, as it reduces transport-related transac- from their village). Therefore, it is still pos- tion costs. However, other types of activities sible that we are missing migration-linked have a wider market area, especially cater- nonagricultural transition. Instead, income ing to urban economic centers such as pro- growth includes agriculture-based growth, vincial centers. These may include higher which, for example, includes improved value added goods, such as bamboo or wood marketing of agricultural products (for products,that are sold in large urban markets. example, vegetables). In this activity, con- Another example is high-quality vegetables necting to larger demand centers seems to for the urban market. In this case, the added be a driving force. value is high enough to cover the transaction costs due to transportation, making distance Policy discussion from the provincial center not an obstacle, This paper is intended to bridge the gap provided that it is connected to economic between academic studies and infrastruc- centers. Better road connectivity to the pro- ture planning. Previous academic studies on vincial center in the form of better local spatial connectivity of rural households were roads may give remote villages the chance to limited in the sense that they perceived con- market such value added products. nectivity only as access to local towns or dis- In the former case, improving the trunk tance from growth centers and were unable roads connecting villages to closer district to discuss the combination of both. But in centers is important, as is improving local actual policy choices, public investment roads that provide access to trunk roads. planners face decisions on the allocation of In the latter, it is important to develop the resources among trunk roads (which lead to network of trunk roads that connect vil- economic centers) and local roads. Public lages to distant economic centers, such as investment planners also face policy choices the provincial capital, as well as to improve regarding the balance between spending on local roads. education and spending on roads. Poverty reduction strategies adopted by The analyses described in this paper sug- low-income countries, especially those in gest that the more educated households Africa, are entering a second stage in which can raise their income with better spatial they are becoming more growth oriented. connectivity at the local level. Better-qual- Compared with the previous generation ity local roads may also improve the access of strategies emphasizing the allocation of of remote villages to trunk roads and thus resources to primary education and health, help the more educated to engage in better the current generation focuses on growth job or business opportunities in the district strategies. Yet little is known about the com- capital (local economy) or provincial capital bination of public investments that induces (larger economic center). growth. This paper suggests that investing However, the effect on income growth is simultaneously in connecting local neigh- augmented both by proximity to the district borhoods spatially with one another as well center, which is significantly positive, and by as in connecting them to distant economic distance from the provincial center, which is centers pays off. It also suggests that invest- significantly negative. Although we cannot ing in both higher education (high school include it in the empirical analysis due to and above) and roads is important.Although data limitations, this difference may be due the actual approach should be country driven to the market space as well as the value added and country specific, such findings can add of different income-generating activities. value to the next generation of growth- First, income-generating activities exist that oriented poverty reduction strategies. focus on the market, with a district capital as the local economic center. These may include Conclusions activities such as food processing with low This paper has examined the impact of spatial value added (such as dried fish or chips and connectivity on household income growth 64 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA and nonagricultural income share, combin- asphalt roads in intervillage roads--the above ing household panel data and village census problem is less important. data in Indonesia. Empirical results show that 2. We use Epanechnikov kernel function and the impact of an improvement in road quality a default bandwidth in the application. We also have performed estimations with alternative in the local area (positively correlated with an bandwidths, but the key messages presented in increase in transportation speed) on income this paper are almost the same. growth and the transition to nonagricultural 3. We use two types of spatial connectivity activities depends on the distance to eco- indicators. First, we calculate the national and nomic centers and household education and provincial road density, in terms of road dis- landholding size. In particular, postprimary tance per area, as a proxy for road network for education significantly increases the benefits each of seven surveyed provinces, using data from an improvement in local connectiv- in JBIC (2004) because development of both a ity in remote areas. Postprimary education road network within an economic region (for and local road quality are complementary, example, a province) as well as a route to eco- increasing income growth. Therefore, the nomic centers is important for measuring spatial connectivity. The road density ranges from 0.04 effectiveness of improved local connectivity in South Kalimantan province to 0.13 in North (measured by household income growth) Sulawesi province. Second, we calculate indica- depends on village remoteness and initial tors of speed (kilometers per minute) to reach household-level endowment. the district or provincial center, using data on time to get there by the most common mode of Notes transportation, as physical distance may have Futoshi Yamauchi is a research fellow and Reno different implications for the connectivity to Dewina is a research analyst with the Interna- economic centers, depending on factors such as tional Food Policy Research Institute (IFPRI); road and traffic conditions. In fact, indicators of Megumi Muto is a senior economist with the speed are not significantly correlated with physi- Japan Bank for International Cooperation (JBIC) cal distance in our data. and Sony Sumaryanto is a senior researcher with 4. The data show that manufacturing activity the Indonesian Center for Agriculture Socio- related to wood (including bamboo) accounts for Economics and Policy Studies (ICASEPS). The more than 25 percent of all manufacturing activ- authors would like to thank JBIC for financial ity in villages located 40­55 kilometers from the support. This study is based on a collaboration of district center, while the main manufacturing JBICI, IFPRI, and ICASEPS. activity in other areas is food processing. 1. 1994­95 PATANAS survey consists of two 5. The high-speed group includes villages subsurveys. Income and production data are where the speed (kilometers per minute) to available from the second part, which contains 34 reach the nearest district center is equal to or villages in six provinces excluding South Kalim- more than 0.56 kilometer in dry season, and the antan. To merge the household panel data with low-speed group includes other villages. spatial data on road quality constructed from 6. In our empirical setting with a small num- PODES (1996­2006), we use the information on ber of villages in each subdistrict, we cannot subdistrict, district, and province identification. identify the effect of a change in the subdistrict- In the analysis, we interact subdistrict-level road level road quality on household-level outcomes. quality variables with household- and village- As figure 3 shows, the effect seems to be positive level variables such as landownership and dis- for income growth and negative for nonagricul- tance to district center. At this stage, we cannot tural income share. In preliminary analyses, we construct road quality data for two subdistricts in could not find significant effects with province- North Sulawesi, as they have missing information level dummies except in a few cases. Therefore, in PODES. When we constructed village panel we focus on intravillage distributional effects data from PODES for other studies to analyze (with village dummies controlling for price village dynamics, we had a problem in linking changes and village-level shocks) in our para- villages across rounds because of village divisions metric estimation. and mergers partly due to the decentralization process in the country. To solve this problem, References we linked subdistricts and then linked villages Binswanger, Hans P., Shahidur R. Khandker, and within each subdistrict by their names. In this Mark R. Rosenzweig. 1993. "How Infrastruc- chapter, however, because we only use subdistrict- ture and Financial Institutions Affect Agri- level information--the average proportion of cultural Output and Investment in India." Spatial networks, incentives and the dynamics of village economies 65 Journal of Development Economics 41 (2): Rural Economy." PIER Working Paper 01-056, 337­66. Penn Institute for Economic Research, Phila- Dewina, Reno, and Futoshi Yamauchi. 2008. delphia, November. "Mobility, Splits, and Household Dynamics: Fujita, Masahisa, and Megumi Muto. 2007. Impacts of Land and Schooling on Income "Development in Developing Countries Dynamics in Indonesia." Unpublished mss. Employing Brand Agriculture: From the View Japan Bank for International Coopera- of Spatial Economics [in Japanese]." Journal tion, Tokyo and International Food Policy of JBIC Institute 33 (February): 97­108. Research Institute, Washington, DC. JBIC (Japan Bank for International Coopera- Fafchamps, Marcel, and Forhad Shilpi. 2003. tion). 2004. Sector Study in the Road Sector in "Spatial Division of Labor in Nepal." Journal Indonesia. JBIC Sector Study Series 2003-2. of Development Studies 39 (6): 23­66. Tokyo: JBIC. ------. 2005. "Cities and Specialization: Evi- Minten, Bart, and Steven Kyle. 1999. "The Effect dence from South Asia." Economic Journal of Distance and Road Quality on Food Col- 115 (503): 477­504. lection, Marketing Margins, and Traders' Wages: Evidence from the Former Zaire." Fafchamps, Marcel, and Jackline Wahba. 2006. Journal of Development Economics 60 (2): "Child Labor, Urban Proximity, and House- 467­95. hold Composition." Journal of Development Economics 79 (2): 374­97. Yamauchi, Futoshi, Megumi Muto, Shya- mal Chowdhury, Reno Dewina, and Sony Fan, Shenggen, Linxiu Zhang, and Xiaobo Sumaryanto. 2008. "Spatial Networks, Con- Zhang. 2004. "Reforms, Investment, and Pov- nectivity, and the Dynamics of Village erty in Rural China." Economic Development Economy: Pathways from Agriculture in and Cultural Change 52 (2): 395­421. Indonesia." Unpublished mss., Japan Bank Foster, Andrew, and Mark Rosenzweig. 2001. for International Cooperation, Tokyo and "Democratization, Decentralization, and the International Food Policy Research Institute, Distribution of Local Public Goods in a Poor Washington, DC. The Iskandar Development Region and Singapore Manu Bhaskaran 5 This chapter argues that both Singapore and Background Malaysia potentially could enjoy consider- The IDR was formally launched by Malay- able synergies if the Iskandar Development sian Prime Minister Abdullah Ahmad Region (IDR) would become integrated Badawi in November 2006.1 The Malaysian more seamlessly with the Singapore econ- government had earlier appointed Kha- omy through the freer movement of people, zanah Malaysia, its holding company for a goods, and capital. However, political bar- large number of state-owned companies, to riers do present a major challenge to realiz- c h a p t e r take the lead in developing the IDR. Given ing these synergies. The fact that Singapore that the IDR is fairly well developed, with and the IDR are in different political juris- significant industrial, logistical, tourist, and dictions is further complicated by several business centers, the aim of the initiative is political difficulties related to the legacy of to take the IDR to an even higher level of Singapore's bitter separation from Malay- development so that it becomes "a strong, sia in 1965, ethnic tensions, the affirmative sustainable conurbation of international action program pursued by Malaysia, and standing" (see Khazanah Malaysia 2006). By the presence of vested interests. 2030, the region is projected to have a per This chapter is structured as follows. It capita income equal to that of a developed begins by sketching the key factors deter- country. mining the relationship between Singapore Encompassing an area of 2,216 square and Malaysia and then assesses their current kilometers, the IDR is located in the south- economic relationship as well as Singapore's ern part of Malaysia's Johor state, just current ties with the IDR. This is followed by across the narrow Johor strait that sepa- an assessment of the key features of the two rates Singapore from Malaysia. It already countries' economic development that will has an international airport, a successful influence the manner in which they could port that has drawn some shipping lines collaborate on the IDR, an analysis of the away from Singapore, good road and rail impact of political differences on their eco- links to other parts of Malaysia as well as nomic relationship, and a discussion of the Singapore, townships, tourist resorts that various channels through which synergies attract visitors from Singapore, and a pop- could be realized for Singapore's economy ulation of 1.4 million. The area embraces through greater collaboration with the IDR. a large manufacturing hub that is globally The chapter concludes with an assessment competitive and has benefited from the of the way forward. The relation between the relocation of manufacturing production focus of this chapter and the World Develop- from Singapore. ment Report 2009 is presented in box 5.1. The Iskandar Development Region and Singapore 67 Key factors driving the relationship between the B O X 5 . 1 Density, distance, and division: Singapore two regions and Johor This section sketches the current state of the This chapter is written in the spirit of the Divisions are preventing the two economic relationship between Singapore World Development Report (WDR) 2009, countries from fully extracting the ben- and the IDR and identifies key features of which focuses on density, distance, and efits of the growing density of economic division: clusters in Singapore.The challenge for the Singapore and Malaysian economies Singapore and Malaysia is to capture the that affect their relationship. · Rising density. Rapid urbanization is benefits of proximity, thereby reducing increasing the density of urban agglom- international fragmentation and increas- erations.More and more activity is being Current economic ties ing regional concentration. Our main concentrated in increasingly denser cities. policy recommendation is for both coun- Singapore is rapidly emerging as a global With only a 1-kilometer-wide strait dividing tries to seek ways to eliminate the barrier center of commerce,attracting clusters Singapore and Malaysia and a long history of distance. of economic activity that are densely The discussion of Singapore and Malay- of close economic integration, the move- located within its limited territory. sia relates to the following issues raised in ment of goods, people, and capital between · Falling distance. The concentration of the WDR: Singapore and Malaysia and between Singa- economic mass in urban agglomerations reduces the distance between economic · Scale economies. Increased integration pore and the IDR is quite substantial. producers, now concentrated more of economic activities between Singa- In terms of trade, Malaysian goods com- proximately to each other in cities.The pore and Johor would yield dividends prise 15 percent of all imports into Singa- flow of goods, services, equity capital, from agglomeration and large-scale economies, which Singapore has not pore, and Singapore exports 13 percent of its direct investment, and technology is higher the shorter the distance between been able to exploit fully due to its exports to Malaysia. The top-traded items two centers of economic activity. small size. include electrical and electronic products · Persisting divisions. The number of bor- · Factor mobility. A key route to achieving and refined petroleum products. The former ders between countries has increased factor mobility is to achieve freer flow reflects the existence of production networks threefold since 1950. Until Singapore and mobility of skills, knowledge and gained independence in 1965, Singa- ideas, labor, and capital. of major electronics firms with collaborat- pore and Malaysia operated mostly as · "Bridge" the distance gap. Facilitat- ing facilities in both Singapore and the IDR. a single economic entity. Before 1965, ing the convergence of the markets Many goods are shipped to southern Malay- other than customs checks (Singapore would reduce transportation costs and sia from the outside world through Singa- was a free port), there were no barriers directly enhance trade between the between the two territories. two countries. pore's port, generating substantial traffic in goods between Singapore and the IDR. Simi- larly, many goods manufactured in the IDR Table 5.1 Share of Singapore foreign direct are shipped through Singapore's port. investment in Asia, by country, select years, 1996­2005 Table 5.1 shows Malaysia's position as a percent destination for outward investment from Recipient 1996 2001 2005 Singapore. Although an important destina- tion, Malaysia has received a declining share ASEAN 32.3 19.8 23.8 Malaysia 17.3 8.4 8.6 of Singapore's investment since the Asian Hong Kong, China 12.0 8.6 6.6 financial crisis. China now receives the lion's Indonesia 7.0 4.2 7.5 share of Singapore's outward direct invest- Philippines 1.8 2.1 1.9 Thailand 2.8 3.4 4.1 ments, and Singapore has started investing India 0.5 0.6 1.0 in India, which had not attracted much Sin- China 11.5 11.8 13.8 gapore investment in the past. Other emerg- Source: Singapore Department of Statistics. a. Year-end stock for all sectors. ing markets like Vietnam have also seen Singapore investments more than double in production when costs rise in Singapore. An the past few years. Malaysia now has to com- example is the shift of Super Foods, a maker pete with newly industrialized countries like of instant coffee and foodstuffs, from its base China and India for its share of Singapore in Singapore to Pasir Gudang in southeast direct investment. Johor, just across the water from Singapore. Singapore companies, especially the The plant currently employs 150 work- smaller ones, see Malaysia and, in particu- ers. Naigai Nitto, a Japanese company with lar, Johor as the natural place to relocate headquarters in Singapore, also runs several 68 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA of its logistics and warehousing operations Malaysia where data for the IDR are not near Tanjung Pelepas in the IDR. available). This brief analysis and accompa- The close ties are also evident in the num- nying tables seek to highlight some of the ber of commuters who cross the border daily. key features of the relationship between Sin- An estimated 205,000 Malaysians work in gapore and the IDR. Singapore, of which 60 percent commute in First, there are strong complementarities. and out through the checkpoints every day. Singapore is rich, well developed, and well Most of the commuters are engaged in the regulated, and its educational system pro- electronics and electrical industry.Almost 60 duces a relatively highly skilled labor force percent of these commuters are Johor born. that is increasingly first world in terms of Most of the workers who traverse the border capacity. However, it is beginning to encoun- daily are part of the unskilled or semiskilled ter constraints in terms of land and labor, workforce, drawn by wages in Singapore, and it needs to do more to unleash entre- which can be between two and three times preneurial energies. Singapore is ranked one the wages they would receive in Johor. This of the most competitive economies in the is one of the reasons why Johor's unemploy- world, whereas Malaysia has seen its com- ment rate is low. Conversely, about 130,000 petitiveness rankings fall over the years; they Singaporeans visit Malaysia every day for are now substantially lower than Singapore's both tourism and work.2 except in business process outsourcing, where the A T Kearney Index ranks Malaysia A Comparison between Singapore third in the world and Singapore eleventh. and Malaysia (IDR) The IDR is well endowed with land, unlike Table 5.2 outlines some of the key features Singapore, where only 15 percent of the ter- differentiating Singapore from the IDR (or ritory is developed. Although its population Table 5.2 Comparison of Singapore and the IDR (continued) Iskandar Development Indicator Singapore Region, Malaysia Area (square kilometers) 692.7 2,216.3 Populationa 4.5 million 1.353 million GDP US$136.9 billion US$20.0 billion Population density (people per square 6,376 631.8 kilometer) GDP per square kilometer US$197.6 million US$9.0 million GDP per capitab US$30,422 US$14,790 Main constraints Land Geological and political limits of land Only 15 percent of land in the IDR is under development, leaving reclamation are being reached; 30 square considerable land available for economic development purposes. kilometers were reclaimed in past 40 years. Labor Against a current population of 4.5 million, Population density is low, and many Malaysians from other parts of Singapore has a "planning parameter" of 6.5 the country migrate to the IDR for work. There is ample flexibility million by a notional year X, representing the to expand and enough room to accommodate inward migration maximum population that its officials believe from other parts of Malaysia to raise IDR's population and work Singapore can accommodate. force, as needed. Education Singapore's education system is highly ranked Malaysia's education system has been criticized. Two of its globally: for example, the National University universities used to be ranked among the top 200 global of Singapore is ranked 33rd in the world.c universities, but no longer enjoy this ranking. Main strengths and weaknesses General competitiveness Singapore is ranked among the world's top in Malaysia is ranked well behind Singapore in general general competitiveness surveys. competitiveness surveys. Logistics Singapore is ranked first in the World Bank Malaysia is relatively well ranked, at 27 in the same survey, just Logistics Performance Index. behind Republic of Korea and Spain, but still far behind Singapore. Ease of doing business Singapore ranks first in World Bank rankings. Malaysia is relatively well ranked at 24, which puts it behind only Hong Kong, China; and Thailand in East Asia. Governance Singapore is above the 90th percentile in all key Malaysia is relatively highly ranked (80th percentile) for government areas except voice and accountability. effectiveness but ranks poorly in other areas. (continued) The Iskandar Development Region and Singapore 69 Table 5.2 Comparison of Singapore and the IDR (continued) Iskandar Development Indicator Singapore Region, Malaysia Affirmative action polices Singapore has adopted a system of meritocracy. Malaysia's affirmative action program, generally known as the No policies favor any particular ethnic group New Economic Policy (NEP), gives preferential treatment to over others, although some modest programs Bumiputeras, the indigenous people of Malaysia. The NEP affects have assisted the Malays as the indigenous the allocation of jobs, scholarships, places in tertiary educational population. institutes, housing, business licenses, and so on. This can be a disincentive for businesses, including domestic and foreign investors, as it raises their costs and reduces their flexibility. The government has conceded that such NEP requirements will be relaxed for foreign companies in the IDR, but it is not clear how well this will be implemented. Labor costs in manufacturingd US$2,360 per month US$414 per month. Although the IDR is not as cheap as India, Indonesia, or many parts of China, its proximity and familiarity to Singapore companies more than make up for its higher costs. Demographicse Singapore is rapidly aging, from 31st in the Malaysia has a youthful population. world in median age in 2005 to the 4th oldest population in 2050. Income inequalityf The Gini coefficient rose from 0.44 in 2000 to 0.47 Malaysia had a higher Gini coefficient of 0.49 in 2005. in 2006. Entrepreneurial capacity Despite considerable deregulation in recent Malaysia's corporate sector continues to demonstrate substantial years, entrepreneurs consider Singapore's verve and vitality. tightly controlled society and regulatory regime as constraints. Infrastructure Excellent Excellent Tourism Singapore is aiming for higher-spending Substantial tourist potential exists, with large tracts of nature such tourists. Its main attractions are traditional as one of the world's most extensive mangrove forests, beaches, tourist products such as a zoo and bird park, and historical sites. but these are well managed and innovatively designed. There is no beach resort of any global standard. Activities requiring large expanses of nature are limited. Source: Collated by Centennial Group from official data. a. July 2007 estimate for Singapore and 2005 estimate for the IDR. b. Singapore: 2006 GDP divided by December 2006 US$ per S$ rate; for IDR, 2005 estimate. c. The Times of London, Higher Education Supplement, November 8, 2007. Its previous rank was 19. d. From International Labour Organisation's LABORSTA Web site: 2006 data for Singapore and 2001 for Malaysia. e. From United National database. f. From Department of Statistics, Singapore and Malaysia, respectively. is small, there is a large, growing, and still- Key features of Singapore's youthful population in the rest of peninsular economy Malaysia that can be tapped by businesses The Singapore economy has gathered sub- operating in the IDR should the need arise. stantial momentum in recent years after a The mobility of labor across the peninsula period of sharply fluctuating GDP growth is sufficiently fluid given that workers from in 1998­2003, a period marked by pain- all over Malaysia already work in the Johor's ful domestic restructuring and significant industrial estates as well as in Singapore. external shocks (recession in the United Second, there are some competitive ele- States, including a downturn in technology ments in the economic relationship between demand in 2001­02, and the outbreak of the two territories. The ports clearly com- severe acute respiratory syndrome [SARS], pete, as seen in the growth of the IDR's Port in 2003). More recently, the economy has of Tanjong Pelepas after a successful effort to been growing rapidly, by 8.8 percent in 2004, lure two large shipping customers away from 6.6 percent in 2005, 7.9 percent in 2006, and the Port of Singapore. Competition is also around 7.5 percent in 2007, according to evident in related logistics businesses, such as preliminary estimates. recent efforts to set up bunkering facilities. 70 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Some characteristics of Singapore's companies underperform foreign-owned growth carry important implications for the companies in terms of profitability. This is relationship with Malaysia and the IDR. creating political pressures that are likely to First, Singapore is in a new phase of persuade policy makers eventually to refine growth in which it is reaping the benefits of Singapore's growth strategy to achieve a the economic restructuring, deregulation, more balanced and equitable growth. This and policy changes carried out in 1996­ chapter argues that improved access to the 2004. Its economy is drawing in substan- IDR for poorer Singaporeans and less profit- tially more foreign professionals, workers, able Singapore companies would allow both and investors than before. New engines of groups to improve their relative position as growth have emerged (high-value manufac- a result of lower costs of living and business turing, such as pharmaceuticals, petrochem- operations in the IDR. icals, and electronic components, as well as Fifth, Singapore is aging rapidly. The pro- new financial services, such as wealth man- portion of the population above 65 years agement and hedge funds). Long-stagnant of age rose from 3.4 percent in 1970 to 4.9 activities, such as construction, are growing percent in 1980, 6 percent in 1990, 7.2 per- strongly again. cent in 2000, and 8.5 percent in 2006. This As a result, Singapore's gross domestic reduced the old-age support ratio from 17 product (GDP) growth averaged close to workers per elderly in 1970 to 8.5 in 2006. 8 percent in 2004­07, above the 3­4 percent These features of Singapore's economy level that the Singapore government estimates raise important questions for its relationship is its long-term potential growth rate. Conse- with the IDR. Can Singapore become a global quently, capacity is becoming constrained: city without expanding its relationship with unemployment fell to a decade low of 1.7 the IDR? percent in the third quarter of 2007, caus- ing wages to accelerate. Rents for homes and Key features of the Malaysian offices are soaring, in many cases doubling economy or even tripling over year-earlier levels. Road Unlike Singapore, which has overcome a congestion has worsened significantly as period of slower growth, Malaysia has not yet well. It is becoming increasingly difficult for regained the vigor it lost after the Asian finan- incoming expatriate families to secure places cial crisis (see table 5.3). The Philippines, in in international schools for their children. contrast, grew 7.5 percent in the first half of Third, a key plank in Singapore's growth 2007, substantially exceeding the growth of strategy is to build itself up as a global Malaysia. Some trends in Malaysia's growth city, one that aspires to be more than just experience stand out. a regional hub for Southeast Asia. Rather, A good part of the deceleration in growth Singapore ambitiously seeks to transform post-crisis is due to a sharp fall in the ratio itself into a city, like London or New York, of investment to GDP, from above 40 per- capable of being a hub for global financial cent in the years just prior to the crisis to market activities and business operations. 20.9 percent in 2006. This is despite fairly Fourth, the composition and distribution strong public sector investment spending, of Singapore's growth remains uneven and showing that the fall has been primarily due may be getting more so. The poorer seg- to desultory private investment. ments of the population are not enjoying Malaysia's competitiveness has taken a the benefits of growth and bore a dispro- hit as measured by general competitiveness portionate share of the burden of economic indexes such as those constructed by the adjustment during the economic restructur- World Economic Forum and the Institute ing in the early 2000s. The bottom quintile for Management Development. This sug- of the income distribution experienced a fall gests that Malaysia has not adjusted to the in real income in 2000­05, with the bottom much more competitive global economy 50 percent seeing a substantial deceleration that has emerged as China, India, and others (see Chan 2007). Data from the Depart- such asVietnam have liberalized and reformed ment of Statistics show that locally owned their economies. The Iskandar Development Region and Singapore 71 More recently, Malaysian export growth Table 5.3 Economic growth before and after the Asian financial crisis, by country, 1991­2006 has been weak, mainly due to the slowdown percent a year in demand for the technology-related prod- Country Precrisis average, 1991­96 Postcrisis average, 2005­06 ucts in which it specializes. Yet the other China 11.9 10.6 Indonesia 7.3 5.6 technology-dependent economies such as Japan 1.7 2.0 Korea, the Philippines, and Singapore have Korea, Rep. of 7.7 4.6 had different experiences. Singapore has Malaysia 9.6 5.5 Philippines 2.8 5.2 found new manufacturing activities that Singapore 8.7 7.2 have been able to offset the technology Thailand 8.2 4.8 slowdown, while the Philippines has found Source: Collated by Centennial Group using CEIC database. sources of growth outside manufacturing. However, Malaysia has made some prog- tially. Government-linked companies have ress in developing a competitive position in been restructured to raise efficiency. Sig- exportable services. It has been consistently nificant efforts have been made to improve ranked by A T Kearney as third in the world the delivery of public services. Most after India and China as a location for busi- recently, the government has implemented ness process outsourcing. Malaysia's exports the policy of setting up economic growth of educational and medical services have also corridors, of which the IDR is just one. Two been rising. Malaysia continues to grow as a others have been announced, with another tourist destination. In all of these areas-- two to follow. education, medical services, and tourism-- Potential synergies between Singapore struggles to attract customers who are highly price sensitive, giving lower- Singapore and the IDR cost Malaysia the edge among this group In this section, we identify areas where Sin- of customers. However, Singapore's more gapore's economy could enjoy synergies rigorous educational and medical standards from greater economic integration between and regulations, coupled with its investment Singapore and the IDR. Synergies are defined in cutting-edge equipment and its success in here as the achievement of greater welfare assembling a critical mass of specialists who for each country individually as a result of have rare skills in these activities, gives it the cooperation than they would enjoy if they edge in providing higher-priced medical did not cooperate. The combination of the and educational services. Thus while Singa- two territories creates more welfare in total pore and Malaysia (IDR) may well compete than the individual sum of the welfare of with one another in the areas of educational each country prior to cooperation. and medical space, they are complementary We see several channels through which in a larger sense. synergies could be realized if Singapore and Within Malaysia, Johor is a major pole of the IDR were to cooperate more together. economic growth and has a manufacturing and services economy that is globally com- · Rational use of resources. Reallocation of petitive in key areas. Johor has enjoyed bet- scarce resources such as land and labor ter flows of foreign direct investment (FDI) to higher-value uses would create higher than any other state in Malaysia for the last economic output. three years. Foreign investments approved · The benefits of scale and diversity. The size in Johor rose sharply from RM 1.4 billion of the combined Singapore and IDR econ- in 2002 to RM 5.9 billion in 2005. In the omies would be substantially larger (table field of education, Johor has some key uni- 5.1), allowing more scope for businesses versities, including the Universiti Teknologi and consumers to exploit economies of Malaysia. scale and other benefits of scale. Malaysian policy makers realize that · Leveraging off each other's competitive they need to create new strategies to reig- advantages. There are important dif- nite Malaysia's economic dynamism. To this ferences in competitiveness between end, they have been reforming the economy. Singapore and the IDR (Malaysia). Each Capital controls have been eased substan- territory could leverage off the relative 72 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA advantage of the other. For instance, the Hong Kong shrunk in the 1980s, and Hong IDR could benefit from being associated Kong became a mostly services-producing with Singapore's higher rankings in gov- economy. By providing the right incentives ernance and ease of doing business. and economic fundamentals, Singapore con- · Dynamic benefits of competition. Increased tinues to attract highly capital-intensive and interaction would bring about more technologically intensive manufacturing, as opportunities for competition, stimulat- seen in the surge to a record high S$20 billion ing efforts to overcome performance gaps in net investments in manufacturing in 2007. and leading to improvements in the lag- Many of the new manufacturing investments ging territory. in Singapore relate to capital-intensive activi- ties such as petrochemicals or pharmaceu- Reallocating land alone would yield ticals (capital intensive as well as requiring significant net gains to Singapore: moving a regime that protects intellectual property) some activities from Singapore to the IDR or in cutting-edge alternative energy (such as would allow this land to be used for even solar cells). However, to accommodate these higher-value activities, resulting in a net new industries without straining the capacity addition to GDP. of industrial zones in Singapore, lower value added manufacturing activity needs to be Reallocation of land and labor to relocated out of Singapore. higher-value uses Second, Malaysia itself has become a labor Several aspects of the successful reallocation importer and is not a labor-surplus econ- of resources would make sense for both Sin- omy, so it might be said that relocation of gapore and the IDR. First, whole business or labor-intensive activities from Singapore to industrial sectors could be moved from Sin- Malaysia does not make sense. But Malaysia gapore to the IDR, if doing so were mutually has maintained relatively liberal immigra- beneficial. In the past, for instance, Singa- tion policies, allowing about 2 million for- pore has made policy changes that resulted eign workers (mainly Indonesians) to work in the elimination of entire sectors. For in Malaysia. Moreover, Malaysia does not example, the removal of tariffs in the early need to offer Singapore companies Chinese- 1970s led to the movement of automobile level or Indian-level labor costs: so long as assembly out of Singapore, and the 1980s Malaysian labor costs are substantially below environmental decision to end support for Singapore's, the greater familiarity that its pig farming resulted in the movement of business environment offers to Singapore activity to neighboring countries. companies (given historical, linguistic, and Second, the more labor- or land-intensive cultural legacies) will make relocation to operations of a particular activity could be Malaysia or the IDR a viable proposition for moved to the IDR from Singapore, leaving many Singapore businesses. high-productivity work that is done more Third, the relocation would not be of profitably in Singapore. For example, a existing activities but of future ones. For watch manufacturing plant could shift to the example, some of the growth in aviation, IDR the manufacture of low-cost watches educational, and other activities needed to that do not need a high level of skilled labor service Singapore's ambitious growth in the or precision engineering capital equipment, future could be located in the IDR. enabling it to concentrate its scarce factory For the IDR, with 85 percent of its area space and expensive workers in Singapore on undeveloped, including 58 percent of land higher-value watches or watch components. now devoted to agriculture, land is not a There are a number of important issues to binding constraint, and the case for relo- consider here. cating activities to it is relatively straight- First, Singapore's economic planners forward. Relocation of activities could be are determined that manufacturing should accommodated and would be beneficial remainakeysegmentof Singapore'seconomy. so long as the value created from reloca- Singapore will therefore not follow Hong tion exceeds the value of agricultural and Kong, China's, example: manufacturing in other rural activities that are replaced. For The Iskandar Development Region and Singapore 73 Table 5.4 Benefits of relocating and undertaking complementary activities Activity Potential for relocation Benefits Manufacturing Substantial manufacturing activity has already shifted from Year-to-date 2007 alone has seen S$20 billion of new manufacturing Singapore to districts such as Johor Baru, Pasir Gudang, Kulai, projects committed to Singapore, against an average in past and Senai in the IDR in the past 20 years. In many cases, the years of less than S$10 billion. Most of this is in very capital- same manufacturer has plants in both the IDR and Singapore, and skill-intensive areas (petrochemicals, alternative energy each focusing on a different component or finished good. technologies, high-end electronics). This will raise input costs, making lower value added per worker in manufacturing activities As land, labor, and other costs continue to rise in Singapore, less viable in Singapore. Shifting such activities to the IDR would manufacturing activity will probably continue to relocate to the be best for Singapore: being close to Singapore would mean that IDR. The policy incentives and infrastructure exist to support such activities would continue to use Singapore-based financial, such relocation. transportation, and logistics services, while releasing resources for the higher-end activities entering Singapore. Port Five port terminals occupy 6 square kilometers of land, Releasing the land from use by the port to use by finance business 0.8 percent of total land in Singapore. Three of these directly activities would result in a net addition to GDP. abut the central business district, where very high-value Road congestion would improve considerably, as container lorries financial and business services are constrained by lack of would not compete with other users for scarce road space. office space. Singapore's marine services sector (specialized legal, logistics, finance for shipping) is already well developed and has critical mass. Like London, the reduction of actual port activities in Singapore will probably not result in such activities moving out of Singapore. Airport The airport occupies 13 square kilometers of land. A third terminal Benefit comes not from relocating aviation services from Singapore will bring capacity to 70 million passengers a year, sufficient to to the IDR but rather from complementing the primary airport in accommodate growth well into the next 20 years. Sufficient land Singapore with a secondary airport in the IDR connected directly has been reclaimed for a third runway and a fourth terminal: to Changi. This would allow Singapore as a metropolis to have the hence the land constraint is not as binding as in ports. number of airports that successful global cities have. Tertiary educational and Singapore has three full-fledged universities with several The IDR plans to devote large areas to educational institutions and research institutions campuses for other tertiary activities (such as business schools is wooing universities to locate campuses there. Provided the and research institutes). As the economy climbs the value added immigration and transportation issues can be eased, locating ladder and is pressed to generate its own innovations and some of Singapore's future tertiary institutions in the IDR could intellectual capital, the need for more tertiary institutions will benefit both territories. grow substantially. Already, a fourth university is planned, even as the new third university is ramping up in Singapore. Warehousing Warehouses currently occupy about 6.1 square kilometers of land If port activities shift out of Singapore, a good part of such land will in Singapore. be released to other, potentially higher-value uses. Tourism Singapore's problem has been the declining duration of average IDR and Singapore can complement one another: (a) Singapore's tourist stay and slow increases in average expenditure per airport connectivity brings in large numbers of tourists, which tourist as well as an insufficient number of repeat visits by IDR can leverage; (b) IDR is developing theme parks, mangrove tourists. An inadequate range of visitor attractions and high nature reserves, and other attractions that can offer tourists in costs probably account for this. Singapore wider choices; (c) IDR offers cheaper accommodations for cost-conscious tourists. Back-office support To be viable, such support services need office space (land) and Malaysia has been consistently ranked third worldwide in business services relatively lower-paid staff. Neither are Singapore's strengths. process outsourcing according to A T Kearney. Skilled workers As wages and rentals rise in Singapore, activities that can are available in Malaysia who can relocate to the IDR. be shifted profitably out of Singapore could include all types of simple transactions processing, basic bookkeeping and accounting, customer service or call centers, draftsmanship for architectural firms, and so forth. Source: Estimates by Centennial Group. Singapore, the analysis is more complex. labor are released to higher-value activi- Table 5.4 assesses the potential benefits to ties. This would apply mostly to manu- Singapore of relocating activity to the IDR. facturing, port, warehousing, and back- This analysis supports the case for sub- room support services; stantial benefits from relocating present or · Greater competitiveness and business future activities from Singapore to the IDR. flexibility would come as Singapore The net benefits would be considerable: businesses are able to provide a range · Singapore as well as the IDR would real- of products and services profitably, the ize a one-off gain in GDP, as land and production of which can be located in 74 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Singapore or the IDR, depending on a limited amount of retirement savings: the where it is more profitable; lower costs of living in the IDR would enable · If additional land and labor from the them to stretch out their retirement funds for IDR were available, these constraints much longer than if they were living in Singa- would ease to some extent in Singapore pore. This is particularly pertinent given the and relieve cost pressures there. It would rising concerns over the adequacy of retire- effectively push out the production pos- ment financing in Singapore. At a broader sibility frontier, enabling Singapore to level, seamless movement of people between make a credible bid to become a global the two territories would enhance standards city of the standing of London or New of living by giving greater choices to consum- York. Without the IDR, Singapore would ers, who can weigh the advantages of living not have the advantage of a secondary in Singapore, such as proximity to workplace airport or the space to build tertiary edu- and more-developed physical, regulatory, and cational institutions capable of support- cultural infrastructure, to the advantages of ing its growth. living in the IDR, such as cheaper living costs, open spaces, and less congestion. The benefits of scale and diversity Leveraging off each other's If there were no political barriers between Singapore and the IDR, a larger joint econ- competitive advantages omy would offer several benefits to economic Table 5.5 expands on some of the factors agents. differentiating Singapore from the IDR, First, with a larger land area, larger pop- focusing specifically on various indicators ulation, larger market, and larger economy, of competitiveness. Although no separate there would be more economies of scale or, measure of competitiveness is available for at least, the promise of a rapidly growing the IDR, Malaysia's competitive positioning economy that would offer considerably will probably reflect the IDR's position in greater economies of scale. This could most instances. attract manufacturers or service produc- The wider the differences in competitive ers who desire a home market of a certain advantage between the two territories, the minimum size and who might otherwise more likely it is that synergies will be released decide to locate operations elsewhere. were they to collaborate more effectively. Second, a larger joint economy with a Singapore is better governed, a place where more diverse mix of skills, types of com- it is easier to do business, and more efficient panies, types of business activities, and a in logistics. Malaysia and the IDR offer one greater variety of business locations (some of the world's best places for business pro- cheap, some expensive) could accommodate cess outsourcing and an environment that the diversity of talents, business activities, is much more supportive of entrepreneur- consumer preferences, and skill sets that ship than Singapore. If Singapore and the make for a successful urban conurbation of IDR could be connected more seamlessly, global scale. each territory could leverage off the other's Third, Singapore's small and medium strengths and reap more benefits. enterprises (SMEs) could operate in the lower-cost IDR. Otherwise, rising costs in Increased competition: costs Singapore would squeeze them out. and benefits Similarly, Singaporeans at the bottom end There is a fear on both sides that there will of the income distribution could benefit if be competition as well as complementarities they had the option of living in the IDR and in the relationship between Singapore and commuting to Singapore,in the same way that the IDR. This may be true, but there are two lower-paid workers in Manhattan live outside arguments against this. First, the likely com- the city and commute to work each day. This plementarities far exceed the areas where would also apply to Singaporean retirees who there might be competition. Second, and are living on fixed pensions or are relying on more important, competition need not be The Iskandar Development Region and Singapore 75 a negative. A real case study is of the impact Table 5.5 Competitiveness indicators for Malaysia and Singapore of Port of Tanjung Pelepas (PTP) on Singa- Name of index or rank Malaysia Singapore pore. The opening of PTP did lure two of the World Economic Forum Global Competitiveness 21 7 Port of Singapore's largest customers. It also Institute for Management Development World Competitiveness 23 2 resulted in a loss of pricing power for the Port A T Kearney Globalization Index (2006) 19 1 of Singapore. However, the net effect is what A T Kearney FDI Confidence Index (2005) -- 18 is important. PTP's emergence forced the AT Kearney Global Services Index 3 11 Port of Singapore to restructure to reduce World Bank Governance Indicators (percentile rank, 2006) costs, improve efficiency, and be more sen- Political stability 58.7 94.7 sitive to customer needs. The overall impact Government effectiveness 80.6 99.5 was that both PTP and the Port of Singapore Regulatory quality 69.8 99.5 grew in revenues and profitability. Control of corruption 68.0 98.1 Similarly, there probably will be further World Bank Logistics Performance Index 27 1 dynamic effects of competition should there World Bank Doing Business Indicators 24 1 be greater economic integration between the Transparency International Corruption Perceptions Index 43 4 two territories. Global Entrepreneurship Monitor Report 2006 Percentage of population involved in early-stage entrepreneurial activity 11.1 4.9 The way forward Percentage of population who are established business owners 7.3 3.4 In this section, we look at what prerequisites Sources: Collated by Centennial Group from World Economic Forum; Institute for Management Development; have to be in place for a more integrated Sin- A T Kearney; World Bank; Transparency International; and Global Entrepreneurship Monitor Web sites. -- Not available. gapore-IDR region to form and succeed. We also assess the political obstacles that make successful collaboration difficult to achieve. nificance. There is no mass transit link that would allow large numbers of people to Economic and other prerequisites move between the two territories at low cost for synergies to be released and with high frequency of services. Conse- Increased collaboration can be structured quently, there is substantial congestion, with in a number of ways: from a collection of delays of up to an hour or more at peak peri- sector-specific collaborations to complete ods for the causeway route. This is enough economic integration in which political to deter the reallocation of many activities boundaries are irrelevant. However this from Singapore to the IDR. Multiple routes collaboration is structured, effective eco- of access involving several modes of trans- nomic collaboration or integration between portation are needed. Another land link to Singapore and the IDR can only happen if supplement the two existing land routes a number of conditions are in place. Four would help. Extending Singapore's mass areas of weakness seriously compromise the transit railway into the IDR would improve ability to expand collaboration between the the ease of access considerably. two territories. Second, border controls for immigra- tion, customs, and security checks are time- Lack of seamless connectivity for people and consuming enough to deter the free flow of goods. However Singapore and the IDR people and goods. Passports are required, collaborate, realizing materially important and forms have to be filled in and checked benefits that come from significant realloca- by immigration officers. Except for a limited tion of activities between the two territories number of individuals who have work per- requires a fairly seamless flow of goods and mits or student passes, there are no quick and people between and within the two territo- easy immigration checks with special passes ries. In concrete terms, this means improve- or smart cards. Security checks on the Singa- ments in the following key areas. pore side are intrusive and time-consuming. First, transportation is inadequate. Cur- Third, taxes and related barriers need to rently, there are only two access routes by be removed. Malaysia imposes special taxes road and one by rail, with the rail service on vehicles transporting cargo into Singa- too limited and infrequent to be of sig- pore from Malaysia, to encourage use of its 76 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA own ports and airports. Similarly, Singapore of foreign professionals. Singapore has liber- imposes restrictions involving timings, entry alized its policies even further, wooing global fees, and daily charges on Malaysian vehi- talent aggressively and offering liberal entry cles entering Singapore as part of its policy to such talent. Malaysia has been very open of restraining the use of motor vehicles to to receiving workers for jobs that Malaysians reduce congestion. no longer want to do, such as construction work, domestic help, plantation work, and Need to improve law and order. Rising low-end factory work. But its approach to crime in Johor has claimed victims among foreign professionals has been mixed. In both Johor residents and travelers from Sin- some cases, Malaysian officials have articu- gapore. Some well-reported cases of grue- lated their willingness to allow the entry of some murders, violent robberies, and car foreign talent in areas such as information thefts have alarmed Singaporeans, who have technology (IT) as part of its ambition to become accustomed to a very low level of excel as an IT hub. However, in practice, crime. The law-and-order situation could actually securing visas for such and related deter Singaporeans and others from visiting professionals has been difficult. Moreover, the IDR region if current trends continue. the mood seems to have soured in recent There is also the perception among Singa- months, with officials now discouraging the poreans that petty corruption (such as offi- inflow of semiprofessional workers, such as cials requiring incentive payments for gov- trained hotel staff. For the IDR to really take ernment services) is common in the IDR. off, a more pragmatic approach to foreign professionals may be necessary. Policy regime: security of investments and savings. Both Singapore and Malaysia What are the obstacles? have generally been investor-friendly juris- Improved political relations between Sin- dictions where the rule of law operates and gapore and Malaysia are a sine qua non where both domestic and foreign investors for effective collaboration. Without a clear conduct business with legal protections that demonstration that a new regime of politi- are effective. However, Singaporean inves- cal relations exists between the two coun- tors remember the Malaysian treatment tries, investors and businesses will not have of the CLOB over-the-counter market for the confidence that a deterioration in politi- Malaysian shares that operated in Singapore cal relations would not result in problems and how these shares were sequestered for for business or discriminatory policies that a number of years after Malaysia imposed would harm the value of their investments. capital controls in September 1998. These This is where the obstacles are serious shares were only released to Singaporean enough to raise questions about whether investors several years later and only after the full synergies between Singapore and the the payment of costly fees to an intermedi- IDR can ever be obtained. ary company set up by Malaysia. This his- tory deters many Singaporeans from whole- A troubled historical legacy. Economic heartedly investing in Malaysia. For their cooperation between Singapore and part, Malaysians were upset by the seemingly Malaysia takes place in the context of a arbitrary decision of Singapore not to allow relationship molded by a troubled political Malaysians from west Malaysia who worked and historical legacy. Singapore is a small in Singapore to take out their contributions island off the southernmost tip of Malay- to the Central Provident Fund when they left sia and was considered for long periods of Singapore, unlike the treatment accorded to its early history to be part of the kingdoms other foreigners and Malaysians from east and sultanates that controlled southern Malaysia. Malaysia. In the years before it was occu- pied by the British in 1819, Singapore was Policy regime: visas for foreign professional seen as belonging to the Johor-Riau-Lingga workers. Singapore and Malaysia have Sultanate, which was split as a result of the taken different approaches to the treatment interventions of the British and Dutch The Iskandar Development Region and Singapore 77 colonial powers that then dominated the disputed ownership of a rocky outcrop, and region. For the better part of its modern the location of customs, immigration, and history since its refounding in 1819 by the quarantine facilities for rail travel between British as a major port and regional center, Singapore and Malaysia. Singapore's economy has been materially integrated with what is now Malaysia. Even Divergent economic policies. Although after Singapore separated from Malaysia in both Malaysia and Singapore are highly August 1965, there was a short period until open economies that have been friendly to 1967 when the economies remained highly foreign investors and foreign trade, there are integrated, with the free flow of labor and some differences in economic policies. capital between the two countries and The first difference is the pursuit of companies incorporated in one territory ethnic-based economic policies in Malaysia. functioning in the other without too much After the racial convulsions of 1969, Malay- trouble. It was only after 1967 that the Sin- sia followed a policy (NEP) of affirmative gapore and Malaysian economies went their action favoring the indigenous communi- separate ways: economic policies diverged, ties of Malaysia, such as the Malays and the commonly owned airline was split into the various east Malaysian ethnic groups, two national airlines in 1972, and the cur- a policy that many Chinese and Indian rency union ended in 1973. Malaysians felt discriminated against them. There are several important reasons why Singapore sought to pursue a policy of political factors might hold back economic meritocracy, but in the early years of rapid ties between Singapore and the IDR. growth, its own Malay community felt mar- ginalized. These experiences helped to nur- Ethnic relations. Ethnic differences unfor- ture resentments on both sides against each tunately affect the relationship between the other. Because the NEP imposed ownership two countries. Singapore is a majority Chi- and other restrictions on businesses operat- nese country, where the Chinese are about ing in Malaysia, it affected the way largely 76 percent of the population, the Malays Chinese-owned Singapore investments are about 15 percent, and the Indians are operated in Malaysia. around 6 percent. Malaysia is the reverse: Second, following Singapore's separation, Malays and other indigenous groups make Malaysian policy makers noticed that sub- up more than 60 percent, while the Chinese stantial economic benefits accrued to Singa- proportion is now around 25 percent, and pore from its economic ties with Malaysia: the Indians account for about 10 percent. Singapore's port and airport handled a con- Singapore left the Federation of Malaysia siderable proportion of goods and people in August 1965 mainly as a result of ethnic being carried in and out of Malaysia, for tensions between the Malays who were the instance. In many cases, Malaysia followed majority in peninsular Malaysia and the Chi- a nationalist policy of developing its own nese who were the majority in Singapore. ports and airports to rival those of Singa- Two major race riots erupted in Singapore pore, using taxes and other economic inter- in 1964 in which many were killed. A bout of ventions to reduce Singapore's role in the racial violence in Malaysia in 1969 in which Malaysian economy. For example, a levy was hundreds died added to the bitter legacy. introduced in the 1980s on lorries carrying Following separation, various bilateral Malaysian-made goods for export through issues became sore points in the relationship Singapore port to divert goods to the port and remain today, sometimes decades after that Malaysia had developed in one part of they first emerged as problems, attesting to the IDR (Pasir Gudang Port). the difficulty involved in settling differences Many of these problems were specific to quickly. Issues included the price of water Johor, where the IDR is located. For instance, supplied to Singapore under what many the agreement on the supply of water by Malaysians deemed to be a one-sided agree- Johor to Singapore is seen by Johor as par- ment, the release of provident fund savings ticularly unfair to it, as it gives Singapore due to Malaysians working in Singapore, exclusive use of Johor's main water source 78 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA up to a certain level of demand and at a without the other. Thus they will have to price that is not indexed to inflation and has sink their political differences or abandon limited scope for adjustment. The issue of their ambitions and settle for a much more sovereignty over Singapore is also particu- mediocre set of goals. larly felt in Johor, where many people feel that Johor was somehow cheated out of Notes its historic ownership of Singapore by the Manu Bhaskaran is senior adjunct research British colonialists as well as by the federal scholar at the Institute of Policy Studies and government, which agreed to the separation partner and head of economic research at the of Singapore from Malaysia in 1965. These Centennial Group in Washington, DC. issues cloud relations with Singapore. This 1. Originally known as the South Johor Eco- means that simple issues of bilateral eco- nomic Region, it was renamed the Iskandar nomic interaction become easily politicized Development Region in November 2006 in honor and thus difficult to resolve. of Sultan Mahmud Iskandar, sultan of Johor, the state in which the IDR is located. Conclusion: can these obstacles 2. Data on the movement of people between Singapore and the IDR were given in a speech be overcome? by Malaysia's International Trade and Industry Ultimately, policy makers have to ask them- Minister Rafidah Aziz, which was reported in selves whether they can achieve their ambi- Singapore Business Times, August 29, 2007. tious goals without finding ways to overcome these political obstacles. Can Singapore References achieve its aim of becoming a global city on Chan, Kenneth. 2007. "Growth with Equity: The the scale of London or New York without Challenge of Income Distribution." Ethos 3 the IDR? And can the IDR be the strong, (October): n.p. sustainable metropolis of global significance Khazanah Malaysia. 2006. "Comprehensive it aims to be without substantial integration Development Plan for South Johor Economic with Singapore? The answers in both cases Region." Report prepared for the Government are clear: neither side can achieve its aims of Malaysia, November 4. Spatial integration and human transformations in the Greater Mekong subregion Jonathan Rigg and Chusak Wittayapak 6 At the core of the World Development Report of marginality and integration in similar 2009 (WDR 2009) is the contention that, ways. We start from the premise, therefore, while the concentration of economic activity that the devil really is in the details. Like is positive because it both stimulates further Ravallion, we are interested in exploring economic growth and is an inevitable out- the "churning that is found under the sur- come of such activity and growth, it also face of the aggregate outcomes" (Ravallion tends to lead to great(er) spatial inequali- 2001: 1812). c h a p t e r ties, which are undesirable. The report We do not seek to challenge the core proposes that the solution to this develop- assumption of the WDR 2009 that regional ment conundrum is the better integration spatial integration and concentration tend of markets. Such integration will enhance to increase aggregate output; the evidence, economic concentration (a "good" thing), both national and international, from the while tempering the tendency to deepen countries of the GMS suggests--strongly-- spatial inequalities (a "bad" thing). otherwise. Rather, we are intent on high- Drawing on the experience of the coun- lighting the inequalities, inconsistencies, tries of the Greater Mekong subregion (the and incongruities that accompany this GMS), which are undergoing just such a process. In particular, we seek to show not spatial transformation in one of the world's only that economic concentration does lead most economically dynamic regions, this to deeper spatial inequalities but also that chapter highlights the practical difficulties spatial integration--the "solution" to such a and policy challenges of achieving such tendency--is accompanied by its own nega- a win-win outcome. We do this, first by tive and undesirable side effects. We end the focusing on spatial integration as a neces- chapter by reflecting on the policy implica- sarily unsettling and destabilizing process; tions of these processes. second, by emphasizing the societal and environmental outcomes and side effects The GMS: an idea becomes of the concentration of economic activities and the integration of markets; and third, a subregion by viewing geographic space not as the mere The GMS encompasses six countries cen- stage on which certain activities occur, but tered on mainland Southeast Asia: Cam- as socially produced and politically charged. bodia, China (originally Yunnan province Regarding the last point, we wish to avoid, only, but since 2005 also including Guangxi for example, simply assuming that remote- Zhuang Autonomous Region), Lao People's ness is a problem, that policies and pro- Democratic Republic, Myanmar, Thailand, grams to stimulate spatial integration are and Vietnam (see figure 6.1). The GMS necessarily beneficial, and that people are program was launched in 1992 and given affected by and respond to the challenges further impetus in November 2001 when 79 80 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 6.1 The Greater Mekong subregion 95° E 100° E 105° E 110° E M ekong CHINA 25° N 25° N in hindw Y U N N A N C Red G U A N G X I MYANMAR S wal Hanoi A een Naypyidaw LAO 20° N yeyaw 20° N and P.D.R. Gulf of y Tonkin Bay of Sittang Vientiane VIETNAM Bengal THAILAND 15° N 15° N Bangkok CAMBODIA Phnom- Andaman Penh Sea Gulf of 10° N Thailand 10° N 5° N 100° E 105° E 110° E 5° N Source: http://www.adb.org/GMS/img/gmsmap.gif. the Strategic Framework for the GMS · Facilitate cross-border trade and invest- was adopted at the Association of South ment; East Asian Nations (ASEAN) ministerial · Enhance private sector participation in conference.1 At the first GMS summit held development and improve its competi- in Phnom Penh at the end of November tiveness; 2002, the leaders of the subregion endorsed Develop human resources and skill com- a 10-year strategic framework with five · strategic thrusts:2 petencies; and · Protect the environment and promote · Strengtheninfrastructurelinkagesthrough sustainable use of the subregion's shared a multisectoral approach; natural resources. Spatial integration and human transformations in the Greater Mekong subregion 81 In a document with the title Linking · Subregional power interconnection and Nations, Connecting People, the rationale for trading arrangements; the GMS program is summarized in the · Cross-border trade and investment; following terms (ADB 2005a: 7): · Private sector participation and compet- [To promote] closer economic ties and coop- itiveness; eration among the six countries. Its vision is · Human resources and skills competencies; to create a more integrated, prosperous, and · Strategic environmental framework; equitable Mekong subregion, complementing national efforts to promote economic growth · Flood control and water resource man- and reduce poverty and augmenting domes- agement; and tic development opportunities to create sub- · Tourism development. regional opportunities. It seeks to encourage trade and investment among GMS countries, From the start, therefore, the GMS had ease the cross-border movement of people a strong infrastructural justification under- and goods, and meet common resource and pinned by a set of assumptions that resonate policy needs. with the WDR 2009. Political rapproche- Underpinning the GMS are different ment and an easing of security tensions in "logics." The economic logic of the GMS the subregion provided the opportunity for lies in the productivity returns that can be cooperation, but the "program's first prior- garnered from the spatial integration of ity was ... to create the vital links within and countries with complementary economies. between countries and promote the devel- Complementarity, here, is rooted in differ- opment of the subregion's resource base" ence; it is because the GMS countries and (ADB 2005a: 9). These linkages connect their economies are different--in wealth as the rural poor to urban-centered services, much as in composition--that economic jobs, and amenities; connect remote regions cooperation is worthwhile (see table 6.1). to the national (and wider) economy; and The geographic logic is founded on the connect backward rural economies with the Mekong, the hydrological thread that links modernizing urban core. There is no doubt the countries of the GMS. And the politi- that the GMS program is ambitious. By mid- cal logic arises from the era of peace and 2006, 26 GMS projects were being funded to rapprochement that saw the six countries the tune of US$6.5 billion.4 In the Kunming of the GMS make the transition from Cold Declaration of July 2005, the GMS countries War foes to post­Cold War friends from the reaffirmed the commitments they had made early 1990s, reflected in their membership at the first GMS summit held in Phnom (with the exception of China) in ASEAN. Penh in November 2002 (ADB 2005b: 17): Of the five strategic thrusts noted above, Cross-border infrastructure is key to eco- the first two relate explicitly to regional spa- nomic development and prosperity in the tial integration in physical terms, the third region. A well-built, seamless, multimodal and fourth relate to private and public sector infrastructure is essential to the facilitation of regional cooperation, and the last relates to trade, movement of people, and the provision regional resource cooperation and manage- of basic services throughout the whole region. ment. The 11 "flagship" programs that will We therefore commit ourselves to fully "con- necting the GMS." To that end, we commit to deliver this strategic framework are all ori- sustained and greater inputs to strengthen the ented toward integration of the subregion, subregional infrastructure linkages through a with a particular focus on three economic multisector and holistic approach. corridors (ADB 2005c):3 · North-south economic corridor; Openness,progress,and · East-west economic corridor; inequality in the GMS · Southern economic corridor; Before we consider in more detail why · Telecommunications backbone and infor- and how spatial integration leads to an mation and communications technology; unsettling of categories and scales and the 82 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 6.1 Openness, progress, and inequality in the GMS, 1990­2006 Indicator and year Cambodia China Lao PDR Myanmar Thailand Vietnam Population (million) 1990 8.6 1,143 4.1 40.8 55.8 66.0 2006 14.2 1,315 5.7 56.5 65.2 84.2 Economic progress Average annual growth in per capita GDP, -- 9.11 3.75 -- 3.52 5.93 1990­2005 (percent) Per capita income (current US$) 1992 220 415 (293) 271 -- 1,945 144 2006 510 1,999 (842)f 601 176d 3,133 724 Percent of GDP (2006) Agriculture 30.1 11.8 44.8 48.4 10.7 20.4 Industry 26.2 48.7 29.5 16.2 44.6 41.6 Services 38.6 39.5 25.7 35.4 44.7 38.1 Development, well-being, and inequality Poverty (percent of population living on less than PPP US$1 a day) 1990 46 32.5 52.7 -- 10.2 50.7 2005 12.7 7.1 21.3 -- 0.0 6.5 Poverty (percent of population living on less than PPP US$2 a day) 1990 76.3 71.5 89.6 -- 43.1 87.0 2005 54.5 29.4 67.7 -- 16.2 39.7 Percent of population living in poverty 34.7f -- 32.7e 26.6c 9.8d 19.5f (percent of national poverty line) Number of US$1-a-day poor (million) 1990 4.0 377 2.2 -- 5.7 33.4 2003 4.5 173 1.6 -- 0.4 7.9 Human development index 1990 0.512 0.627 0.449 -- 0.707 0.610 2004 0.583 0.768 0.553 0.581 0.784 0.709 Gini coefficient 1993 31.80 40.70 30.40a -- 46.22a 34.91 2004 38.05 45.50 34.68d -- 41.96d 37.08f Openness and integration Openness ratio (ratio of total trade to GDP at current market prices) 1992 35.8 27.7 33.8 2.8 64.9 50.8 2006 117.2 65.7 56.7 -- 123.3 136.4 Foreign direct investment (US$ million) 1992 33 11,008 280b 149 2,151 474 2006 318g 64,468 650 128e 8,837 4,100 Investment rate (ratio of gross domestic investment to GDP) 1992 11.3 36.2 -- 1.3 40.0 29.6b 2006 20.8 44.9 -- 11.0e 27.9 35.4g Tourist arrivals (million) 1995 0.22 1.02 0.35 0.12 6.95 1.35 2006 1.70 3.38 1.26 0.21 13.82 3.58 Registered migrants in Thailand 1998 9,492 -- 1,164 89,318 n.a. -- 2004 104,789 -- 99,352 610,106 n.a. -- Sources: ADB (2007a, 2007b); Ali and Zhuang (2007); Caouette and others (2007). Note: The per capita GDP figures in parentheses for China are for Yunnan province and Guangxi Zhuang Autonomous Region; the data on registered migrants in Thailand should be treated with caution because of the large number of unrecorded migrants. a. 1992. b. 2000. c. 2001. d. 2002. e. 2003. f. 2004. g. 2005. n.a. Not applicable. -- Not available. Spatial integration and human transformations in the Greater Mekong subregion 83 destabilization of lives and livelihoods, we and inclusive development (ADB 2007b: wish to set out the positive effects that have 12­13). Three issues stand out. First, there flowed, directly and indirectly, from GMS is the question of the quality of integration. integration. Regional road arteries such as the GMS eco- A mid-term review of the GMS pro- nomic corridors need to be accompanied gram, published in June 2007, observes that by rural feeder roads and improvements in the "GMS economies have grown at one of domestic infrastructure if they are to deliver the fastest rates in the world since the early benefits that are broad based in both their 1990s, as many of them started the transi- social and spatial impacts. In an economet- tion from central planning to market-based ric analysis of the impacts of cross-border systems and began opening up and integrat- road infrastructure on trade and foreign ing their economies with the other countries direct investment in the GMS, Edmonds and in the subregion, the rest of Asia, and the Fujimura (2006) conclude that improve- world" (ADB 2007b: 4). Table 6.1 presents ments in road infrastructure boost trade, the empirical indicators of economic and particularly when domestic road infrastruc- social development among the countries of ture is good.6 In other words, the focus on the GMS since the grouping was formally transnational (cross-border) links must be established in 1992. There are a number of accompanied by payment of an equal level points to note: first, the clear progress that of attention to the national infrastructural has been achieved by the countries of the dimension.A related point is evident in Warr GMS over the period since 1992, particularly and Menon's (2006) general equilibrium in terms of poverty reduction and per capita model of road improvement and poverty GDP (all the more remarkable bearing in reduction in Lao PDR, in which they note mind the Asian economic crisis of 1997­99);5 that the pro-poor impacts are significantly second, the degree to which the region has greater when households without road become more open, whether measured in access are provided with dry-season access terms of the openness ratio, foreign direct than when dry-season roads are upgraded investment (FDI), tourist arrivals, migra- to all-weather status. Second, it cannot be tion streams, or investment rate; third, the assumed that the poor and the vulnerable-- manner in which economic progress and the destitute, the elderly, ethnic minorities, greater openness have been accompanied, women--will find equal and equivalent particularly in the reform economies, by benefits from regional and national integra- widening inequalities (and also see table tion. Indeed, they may counterintuitively 6.4); and finally, the large number of people be "crowded out" by the process of integra- who are living close to poverty, reflected in tion (ADB 2007b: 12). And third, regional the differential between the incidence of integration can be accompanied by certain US$1-a-day and US$2-a-day poverty. adverse effects, most obviously in the realm Of course, in considering historical cau- of environmental degradation. We explore sality we must be wary of post hoc rational- the last two issues in more detail below. ization--the logical fallacy of assuming that, because one thing follows another, they must Scales and sites: the empirics be causally linked. With this in mind, the of spatial transformations following questions are relevant: Has greater openness driven the economic gains of the in the GMS last 10­15 years? Have deepening inequali- The core of this chapter focuses on the effects ties been an outcome of market reforms? of spatial transformations on people and And what is the relationship among eco- places. But to get to that point, it is valuable nomic reform in general, the GMS initiative to set out the spatial, policy, and historical in particular, and economic progress, on the contexts within which, and on which, those one hand, and inequality, on the other? effects are set. In doing this, it will become The GMS mid-term review notes that clear why the GMS offers such a rich ground openness and integration per se are not suf- for reflecting on the issues debated in the ficient, in themselves, to deliver broad-based WDR 2009. 84 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Lagging countries, lagging regions, have incomes grown faster for the poor(er) lagging people than for the rich(er). In drawing (relatively) unconnected The GMS region shows marked inequalities regions and areas into the mainstream, GMS in income and poverty at the international policies are likely to be narrowing inequalities and interregional levels. More than half of at some scales, while widening inequalities at the population of Lao PDR and Cambodia others.More precisely,interregional inequal- live below the PPP (purchasing power par- ities are likely to narrow, while intraregional ity) US$2-a-day poverty line, compared inequalities will widen. This is because of the with less than one-fifth of the population way in which better access bestows differen- in Thailand (table 6.1). More significantly, tial benefits on social classes and population in Thailand, more than half of the coun- groups. Generally, men are in a better posi- try's total poor population of a little over tion to benefit than women, young(er) than 7 million are concentrated in the north- old(er), rich than poor, majority populations east region; in Lao PDR and Vietnam, the than minority groups, and the educated than poor are disproportionately concentrated in the less well educated. This is not a reason to upland areas and among ethnic minorities curtail further integration, but it does high- (see table 6.2), while in all the countries of light the existence of a variegated landscape the GMS, poverty is to a large extent a rural of opportunity that represents both a devel- phenomenon (see table 6.3). The GMS proj- opment challenge and a political dilemma. ect is directed, in no small way, at address- ing these spatial manifestations of poverty, Border sites and cross-border whether they reflect inequalities among interactions: economic and countries, among regions, or between rural environmental and urban areas. There is good evidence that income-expenditure inequalities have wid- Research undertaken in borderland areas ened significantly in the transition econo- of the GMS notes the increase in economic mies of the GMS over the course of the last activity made possible by improving trans- 10­15 years (see table 6.4). Only in Thailand port links, receding political and bureau- cratic barriers to exchange, and a shared economic vision. The deepening of trans- Table 6.2 Incidence of poverty in Lao PDR, by ethnolinguistic family, 2001 boundary economic relations is seen by gov- Percent of poor Percent of total ernments, businesspeople, and multilateral Family in sample sites population agencies as providing considerable scope for Mon-Khmer 56 23.5 local development. These borderland sites, Hmong-Mien 15 7.5 therefore, take on particular qualities that Tibeto-Burman 9 2.5 are, in part, a product of their geographic Tai-Kadai location. They may develop in such a way Thai-Thay 13 36.5 Lao 7 30.0 that they become "enclave" zones relative Total 100 100.0 to other areas, sites of particular economic Source: ADB (2001: 25). dynamism and, also, social tensions. For Swe Note: Column 1 shows the percentage of poor by ethnic group in the sampled poor sites; and Chambers, "Frontier towns represent column 2 shows the estimated representation of each ethnic group in the total popula- tion. So, while ethnic Lao comprise 30 percent of the population of the Lao PDR, they a nexus where opportunities for profit ... make up only 7 percent of the population of poor sites in this survey. abound" and which are "increasingly serv- ing as strategic nodes for commerce and Table 6.3 Rural and urban distribution of poverty based on national growth in a singular segment of a multi- poverty lines, various years segmented region state [that is, the GMS]" Country and year Total Urban Rural (Swe and Chambers 2008: 2). Moreover, China (1998) 4.6 2.0 4.6 while obstacles to trade persist, from transit Cambodia (1999) 35.9 18.2 40.1 taxes (including bribes) to stifling bureau- Lao PDR (1997) 38.6 26.9 41.0 cratic inefficiencies, the growth of com- Myanmar (1997) 22.9 23.9 22.4 Thailand (2002) 9.8 4.0 12.6 merce is seen as a"positive-sum game for all Vietnam (2002) 28.9 6.6 35.6 countries in terms of profits gained" (Swe Source: http://www.adb.org/documents/books/key_indicators/2005/xls/rt01.xls. and Chambers 2008: 3). Spatial integration and human transformations in the Greater Mekong subregion 85 Table 6.4 Annualized growth rates of per capita expenditure and income, by country and quintile Q1 (poorest Q5 (richest Country 20 percent) Q2 Q3 Q4 20 percent) Cambodia (1993­2004) 0.69 1.27 1.84 2.39 3.38 China (1993­2004) 3.40 4.46 5.42 6.19 7.10 Laos (1992­2002) 1.47 2.22 2.85 3.40 3.82 Myanmar -- -- -- -- -- Thailand (1992­2002) 2.35 2.27 1.96 1.51 0.38 Vietnam (1993­2004) 3.37 3.92 4.29 4.61 4.69 Source: Extracted from ADB (2007a: 35). -- Not available. Gainsborough (2007) undertook research the state. However, while there has been an in two Vietnamese borderland sites, in Lao increase in the number of private actors in Cai in the north, on the Vietnam-China his two border case studies, their success is border, and in Tay Ninh in the south, on the contingent on their close links with the state Vietnam-Cambodia border. His interest lay and state enterprises and agencies. He says, in unpicking the relative roles of the state, "I have ... argued in this article that there private enterprise, and multilateral institu- are important ways in which the growth tions in orchestrating trade flows across of private and transnational actors may these frontiers. In the context of this chap- be associated with a strengthening--not ter, Gainsborough wonders whether the a decline--of state power in some areas" policies and programs of the GMS have had (Gainsborough 2007: 15).7 any effect on cross-border flows or whether Border zones are, self-evidently, political this is just wishful thinking. He concludes, and politicized spaces: the frontier makes "In relation to the GMS it is hard to argue them so. As Sturgeon writes in her book against the view that there has been a signifi- on border landscapes in China and Thai- cant increase in the intensity of cross-border land, "Borders are processes replete with flows of goods, people, money, and informa- politics, both as margins of the nation state tion since the early 1990s" (Gainsborough (border-as-margin) and as cross-border 2007: 8; see also table 6.5). He also suggests social relations (border-as-line)" (Sturgeon that Asian Development Bank (ADB) invest- 2005: 201). What is less often investigated is ments linked to the GMS program have the way in which the politics of access fall played a defining role in delivering improved unequally on groups living in and outside infrastructure and raising prosperity that, in the border zone. The GMS may be trying to turn, lie behind the increased trade flows. go "beyond borders" (ADB 2005b), but this A concern in Gainsborough's study is to must be seen--for the time being--as just ascertain who the "actors" are in cross-bor- an articulated desire. Borders matter, with der trade and what their relationships are the result that there is a quite distinctive with the state. For him, there is a tendency border or frontier geography. among scholars and officials to assume that The increase in economic activity in private enterprises are taking the place of borderlands arises partly because integra- tion permits the exploitation of space and Table 6.5 Expanded trade flows in the Greater Mekong subregion two-way trade (US$ million) the more efficient and intensive use and extraction of natural resources. Cross- Vietnam and Vietnam Time period Cambodia and China border market forces are shaping the trans- formation of the agricultural sector in the Early 1990s 30­40 300a 1995 118.1 691.6 Mekong corridor as production is oriented 2000 178.9 2,937.5 toward the demand profiles of China, Viet- 2003 300 4,800 nam, and Thailand (Lao PDR 1999: 35). Annual increase, 19.3 73.0 Sometimes acute environmental pressures 1995­2003 (percent) have arisen from such spatial integration. Source: Gainsborough (2007: 8). a. 1991. Improvements to the east-west economic 86 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA corridor, for example, led to the "massive is based on the premise that the most effec- illegal movement of live animals [from Lao tive means with which to narrow spatial eco- PDR] into neighboring countries [Thailand nomic inequalities is by drawing people and and Vietnam]"(UNEP 2001: 55). Consumer places into the market mainstream. Poverty demand in China has fueled an unsustain- has a strong spatial component, and the able harvesting of nontimber forest prod- poor are concentrated in those areas where ucts in provinces like Luang Prabang and the market has a weak presence. Roads can Luang Namtha and their funneling, along bring both the market to the people and the the valley of the Nam Ou and the Nam Tha, people to the market, thus becoming arteries to markets and consumers in China (ADB through and along which spatial inequalities 2000: 8; see also Lao PDR 2000b: 44; Hoang in development and service provision can 2007). In northern Lao, the rapid expansion be bridged. That roads are developmental is since 2000 of Chinese rubber concessions taken, often, as both obvious and unprob- has threatened the sustainability of ethnic lematic: "Remoteness is an important cause minority­operated ecotourism activities of rural poverty" (World Bank 1999: 7), and (Schipani 2007). a "well-managed road network is one of the Our view of the livelihood and environ- essential prerequisites for economic growth, mental effects of spatial transformations and, given the growing focus on developing such as those linked to (but not limited to) rural areas, it is a sine qua non for balanced the GMS project entertains the possibil- and equitable growth for all sectors of the ity, indeed the likelihood, that they will be community" (Lao PDR 2000a: 64; see also mixed in general and will vary across popu- UNDP 1996: 3; Lao PDR 2000c: 9). lation groups. This latter issue is explored in There is also strong evidence that road greater depth below. It is also the case that, improvements help in delivering social while environmental tensions may be par- development and reducing poverty. Using ticularly noticeable across borders, they are the 1997­98 and 2002­03 Lao Expenditure not limited to border zones, and the market and Consumption Surveys, Menon has stud- effects that more open borders engender ied the impact of road improvements on may have considerable spatial "reach." household well-being. He concludes, "Road improvement in rural areas can contribute Agents,agency,and impacts to lowering poverty incidence, improving of spatial transformation educational participation of primary school- age children, and reducing the rate of illness" in the GMS and calculates that around one-quarter of Of all the investments in physical infra- the reduction in poverty over the period structure, none, arguably, has done more to between the two surveys could be attrib- transform regional economic landscapes, uted to the conversion of dry-season access spatialities of production and consump- roads into all-weather roads (cited in ADB tion, household livelihoods, and individual 2007b: 10). Road improvements increase mind-sets than has investment in roads. This access to opportunities beyond the local section, therefore, deals specifically with the area, boost the potential for in situ local eco- impact of roads and the movement of people. nomic development by reducing transaction costs, and bring services such as schools and Roads health centers within easier reach of people, For many analysts, the benefits of spatial particularly in rural areas. This is also con- integration--and the costs of isolation--are firmed in a second study of Lao PDR, which self-evident: "Investment in physical infra- concludes that "reducing transport costs structure will significantly contribute to the through rural road improvement generates pursuit of socially inclusive development. ... significant reductions in poverty incidence" Roads appear to have strong indirect and (Warr and Menon 2006: 16). direct effects on poverty reduction" (Ali That roads change things in profound and Pernia 2003: 2, 10). The road-building and significant ways is without question. But imperative that informs the GMS initiative two additional questions have to be asked: Spatial integration and human transformations in the Greater Mekong subregion 87 How? and For whom? As Leinbach says, integration in the GMS. An ADB report, for "We still know all too little about the ways in example, claims in the context of Lao PDR which rural transport should be improved that the "penetration of the market may be and how to deliver benefits to more needy aggravating ... social differentiation with the populations" (Leinbach 2000: 2). This emergence of an entrepreneurial (capitalist) extends from their effects on individual group of farm households, on the one hand, mobility to their distributional implications and a dispossessed labor-selling group of and their direct and indirect effects on agri- households, on the other" (ADB 1999: 6). cultural and nonagricultural productivity. In 1999 scholars at the National Uni- For Johnston (2007: 171­72; see also van versity of Laos studied the impacts of the de Walle 2002), the three main fallacies that upgrading of Route 7 on 227 households have dogged work on transportation and in six villages in the provinces of Houa development are (a) the presumed direct Phanh and Xieng Khouang (NUOL 1999). causal link between transport improve- The study lists a large number of positive ments and economic growth, (b) the belief impacts of road upgrading but also notes that improved transport will inevitably lead that, in all the study villages, poor house- to higher agricultural output and better ser- holds had a markedly lower level of engage- vice provision and use, and (c) the fact that ment with the sorts of new market-based the benefits of such improvements will hold activities that road upgrading encourages: attractions for all and be distributed equally "The lack of capital available to the poorest through a population. group, and their related lower participation in current economic activities, suggests that Roads and spaces of inequality. There these households will be at a disadvantage is little doubt that building or upgrading in relation to the economic opportunities roads--in general--increases aggregate out- afforded by road improvement ... Potential put and has a positive effect on poverty. It is benefits from increased market access will also true that people in areas without easy be relatively lower ... In this way, road devel- road access often seem to crave better trans- opment may indirectly lead to increased port. It is one of the interventions that local differences between wealth groups" (NUOL people mention most often and prioritize 1999: 55­56). most highly. So roads are not interventions The big winners from road construc- "imposed" on local people from above; they tion are, almost inevitably, the wealthy are usually enthusiastically welcomed from and middle-income households who have below as well. the resources to exploit a latent resource The first general point to make is that (ILO 1997: 6). Poorer households often find market integration tends to accentuate social themselves unable to exploit and therefore differentiation by giving certain groups the benefit from the economic potential of ability to accumulate wealth. Where com- roads. At the same time, because roads can munity regulation is weak or where power disturb established patterns of activity--by, imbalances are great, natural resources for example, increasing logging or acceler- may be appropriated whether by the state, ating the exploitation of natural resources by outsiders, or by wealthy and influential more generally--they can harm those local people. Lowlanders entering upland groups (tribal), households (poor), and areas, using roads as access conduits, often individuals (women) who depend on the have advantages over local people in terms natural environment for their livelihood of language, financial resources, contacts, and well-being. and business acumen. Rather more conten- The spatial poverty traps facing women tiously, some analysts believe that, although are different from those facing men. This is road-led market integration benefits some, a product not only of poverty per se (or only it may harm others. Furthermore, one does of poverty) but also of cultural norms, eco- not have to look at the more radical literature nomic circumstances, and productive and to unearth concerns about the marginaliz- reproductive demands and needs. "To most ing effects of road construction and spatial women it does not really matter that much if 88 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA they are able to make the once-a-month trip the poorest people in a very poor country, to Vientiane [Lao's capital] in one and a half and the construction of roads into the area hours instead of three or four compared to has played a central role in driving environ- the time-consuming daily necessities of car- mental deterioration and livelihood col- rying water and fuel for household needs" lapse. The sequence of changes is outlined in (Trankell 1993: 84). For some women, the a study of three villages on the Nakai plateau key spatial development needs are not, undertaken in November 2000 (Culas 2001). therefore, off-farm and extra-village, but In 1995 a laterite logging road was cut to the on-farm and intra-village. It is improved villages of Ban Makfeuang, Ban Navang, and transport of water and firewood from the Ban Theung. This was used as such for only river and forest to home that would do most two or three years, until 1997­98. The road to revolutionize women's lives, not the abil- did not become an axis of development for ity to access a local urban center more easily villagers because no one owned a truck or (see table 6.6). even a motorbike. Instead, the road became Transport issues related to gender are the means by which outsiders could pen- underpinned and overlain by class- or etrate the area. Lowland Lao and Vietnam- wealth-based inequalities. Poor families do ese traders created higher demand for rare not have the means or the time to travel. woods, endangered species, and nontimber This applies to women and men. Moreover, forest products more generally (Culas 2001: we know from the experience of Thailand 29). Some of this wealth did trickle down to that cultural and social change can very the largely Brou and Sek inhabitants of the quickly undermine our assumptions about three villages, but only to some households the gendered nature of mobility. In the 1970s and usually in small quantities. The great women in Thailand were relatively immobile benefits accrued to outsiders, leaving the vil- compared with men; cultural norms about lages with a degraded resource and villagers, seemly behavior militated against female particularly poor villagers, with a yet more mobility, and there were, in any case, few off- tenuous existence. farm employment opportunities available It is from grounded, local-level, and for women. By the early 1990s, the moral often qualitative studies such as these that envelope of accepted practice had been it is possible to build an understanding of torn open, and modern factories selectively the pattern of data revealed in table 6.4. employing young women had blossomed We can surmise that, without the reforms so that women became, often, more mobile and the market and spatial integration of than men. Beyond class and gender, ethnic- the last 10­15 years, the countries and the ity and generation can also have a determin- people of the GMS, in general, would almost ing effect on patterns and impacts of spatial certainly be poorer, but they would also be integration. The minority inhabitants of more equal. Lao's Nakai plateau, for instance, are among Roads: creating new spaces of isolation. One of the less understood aspects of spa- Table 6.6 Effects of improved roads and transport tial integration is the manner in which Category Effects of improved roads or transport improving access can actually, and counter- Women Women are less likely to be able to take advantage of improving transport intuitively, increase isolation.At the regional facilities, even when cost is not an issue, because they face social level, fears have been expressed that, in barriers to mobility such as the stigma of riding a bicycle or traveling connecting centers of economic activity alone outside their community. Women and men Women and men have different transport needs. Women's needs tend to in the GMS, those outside the corridors be for frequent, local journeys; men's tend to be for less frequent, longer of connection and nodes of activity may trips. Women's trips are directed at meeting household consumption requirements; men's trips are for income generation and production. actually find themselves more isolated: Rich and poor Richer families have the time to travel, the products to sell, and the money to "With current focus only on transborder purchase goods. The poor are short of time and money, and better roads economic flows, it [the GMS program] is often do not increase incomes because they have nothing to sell. at risk of doing little beyond fostering an Very poor The very poor usually walk and "inhabit a localized walking world" (Hettige 2006: 18); roads deliver little for this marginal and marginalized group. entrepōt region, defined by increasingly Source: Information extracted from Hettige (2006). complex corridors linking poles of activity, Spatial integration and human transformations in the Greater Mekong subregion 89 but with very little else within or between" have few livelihood alternatives and may (Oehlers 2006: 467). Local-level studies over-extract resources in an attempt to sur- demonstrate most convincingly how and vive" (ADB 2000: 18). But this, the report why increased isolation may result from argues, only occurs when land and resources programs of infrastructural development become scarce, and resources become scarce and market integration. largely because of pressures that are nonlocal It seems that two processes are under way. in origin, including in-migration, establish- On the one hand, when roads are upgraded, ment of new protected areas or hydropower this can accentuate "tarmac bias," making developments, and unsustainable commer- off-road communities more cut off and cial logging. Furthermore, a "power imbal- isolated rather than less so, as market activi- ance leads to a fundamental inequity in ties are concentrated along the roadside and the flow of ecological goods and services traders restrict their activities to near-road between the uplands and lowlands" (ADB locations.At the same time, roads operate as 2000: 5). axes of attraction for people living off-road, These pressures are brought to bear encouraging spontaneous migration to the through processes of spatial integration. In roadside. There is little work from Southeast Saravan, Lao PDR, the increasing presence Asia on the spatially marginalizing effects of of Vietnamese traders is raising fears that road improvements--anecdotal evidence traditional livelihood systems will collapse aside--but Porter's research on the Jos pla- (Denes 1998: 11). In the Sii Phan Done area teau in Nigeria is instructive (Porter 1995: of Champassak province close to the bor- 10­12; see also Porter 2002). Bush village der between Lao PDR and Cambodia, Bush populations declined as individuals, house- (2004) suggests that the general decline holds, and sometimes entire villages relo- in fish stocks is caused by market integra- cated to the roadside. Loss of population led tion, driven by political rapprochement and to a loss of local markets in off-road areas, infrastructural improvements, which are and this, in turn, led to a loss of market con- tying the area into wider regional trading nection. Dirt roads fell into disrepair, and networks. In the 10 years between 1989 and bush transport services dwindled.The effects 1999, the availability of fish and rattan for were felt particularly keenly by women and the residents of Ban Nong Hin in Champas- the poor, whose livelihood interests tended sak declined precipitously (UNDP 2002; see to be local rather than extra-local and also table 6.7). who could not afford, either financially or There is little doubt that, in the end, spa- in terms of time, to travel to the roadside. tial integration delivers greater economic Cultural and reproductive impediments to returns; but, and here we depart from the travel also constrained women's mobility. WDR 2009, we question whether market Porter concludes that many bush markets integration will moderate the inequality- on the Jos plateau were "in a stage of termi- widening effects of the concentration of nal decline, kept going only by local women economic activity, at least initially. Instead, who patronize the market both as traders we propose a temporal sequence to the and purchasers" (Porter 1995: 11). triangular relationship among infrastruc- ture-integration, environment-livelihoods, Roads, spatial integration, and the environ- and inequality. To being with, better access ment. As indicated in the discussion of the Nakai plateau, it is often the deleterious environmental effects of spatial integration Table 6.7 Decline in the availability of nontimber forest products in Ban Nong Hin, Champassak province, Lao PDR, 1989­99 that do most to undermine the livelihoods of some of the poor.An ADB report on envi- Product 1989 1999 ronmental management in the remote GMS Wildlife An abundance of animals are available Many species have disappeared, and a watersheds of Cambodia, China, Lao PDR, in "your own backyard." two-day trek may yield nothing. Fish One hour's fishing yields 4­5 kilograms One hour's fishing yields 0.5 kilogram and Vietnam explores poverty-environment of fish. of fish. linkages and states, "The conventional wis- Rattan One day's collecting yields 300 stems. One day's collecting yields 20­30 stems. dom is that poor people in remote areas Source: Adapted from UNDP (2002: 82). 90 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA leads to accelerated exploitation of natural spatial integration effects that are beyond resources and environmental decline. In their control. This, though, overlooks the extreme instances, this may take the form agency of individuals and the surprising of a boom-and-bust cycle. Returns to this and unexpected ways in which they take boom are unequally allocated, both socially advantage of opportunities afforded by (richer and better connected groups and spatial integration policies. Nowhere is this individuals) and spatially (nonlocal groups more dramatic, at least for the GMS, than in and individuals). In time, the local economy the revolution that has occurred in people's is reoriented and restructured as connec- mobility. The discussion here serves to tem- tions permit new activities to colonize local per the tone of the last section, while also spaces and local people to access nonlocal injecting a destabilizing human component opportunities. This end point, however, can into what we regard as the rather too neat- be seen as arising from the environmentally and-tidy depiction of the economics of spa- destructive and selectively immiserizing tial integration depicted in the WDR 2009. earlier stage in the process of spatial inte- Since the mid-1970s in Thailand,from the gration. The best example from the GMS of mid-1980s in China, Lao PDR, and Vietnam, how spatial integration can create poverty and from the early 1990s in Cambodia and and erode livelihoods is reflected in the Myanmar, the GMS has become a region, debate over the focal site strategy of the Lao increasingly, "on the move" (see figure 6.2). government (see box 6.1). This includes patterns of daily mobility and longer-term and longer-distance migration, People, migration, and mobility both national and international. The work The discussion thus far has tended to pres- on migration in the GMS is quite extensive, ent "ordinary" people in the GMS as being although there is variation in knowledge squeezed, molded, and incentivized by among countries. For Thailand, we know a B O X 6 . 1 Development through concentration? The Lao PDR government's focal site strategy The Lao government's rural development policy growing volume of evidence demonstrating that concentrates resources and services in particular internal resettlement in Laos is having a major areas, bringing people to these development cen- and mainly negative impact on the social systems, ters, rather than vice versa.The focal site strategy in livelihoods, and cultures of many indigenous its current form was formally endorsed in February ethnic communities and people"(Baird and Shoe- 1998 and has become a central plank in the govern- maker 2007: 868; see, for example, ADB 2001; Baird ment's Rural Development Program.Focal sites are and Shoemaker 2005; Ducourtieux, Laffort, and "integrated rural development clusters par excel- Sacklokham 2005; Evrard and Goudineau 2004; lence, located in the most deprived areas where Rigg 2005;Thapa 1998; UNDP 1997;Vandergeest presently there are no or only minimum develop- 2003). Some of these problems are connected ment activities taking place"(Lao PDR 1998: 5) with with the way in which resettlement has been car- the intention of creating"development centers"or ried out, but others relate to assumptions about "growth poles"for rural areas"that will thwart or at the positive effects of integration and concentra- least slow down the present trend toward widen- tion. Besides the implications of these policies for ing gaps between rural and urban areas, but also human well-being, area-based development has within the rural areas themselves"(Lao PDR 1998: had a profound effect on economic geographies 6 [emphasis in original]).(Closely allied to the focal in the uplands by concentrating populations in site strategy are two other initiatives: the Land-For- particular sites (close to roads), barring access to est Allocation Program and Village Consolidation.) traditional shifting-cultivation fields, encouraging Under the program, upland (minority) villagers permanent-field agriculture and, in particular, wet practicing shifting cultivation are resettled in focal rice cultivation, and capturing the forested spaces sites where government services--schools, health and their value for the state and its associates. centers (souk sala), and so on--are provided as well The focal site strategy illustrates what can happen as market access through better roads. when the rationale of development collides with The focal site strategy has become highly the messy reality of local cultural and economic contentious because there"is a compelling and geographies. Spatial integration and human transformations in the Greater Mekong subregion 91 Figure 6.2 International migrant flows in the GMS CHINA Jianghong Mandalay Phongsali MYANMAR Kong HANOI Tung Mae Sai Luang Prabang Chiang Rai LAO Mae Gulf of Mae Suai P.D.R. Tonkin Hong Son Chiang Mai VIENTIANE VIETNAM Pagan Nong Khai Mae Sot Udon Pegu Nakhon Phanom Tak Thani Yangon Mukdahan Savannakhet (Rangoon) THAILAND Three Pagodas Ye Pakse Pass Surin Champassak Kanchanaburi Tavoy BANGKOK Poi Pet Siem Reap Rat Buri Battambang CAMBODIA Mergui Rayong Kampong Chan Trat Tenasserim Koh Kong PHNOM- PENH Ho Chi Minh City Gulf of Koh Song Ranong Thailand Ca-Mau Andaman Sea Phuket Hat Yai Source: Rigg (2005). reasonable amount (although the data are and probably between 2.0 million and 2.5 poor for international migrants); for Myan- million across the subregion as a whole.9 The mar and Lao PDR, we know rather less.8 main GMS migrant flows are as follows: In terms of international flows among the GMS countries, Thailand acts as the ful- · Unskilled migrants from Myanmar, crum in an emerging regional labor market. Cambodia, and Lao PDR to Thailand, of While many moves are undocumented, it is which the Myanmar migrant stream is thought that there are between 1.5 million easily the most substantial, followed by and 2.0 million GMS migrants in Thailand Lao PDR and Cambodia; 92 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA · Unskilled migrants from Myanmar to migration in general. As spatial integra- China; tion proceeds, the incidence of migration · Skilled migrants from Vietnam to Lao increases and spreads to other classes (that PDR and Cambodia; and is, the poor as well as the nonpoor) and to · Skilled migrants from China to Myanmar women, although it will remain a young(er) and Lao PDR. person's prerogative.With high levels of spa- tial integration, the opportunity will arise Studies show that the majority of interna- for a partial re-localization of life (rather tional migrants in Thailand are employed in than livelihoods), as daily mobility replaces manual, unskilled activities, usually in those longer-term migration. This phase may also 3-D jobs that Thais resolutely avoid (those see the permanent dislocation of some peo- that are dirty, dangerous, and difficult). ple from their natal villages, as they make These include agricultural work, fishing and the decision to pursue their lives elsewhere. fish processing,construction,domestic work, and the commercial sex industry. Despite Policies and politics of spatial the ease of transport in Thailand, the loca- tion of these international labor migrants transformation reflects their geographies of origin. Thus Lao The GMS program has made an imagined migrants are disproportionately employed in region, increasingly real (Kaosa-ard and the border provinces of the northeast; those Dore 2003), and it is seen by many as a role from Cambodia are employed in the eastern model of successful cross-border, regional provinces close to Cambodia; and migrants cooperation.10 One attempt to tease out from Myanmar are employed in provinces in the "success factors" distinguishes between the north and south of the country, usually those characteristics that are inherent to the abutting the border with Myanmar (World region and therefore exceptional (geogra- Bank 2006: 37). phy) and those that focus on the manner in The creation of an increasingly vital space which the GMS regional cooperation objec- of human activity across the countries of the tives have been structured (institutions) GMS has forged the context within which and achieved (strategies, sponsorship) and a whole series of development processes which are therefore repeatable (see table have been initiated, molded, augmented, or 6.8). There is little doubt that, in headline accelerated, including the delocalization of terms, the GMS has been a success. How- work, de-agrarianization, livelihood diver- ever, in this chapter we have been intent on sification, household reconfiguration, and excavating between the lines of the "text" cultural re-identification. Notwithstanding of GMS cooperation and, in particular, its the evident importance of migration and impacts on people (well-being, equity) and mobility, it is striking how far rural develop- places (environment). A recent mid-term ment studies, particularly by development review of the GMS strategic framework economists, have tended to skirt the issue. (ADB 2007b) accepts that an important dis- Dercon states, rhetorically: "Surely, study- tinction can be drawn between the impres- ing these [population] movements must be sive progress that has been made in terms at the core of understanding rural poverty of the "hardware" aspects of cooperation, and policies to reduce it" (Dercon 2006: 8). as opposed to the "software" components, Given that such movements are so much a where progress has been more problematic. part of the processes that arise from spatial This relates to a number of the issues high- integration, this must be counted a signifi- lighted in this chapter.11 cant omission. Those critics of the GMS outside the We hypothesize that when levels of spa- ADB, not surprisingly, have been more tial integration are low, migration will be strident in their views, seeing many of the restricted to a small number of the non- individual projects as typically "poorly con- poor in rural areas, mainly young(er) men, ceived," "disastrously implemented," and who move primarily for economic reasons. ultimately designed to serve the sectional The costs and risks of migration will limit interests of an elite few: "Perhaps more Spatial integration and human transformations in the Greater Mekong subregion 93 Table 6.8 Distilling the GMS "success factors" Factor Perceived benefit Geography A reasonably compact land mass, where every member shares land borders with at least three other members A location within an economically dynamic region where markets are growing and where the development of subregional trading links is seen as beneficial by all No great differences in size, population, or power (note that it is not China but regions of China that belong to the GMS) Strategy Broad long-term vision Focus, particularly initially, on small concrete, confidence-building measures Realistic number of well-defined, sometimes modest, targets Institutional structure A framework to build a shared identity and purpose but one which is not overly rigid and institutionalized (for example, the GMS opt-in, opt-out approach for infrastructure initiatives) An incremental approach that does not require unanimity (But) a well-defined program that enjoys the support of all members Sponsorship External support through an "honest broker," to facilitate cooperation and provide assistance (in this case, the ADB, with its own substantial budget) Timing Patience with mid- to long-term planning horizons Source: Extracted and adapted from DAC (2005: 10). importantly, however, the ADB has been at headed households struggling to find a way the center of driving a broader process of to get water back to their homes or with economic change that is rapidly unravel- tribal groups who depend for their sub- ing the social fabric of Mekong communi- sistence and meager incomes on their sur- ties and disconnecting human economies rounding natural environment. from their relationship to the environment" In writing this, however, we do not sub- (Cornford and Simon 2001:7). scribe to the view that these individuals are living in a state of subsistence affluence and Policy implications: humanizing should be insulated from change. Rather, the spaces of development in we highlight their multiple vulnerabilities the GMS and different capacities and the need to This chapter has sought to question the be cognizant of and sensitive to these vul- assumption that spatial integration is nerabilities and capacities. A focus on the unalloyed in terms of its effects. We have human context clearly shows that people are unpicked the view that investing in infra- not, for example, equally mobile or equally structure, improving access, and drawing able to take advantage of market integration people and places more fully into the main- (an obvious point perhaps, but often lost stream of national, regional, and interna- in the big picture). There is a rider to this tional life will deliver developmental benefits that is all too easy to ignore when studies in an unproblematic manner. In making this are snapshots in time: things change. There case, however, we do not wish to suggest that is a mobility transition, for example, where such processes are anti-developmental. The "immobile" groups can become mobile in a aggregate outcomes are invariably positive, surprisingly short space of time. As recently at least in terms of economic development as the end of the 1990s, more than two- and with attendant positive effects on the thirds (69 percent) of registered migrants depth and incidence of poverty. But it is to Thailand from Cambodia, Lao PDR, and important not to be completely seduced Myanmar were male. By 2005, however, the by the big picture. Up close, it is clear that figure had dropped to barely half (53 percent). this large-scale image consists of a mosaic There are few rules about patterns of human of sometimes contradictory processes and behavior that hold fast in the vortex of social effects that are best illuminated through and economic change in the GMS. grounded, micro studies. The grand, strate- While the debate over "openness," gic market­integrating aims of the GMS, for "reform," and "transition" in the former example, resonate little with poor female- socialist countries of the GMS tends to take 94 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA place and be orchestrated at the national ment, where inherited structures, prevailing and macroeconomic levels and can be traced power hierarchies, cultures of engagement, through statistics such as those in table 6.1, and so forth, give those "spaces" particular the effects of these policies are played out and unique qualities that have a bearing at the local, household, and individual lev- on how spatial interactions and dynamics els. This is reflected in how the "livelihood operate and evolve. There is evidence from footprints" (Rigg 2005; Bouahom, Douang- Xishuangbanna in China's Yunnan province, savanh, and Rigg 2004) of households are for instance, that minority groups have been being respatialized as integration delivers surprisingly astute and effective in taking new possibilities for making a living and advantage of the new opportunities that new ways of escaping from poverty. have arisen from market and cross-border Consider the story of Mrs. Chandaeng, integration. Janet Sturgeon speaks of them who lives close to the Mekong River around as willing and successful "neo-liberal sub- 60 kilometers west of Vientiane, the capital jects" who have taken a lead in outsourcing of Lao PDR. She was born and raised in the rubber from Xishuangbanna to Lao's Sing war-shattered province of Xieng Khouang, district in Luang Namtha province. While several hundred kilometers to the east. The in Lao, some of the minority groups may death of her husband while she was a young be vulnerable and marginalized, the story mother and a family dispute forced her to from this part of China is of minority farm- leave her home in Xieng Khouang. She even- ers taking advantage of cross-border possi- tually settled in a village on the banks of the bilities, permitting them to transcend their Mekong in 1991. At that time her prospects backward status and peripheral location were not good. She had six young children (Janet Sturgeon and Nick Menzies, personal to raise and support, no land beyond her communication, 2008). house plot, no education, few skills, and no Solutions to the "problem" of remote- husband. A decade later, in 2001, she was ness and isolation invariably have social and building a new house and was comfortably political consequences.As recent papers (for off in village terms. Her ability to survive-- example, Ali and Zhuang 2007; ADB 2007a) and, indeed, finally to prosper--as a land- have (re)emphasized, the benefits of Asia's less, widowed mother of six was linked, ulti- growth are not being distributed equally, mately, to the fact that four of her children whether over space, across population managed to secure work in neighboring groups, between the genders and genera- Thailand, remitting home around US$25­ tions, or according to ethnic group. Access US$50 a month. Her son was working as a to economic opportunities is linked to laborer on a shrimp farm in southern Thai- social structures. Market access has, simul- land, while her three daughters, Wan (19 taneously, positive and negative effects, years old), Lot (17), and Daeng (15), were which are unequally distributed. Income employed as housekeepers in Bangkok. She inequality does not map neatly onto other may have explained her children's sojourns inequalities. A focus on the modern (urban) in Thailand in terms of "when you are poor, sector--and the opportunities that are seen you have to go," but the outcome was a to reside there--is as likely to deliver new degree of economic prosperity and security. and sometimes deeper inequalities as to There is a direct link between the ability of deliver growth. Too often the "take-away" Mrs. Chandaeng to break out of her struc- point is reduced to the assertion that spatial turally ordained status as a poor, vulnerable integration delivers economic benefits; this, woman and the spatial integration policies however, should not be separated from the of the GMS and the Lao government. riders "not everywhere," "not equally," and A strong temporal dimension shapes "rarely in similar ways." the transformations brought about by (and The broader policy implications that flow through)spatialchanges;thereisalsoa"geog- from this discussion can be distilled down raphy" to space. By this we mean that the to three. First, the hardware-driven logic of dehumanized "spaces" of integration need the GMS needs to be allied to and integrated to seen as humanized "places" of engage- with the rather softer pro-poor policies and Spatial integration and human transformations in the Greater Mekong subregion 95 imperatives being pursued and encouraged two ongoing research projects, one funded by by the ADB and other agencies and institu- the Danish Council for Development Research tions. Even if we assume that income and (grant #91206) on rural-urban dynamics in four expenditure are adequate and appropriate countries of Asia and Africa (see http://www. geogr.ku.dk/projects/ecosoc/rud/) and the sec- indicators of well-being, "The behavior of ond funded by the Canadian Social Sciences and average incomes may tell us little about the Humanities Research Council on the challenges economic well-being of different subgroups of the agrarian transition in Southeast Asia (see of the population"(ADB 2007a: 8). To put it http://www.caac.umontreal.ca/en/chatsea_intro. another way, we need to be concerned about html). The authors would like to thank those what happens at the margins--social and scholars who have generously permitted us to spatial--when subregional integration is refer to their unpublished work: Jytte Agergaard pursued.12 Second, it is clear that the bene- (University of Copenhagen), Paul Chambers fits of subregional integration are tied to the (Payap University), Nick Menzies (University nature of national integration. Thailand's of California, Los Angeles), Robin Roth (York excellent physical infrastructure creates the University), Janet Sturgeon (Simon Fraser Uni- versity), Sarah Turner (McGill University), and national context through which additional Thein Swe (Payap University). regional benefits can be leveraged; the same 1. See http://www.adb.org/GMS/Program/ does not apply to Lao PDR or Myanmar. default.asp. In addition, the national policy environ- 2. See http://www.adb.org/GMS/devt-matrix. ments create the context in which regional asp#background. initiatives bite. It is not, therefore, only the 3. "The basic idea of economic corridors is regional dimension that counts, but the that, by focusing on the same geographic space, national one too. The third policy implica- investments in priority infrastructure sectors, tion is that, while the balance of effects aris- such as transport, energy, telecommunications, ing from the GMS project may be positive, and tourism, will maximize development impact the negative side effects are far from neg- while minimizing development costs. The objec- tive of the transport corridors is to develop a ligible. The "do nothing" status quo is not highly efficient system--allowing for easy circu- tenable in two respects: integration is hap- lation of goods and people around the Mekong pening, will happen, and should be encour- subregion. At the same time, they are expected to aged; at the same time, this integration will form the basis of corridors of economic growth inevitably lead to negative side effects, and and social development in the subregion, attract- these need to be identified, managed, and ing investment and skills" (ADB 2005b: 17). ameliorated. 4. See http://www.adb.org/GMS/Program/ In an important recent book on devel- default.asp. opment and governmentality in Indonesia, 5. Although note the general lack of data on Tania Li (2007) seeks to challenge those social and economic conditions in Myanmar. analysts who "separate the study of govern- 6. "From this analysis, we conclude that the development of cross-border road infrastruc- ment rationalities from the study of situ- ture in the GMS has had a positive effect on ated practices." For her, like us in this chap- the regional trade. The result that cross-border ter, "engaging with the `messy actualities' of roads have distinct effects from domestic road rule in practice is not merely an adjunct to infrastructure suggests promotion of regional the study of government--it is intrinsic to trade may require deliberate policy shifts toward it" (Li 2007: 283). To ignore the inequalities, investments in roads in border areas" (Edmonds inconsistencies, and incongruities that are and Fujimura 2006: 14). part and parcel of strategies of integration 7. Sarah Turner (2007) has also conducted and concentration is a notable oversight, not research in Lao Cai, but in her case focusing on a trifling thing. ethnic minorities, the trade in textiles, and cross- border relations. The politics of access bestows advantages on those living at the border, because Notes only border residents can cross at the minor Jonathan Rigg is a professor at Durham Uni- "open entrance crossings" with a permit (that versity. Chusak Wittayapak is assistant profes- is, without a passport) and without any taxes sor in the geography department at Chiang being levied. Others are required to use the for- Mai University. This chapter benefited from mal national-level border crossing points. Thus 96 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Hmong traders (mostly women) living close to inequality in most developing countries. Despite the border are at an advantage over other Hmong the ambiguities involved in identifying the rela- traders. Ethnic Han Chinese and Vietnamese tionship between openness and distributional (Kinh) traders are able to use the national-level changes, it seems fair to say that the evidence crossing points, and their extra-local networks has provided little support for the conventional facilitated this process. Turner shows how state wisdom that trade openness in developing coun- policies have created a variegated landscape of tries would favor the least fortunate (at least in access, which influences the livelihood options relative terms)" (Goldberg and Pavcnik 2007: open to different groups. 76­77). The paper notes that the relationship 8. For two recent summary reports, see between openness and inequality is country, World Bank (2006); Caouette and others (2007). time, and case specific and needs to be analyzed For data, see http://siteresources.worldbank. in the context of prevailing policies. org/INTTHAILAND/Resources/333200- 1089943634036/475256-1151398858396/LM_ in_GMSs_Nov06.pdf; Huguet and Punpuing References (2005: 79). ADB (Asian Development Bank). 1999. Evalua- 9. Caouette and others (2007: 19) provide a tion Studies in the Bank's Developing Member higher range of 1.8 million to 4 million intrare- Countries: Lao Poverty Reduction Evaluation. gional, cross-border migrants in the GMS. For Manila: ADB, August. (ADB Library ref: 16A/ more detailed maps of migrant flows between the POV 0040.) countries of the GMS, see http://www.rockme- ------. 2000."Poverty Reduction and Envi- kong.org/pubs/Year2005/Migration_Mekong/ ronmental Management in Remote Greater map.pdf. Mekong Subregion Watersheds, Phase II: 10. "The Greater Mekong subregion provides Draft Final Report."Vol. 1. ADB, Manila, perhaps the benchmark for successful subre- December. gional and cross-border cooperation in South- east Asia. Over the 12-year course of its existence, ------. 2001. Participatory Poverty Assessment: it has steadily evolved from a disparate collection Lao People's Democratic Republic. Manila: of wary neighbors into a highly effective collabo- ADB. ration that can now point to numerous infra- ------. 2005a. Connecting Nations, Linking structure investments directly attributable to the People. Manila: ADB. GMS initiative" (DAC 2005: 40). ------. 2005b. The Greater Mekong Subregion: 11. The mid-term review concludes,"Placing Beyond Borders; Regional Cooperation Strat- more emphasis on the `soft' aspects of subre- egy and Program Update, 2006­2008. Manila: gional cooperation will be critical to achieving ADB. (http://209.225.62.100/Documents/ the goals and objectives of the GMS. ... Com- Reports/Beyond-Borders/.) plementary measures are needed to translate advances in physical connectivity into acceler- ------. 2005c. Regional Study on Rural, Urban, ated improvements in livelihoods and poverty and Sub-Regional Linkages in the Greater reduction. ... [In addition] improved physical Mekong Sub-Region (Viet Nam, Lao PDR, connectivity and mobility of people and goods Cambodia): A Holistic Approach to Develop- can have undesirable consequences, such as the ment and Poverty Reduction. RETA 6121, final transmission of communicable diseases, illegal report. Manila: ADB, December. migration of workers, and environmental deg- ------. 2007a. Key Indicators 2007. Vol. 38, with radation, which need to be contained and miti- a special section on inequality in Asia. Manila: gated. This proposed shift in emphasis does not ADB. (http://www.adb.org/documents/books/ mean less concern for developing subregional key_indicators/2007/default.asp.) infrastructure, as unmet needs for infrastruc- ------. 2007b. Mid-term Review of the Greater ture investments in the GMS are huge. What Mekong Subregion Strategic Framework 2002­ it calls for is a more balanced approach which 2012. Manila: ADB. (http://www.adb.org/ ensures that benefits from subregional eco- documents/reports/mid-term-review-gms/ nomic cooperation and integration are maxi- mid-term-review-gms.pdf.) mized and far outweigh the costs involved" (ADB 2007b: 34). Ali, Ifzal, and Ernesto M. Pernia. 2003."Infra- 12. There is good reason to think that eco- structure and Poverty Reduction: What Is nomic reforms and trade openness lead to greater the Connection?" Economics and Research inequality. The available evidence "suggests a Department Policy Brief 13, ADB, Manila, contemporaneous increase in globalization and January. Spatial integration and human transformations in the Greater Mekong subregion 97 Ali, Ifzal, and Juzhong Zhuang. 2007."Inclusive Ducourtieux, Olivier, Jean-Richard Laffort, and Growth toward a Prosperous Asia: Policy Silinthone Sacklokham. 2005."Land Policy Implications." Economics and Research and Farming Practices in Laos." Development Department Working Paper 97, ADB, Manila. and Change 36 (3): 499­526. Baird, Ian, and Bruce Shoemaker. 2005. Aiding Edmonds, Christopher, and Manabu Fujimura. or Abetting? Internal Resettlement and Interna- 2006. Impact of Cross-Border Road Infra- tional Aid Agencies in the Lao PDR. Ontario: structure on Trade and Investment in the Probe International. Greater Mekong Subregion. Discussion Paper 48. Manila: ADB Institute, October. (http:// ------. 2007."Unsettling Experiences: Internal www.iadb.org/intal/aplicaciones/uploads/ Resettlement and International Aid Agencies ponencias/i_Foro_LAEBA_2006_11_1_ in Laos." Development and Change 38 (5): EdmondsFujimura.pdf.) 865­88. Evrard, Olivier, and Yves Goudineau. 2004. Bouahom, Bounthong, Linkham Douangsa- "Planned Resettlement, Unexpected Migra- vanh, and Jonathan Rigg. 2004."Building tions, and Cultural Trauma in Laos." Develop- Sustainable Livelihoods in the Lao PDR: ment and Change 35 (5): 937­62. Untangling Farm and Non-farm." Geoforum 35 (5): 607­19. Gainsborough, Martin. 2007."Globalisation and the State Revisited: A View from Provincial Bush, Simon R. 2004."Scales and Sales: Changing Vietnam." Journal of Contemporary Asia 37 Social and Spatial Fish Trading Networks in (1): 1­18. the Siiphandone Fishery, Lao PDR." Singapore Goldberg, Pinelope, and Nina Pavcnik. 2007. Journal of Tropical Geography 25 (1): 32­50. "Distributional Effects of Globalization in Caouette, Therese, Rosalina Sciortino, Philip Developing Countries." Journal of Economic Guest, and Alan Feinstein. 2007. Labor Migra- Literature 45 (1): 39­82. tion in the Greater Mekong Sub-Region. Report Hettige, Hemamala. 2006."When Do Rural submitted to the Rockefeller Foundation. Roads Benefit the Poor and How? An In- (http://apmrn.anu.edu.au/regional_members/ Depth Analysis Based on Case Studies." ADB, LaborMigration%20in%20GMS.pdf.) Operations Evaluation Department, Manila. Cornford, Jonathan, and Michael Simon. 2001. (http://www.adb.org/Documents/Books/rur- Breaking the Banks: The Impacts of the Asian alroad_benefits/rural-roads.pdf.) Development Bank and Australia's Role in the Hoang, Cam. 2007."On Being `Forest Thieves': Mekong Region. Victoria, Australia: Oxfam. State Resource Policies, Market Forces, and Culas, Christian. 2001. "Anthropologist Struggles over Livelihood and Meaning of Report." District Upland Development and Nature in a Northwestern Valley of Vietnam." Conservation Project, Khammouane, Lao Working Paper Series 6, Chiang Mai Univer- PDR, February. sity, Regional Center for Social Science and DAC (Development Assistance Committee). Sustainable Development, Chiang Mai. 2005."Regional and Cross-Border Infra- Huguet, Jerrold W., and Sureeporn Punpu- structure and Its Role in Trade, Pro-Poor ing. 2005. International Migration in Thai- Economic Growth, and Poverty Reduction: land. Bangkok: International Organization Background Paper." DAC Poverty Reduc- for Migration. (http://www.iom-seasia.org/ tion Network (povnet),"Third Workshop resource/pdf/SituationReport.PDF.) on Developing the DAC Guiding Principles," ILO (International Labour Organisation). 1997. Organisation for Economic Co-operation and "Socio-Economic Survey on Short-Term Development, Tokyo. (http://www.oecd.org/ Impact of Rural Roads Construction." (Con- dataoecd/34/22/36568157.pdf.) sultant: Ulf Johanson, A.G.) Employment- Denes, Alexandra. 1998."Exploring the Links Intensive Rural Roads Construction and between Foraging and Household Food Maintenance Project, Vientiane, April­June. Security: A Gender-Based Study of Foraging Johnston, Doug. 2007."These Roads Were Made Activities in Salavan Province." Australian for Walking? The Nature and Use of Rural Embassy, Vientiane, April. Public Transport Services in Garut Regency, Dercon, Stefan. 2006."Rural Poverty: Old Chal- West Java, Indonesia." Singapore Journal of lenges in New Contexts." GPRG-WPS-072, Tropical Geography 28 (2): 171­87. Global Poverty Research Group, University of Kaosa-ard, Mingsarn, and John Dore. 2003. Oxford. (www.gprg.org.) Social Challenges for the Mekong Region. 98 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Chiang Mai: Chiang Mai University, Social Rigg, Jonathan. 2005. Living with Transition in Research Institute. Laos: Market Integration in Southeast Asia. London: Routledge. Lao PDR. 1998."The Rural Development Pro- gramme 1998­2002: The `Focal Site' Strategy." Schipani, Steven. 2007."Ecotourism as an Alter- Report prepared for the "Sixth Roundtable native to Upland Rubber Cultivation in the Follow-up Meeting,"Vientiane, May 13. Nam Ha National Protected Area, Luang Namtha." Jut Pakai 8 (April): 5­17. ------. 1999."The Government's Strategic Vision for the Agricultural Sector." Ministry of Sturgeon, Janet C. 2005. Border Landscapes: Agriculture and Forestry,Vientiane, December. The Politics of Akha Land Use in China and Thailand. Chiang Mai: Silkworm Books. ------. 2000a."Fighting Poverty through Human Resource Development, Rural Swe, Thein, and Paul Chambers. 2008."Political Development, and People's Participation." Economy on the Perimeter: State Policy and Unpublished mss. Government report to the Trade on Thailand's Border with Myanmar "Seventh Roundtable Meeting,"Vientiane, and Lao PDR." Paper presented at the "Tenth November 21­23. International Thai Studies Conference," Thammasat University, Bangkok, January ------. 2000b."Oudomxay Province: Envi- 9­11. ronmental Inventory." Report prepared by the Ministry of Communication, Post, and Thapa, Gopal B. 1998."Issues in the Conserva- Construction and the International Union tion and Management of Forests in Laos: The for Conservation of Nature with assistance Case of Sangthong District." Singapore Jour- from the Swedish International Development nal of Tropical Geography 19 (1): 71­91. Cooperation Agency, March. Trankell, Ing-Britt. 1993. On the Road in Laos: ------. 2000c. Road Infrastructure for Rural An Anthropological Study of Road Con- Development, Final Report. Vientiane: Minis- struction and Rural Communities. Uppsala try of Communication, Transport, Post, and Research Reports in Cultural Anthropology Construction, April. 12. Uppsala: Uppsala University. Turner, Sarah. 2007."Trading Old Textiles: The Leinbach, T. R. 2000."Mobility in Development Selective Diversification of Highland Liveli- Context: Changing Perspectives, New Inter- hoods in Northern Vietnam." Human Organi- pretations, and the Real Issues." Journal of zation 66 (4): 389­404. Transport Geography 8 (1): 1­9. UNDP (United Nations Development Pro- Li, Tania. 2007. The Will to Improve: Governmen- gramme). 1996."Accessibility, Rural Roads, tality, Development, and the Practice of Politics. and Sustainable Rural Development." Back- Durham: Duke University Press. ground paper for the "Road Sector Donor NUOL (National University of Laos). 1999. Coordination Meeting,"Vientiane, February "Assessment of Road Development Impacts 6­7. on Land Use in the Nam Mat Watershed, Lao ------. 1997. Basic Needs for Resettled PDR." Final report, National University of Communities in the Lao PDR. Vol. 1: Main Laos, Vientiane. Report. Vol. 2: Provincial Surveys. Vientiane: Oehlers, Alfred. 2006."A Critique of ADB UNDP, June. Policies toward the Greater Mekong ------. 2002. National Human Development Sub-Region." Journal of Contemporary Asia 36 Report Lao PDR 2001: Advancing Rural (4): 464­78. Development. Vientiane: UNDP. Porter, Gina. 1995."The Impact of Road Con- UNEP (United Nations Environment Pro- struction on Women's Trade in Rural Nige- gramme). 2001. State of the Environment: Lao ria." Journal of Transport Geography 3 (1): PDR 2001. Bangkok: UNEP with the Norwe- 3­14. gian Agency for Development Cooperation. ------. 2002."Living in a Walking World: Rural van de Walle, Dominique. 2002."Choosing Mobility and Social Equity Issues in Sub- Rural Road Investments to Help Reduce Pov- Saharan Africa." World Development 30 (2): erty." World Development 30 (4): 575­89. 285­300. Vandergeest, Peter. 2003."Land to Some Til- Ravallion, Martin. 2001."Growth, Inequal- lers: Development-Induced Displacement in ity, and Poverty: Looking beyond Averages." Laos." International Social Science Journal 175 World Development 29 (11): 1803­15. (March): 47­56. Spatial integration and human transformations in the Greater Mekong subregion 99 Warr, Peter, and Jayant Menon. 2006."Does Project."Aide-mémoire, World Road Improvement Reduce Poverty? Bank identification mission, A General Equilibrium Analysis for Lao September­October. PDR." ADB Institute and Australian ------. 2006."Labor Migration in the Greater National University, October. (http:// Mekong Sub-Region, Synthesis Report: www.iadb.org/intal/aplicaciones/uploads/ Phase I." World Bank, Washington, DC, ponencias/i_Foro_LAEBA_2006_11_1_ November. (http://siteresources.worldbank. MenonWarr.pdf.) org/INTTHAILAND/Resources/333200- World Bank. 1999."Lao People's Democratic 1089943634036/475256-1151398858396/ Republic: Proposed Agricultural Development LM_in_GMSs_Nov06.pdf.) SECTION II Rural development issues in Vietnam: spatial disparities Southeast Asia: Vietnam, the and some recommendations Philippines, Indonesia, and Thailand Dang Kim Son 7 Before 1945, Vietnam was an agricultural 1993 to 19.5 percent in 2004 and to 16.0 per- economy with a large gap in income and liv- cent in 2006. ing standards between the urban and rural Despite these significant successes, con- population. Even in rural areas, the farmers cerns have arisen regarding the socioeco- who made up 90 percent of the population nomic disparities between the rich and the only managed 25 percent of the farming poor, between the urban and the rural areas, area. In 1945 Vietnam gained independence, between the plains and the highlands, and and the government instituted land reform, among ethnic groups. These disparities are c h a p t e r redistributing 800,000 hectares of land to having many negative impacts on the eco- more than 2 million farmer households. The nomy, society, and environment and on the state reduced land rent by 25 percent and efficiency and sustainability of the develop- cut most taxes for farmers, which helped to ment process. As Vietnam becomes increas- improve social equality in rural areas, pro- ingly globalized, industrialized, and market mote agricultural production, and narrow oriented, these socioeconomic disparities substantially the gap in income and living may threaten the sustainability of the coun- standards between rural and urban areas. try's renovation and development. In the years between the 1960s and the This paper is based on published data of early 1980s, Vietnam's agricultural produc- official studies and surveys of the General tion declined considerably due to war and Statistical Office (GSO), some ministries a central planning economy. Since the mid- and sectors, and researchers inside and out- 1980s, Vietnam has embarked on an era side the country. It seeks to describe the cur- of market-oriented reform. The land and rent disparities in Vietnam's agriculture and production materials of cooperatives were rural areas and to offer recommendations distributed to farmer households. The pri- for improving the situation. vate sector has been encouraged, and trade Current spatial disparities and liberalization has been pursued. As a result, agricultural production, in particular, and policy issues economic growth, in general, improved. Since Vietnam shifted to a market economy In the past 30 years, Vietnam has experi- in the mid-1980s, industrialization and enced a period of robust development and urbanization have accelerated rapidly. From renovation. For the past 20 years, the annual 1992 to 1997 (before the Asian financial cri- growth rate of the Vietnamese economy sis), average gross domestic product (GDP) averaged 7 percent. The human develop- growth was 9 percent a year.By 2007 the share ment index (HDI) increased more than 6 of agriculture in GDP declined to 20 percent, percent, from under 0.69 in 1999 to 0.73 while that of industry and services was 42 in 2004. According to the World Bank, the and 38 percent, respectively. Foreign invest- poverty rate dropped from 58.2 percent in ment increased dramatically. In the period of Rural development issues in Vietnam: spatial disparities and some recommendations 101 2001­05, the revenue of foreign enterprises purposes and managed by enterprises or rose 30 percent a year,and the revenue of pri- localities, rather than individuals, the price vate enterprises rose 34 percent a year. Based of land rose dramatically. This price differ- on the five living standard surveys conducted ence created social inequality and gave rise by the GSO in the 13 years between 1993 and to disputes in many regions. 2006, the poverty rate declined 42 percent, A large number of rural workers have moving 35 million people out of poverty.1 migrated to urban areas, and migration has Vietnam's agricultural structure changed helped to narrow the income and employ- substantially during this time period. ment gap between rural and urban areas. In According to the GSO, from 2001 to 2006, 1995 average income per capita in current the area devoted to rice paddy declined by prices was 2.6 times higher in urban than in nearly 341,900 hectares, while production rural areas. In 2006 the difference was nearly output increased from 32.5 million tons in 2.1 times (GSO 2006, 2007b). 2000 to 35.8 million tons in 2006, due to In 2007 Vietnam became a member of the use of new rice varieties and the appli- the World Trade Organization (WTO). cation of modern cultivation techniques. Before that, Vietnam was a member of the Productivity increased from 42.4 quintals Association of South East Asian Nations per hectare in 2000 to 48.9 quintals per (ASEAN), Asia-Pacific Economic Coopera- hectare in 2006. The production output of tion (APEC), and other bilateral and multi- export crops, such as coffee, rubber, tea, and lateral agreements. The process of interna- pepper, also rose. The production value of tional integration has helped to narrow the livestock increased 7.1 percent annually. The economic, cultural, and social gaps between export of agroforestry and fishery products Vietnam and other countries. increased rapidly, to US$9.6 billion in 2006, All of these processes have concentrated 2.3 times higher than in 2000. Forest cov- investment, infrastructure, labor, land, erage expanded from 34.2 percent in 2000 and capital in urban areas. They also have to 37.5 percent in 2004, an annual increase improved spatial linkages, expanding infra- of 0.7 percent (MARD 2005). From 2001 to structure and services and improving mar- 2006, aquacultural yield tripled to 1.7 mil- ket accessibility for many regions. In general, lion tons, and fishery catch increased 20 the socioeconomy of the country has been percent to more than 2 million tons, for a developing, but Vietnam's urban, industrial combined catch of 3.7 million tons. areas have been developing much faster than The agroforestry and fishery product­ its rural, agricultural areas (see table 7.1). In processing industry grew 14 percent annu- other words, spatial disparity has become a ally, focusing on vegetables, fruits, and big concern. timber. Rural crafts increased 15 percent Disparities between rural and annually. Currently, the country has around 2,020 craft villages with more than 1.5 mil- urban areas lion households, creating more than 10 By the end of 2004, there were about 14.2 million jobs and playing an important million rural households, of which 10 million role in addressing unemployment in rural areas (MARD 2007). Since 1996, average Table 7.1 GDP growth rate and GDP per capita in Vietnam, by ecological zone income per capita has tripled in rural areas. Zone GDP per capita (US$) GDP growth ratea The living conditions of rural people have improved in terms of the supply of clean Red River delta 753 8.68 Northeast 417 8.08 water and electricity, means of communica- Northwest 300 7.98 tion, development of small enterprises and North central coast 388 7.62 rural crafts, and incidence of poverty. South central coast 585 7.59 Urban and industrial areas grew rapidly Central highlands 420 9.38 Southeast 1,967 8.46 as agricultural land was claimed from rural Mekong River delta 826 8.71 areas. Farmers received low prices for their Source: Provincial Statistics Office (2008). land, but, once the land was used for other a. Adjusted based on the official national growth rate. 102 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA worked in agriculture, accounting for 74.3 capita in rural areas rose from 2.1 million percent of total employment (see figure 7.1). dong in 1995 to 6.1 million dong in 2006. 2 The household production and business Despite the remarkable improvement in structure has become increasingly diversi- living standards of the Vietnamese people fied. The share of income from livestock and generally, and rural people particularly, the nonagricultural activities in total household expenditure and income gap between rural income is increasing. Average income per and urban areas is still wide, as shown in the Vietnam household living standards surveys (see figure 7.2). The difference in monthly Figure 7.1 Employment in Vietnam, by sector, 1988­2005 income per capita between urban and rural Employed labor force as of July 1 areas continues to grow, from 1.8 times in 45.0 1993 to 2.1 times in 2006. 40.0 Besides, there is a difference in accessibil- 35.0 ity to information, markets, employment, 30.0 and education, which is difficult to measure. For example, according to the GSO's Enter- 25.0 prise Survey 2006 (GSO 2008a), 21 percent millions20.0 of enterprises in rural areas said that the 15.0 quality of rural roads is poor compared with 10.0 13 percent of enterprises in urban areas; 25­30 percent of rural enterprises consid- 5.0 ered traffic a hindrance to development, 0.0 compared with only 14­20 percent of urban 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 enterprises; and 8 percent of rural enter- prises complained about the condition of Others Manufacturing and mining Agriculture telephone service, compared with 0 percent Source: GSO (2007a). of urban enterprises. Twice as many rural as urban enterprises thought that the supply of electricity is unreliable and a hindrance to Figure 7.2 Urban and rural expenditure in Vietnam, 2002­06 their business, while 9 percent of rural enter- prises complained about the poor quality 900.0 of education and training, compared with 811.8 only 1 percent of urban enterprises. Finally, 800.0 31 percent of rural enterprises complained about the quality of vocational education, 700.0 twice the percentage of urban enterprises 652.0 voicing this complaint. 600.0 Poor living standards 497.5 dong 500.0 Rural households have low income and thus 401.7 spend much of it on basic living expenses, 400.0 such as food and drink (see table 7.2). thousand 314.3 Meanwhile, urban households have higher 300.0 income and thus more disposable income. 232.1 The disparity in living standards is evi- 200.0 dent in the nutritional value of food con- sumed (see table 7.3). The main source of 100.0 energy for rural people is starch (rice), while that for urban people is protein (livestock 0.0 urban expenditure rural expenditure and fishery products). Urban dwellers eat twice the meat per capita as rural dwellers. 2002 2004 2006 Rural people mainly consume pork, while Source: GSO (2007b). urban people consume a diverse menu of Rural development issues in Vietnam: spatial disparities and some recommendations 103 Table 7.2 Average living expenditure in urban and Table 7.4 Income and health indicators in the plains and highlands of Vietnam, 2004 rural areas of Vietnam, 2006 Income per capita Malnutrition rate dong thousand per person per month of the lowest GDP per capita Infant mortality of children under Item Urban Rural Region group (US$ PPP) (US$ PPP) (percent) 5 (percent) Food, drink, and cigarettes 291.0 160.6 Highlands Food 43.2 46.5 Northeast 455 308 31.4 29.8 Foodstuff 151.1 81.8 Northwest 329 225 33.9 32.0 Fuel 13.6 8.8 Central highlands 406 275 34.1 35.8 Meal outside home 65.9 15.2 Plains Drink and smokes 17.2 8.4 Red River delta 616 530 14.8 22.8 Other 304.5 122.9 Southeast 806 1,385 14.4 19.9 Clothes, hats, shoes 24.6 13.7 Mekong River delta 563 451 19.8 25.1 House, electricity, water, 33.9 8.7 Source: VASS-UNDP (2006). sanitation Household equipment and 53.9 25.8 devices Health care 38.0 21.2 of rural households was 6.7 million dong, Travel and post 74.5 27.3 an increase of 3.5 million dong compared Education 42.9 16.2 Culture, sports, and 12.9 1.8 with October 1, 2001. Despite the improve- entertainment ment, such levels of savings are too low for Other effects and services 23.8 8.1 productive investment. Total 595.4 283.5 While saving is low, most farmers, farm Source: GSO (2007b). owners, household heads, and private enter- prises usually mobilize their own captial for Table 7.3 Energy in the Vietnamese diet in rural and investment; very few turn to banks for credit. urban areas, 2000 Therefore,their ability to buy land,hire labor, percent purchase equipment, renovate technology, Region Protein Fat Sugar change production structure, and industri- Rural 14.6 15.64 69.8 alize agriculture and rural areas is limited. Urban 12.7 10. 9 76.4 Most farmers use backward technology, and Source: Ministry of Health, Institute of Nutrition (2004). few households have fixed assets. About 30 percent have breeding facilities as their main livestock products (for example, poultry, asset. And 14­18 percent have other small cattle, eggs, and milk). fixed assets such as gardens, cattle, sows, Living standards and the incidence of breeding boars, and pesticide pots. Few have hunger and poverty have improved sig- high-value fixed assets such as warehouses, nificantly in recent years. However, the gap cars,and trailers.When income rises,farmers between the plains and the highlands is tend to purchase durable equipment. In 2005 still wide (see table 7.4). Infant mortality is each rural household invested an average of between 31 and 34 percent in the highlands, 1.2 million dong in durable equipment, up compared with between 14 and 20 percent slightly from 1 million dong a year in the in the plains. The malnutrition rate follows period of 1998­2000 (GSO 2002, 2007a). the same pattern. In general, backward technology and small-scale production lead to lower labor Low rate of savings and investment productivity in agriculture than in society According to the Ministry of Planning and as a whole, where labor productivity has Investment, in the period between 1996 increased rapidly thanks largely to interna- and 1998, rural dwellers made up 80 per- tional investment, technology, and modern cent of the total population but only 29­33 management (see figure 7.3). percent of social consumption expendi- ture (Saigon Economics Times, June 10, Regional disparities 1999). Rural people were able to save, on To narrow these types of disparities,Vietnam average, only 172,000 dong a year in 1994 has adopted many policies. For example, the and 700,000 dong a year between 2001 and government has constructed clinics, schools, 2006. On July 1, 2006, the average savings markets, and cultural centers and upgraded 104 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 7.3 Labor productivity in Vietnam, 1995­2006 2000, the agricultural sector contributed almost 37 percent of the newly created jobs. 12 The indexes for the industrial sector and the prices service sector were 19.8 and 43.5 percent, 10 respectively. The situation was totally differ- 1994 ent in 2000­06, when 5.7 million jobs were year, 8 created. The industrial sector employed 57 per percent and the service sector employed 49 6 percent of workers in these new jobs, cap- person 4 turing the jobs lost in the agricultural sector. per The number of laborers in the agricultural 2 sector declined 1.5 percent during this million period (2006 compared with 2000). dong 0 It is becoming more and more difficult 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 to create jobs in rural areas. The annual Overall productivity (million dong/labor) Agriculture productivity (million dong/labor) growth of new rural jobs is 1.7 percent, which is lower than that of the country as a Source: GSO (2007a). whole. The proportion of employed workers in rural areas has decreased gradually. From Table 7.5 Living standards in the highlands and plains of Vietnam, 2004 1996­2005, the percentage of unemployed School who were in rural areas was more than 80 attendance rate percent a year. In 2005, 3.53 million people of all school- Number of Local budget for Adult literacy age children doctors per health care (dong were unemployed, and nearly 87 percent of Region rate (percent) (percent) 10,000 people per person) the unemployed were in rural areas. In addi- Highlands tion, the unemployment rate was higher in Northeast 90.4 73.6 61 65,080 rural than in urban areas (10.2 and 9.8 per- Northwest 76.0 65.3 39 62,582 cent, respectively, in 1996, 9.1 and 6.6 per- Central highlands 88.2 73.9 46 75,891 cent, respectively, in 2000, and 9.3 and 4.5 Plains Red River delta 96.2 76.0 51 48,089 percent, respectively, in 2005). Southeast 93.6 71.1 56 97,559 Compared with the situation in other Mekong River delta 89.8 61.5 41 45,320 Asian economies and the world, in Vietnam Source: VASS-UNDP (2006). labor is highly concentrated in rural areas. This partly reflects the current trend toward roads, electricity, communication, and capital-intensive rather than labor-intensive irrigation systems in highland provinces. use of technology in industry. Consequently, the index of health care and Labor productivity education is similar in the plains and the highlands (see table 7.5). However, the eco- Labor productivity is much lower in rural nomic disparities among groups of people than in urban areas. Rural workers use and areas have had many negative impacts backward tools and technology, resulting in on the economy, society, and environment. low productivity and income. Urban work- ers in the industrial and service sectors use Uneven job opportunities advanced technology and apply modern The rural population was 60.7 million in management techniques, resulting in high 2005, accounting for 73 percent of the total productivity and income. population of the country. The rural labor force was approximately 33.3 million, mak- Food security ing up more than 75 percent of the coun- For the past 15 years, Vietnam has enjoyed try's labor force.3 The number of employed stable growth of agricultural production, workers in rural areas increased from 28.6 which has ensured food security for the million (80 percent of total employment) in majority of the population. Cereal pro- 1996 to 32.9 million in 2005 (76 percent of duction increased from 301 kilograms per total employment). In the period of 1990­ capita in 1990 to 479 kilograms per capita Rural development issues in Vietnam: spatial disparities and some recommendations 105 in 2004 (see figure 7.4). Vietnam exports a Figure 7.4 Cereal production per capita in Vietnam, by region, 2006 large quantity of rice and retains its posi- tion as the second-largest rice exporter in 1,200 the world. The consumption of many agri- 1,000 cultural products such as vegetables, fruits, person 800 meat, seafood, and forestry products as well per 600 as the volume of agricultural exports have 400 expanded constantly. Production of food per capita is as follows: vegetables, 67 kilograms, kilograms 200 fruit, 29 kilograms, seafood, 28 kilograms, 0 Whole Red river North North North South Central South Mekong and meat, 25 kilograms. country delta east west central central highlands east river delta At the local level, only the Red River coast coast delta and the Mekong delta grow more than 2004 2005 2006 enough rice to satisfy local demand. Other Source: GSO (2008b). areas, especially the highlands and some coastal areas, are not self-sufficient in food production. They produce other products Figure 7.5 Distribution of malnourished children under five years of age in Vietnam and buy rice for consumption. When natural disasters, epidemics, and unemployment depress income in an area that does not produce sufficient rice to meet demand, residents cannot afford to buy food, leading to food insecurity at the local level despite a surplus of food in the coun- try as a whole. To obtain food security at the local and household levels requires outside sources of income and access to markets. It is also important for the state to maintain a reserve of food and to offer other forms of support during emergency situations. Nutritional structure Although the country does not face starva- tion, disparities still exist in the quality and 33% and higher 30­33% makeup of food as a result of income dif- 26­30% ferences among groups of people. Food still 26% and lower absorbs most of the poor's income, prevent- ing them from investing in production and improving their standard of living. As Vietnam's food security has improved, so has its nutritional status. There were approximately 10.9 million malnourished Vietnamese in 2004 compared with 15.9 mil- lion in 1994.However,nutritional status mir- rors income,with the percentage of malnour- ished people much higher in rural than in urban areas.The rate of underweight children less than five years of age in the highlands is Source: Ministry of Health, Institute of Nutrition (2004). twice that in urban areas. Also, 42.3 percent of children in remote areas were malnour- west, and central highlands have the most ished in 2004, compared with 23.4 percent in malnourished children in the country. They urban areas. As shown in figure 7.5, moun- also have the highest rate of children whose tainous areas such as the northeast, north- diets lack protein, fat, vitamin A, iodine, and 106 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA other essential ingredients. At the same time, tion) often have lower economic development children in cities are at risk of obesity due to indexes than those with a high share of indus- excessive amounts of food. Such differences try in GDP (see table 7.7).Industrialized prov- lead to significantly different levels of physical inces contribute more to the national budget and intellectual development between areas. and thus have more political power.This is the reason why local authorities tend to give pri- Migration ority to accelerating industrial development The economic disparities between areas push and reducing the contribution of agricultural labor out of underdeveloped areas and pull production to the local economy. In this race, labor into developed ones.Vietnam has expe- provinces without advantages for industrial rienced waves of migration both of jobs and development will certainly lag behind in eco- of workers from rural to industrial, urban nomic growth. areas. In Vietnam, the southeast is the main source of jobs, in addition to Hanoi and Ho Poverty rate Chi Minh City. The Red River delta and the Vietnam has achieved significant progress Mekong River delta are the main source of in poverty alleviation; however, there is still workers. As shown on table 7.6, most of the a considerable difference in poverty den- Vietnamese labor force is unskilled. The flow sity among regions. Regions with a high of immigration is to the southeast region proportion of poor households often lack (including Ho Chi Minh City). The number natural and socioeconomic resources, have of people moving within the Red River delta few advantages for economic development, is huge, but not shown clearly in the table. and lack accessibility to markets. They are The reason is that most of the immigrants to often the mountainous, island, remote, or Hanoi are included in the Red River delta. ethnic minority areas, which are far from According to the statistics of the Viet- urban centers and are underdeveloped in namese Household Living Standard Survey terms of industry and services. Although in 1999 (GSO 2000), most of the immigrants from 2000 to present, the absolute density of to Hanoi were from the central north and poverty has changed, the relative density has the Red River delta, and some were from the remained nearly stable. The next population northern mountainous area, while most of census will be conducted in 2010. the immigrants to Ho Chi Minh City came As shown in figure 7.7, regions with from the Mekong River delta, the southern a high poverty rate are in the northwest, area, the central highlands, the central area, northeast (mountainous areas in the north), and the Red River delta (see figure 7.6). and along the range of mountains in the west (west of the north central coast and central Economic growth rate highlands). The region with the lowest pov- In general,provinces with a high share of agri- erty rate is the southeast, which is the most culture in their economic structure (which industrial region in the country. The delta means those with a low level of industrializa- and coastal areas have a low poverty rate. Table 7.6 Labor and migration situation in Vietnam, 2004 Percent of the Percent of people national labor force Percent of workers living in urban areas Net immigration Region (percent) who are unskilled (percent) (number of persons) Red River delta 22.6 68.5 24.5 -4,656 Southeast 14.9 67.8 53.7 94,424 Mekong River delta 21.3 85.7 20.2 -38,632 North central coast 12.4 84.1 6.0 114 South central coast 8.2 74.9 29.0 -8,017 Northeast 11.8 81.8 18.7 -11,717 Northwest 3.2 88.8 13.8 -379 Central highlands 5.6 84.3 28 -1,676 Vietnam 100 77.3 26.5 0 Source: GSO (2006a). Rural development issues in Vietnam: spatial disparities and some recommendations 107 Figure 7.6 Source of migrants to Hanoi and Ho Chi Minh City, 1999 MIGRATION TO HANOI, MIGRATION TO HO CHI MINH CITY, 1994­99 1994­99 Hanoi Total 196,643 Number of Migrants Over 5 Years 20,000­30,000 10,000­20,000 5,000­10,000 1,000­5,000 250­1,000 fewer than 250 Ho Chi Minh City Total 433,662 Source: Based on migration survey data, GSO (2006a). Table 7.7 Economic growth rate in Vietnam, by province, 2006 Contribution to national Proportion of agricultural Provinces GDP per capita (US$) GDP (percent) GDP in total GDP (percent) Provinces with low share of agriculture Quang Ninh 908 1.41 7.46 Da Nang 1,099 1.24 4.17 Ba Ria, Vung Tau 10,572 13.97 1.56 Binh Duong 1,189 1.64 7.00 Provinces with high share of agriculture Thai Binh 422 1.12 39.91 Ha Tinh 307 0.57 43.47 Ben Tre 517 1.00 54.67 Vinh Long 564 0.85 53.00 Source: Provincial Statistics Offices (2008). However, regions with the highest popu- central coast, and the Mekong delta. Only lation density (which are the most devel- the southeast has a high level of economic oped regions) also have the highest poverty development but a small absolute number density. They are the Red River delta, the of poor households. 108 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 7.7 Poverty rate and density in Vietnam Rural Poverty Rate (PO) Poverty Density 0­20 20­40 1 dot = 1000 persons 40­60 below poverty line 60­80 80­100 Sources: Swiss National Centre of Competence in Research (NCCR) North-South Switzerland; Department of Population and Labour Statistics,General Statistics Office (GSO); Informatics Center for Agriculture and Rural Development (ICARD), Ministry of Agriculture and Rural Development (MARD). GNI and other equality indexes considerably among regions in the country While Vietnam has been developing eco- (see table 7.8). nomically, social and economic inequal- According to statistics of the population ity has been lower than in other countries and housing census in 1999, poor areas in in the region. From 1993­2004, Vietnam's the mountains of the north, the north cen- GNI (gross national income) was almost tral coast, the south central coast, and the unchanged, while the economy grew rap- central highlands have a high GNI com- idly. From 2004­06, GNI went down a little. pared to the national average. In the future, Vietnam's GNI has remained stable thanks Vietnam needs to address not only poverty largely to improvements experienced by the but also inequality (see figure 7.8). three middle quintiles of income. In 2006 these three groups accounted for 50 per- Investment cent of total expenditures. However, the gap Due to differences in natural and socioeco- between the richest and the poorest group nomic conditions and in investment incen- widened during this period, and GNI varied tives, foreign investment varies by region, as Rural development issues in Vietnam: spatial disparities and some recommendations 109 Table 7.8 Poverty in Vietnam, by region, 2004 Percent of Gap between 20 Income per capita of population below the Human poverty percent richest and 20 percent poorest Region food poverty line index (HPI) 20 percent poorest (US$ PPP) Red River delta 4.59 8.4 7.0 164 Southeast 1.82 9.6 8.7 233 Mekong River delta 5.22 20.0 6.7 159 North central coast 12.24 15.8 6.0 114 South central coast 7.58 13.8 6.5 141 Northeast 9.35 18.6 7.0 124 Northwest 21.81 32.7 6.4 95 Central highlands 12.30 20.5 7.6 119 Vietnam 6.92 15.1 8.3 142 Source: GSO (2006c). does the creation of jobs and income (see Figure 7.8 Inequality in Vietnam table 7.9). Capital, technology, management skills, and job creation will, in turn, alter the local infrastructure, service sector, and urbanization of the localities, further widen- ing the gap among localities. Causes of disparities Spatial disparities are caused by many fac- tors. First, differences in natural condi- tions such as topography, land, climate, and resources create differences in the ability of each region to adapt to different modes of production and to make use of comparative advantages. In addition, the socioeconomic characteristics of each region play a role, including population density, language, Gini Coefficient culture, quality of blue- and white-collar 0.2­0.25 skills, and business culture (such as the 0.25­0.3 ability to cooperate, entrepreneurial mind- 0.3­0.35 set, interest in education, and understand- ing of the need to save and invest). Geographic factors also play an impor- tant role in the economic development of each region. Coastal regions have advantages of fishery and accessibility to international traffic. Similarly, border regions, regions that are near an international airport or a big city, especially those near an interna- tional route (land, sea, air), or regions that have cultural and natural attractions can use Source: Based GSO (2000). their geographic advantages. Unequal investment in sive but highly effective. Therefore, these infrastructure regions are given priority in the develop- Investment in infrastructure in favorable ment of infrastructure. Once again, the regions (plains, urban areas) and eco- unequal investment in infrastructure wid- nomically advantaged regions (rich natural ens the economic gap between localities. resources, fertile land) is relatively inexpen- Table 7.10 shows a clear relation between 110 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 7.9 Investment in Vietnam, by region Proportion of local Proportion of Foreign direct budget income in GDP investment capital in investment, 1998­2004 Region (percent) GDP (percent) (US$ million) Red River delta 28.6 44.8 12,183 Southeast 35.9 32.8 27,518 Mekong River delta 10.9 32.4 1,136 North central coast 19.4 41.9 1,210 South central coast 29.7 45.3 3,315 Northeast 17.5 43.9 1,939 Northwest 8.5 58.4 59 Central highlands 15.5 43.6 966 Vietnam 26.7 37.9 48,341 Source: GSO (2008a). Table 7.10 Accessibility of infrastructure in Vietnam, by region, 2004 percent Access to a No access to clean No access to Access to the Region telephone at home water electricity Internet at home Red River delta 24.3 0.9 0.4 1.1 Southeast 45.6 6.6 3.6 5.1 Mekong River delta 17.3 31.5 14.4 0.5 Central north 12.6 13.2 3.4 0.2 South central coast 21.3 10.9 2.6 0.5 Northeast 16.1 22.5 10.9 0.2 West east 8.8 57.7 32.7 0.1 Central highlands 18.8 20.0 12.6 0.5 Vietnam 22.7 15.9 7.3 1.3 Source: GSO (2006c). infrastructure and economic disparities. of this province are well developed, raising The people in the most developed regions the average value of the region. (the southeast) have the best access to infrastructure and services. In contrast, the Macro policies people in the most isolated areas (north- Apart from the natural characteristics of east, northwest, central highlands) have the regions, state policies also result in economic worst access. disparities. These policies seek to promote growth and create favorable conditions for Economic structure advantaged sectors, regions, and groups by Seemingly, the localities with a low share of offering incentives and according them pri- agriculture and a high share of industry in ority in accessing resources. One example is GDP enjoy a higher growth rate (see figure trade protection policy, which treats com- 7.9 and table 7.11). The regions with a high modities differently depending on the sector (see table 7.12). share of industry can create high export The following are some other macro pol- value and make a large contribution to the icies with the same impact: state and national budgets. A special case is the northeastern region. While many prov- · In some periods, the domestic cur- inces in the highlands have underdeveloped rency was kept at an inflated value. This economies and conditions that are not exchange rate policy favored industrial conducive to the development of industry, products with low competitiveness,while Quang Ninh province located on the eastern it disfavored export-oriented industries border has a significant gateway to China, such as rice, coffee, cashews, and rubber. a deepwater port; the Ha Long World Heri- · For a long time, the terms of trade were tage site; and the biggest coal mine in the disadvantageous for agricultural prod- country. Hence, the industry and economy ucts. The price of agricultural products Rural development issues in Vietnam: spatial disparities and some recommendations 111 Figure 7.9 Relation between GDP per capita and agricultural value added in GDP in Vietnam 180 160 140 VND) 120 (mil 100 capita 80 per 60 GDP 40 20 0 0 10 20 30 40 50 60 70 Share of agriculture in GDP (%) Source: Provincial Statistics Offices (2008). Table 7.11 Structure of GDP and economic development in Vietnam, by region, 2006 percent Share of service in Share of industry in Contribution to total Region GDP GDP GDP Red River delta 44.43 40.32 19.56 Southeast 30.62 63.95 38.73 Mekong River delta 28.41 31.30 20.53 Central north 36.57 32.27 5.90 South central coast 38.55 37.81 5.95 Northeast 36.34 36.89 5.29 West east 35.99 21.84 1.12 Central highlands 28.21 24.85 2.92 Vietnam 38.07 41.56 100 Source: Provincial Statistics Offices (2008). Table 7.12 Effective rate of protection in Vietnam, by sector, 1997­2003 benefits enterprises implementing indus- percent trial and urban development projects Sector 1997 2001 2003 and creates an opportunity for specula- Agriculture 7.74 7.43 12.52 tion, placing farmers at a disadvantage. Manufacturing 121.47 95.97 43.94 Mining 6.05 16.39 -0.03 There are many other examples of incen- Average 59.54 54.1 26.23 tives for urban development, such as infra- Source: Athukorala (2005). structure investment, high-quality service, price stabilization, intellectual attraction often rises more slowly than the price of to cities and industrial zones, and many inputs (fertilizers, chemicals, gasoline, examples of unfavorable policies for farmers, machines) as well as the price of con- such as prohibiting the use of outdated vehi- sumer goods and services for farmers cles in rural industry and restricting the use (medicine, health care, education). of unofficial services (which are performed · One of the most controversial policies is mostly by migrants from rural areas). compensation for the transfer of agricul- tural land to industrial, urban land. Nor- Investment orientation mally, the price paid to farmers for land Although agriculture is creating jobs and is different than the price paid by enter- income for the majority of the population prises and landowners who will convert it and contributes 22­23 percent of total to industrial or urban use. This situation GDP, in recent years the portion of total 112 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 7.10 Total investment in the economy and in agriculture in Vietnam, 1995­2005 400,000 350,000 prices) 300,000 current 250,000 in dong 200,000 150,000 (million 100,000 50,000 investment 0 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Investment in agriculture Total investment Source: GSO (2007a). investment in agriculture has been low The cause of this situation includes both (see figure 7.10). Public expenditure for market mechanisms that direct resources to agriculture only accounts for 5­6 percent highly profitable fields and areas and public of the government budget, equivalent to policies that focus attention and resources 3­6 percent of total agricultural produc- on industry, services, and urban areas in an tion, and 1­1.5 percent of total GDP. As a effort to accelerate economic growth and result, the development of infrastructure development. and many public services (animal health, The solutions to this complicated prob- plant protection, information, research, and lem should begin with research on a theory technical transfer) have not kept pace with of sustainable development. All strate- the demand, especially in rural areas. gies and policies should be formulated to The proportion of agricultural invest- ensure this objective. In principle, key link- ment in total social investment decreased ages in the economy should be improved as from 7.1 percent in 1996­99 to 6.4 percent follows: in 2000­03.Private investment in agriculture and in rural areas only makes up 13­14 per- · Link production, processing, busi- cent of private investment in general.Despite ness, and consumption (value chain, the role of foreign investment in the econ- commodity chain) so that stakehold- omy overall, the share of foreign investment ers can communicate with one another in agriculture is low, at only 3­4 percent. As throughout the process from production a result, the socioeconomic gaps between the to consumption through associations, rural and urban areas and between agricul- economic contracts, and community ture and industry are inevitable. commitment and so that commodity associations can provide science and Recommendations technology, market information, credit, Although they are not serious yet, economic extension services, and food safety. Such disparities are increasing in Vietnam. The linkages are needed to allow the sharing disparity is not only in space and growth of profit and risk and allow production rates but also in accessibility of resources and households to participate more in the services, not only in the economy but also in value chain, raising their income and culture, society, and the environment. reducing their risks. Rural development issues in Vietnam: spatial disparities and some recommendations 113 ·Link the rural and urban areas by remov- · Attract foreign investment to develop ing barriers to the migration of rural science, technology, and management; laborers to urban areas and create a favor- promote trade liberalization; and enter able environment in rural areas to attract integration commitments. investment, develop small and medium · Through the market, media, civil soci- enterprises (SMEs), upgrade services and ety, education, and scientific and tech- infrastructure, and improve education. nological research, give all social classes, Build a system of satellite cities and indus- including vulnerable groups, the oppor- trial zones to decentralize urban areas and tunity to participate in the national eco- attract industry to rural areas. nomic development process, to join and ·Link key labor-supplying areas with benefit from globalization, and to raise urban areas and industry so that workers their voice, self-defense capacity, and can return home after work and urban economic, political, and social position. dwellers can enjoy life in rural areas. Give priority to spillover industries that Socioeconomic disparity is an inevitable mobilize rural laborers in industry and consequence of fair market competition. the service sector in urban, industrial However, in the current context of higher zones. This would help to create jobs, growth and untenable solutions, such as raise income, and improve living stan- destroying the environment now and repro- dards for rural people. ducing it later or pauperizing farmers and ·Link regions by mobilizing private and mobilizing them as unskilled labor in indus- public capital to build infrastructure, try, disparities are leading directly to ecolog- especially key traffic axes among regions, ical imbalance and social crisis. Therefore, focusing on difficult regions such as solving the problem of disparity is as crucial mountainous, remote areas and minority as pursuing economic growth. ethnic areas and giving priority to agri- culture and rural development to ensure Notes equality between favorable and unfavor- Dr. Dang Kim Son is director general of the able areas for industrial development. Institute of Policy and Strategy for Agriculture ·Deal with the issue of an aging rural and Rural Development in the Vietnam Ministry population by gradually setting up a sys- of Agriculture and Rural Development. tem of social safety nets and transferring 1. According to the official poverty line set by the Vietnamese government. this system from the household level to 2. Vietnam's currency is the dong. Vinh, the community level and the state. Hoang Ngoc, estimation data based on total ·Apply incentives for ethnic groups not agricultural labor force in 2005. favored by natural and social conditions. 3. The rural labor force is the number of peo- At the same time,preserve ethnic cultural, ple in rural areas who are more than 15 years old economic, and social values and encour- and are involved in economic activities. It does age the community to manage its own not include people who are unemployed. life and development and not depend passively on the state and outsiders. References ·Build key development corridors (roads, Athukorala, Prema-chandra. 2005."Trade Policy railways) to link the localities within Reforms and the Structure of Protection in a region, connect foreign localities in Vietnam." Division of Economics, Research Cambodia, Lao, and Yun Nan, China, School of Pacific and Asian Studies, Austra- for example, to the sea through Viet- lian National University. namese territory, develop key deepwater GSO (General Statistical Office of Vietnam). ports in the north, the center, and the 2000. Vietnamese Household Living Standard south, and build international airports Survey 1999. Hanoi: GSO. linking Vietnam to international ports ------. 2002. Results of Agriculture, Rural, and and airlines. Fishery Census 2001. Hanoi: GSO. 114 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA ------. 2003. Vietnamese Household Living MARD (Ministry of Agriculture and Rural Standard Survey 2000. Hanoi: GSO. Development). 2005. National Report on ------. 2006a. Migration Survey in 2004. Hanoi: Forest Situation. Hanoi: MARD. GSO. ------. 2007."Preliminary Report on Five- ------. 2006b. Vietnam, 20 Years of Renovation Year Implementation of Resolution 15-NQ/ and Development. Hanoi: Statistical Publish- TW-IX on Accelerating Agricultural and ing House. Rural Industrialization and Urbanization." MARD, Hanoi. ------. 2006c. Vietnamese Household Living Standard Survey 2004. Hanoi: GSO. Ministry of Health, Institute of Nutrition. 2004. Result of Nutrition Survey for 2000. Hanoi: ------. 2007a. Results of Agriculture, Rural, and Ministry of Health. Fishery Census 2006. Hanoi: GSO. Provincial Statistics Offices. 2008. Provincial ------. 2007b. Vietnamese Household Living Statistical Year Books 2006. Provinces. Standard Survey 2006. Hanoi: GSO. VASS (Vietnam Academy of Social Science)- ------. 2008a. Enterprise Survey 2006. Hanoi: UNDP (United Nations Development Pro- GSO. gramme). 2006. Human Development in ------. 2008b. Statistical Year Book 2006. Vietnam, 1999­2004: Main Changes and Hanoi: GSO. Trends. Hanoi: UNDP. Economic geography of Indonesia: location, connectivity, and resources Hal Hill, Budy P. Resosudarmo, and Yogi Vidyattama 8 With its 13,000 islands, Indonesia is the mainly concentrated on Java and Bali, which world's largest archipelagic state and one of in turn boosted the economic fortunes of the most spatially diverse nations on earth in these islands. Third, the economic crisis of its resource endowments, population settle- 1997­98 particularly affected construction ments,location of economic activity,ecology, of the modern sector, finance, and import- and ethnicity. There are about 350 identified substituting manufacturing sectors, and, ethnic groups. In the early 2000s, per capita because these are mainly located on Java, this regional product in the richest province,East region experienced the sharpest decline in c h a p t e r Kalimantan, was around 16 times that in the economicactivity.Fourth,thedecentralization poorest, Maluku. The range of poverty inci- program has transferred considerable finan- dence was from 3.4 percent of the popula- cial resources and administrative authority tion in Jakarta to 42 percent in Papua. from the central government to the second- The country's regional development leveltiersof government(kabupaten and kota) patterns are therefore of great analytical and,in the process,is likely to alter Indonesia's and policy interest. Indonesia is formally a economic geography significantly. unitary state, but all national governments While much has been written on various have had to deal with major regional devel- aspects of regional development in Indone- opment challenges. The country's inter- sia, there are two reasons to revisit the issue. national boundaries have changed twice First, it has only been possible to measure since independence, with the formal entry accurately and quantify regional trends since of Papua (then Irian) in 1969 and the entry the mid-1970s. Development dynamics are a and later exit of East Timor in 1976 and long-term phenomenon, involving decades 1999, respectively. Subnational boundaries rather than years, and we are only now in have changed frequently. a position to analyze Indonesia's regional While national economic fortunes and economic, social, and demographic devel- policies explain much of the local devel- opment over a period of 30 years. opment outcomes, regional responses to The second motivation has to do with international and domestic events inevi- the renaissance of regional economics and tably vary. Four examples briefly illustrate science. Traditionally regarded as inhabit- this proposition. ing the backwaters of the profession, "new First, the 1970s oil boom disproportion- economic geography has come of age" in ately benefited the country's four resource- the words of Neary (2001). This has arisen rich provinces, even though much of the principally owing to the intellectual fusion windfall gains accrued to the central gov- between international trade and geography ernment and oil companies. Second, the articulated by Krugman (1991). major policy reforms of the 1980s resulted A key insight from this literature concerns in rapid, export-oriented industrialization, the interaction between the international 115 116 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA economy and local development patterns. analysis is conducted at the provincial level As countries remove regulatory impedi- and is based on a standard set of 26 prov- ments to the cross-border flow of goods, ser- inces. These are the 27 provinces that existed vices, capital, technology, and people, those for most of the Soeharto era, excluding the regions most connected to the global econo- special case of East Timor. Since 2000, there my--by dint of location, infrastructure, and has been considerable fragmentation (peme- enabling institutions--are likely to grow karan) of provincial boundaries, and so it is the most quickly. In cases where domestic necessary to adjust the published data back infrastructure lags or there are regulatory to the pre-2000 provincial boundaries.1 barriers to domestic commerce, these inter- The second general point to note is that, nationally oriented regions may become in reflecting data constraints, our story com- effect enclaves, more connected to the global mences in the 1970s,the period when reliable economy than to the hinterland. regionalsocioeconomicdatabecameavailable As a corollary, to the extent that national (see Arndt 1973). In the case of demographic economic policies--openness, macroeco- and related data, the starting point is the nomic management, and so forth--are 1971 population census, while the regional tending to converge around the world, local accounts effectively commence in 1975. governance and institutions are likely to become increasingly important determi- Economic geography nants of regional development outcomes. As is well known in the Indonesian context, In the search for markets and mobile fac- there are two relevant measures of regional tors, for example, Jakarta is competing with economic activity and three indicators of both Surabaya and Shanghai, albeit in dif- economic welfare. There is no "true" mea- ferent dimensions. sure of economic activity and welfare, as This paper draws on this rapidly expand- each one measures a different concept. ing literature and the rich Indonesian We therefore present and examine the regional database to address the following three series. issues, each of which constitutes a section The activity measures are regional gross of the paper. First, we provide an overview domestic product (GDP) and regional GDP of Indonesia's changing regional economic excluding mining, in particular oil and gas. geography, examining how the location of The latter measure is frequently employed in economic activity and provincial economic Indonesia owing to the presence of extrac- rankings have changed since the 1970s.Next, tive activities, which significantly affect we investigate patterns of regional economic measured local economic activity but have growth and structural change, examining much less effect on local economic and regional growth dynamics, followed by the social welfare. This difference between the interrelationships among growth, structural two series arises because a large proportion change, and demographic dynamics. Next of the returns to extractive activities accrue we examine convergence and inequality, to extra-provincial entities, principally the both in terms of the "four-quadrants" story central government and foreign and domes- of initial incomes and subsequent growth tically owned mining companies. With the and the various measures of convergence. introduction of decentralization measures in These results are compared with Indonesia's January 2001, regions now receive a higher provincial social indicators. We also discuss proportion of mining revenue, and thus the conflict at the regional level and assess vari- differences between the welfare measures ous explanatory hypotheses. In a final sec- might be expected to narrow gradually over tion, we summarize our main findings. time (Resosudarmo and Vidyattama 2007). To address these issues,we have assembled In principle, the output of any "enclave" a large regional database from various series activity might be deducted from regional of Indonesia's Central Board of Statistics GDP to provide a better indication of local (Badan Pusat Statistik). These data are dis- economic activity and welfare. In practice, cussed in detail in the relevant sections, but the choice is between oil and gas, on the we note here two general points. First, the one hand, and mining, on the other. Other Economic geography of Indonesia: location, connectivity, and resources 117 resource-based activities, notably forestry, the country's total GDP, non-mining GDP, are substantially more labor intensive and and household expenditure, respectively, in therefore have larger local employment and 2004 (see table 8.1). Sumatra comes next, income spin-offs. Some mining activities with 22, 20, and 20 percent. Kalimantan has are also quite labor intensive (for example, 9, 8, and 5 percent, Sulawesi has 4 percent on small-scale gold mining) and perhaps do not all measures, and the eastern provinces have need to be deducted from regional GDP. In around 3 percent on all measures. We exam- practice, the distinction is inevitably some- ine the factors underlying these regional what arbitrary. dynamics in the following section. In this paper, we employ regional GDP Over time and regardless of the measure and regional non-mining GDP. The latter is used, there has been a clear shift of economic selectedfortworeasons.First,thenon-mining activity toward Java-Bali and, in particular, series is available for a longer period of time the national capital Jakarta.Jakarta generated (since 1975) than the non-oil and gas series one-sixth of Indonesian GDP in 2004,double (since 1983). Second, the difference between that of 1975. Its share of non-mining GDP the non-mining and non-oil and gas series is also has increased significantly,though not as not large,as oil and gas are the major compo- fast.It accounts for virtually all of the increase nent of Indonesian mining output, account- in the Java-Bali share of GDP and more than ing for 68 percent of mining value added in 100 percent of the increase in non-mining 2004. The only regional exception--that is, a GDP. That is, the Java-Bali share excluding very large non-oil and gas mining sector--is Jakarta is stable for the total regional GDP Papua (Manning and Rumbiak 1989). series, while declining slightly for the other In addition to total and non-mining two series. In fact, the increase in Jakarta's regional GDP, there are estimates of house- share is understated,as some of its growth has hold consumption expenditure (CE) per spilled over the border to West Java, the only capita. The latter are available for a shorter other province in the group with an increased time period,since 1983.They are particularly share of GDP. The three big Java provinces-- useful for computing poverty estimates.They these two and East Java--account for half of are not a superior indicator of economic wel- Indonesia's GDP and a slightly higher share fare--by definition they exclude household of its non-mining GDP. saving and government consumption and Sumatra's share of non-mining GDP saving--but they do provide an additional and household expenditure has been sta- dimension. This series would be expected ble at 20­21 percent. Its share of GDP has to correlate more closely with non-mining been declining, owing to the falling share regional GDP. of oil and gas in the national economy We present the regional accounts data at and reflected in the declining shares of the three points in time, 1975 (1983 for the CE island's main producers, Riau and Aceh. data), 1990, and 2004. These correspond to The two largest economies have been Riau important time periods in Indonesia's recent with mining included and North Sumatra economic history. These are, respectively, with mining excluded. Riau is a particularly the early years of the oil boom, the year unusual regional economy, with a large oil in which the major post­oil boom policy enclave, a cash crop economy, a relatively reforms were introduced, and the year in wealthy capital city, and a strong export-ori- which national income per capita returned ented manufacturing and service economy to pre-crisis levels. in the islands adjacent to Singapore. Thus, although its share of national GDP has Major concentrations of declined since 1975 owing to the oil effects, Economic activity its share of national non-mining GDP (and It is convenient initially to divide the country household expenditure) has more than into five major island groupings: Java-Bali, doubled since 1990, the fastest increase in Sumatra, Kalimantan, Sulawesi, and"eastern the country for this period. Indonesia."Java dominates Indonesia's econ- Of note is the fact that the three south- omy, contributing 61, 66, and 67 percent of ern provinces of Sumatra--South Sumatra, 118 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 8.1 Shares of regional GDP with and without mining and household consumption expenditure in Indonesia, by province, various years, 1975­2004 Indonesia = 100 percent Regional GDP Non-mining regional GDP CE Province 1975 1990 2004 1975 1990 2004 1983 1990 2004 Sumatra 32.2 24.9 22.2 21.0 20.1 20.0 20.6 20.1 20.2 Aceh 1.6 3.8 2.2 1.7 2.8 1.7 2.1 2.1 0.9 North Sumatra 5.7 5.7 5.4 6.6 6.3 5.8 6.4 6.0 5.4 West Sumatra 1.8 1.8 1.7 2.3 2.0 1.8 2.2 2.2 1.8 Riau 15.1 6.5 6.8 2.1 1.9 5.0 1.9 2.0 5.5 Jambi 0.8 0.7 0.8 0.9 0.8 0.8 0.6 0.8 0.9 South Sumatra 4.8 4.2 3.3 4.5 3.8 2.8 4.7 4.2 3.6 Bengkulu 0.3 0.4 0.4 0.4 0.5 0.4 0.5 0.5 0.4 Lampung 1.9 1.7 1.6 2.4 1.9 1.7 2.2 2.4 1.6 Java-Bali 51.5 58.6 61.0 62.8 64.5 65.7 64.4 65.8 67.4 Jakarta 8.7 12.1 17.1 11.0 13.8 18.8 10.4 9.9 16.5 West Java 14.5 16.8 17.2 16.3 17.1 18.0 17.2 19.4 19.0 Central Java 9.9 11.5 8.8 12.5 13.1 9.6 14.5 12.2 10.4 Yogyakarta 1.2 1.0 1.0 1.5 1.1 1.1 1.6 1.3 0.9 East Java 15.8 15.5 15.5 19.9 17.5 16.8 18.7 20.8 19.3 Bali 1.3 1.6 1.3 1.6 1.8 1.4 2.0 2.2 1.3 Java-Bali without Jakarta 42.8 46.4 43.8 51.8 50.7 46.9 54.0 55.9 51.0 Kalimantan 7.1 9.1 9.3 6.1 7.9 7.5 5.4 5.4 4.6 West Kalimantan 1.4 1.5 1.3 1.8 1.7 1.5 1.7 2.0 1.3 Central Kalimantan 0.5 0.7 0.8 0.7 0.8 0.9 0.9 1.0 0.9 South Kalimantan 1.0 1.2 1.2 1.3 1.4 1.1 1.5 1.3 0.9 East Kalimantan 4.1 5.7 6.0 2.3 4.0 4.0 1.2 1.1 1.6 Sulawesi 5.0 4.1 4.2 6.3 4.5 4.3 6.2 5.3 4.4 North Sulawesi 1.3 0.8 0.8 1.6 0.9 0.8 1.3 1.0 0.7 Central Sulawesi 0.4 0.5 0.7 0.6 0.6 0.7 0.8 0.8 0.8 South Sulawesi 3.0 2.4 2.2 3.8 2.6 2.2 3.5 2.9 2.4 Southeast Sulawesi 0.3 0.4 0.5 0.3 0.5 0.5 0.6 0.6 0.5 Eastern Indonesia 4.3 3.3 3.3 4.0 3.0 2.5 3.5 3.3 3.3 West Nusa Tenggara 0.8 0.7 1.0 1.0 0.8 0.7 1.0 1.0 0.7 East Nusa Tenggara 0.8 0.6 0.6 1.0 0.7 0.6 1.0 1.0 0.7 Maluku 0.9 0.8 0.3 1.1 0.9 0.3 0.9 0.9 0.4 Papua 1.8 1.2 1.4 0.9 0.7 0.8 0.7 0.5 1.5 Indonesia (current Rp trillion) 12 188 2,203 10 165 1,996 34 83 1,182 Source: Central Board of Statistics (various years). Note: All numbers are in percentages. Based on current prices. West Sumatra, and Lampung--have been has enriched kabupaten Kutai Kartanegara, slipping. In 2004 their share of non-mining which has the nation's highest regional GDP GDP was about two-thirds of that in 1975. per capita.2 Both regional GDP series are Evidently, their proximity to stronger econ- misleading indicators of the region's living omies to their south and north has not had a standards, as indicated by the much lower growth spillover effect. Lampung, in partic- share of household expenditure compared ular, was seen as a solution to Java's alleged to their shares of regional GDP with and problems of over-population and poverty, without mining. Nevertheless, the latter is but since the 1970s its economic perfor- growing quickly, rising 50 percent as a pro- mance has lagged behind that of Java. portion of the national total since 1990. The largest and most dynamic regional The share of the eight eastern prov- economy in Kalimantan is East Kalimantan, inces in the national economy is gradually with its large oil and gas resources. In fact, it declining. This generalization applies to has experienced "twin booms" in the words the largest regional economy in the east, of Pangestu (1989),from both hydrocarbons South Sulawesi, and its traditionally most and timber. Downstream industrial pro- prosperous region, North Sulawesi. The cessing has provided a further boost, while share of Maluku, the site of the country's since 2001 the decentralization program most serious religious conflict, is now less Economic geography of Indonesia: location, connectivity, and resources 119 than one-third of the 1975 figure. The income and welfare differences and evidence only exceptions to this picture of declining of both continuity and change in these shares are the two small Sulawesi provinces rankings (see table 8.2). In 2004, the gap (which were boosted by in-migration), between the richest and poorest provinces West Nusa Tenggara (which recently expe- was very large, depending on which series rienced a major expansion in mining) and is used. The ratio of the richest to poorest Papua (in the case of household expen- was 15.9 for regional GDP per capita (East diture since 1990). The latter reflects the Kalimantan:Maluku), 14.7 for non-mining combined effects of the mining boom and regional GDP per capita (Jakarta:Maluku), special government programs.3 and 11.3 for household expenditure (Jakarta: West Nusa Tenggara). Provincial economic rankings The first three columns indicate how We examine these rankings with reference to the inclusion of mining inflates the esti- the three measures discussed above. All data mated regional GDP per capita for the are normalized around the national aver- resource-rich regions, especially in the ear- age of 100. There are large interprovincial lier years. For example, in the case of Riau, Table 8.2 Regional GDP with and without mining and household consumption expenditure per capita in Indonesia, by province, various years, 1975­2004 Indonesia = 100 (index) Regional GDP per Non-mining regional capita GDP per capita PCE Province 1975 1990 2004 1975 1990 2004 1983 1990 2004 Sumatra 177.0 121.7 103.1 115.3 98.1 92.9 104.8 98.4 93.9 Aceh 93.3 200.7 114.5 97.9 147.4 92.0 114.4 108.9 49.5 North Sumatra 101.9 99.6 92.2 116.7 110.1 100.5 111.0 104.9 92.3 West Sumatra 79.1 78.3 81.6 99.2 88.0 86.8 96.8 96.1 87.6 Riau 1,061.5 352.0 245.2 150.2 103.9 178.6 128.8 106.0 198.0 Jambi 87.1 65.5 67.0 101.5 72.0 62.2 62.0 72.5 75.9 South Sumatra 160.6 118.5 92.8 150.1 107.5 77.2 144.8 119.2 100.5 Bengkulu 61.9 64.6 49.0 77.6 70.0 52.4 90.5 75.7 56.3 Lampung 72.9 50.8 48.4 91.6 57.8 50.9 62.2 70.2 48.4 Java-Bali 79.4 94.9 103.3 96.9 104.4 111.3 101.9 106.5 114.2 Jakarta 212.1 262.9 419.1 267.1 299.9 460.9 224.9 214.3 403.0 West Java 78.7 84.9 85.9 88.6 86.2 89.6 91.3 97.7 94.8 Central Java 55.6 72.2 58.4 69.6 81.9 63.9 85.9 76.7 69.4 Yogyakarta 61.6 62.0 64.5 77.4 70.3 70.6 88.1 78.2 59.7 East Java 76.3 85.1 92.7 95.9 96.5 100.3 96.7 114.3 115.2 Bali 77.6 103.2 83.4 97.1 117.3 91.4 119.0 143.9 82.5 Java-Bali without Jakarta 70.5 81.3 79.8 85.4 88.7 85.3 92.2 97.8 92.7 Kalimantan 159.2 178.4 159.8 136.6 154.0 128.2 114.7 106.3 79.2 West Kalimantan 84.2 80.3 65.8 105.9 91.1 71.8 101.9 113.0 62.2 Central Kalimantan 88.3 93.9 83.9 110.9 106.7 91.9 132.7 122.5 86.7 South Kalimantan 72.2 85.3 77.0 90.5 93.7 70.8 110.6 90.9 59.3 East Kalimantan 576.5 538.2 462.3 325.9 380.4 311.8 131.5 104.0 123.3 Sulawesi 70.6 58.8 55.9 87.7 64.5 57.4 87.4 76.3 59.0 North Sulawesi 86.9 57.7 59.6 109.0 65.2 59.9 89.6 75.6 51.9 Central Sulawesi 55.1 53.2 60.0 69.1 59.1 65.0 91.4 79.9 67.5 South Sulawesi 70.7 60.9 55.3 89.0 66.6 56.0 85.7 75.3 61.4 Southeast Sulawesi 52.7 57.6 48.5 52.8 59.6 50.8 87.6 78.6 49.8 Eastern Indonesia 78.1 58.2 54.6 72.5 53.6 40.8 64.1 58.5 54.3 West Nusa Tenggara 45.5 37.5 50.6 56.6 42.1 36.2 53.9 51.5 35.8 East Nusa Tenggara 41.5 34.7 30.5 52.1 39.4 33.2 52.0 53.2 38.5 Maluku 91.9 76.6 29.0 113.1 82.6 31.3 89.6 84.6 38.5 Papua 226.8 126.8 123.5 111.1 72.8 69.7 84.3 54.0 126.2 Indonesia (current Rp 91 1,051 10,421 72 922 9,443 216 461 5,592 thousands) Source: Central Board of Statistics (various years). Note: All provincial numbers are relative to Indonesia, which is set to 100. Based on current prices. 120 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA regional GDP per capita was seven times declined sharply during the 1980s in the wake higher than non-mining regional GDP in of the fading oil boom,resulting in its income 1975. By 2004, these effects were much and expenditures being close to the national smaller. The three series were about 37 average. However, as noted, strong growth in percent higher in Aceh, 52 percent in Riau the islands close to Singapore,combined with (and also in West Nusa Tenggara, owing to export-oriented cash crops on the mainland, its recent mining expansion), 63 percent resulted in it being the third-richest province in East Kalimantan, and almost double in in 2004 according to both series. Papua. In the first and last of these prov- inces, non-mining regional GDP had fallen Consistently nonpoor. A second group below the national average. of provinces may be termed consistently We therefore develop our main story well-off, with non-mining regional GDP aroundthenon-miningseries,whichexcludes per capita at least 85 percent of the national the enclave mining effects. We identify what average. This includes the traditionally may be termed consistently "wealthy" and strongest agricultural exporter, North "poor" regions, those close to the national Sumatra; the frontier province of Central average, and those that have experienced a Kalimantan (initially driven by timber, but significant change in relative incomes. with cash crops now the major agricultural activity); the country's two major industrial Consistently wealthy. There are two really provinces, West and East Java (the latter's wealthy provinces, Jakarta and East Kalim- ranking rising appreciably); the major tour- antan. Jakarta is by far the richest province ist region, Bali;5 and West Sumatra (where as measured by non-mining regional GDP both agriculture and a range of services are per capita, at about four times the national important). Aceh would have belonged in average and double that of the next rich- this group until recently, but the protracted est province. It has been getting relatively conflict (at least until 2005) has resulted in richer, especially since 1990. This is not- sharply lower living standards. withstanding, first, the 1980s liberalizations, which reduced the regulatory powers of the Very poor. At the other extreme are the capital; second, the decentralization of 2001, poor provinces, with a ratio of about half the which transferred resources and funds to the national average or less. They are all located regions; and third, the 1997­98 crisis, which in eastern Indonesia. The two Nusa Tenggara affected it more severely than any other provinces are consistently poor and evidently province apart from West Java.4 However, slipping further behind, falling from just it also recovered more quickly than most over half the national average in both series provinces. In spite of its role as the national to 35­40 percent. Maluku, which has expe- capital, the public sector is one of the small- rienced the most serious conflict since 1998, est in the country. has fallen sharply, from above the national East Kalimantan's per capita non-mining average (in non-mining regional GDP per regional GDP has consistently been at least capita) to one-third of it. Southeast Sulawesi, three times the national average, indicating the poorest province on this island, is about that its economic wealth extends well beyond half the national figure in all series. the mining enclaves. However, its household expenditure suggests that community living Slipping behind. A number of provinces standards are much closer to the national have slipped significantly in their rankings average. About 60 percent of East Kaliman- in both the non-mining regional GDP and tan's non-mining regional GDP comes from expenditure series. These are mainly tra- oil- and gas-processing industries. These ditional agricultural exporters that have are relatively capital-intensive activities, and not been able to capitalize on their ini- much of the return on these investments tial advantages. Examples include South accrues to entities outside the province. Sumatra,6 Jambi, Bengkulu (all in Suma- A third province, Riau, is generally well tra), West and South Kalimantan, North above the national averages. Its fortunes and South Sulawesi, and resource-rich Economic geography of Indonesia: location, connectivity, and resources 121 Papua (although its household expenditure Java-Bali in general was the fastest-growing has risen). It is notable that Central Java and region, followed by Sulawesi (see table 8.3). Yogyakarta have slipped according to both Sumatra was pulled down by the oil sector. series, although not as much as the others in In the case of growth in non-mining regional this group. The latter case is puzzling given GDP per capita, there is the same relatively its traditional importance as a major center even pattern of growth across island group- of higher education. This is such a heteroge- ings: Java-Bali was the only major region to neous group of provinces as to render haz- grow (slightly) faster than the national fig- ardous any attempt to find a common set ure of 4.6 percent. Kalimantan and Sulawesi of explanations. Perhaps the most impor- were just below it, followed by Sumatra and tant observation is that they generally lack by eastern Indonesia, 0.9 percentage points a major, internationally oriented engine of below the average.The growth rates of house- growth. We return to this issue shortly. hold expenditure are similarly quite even. These interprovincial rankings shed much It is not easy to identify obvious group- light on Indonesian regional dynamics. ings and characteristics of provinces based In the first three decades of Indonesian inde- on growth rates. The fastest growth rates (in pendence, Java was regarded as the country's regional GDP per capita) over the period most serious development challenge, with of 1976­2004 occurred in Bali, West Nusa the island "asphyxiating for want of land," in Tenggara, West Sumatra, Jakarta, Central the words of Keyfitz (1965: 503). By contrast, Java, and North Sulawesi. The slowest rates in spite of their poorer human and physical were recorded in Riau, Papua, South Suma- infrastructure, the resource-rich regions in tra, Maluku, East Kalimantan, and Jambi. the Outer Islands were considered to have less Thus the fast growers included both small poverty and better development prospects. and large provinces, "central" and remote However, a different picture emerged locations, and areas with initially high and in the 1980s. The major economic policy low per capita incomes. The only common reforms increased the relative profitability of element appears to be the absence of a major export-oriented manufacturing and related resource sector, whereas this is a feature of higher-value services, which are located all but one of the slow growers. mainly on Java-Bali. Declining commod- For a more detailed examination, we ity prices adversely affected many off-Java focus on the non-mining regional GDP per regions. Thus Sumatra's ranking on all three capita series, which is arguably the most series declined significantly. It was over- accurate indicator of provincial economic taken by Java-Bali by 1990 and was below performance. The fastest-growing provincial the national average for both series in 1990 economy by a significant margin (1.1 per- and 2004. A particularly notable decline is centage point over number two) was Bali. Lampung, historically seen as the solution Also in the high-growth group (at least 5 to Java's "population problem." In 2004 its percent) are Southeast Sulawesi,7 Jakarta, income and expenditures were less than half and West Sumatra. A further four provinces those of Java-Bali in all three series. are just above the average: North Sumatra, Kalimantan displays above-average West and Central Java, and North Sulawesi. income but below-average expenditure, Conversely, a number of provinces grew at a owing to the distributional effects of the rate at least a percentage point slower than natural resource sectors. The eight provinces the national average. These are Papua and of eastern Indonesia are both poor and slip- Maluku in the east and Riau, Jambi, and ping further behind, with the partial excep- South Sumatra in Sumatra. The remaining tion of Papua's enclaves. 13 provinces grew close to, but slower than, the national average. Regional economic growth and The story differs for the shorter (and not change directly comparable) household consumption We now examine provincial economic expenditure per capita (PCE) series. The fast- growth rates on a per capita basis over est-growing provinces from 1984 to 2004 were the same periods and for the same series. East Java,North Sulawesi,East Nusa Tenggara, 122 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 8.3 Annual growth rates of regional GDP with and without mining and household consumption expenditure per capita in Indonesia, by province, 1976­2004 Regional GDP per capita Non-mining regional GDP per capita CE Province 1976­90 1991­2004 1976­2004 1976­90 1991­2004 1976­2004 1984­90 1991­2004 1984­2004 Sumatra 1.0 2.2 1.6 4.7 3.2 4.0 2.6 3.7 3.3 Aceh 9.8 -1.8 4.0 7.2 0.8 4.1 3.3 2.7 2.9 North Sumatra 5.5 3.6 4.6 5.9 3.8 4.9 1.8 4.2 3.4 West Sumatra 6.3 4.0 5.2 5.9 4.1 5.0 3.5 4.5 4.2 Riau -5.3 -0.5 -3.0 2.6 1.7 2.2 0.8 2.7 2.1 Jambi 2.7 3.1 2.8 2.9 2.6 2.8 5.3 3.3 3.9 South Sumatra 2.4 2.7 2.5 2.9 3.5 3.2 2.2 3.9 3.3 Bengkulu 6.0 2.5 4.3 5.5 2.7 4.2 0.5 3.1 2.2 Lampung 4.3 3.9 4.1 4.2 3.8 4.0 4.7 3.1 3.6 Java-Bali 6.5 3.1 4.9 6.5 3.2 4.9 3.8 4.0 3.9 West Java 5.6 3.7 4.6 5.8 4.0 4.9 3.7 3.9 3.9 Central Java 6.7 3.2 5.0 6.6 3.2 4.9 2.5 4.8 4.0 Yogyakarta 4.4 2.9 3.7 4.3 3.0 3.7 1.3 2.0 1.8 East Java 6.8 2.3 4.6 6.7 2.2 4.5 5.7 3.9 4.5 Bali 8.7 3.6 6.2 8.7 3.6 6.2 2.4 1.9 2.1 Java-Bali without Jakarta 6.3 3.1 4.8 6.4 3.2 4.8 4.0 4.0 4.0 Kalimantan 5.1 2.7 3.9 6.2 2.5 4.4 1.9 3.9 3.2 West Kalimantan 5.7 2.5 4.1 5.6 2.5 4.1 4.3 2.1 2.9 Central Kalimantan 5.6 1.8 3.7 5.5 1.8 3.7 1.5 3.1 2.5 South Kalimantan 4.7 4.0 4.3 4.5 2.8 3.7 1.5 4.4 3.4 East Kalimantan 3.4 2.1 2.8 6.3 2.2 4.3 -0.2 4.9 3.2 Sulawesi 5.2 3.9 4.5 5.1 3.8 4.4 2.8 3.8 3.5 North Sulawesi 5.1 4.8 5.0 5.0 4.7 4.8 3.1 5.1 4.4 Central Sulawesi 5.1 3.4 4.2 4.9 3.4 4.2 1.3 3.7 2.9 South Sulawesi 5.1 4.0 4.6 4.9 3.9 4.4 3.3 3.7 3.6 Southeast Sulawesi 6.3 2.5 4.4 7.7 2.4 5.1 2.1 1.9 1.9 Eastern Indonesia 3.2 3.7 3.4 4.5 2.7 3.7 2.7 4.1 3.6 East Nusa Tenggara 4.9 4.2 4.5 4.8 4.2 4.5 3.0 5.0 4.3 Maluku 5.4 0.0 2.8 5.3 0.3 2.8 2.3 2.5 2.4 Papua 0.3 3.1 1.6 2.7 2.6 2.6 1.5 3.7 2.9 Indonesia 4.8 3.0 3.9 6.0 3.1 4.6 3.4 3.9 3.7 Source: Central Board of Statistics (various years). Note: All numbers are in percentages. Based on 1993 prices. West Sumatra, and Central Java. The slowest for a number of export-oriented economies, growth was recorded inYogyakarta,Southeast which benefited from the 1980s reforms and Sulawesi, Riau, Bali, and Bengkulu. which weathered the economic crisis bet- The story also differs by subperiods. Aceh ter than other regions. Examples include grew very fast over the period 1976­90, as the predominantly agricultural producers, its gas production came on stream, but very North, West, and South Sumatra (the lat- slowly since 1990 in an era of (mostly) lower ter the only province to grow faster in the energy prices and conflict that increasingly second period than in the first), Lampung, affected economic activity.Similarly,East Kali- all of Sulawesi except the Southeast, and the mantan's growth slowed in the second period industrial province of West Java. as a result of lower energy prices and slower Although in aggregate growing more timber exploitation.In fact,Kalimantan expe- slowly than the national average, the four rienced the greatest deceleration in growth eastern provinces experienced mixed for- among the major island groupings, mainly tunes. Maluku, as noted, was severely owing to these factors, principally the former. affected by the post-crisis conflict. East Nusa Bali also slowed from its exceptionally rapid Tenggara grew a percentage point faster than growth, but was still above average after 1990. the national average in the second period, By contrast, some provinces that grew and the West grew at about the average. Pap- slower than the national average in the first ua's growth was dependent on commodity period recorded above-average rates in the prices, but its household expenditure grew second. This appears to be especially the case at almost the national rate. Economic geography of Indonesia: location, connectivity, and resources 123 There are several cases of provinces Are these differences in regional growth growing faster than the national average but amenable to quantitative explanation? As a slipping in the relative income rankings. For growing literature has argued, the growth example,North Sulawesi grew faster than the literature can be productively employed, in Indonesian (non-mining) average during a modified form (Barro and Sala-I-Martin 1976­2004, but its relative regional GDP per 1991). That is, openness can be redefined to capita fell very sharply, from 109 to 60. East mean "connected" (to the global economy); Nusa Tenggara grew at the national average, institutions clearly do differ among regions but its income fell from 52 percent of the in many countries; and factor and product national average to just 33 percent. There are markets in developing countries are often also converse cases, such as Riau, where non- poorly integrated. mining per capita growth was less than half The international evidence suggests, first, the national average, but its relative income that regions which are the most connected rose. These are presumably the result of to the global economy (in the sense of loca- local terms of trade effects--that is, of local tion, infrastructure, and trade regime) are economies specializing in the production of likely to grow more quickly, as is the case goods and services whose prices have risen of Jakarta, Bali, and in recent times Riau faster or slower than the general price level (at least the islands adjacent to Singapore). (or specifically the national accounts defla- These are arguably the regions most con- tor). This is confirmed, for example, in the nected to the global economy, in terms of case of North Sulawesi: using constant rather facilitating physical infrastructure, trade in than current prices, its per capita income goods and services, and the movement of ranking rose considerably. people. There are no obvious correlates among A second factor is clustering and increas- the fast growers.The explanations for Jakarta ing returns to scale, as forward and back- and Bali are relatively straightforward--the ward linkages develop and spill over from seat of government, global connections, and growth centers. The best example in the high-value services and industry in the for- Indonesian context is probably the rapid mer and the tourism success story and resul- industrialization in West Java since 1980 tant spillovers in the latter. In West and Cen- around the periphery of Jakarta. This region tral Java, export-oriented industrialization, has now become the industrial heartland especially in West Java from the mid-1980s, of Indonesia. and the earlier agricultural successes, espe- The evidence regarding regional institu- cially in Central Java, were important. North tions and governance is mixed and incom- Sumatra has a strong agricultural base and plete.We lack reliable long-term estimates of was traditionally the most industrial prov- any "quality" variables, and in any case the ince outside Java. provinces have enjoyed significant political West Sumatra and North Sulawesi had authority only since the decentralization of traditionally strong agricultural bases and 2001,while local-level democracy has arrived quite good education records. But both are even more recently. There is some anecdotal somewhat distant from the main centers of evidence to suggest that the higher-growth commerce, and neither has had a "boom- regions have been quite well governed. ing sector." West Sumatra's service sector The indifferent record of the resource- growth is probably connected to high levels rich provinces is suggestive of a Sachs and of inward remittances, as a result of its long Warner (2001)"resource curse"at work.Two history of mainly male out-migration (mer- of the four provinces have experienced very antau). In the case of North Sulawesi, tour- serious conflict, and most of the resource ism, shipping, and agroprocessing (mainly wealth (at least until 2001) accrued to enti- based on coconuts and fisheries) have all ties outside the province. However, there is done quite well. More recently, its toler- sufficient diversity within this group to cau- ance of diverse religions and ethnicities has tion against sweeping generalizations. Two reportedly attracted investment from neigh- of the provinces, East Kalimantan and Riau, boring conflict-prone regions.8 have become increasingly prosperous. 124 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Structure of regional economy There are high shares in Bali andYogyakarta, Indonesia was a predominantly agrarian reflecting their status as leading tourism and economy in the mid-1970s. Reflecting this, education centers, respectively. The share is agriculture was more than one-third of also high in West Sumatra, reflecting the regional GDP in 21 out of the 26 provinces traditional importance of remittances. But in 1975. In 10 it was at least half. By 2004, the share is also high in poorer, more remote only 8 were above this threshold. Thus, regions, including Maluku, East Nusa Teng- consistent with the well-known hypothesis gara, and North Sulawesi. For the poorer linking economic growth and structural regions, the explanation has more to do with change, there has been a rapid shift out of a relatively large government sector, as fiscal agriculture. The provinces that have been transfers have been weighted in their favor. slow to make this transition either are Higher transport shares in remote regions among the poorest in the country (Maluku, are also a factor. Southeast and Central Sulawesi, East Nusa Theory predicts that there is a positive Tenggara) or have a very strong compara- association between economic growth and tive advantage in agriculture (Central Kali- the speed of structural change. We test this mantan, Jambi) or a combination of both by calculating a simple index of structural (Lampung, Bengkulu). change among the agriculture, non-mining Industrialization is the flip side of the industry, and service sectors for each prov- coin: no province had a share of manufactur- ince. The estimates and growth of non- ing in regional GDP in excess of 20 percent mining regional GDP per capita are plotted in 1975. By 2004, seven provinces registered in figure 8.1. shares greater than 20 percent: the three big There appears to be quite a weak correla- Java provinces dominated, particularly West tion between growth and structural change. Java with 43 percent. Off Java, the higher The fastest structural change has occurred in shares are found in Riau, owing principally a diverse group of provinces: East Kaliman- to Singapore industrial spillover, the two tan (reflecting the resource boom and spill- Sumatran provinces with large agricultural overs), West Java (rising industrialization), or industrial processing sectors (North and Riau (resource boom plus Singapore-related South Sumatra), and East Kalimantan with industrialization), Maluku, Bali (tourism its timber processing and oil-related fertil- growth), and Central Java. Structural change izer and heavy industries. There has been has been relatively slow in many of the agri- only one significant case of "deindustrializa- cultural provinces (Kalimantan, Sulawesi, tion," in Jakarta, where the manufacturing and Sumatra), reflecting the slow movement share is a little over half the 1985 figure, as out of this sector in many of them. It is sur- factories have migrated across the border to prisingly low in Jakarta, presumably because West Java­Banten. the classification is too aggregated to pick up There has also been a general increase in many of the new service sector activities. the services sector share. In 1975 there were Demographic dynamics just two provinces in which services contrib- We are interested to know how closely Indo- uted at least half of regional GDP. By 2004, nesia's regional demographics correlate with five provinces were in this group, and sev- these economic changes.9 The country's eral more were close to it. Only resource- demographics reflect the interplay of four rich Riau, Papua, and East Kalimantan main factors: highly uneven "initial condi- recorded a share below 25 percent of GDP. tions"(in the pattern of spatial settlements); A high or increased share of the services the uneven location of opportunities for sector occurred in a variety of development employment, economic advancement, and contexts. Land-scarce Jakarta has always education, which in turn triggers migration; had the highest service sector share, as the official migration policy (a factor especially seat of national government, the provider in the period of 1970­85); and the speed of of high-value commercial services, and the the demographic transition toward low fer- national transport and communication hub. tility and mortality. Economic geography of Indonesia: location, connectivity, and resources 125 Table 8.4 highlights these patterns over Figure 8.1 Structural change and growth in Indonesia, 1975­2004 the period 1971­2000. First, the population 7 is heavily concentrated on Java-Bali, though Bali becoming less so, especially outside Jakarta­ 6 West Java. Sumatra and Kalimantan have (%) been gaining most of the declining Java-Bali 5 population share,while the share of Sulawesi growth Jakarta W.Jav E.Kal and Eastern Indonesia (excluding Papua) 4 has been constant over the three decades. capita per Provincial population growth is a combi- 3 S.Sum nation of natural increase and net migration. GDP Papua Riau No recent decomposition of these two ele- 2 ments is available, and thus low population growth could be the result of a very rapid non-mining 1 decline in fertility, continuing high mor- tality, or out-migration. These factors have 0 very different economic and demographic 0 20 40 60 80 100 120 structural change (3 sectors) implications. However, the percentage of the population born outside the province Source: Authors' calculations. Note: Structural change = i sharei,2004 ­ sharei,1975; i = {agriculture, non-mining industry, services}. C.Jav is Central gives a reasonably accurate indication of the Java. E.Kal is East Kalimantan. Jkt is Jakarta. Mlk is Maluku. S.Sum is South Sumatra. W.Jav is West Java. extent of in-migration (see table 8.5). The major magnets are those provinces most provinces were in either the "above- that offer opportunities for socioeconomic average growth and below-average income" advancement. Thus they tend to be the category or the converse, suggesting that richer ones or the frontier regions. Jakarta interprovincial inequality was declining is quintessentially a migrant city, as it always over this period. Many were very close to has been (Castles 1989), with by far the the national average growth rate.We shortly highest proportion. There are also very high test this formally with reference to conver- shares in resource-rich, frontier East and gence estimates. Central Kalimantan, almost all provinces When mining is excluded, the story in the southern part of Sumatra (proxim- changes somewhat (figure 8.2, panel B). Two ity to Java and employment opportunities), of the seven provinces with above-average Central and Southeast Sulawesi,and Papua.10 non-mining regional GDP in 1975 also reg- Yogyakarta,a major education center,has the istered above-average growth in 1976­2004. highest figure for Java-Bali outside Jakarta. These were Jakarta and East Kalimantan. Reassuring from the point of view of inter- regional equity, although there are seven Regional inequality and provinces in the below-average income and convergence slow-growth quadrant for the non-mining We first extend this analysis with reference regional GDP series, all but one is close to to the "four-quadrant" story relating initial one or other of the national averages. The (that is, 1975) levels of regional GDP per one exception is the special and recent case capita to income growth per capita over of Maluku. In the case of the expenditure the period of 1976­2004 (see figure 8.2, series, six provinces are in the bottom-left panel A). In 1975 only 4 provinces had quadrant--that is, they are poor and appar- above-average income: East Kalimantan, ently slipping behind: Bengkulu,Yogyakarta, Jakarta, Papua, and Riau. Subsequently, West Kalimantan, Maluku, Central Sulawesi, only Jakarta grew at above the national and Southeast Sulawesi (see figure 8.2, panel average. Conversely, of the 22 provinces C). Here, too, most of these are very close to with below-average income in 1975, only one or the other national average. The latter 5--Jambi, South Sumatra,Yogyakarta, Cen- three are farthest inside the quadrant and tral Kalimantan, and Maluku--grew at a therefore are regions of concern from the slower rate than the national average. Thus point of view of regional equity. 126 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 8.4 Social and demographic indicators in Indonesia, by province, 1971 and 2000 Annual growth in population Infant mortality Average schooling Poverty Population (millions) (percent) Province 1971 2000 1971 2000 1984 2002 1971 2000 1971­2000 Sumatra Aceh 143 40 2.3 6.0 14.3 29.8 2.0 4.0 2.4 North Sumatra 121 44 2.7 6.1 22.6 15.8 6.6 11.5 1.9 West Sumatra 152 53 2.6 5.6 23.8 11.6 2.8 4.2 1.5 Riau 146 48 1.8 6.0 29.1 13.6 1.6 4.8 3.7 Jambi 154 53 1.9 5.3 27.7 13.2 1.0 2.4 3.1 South Sumatra 155 53 1.9 5.3 34.1 21.1 3.4 7.8 2.8 Bengkulu 167 53 1.6 5.5 16.7 22.7 0.5 1.6 3.9 Lampung 146 48 1.6 5.1 54.5 24.1 2.8 6.6 3.1 Java-Bali Jakarta 129 25 4.0 8.4 13.7 3.4 4.6 8.3 2.1 West Java 167 59 1.9 5.5 19.4 12.6 21.7 43.8 2.5 Central Java 144 44 1.4 5.0 37.9 23.1 21.9 30.9 1.2 Yogyakarta 102 25 2.3 6.6 30.1 20.1 2.5 3.1 0.8 East Java 120 48 1.6 5.1 29.1 21.9 25.6 34.8 1.1 Bali 130 36 1.4 5.9 34.4 6.9 2.1 3.1 1.4 Kalimantan West Kalimantan 144 57 1.1 4.3 47.0 15.5 2.0 3.7 2.1 Central Kalimantan 129 48 2.3 5.4 29.4 11.9 0.7 1.8 3.3 South Kalimantan 165 70 1.9 5.1 22.4 8.5 1.7 3.0 2.0 East Kalimantan 104 40 2.0 6.3 37.7 12.2 0.7 2.4 4.2 Sulawesi North Sulawesi 114 37 2.9 6.0 26.7 17.4 1.7 3.8 1.7 Central Sulawesi 150 66 2.4 5.3 45.7 24.9 0.9 3.8 2.8 South Sulawesi 161 57 1.9 4.9 24.7 15.9 5.2 1.8 1.4 Southeast Sulawesi 167 53 1.4 4.9 29.1 24.2 0.7 2.1 3.2 Eastern Indonesia West Nusa Tenggara 221 89 1.0 3.9 53.8 27.8 2.2 3.8 1.9 East Nusa Tenggara 154 57 1.9 4.0 52.9 30.7 2.3 3.8 1.7 Maluku 143 66 2.7 5.6 31.7 26.6 1.1 1.8 1.8 Papua 86a 57 4.2a 4.3 27.2 41.8 0.9 2.1 2.9 Indonesia 145 47 1.9 5.4 29.5 18.2 119.3 203.9 1.9 Coefficient of variation 0.184 0.262 0.357 0.164 0.362 0.439 1.490 1.375 0.386 Source: Central Board of Statistics (various years). Note: Infant mortality rate is defined as the number of deaths of infants (one year of age or younger) per 1,000 live births. Average schooling year is the average schooling year among those above 10 years old. Poverty is the percentage of poor people defined by the Central Board of Statistics in the province. a. Urban areas only. Convergence across regions than among countries, prin- We now examine the evidence on inequality cipally because there are fewer barriers to and convergence, with reference to the two mobility in the former and less variation in usual measures, absolute convergence, that policies and institutions. However, much is whether poorer provinces are catching up depends on center-region policies, particu- to richer ones, and convergence, an overall larly concerning fiscal arrangements (Sala- measure of inequality. Furthermore, there I-Martin 1996). are two types of convergence, absolute and convergence is a necessary, but not a conditional.The former refers to the absence sufficient, condition to achieve conver- of any of the control variables presumed gence.That is,the presence of poorer regions likely to influence convergence. In this paper catching up to richer ones is necessary for we focus just on this concept, because an aggregate inequality to decline. But catch-up analysis of conditional convergence entails does not guarantee reduced inequality. For a much larger and more complex exercise. example, the catch-up process may involve Furthermore, growth theory predicts that the once poorer provinces overtaking the absolute convergence is more likely to apply once richer ones; if the margin between Economic geography of Indonesia: location, connectivity, and resources 127 Table 8.5 Indicators of social vulnerability in Indonesia, by province, various years, 1971­2004 CV of growth per capita Non-mining Percent of population Regional regional Gini coefficient Religious diversity born outside region GDP, GDP, CE, Province 1976­2004 1976­2004 1984­2004 1984 2002 1971 2004 1971 2000 Sumatra Aceh 2.4 1.6 1.5 0.26 0.28 97.0 97.3 3.1 5.8 North Sumatra 0.9 0.8 0.7 0.26 0.29 60.3 65.4 8.3 3.9 West Sumatra 0.8 0.8 0.9 0.26 0.29 98.7 97.8 1.0 5.8 Riau 2.7 3.5 1.8 0.26 0.34 83.4 88.6 13.0 32.3 Jambi 1.4 1.4 0.8 0.20 0.27 97.2 96.2 15.9 23.5 South Sumatra 2.1 2.0 1.6 0.27 0.30 94.2 95.8 9.7 13.9 Bengkulu 0.9 1.0 1.3 0.21 0.30 97.4 97.5 7.0 22.7 Lampung 1.4 1.5 1.7 0.29 0.27 94.4 95.6 36.2 22.3 Java-Bali Jakarta 1.2 1.2 1.6 0.29 0.39 84.3 85.7 40.1 42.4 West Java 1.2 1.2 1.9 0.30 0.32 97.8 97.3 1.8 11.5 Central Java 0.9 0.9 1.0 0.31 0.29 96.4 96.8 1.2 2.3 Yogyakarta 1.1 1.1 1.6 0.34 0.41 93.5 91.8 4.1 12.3 East Java 1.3 1.3 0.8 0.31 0.32 96.9 97.1 1.2 2.2 Bali 0.6 0.6 1.2 0.29 0.33 93.3 87.4 1.1 7.0 Kalimantan West Kalimantan 1.0 1.0 1.0 0.25 0.32 42.7 57.6 1.2 7.2 Central Kalimantan 1.4 1.4 2.1 0.29 0.27 54.7 74.1 5.6 23.5 South Kalimantan 0.7 1.0 1.5 0.26 0.30 96.2 97.1 3.9 12.1 East Kalimantan 3.2 1.7 2.1 0.36 0.33 68.4 85.0 7.2 35.0 Sulawesi North Sulawesi 2.1 2.2 2.4 0.35 0.29 48.3 49.8 2.9 6.2 Central Sulawesi 0.9 0.9 1.1 0.30 0.30 72.4 78.4 5.6 18.4 South Sulawesi 0.8 0.9 0.7 0.35 0.30 88.8 89.2 1.4 3.5 Southeast Sulawesi 1.1 0.9 1.2 0.32 0.29 98.0 95.3 3.6 20.7 Eastern Indonesia West Nusa Tenggara 0.9 0.7 1.1 0.30 0.28 99.5 96.6 1.6 2.8 East Nusa Tenggara 0.7 0.7 1.3 0.31 0.29 52.0 53.9 0.6 2.8 Maluku 3.7 3.7 4.7 0.30 0.25 49.9 62.4 4.0 7.5 Papua 4.1 3.1 1.2 0.37 0.38 56.3 59.9 22.5 19.6 Indonesia 1.0 0.9 1.0 0.32 0.35 87.5 88.2 4.9 10.1 Source: Central Board of Statistics (various years). Note: Religious diversity is defined as the percentage of people with the majority religion in the province. them remains the same, convergence has the years 1975­81, coinciding with the oil occurred, but there is no convergence. boom, the absolute convergence was even We report here estimates of provincial higher (2.0 percent) and significant at 5 per- growth rates relative to initial (that is, 1975) cent. In fact, excluding mining, the absolute incomes, that is, whether absolute conver- convergence for the whole period falls to gence is present (see table 8.6). For regional 0.4 percent, and it is insignificant. In the case GDP per capita, we find a coefficient of of household consumption, available only 1.5 percent for the period since 1975, sug- since 1983, the coefficient is also low, 0.2 gesting that the observed disparity would percent, and statistically insignificant. halve over 46 years.11 The results are statis- The pace of convergence varies signifi- tically very significant. cantly across development periods. It was However, these findings are sensitive quite rapid (2 percent) during the oil boom, to the period of analysis, as they are heav- 1975­81, with the coefficient significant at ily influenced by the very high incomes of 5 percent. This is to be expected, with the the resource-rich provinces in 1975 and the oil-rich provinces such as Riau and East declining relativities as the oil and gas sector Kalimantan having high initial income but has become less important. For example, for slower growth over the period. Moreover, 128 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 8.2 Initial regional GDP with and without mining and household consumption expenditure per capita vs. growth in Indonesia, by province, 1975­2004 Panel A: Regional GDP per capita 8 Bali 6 (%) Jkt 2004 4 ­ Ygkt C.Kal Jmb 1976 E.Kal 2 Mlk S.Sum Papua growth 0 annual ­2 Riau ­4 0 2 4 6 8 10 12 level in 1975 (million Rp.) Panel B: Non-mining regional GDP per capita 7 Bali 6 (%) Jkt 5 2004 ­ E.Kal 4 1976 W.N.T Lmp Ygkt 3 Mlk growth Jmb Riau 2 annual 1 0 0 0.5 1 1.5 2 2.5 level in 1975 (million Rp.) Panel C: Household consumption expenditure per capita 5 E. Jav (%) 4 W. Jav Jkt N. Sum 2004 S. Sum E. Kal ­ 3 1984 W. Kal Papua Mlk 2 Bkl growth Bali S. E. Sul Ygkt annual 1 0 0 0.5 1 1.5 2 level in 1983 (million Rp.) Source: Authors' calculations. Note: Blk is Bengkulu. C.Kal is Central Kalimantan. C.Sul is Central Sulawesi. E.Jav is East Java. E.Kal is East Kalimantan. Jkt is Jakarta. Jmb is Jambi. Lmp is Lampung. Mlk is Maluku. N.Sum is North Sumatra. S.Sum is South Sumatra. W.Jav is West Java. W.Kal is West Kalimantan. W.N.T is West Nusa Tenggara. Ygkt is Yogyakarta. Economic geography of Indonesia: location, connectivity, and resources 129 central government grants to the regions and more or less continuously,until the crisis became increasingly important toward the period, after which it slightly increases again. end of this period. The coefficients for non-mining regional The process of convergence accelerated GDP and household expenditure are ini- still further in the wake of the oil boom, tially much lower, less than half the value of with a coefficient of 2.8 percent for 1981­86, the regional GDP series. They remain fairly reflecting the impact of the major policy stable during the 1980s reform period, but reforms. It was also positive, though slower, both begin to rise after the crisis, again only for the other series. As the export-oriented slightly. By 2004, reflecting the declining reforms took hold, the speed of convergence share of the mining sector, the two regional slowed, to 1.7 percent for the period of GDP series had almost converged. 1986­92, and further still during the 1990s, Additional insights are obtained by to just 1 percent. During the crisis period, decomposing the variations in provincial no significant convergence occurred. This income by sector. The results of this analy- may appear surprising, given the widely held sis are not shown here, but the broad sum- presumption that this event particularly mary is as follows. Overall, and as would be affected the country's richer regions, such as expected, regional inequality in agricultural Jakarta. However, it needs to be remembered and services output is much lower than that that some poorer regions were very badly of mining and manufacturing. Regional affected by post-crisis conflicts (for example, inequality in mining is, of course, the high- Maluku) and that some strong agricultural est, owing to the uneven spatial distribution exporters off Java capitalized on the sharp of major mineral deposits. The inequality depreciation of the exchange rate. for the aggregate industrial sector (that is, For convergence, measured as coeffi- mining, manufacturing, construction, and cients of variation,the estimates are similarly utilities) has therefore always been high, highly sensitive to whether the mining sec- although it fell for most of the period, tor is included (see figure 8.3). With mining, reflecting mainly the declining share of min- inequality is high and variable during the oil ing since the late 1970s. Regional inequality boom period. It then declines significantly, in agricultural output rose for most of the period, but this sector's share of GDP fell Table 8.6 Absolute convergence rapidly, hence the increase had little overall Absolute impact. By contrast, inequality in services convergence declined, and this sector's share rose. Time period and proxy of income (-) Regional GDP per capita 1975­2002 0.015*** Figure 8.3 Provincial income inequality in Indonesia, 1975­2003 1975­81 0.020** 1981­86 0.028*** 2 1986­92 0.017*** 1992­97 0.010* 1997­2002 0.007 Non-mining regional GDP per capita 1.5 1975­2002 0.004 1975­81 0.010 variation 1981­86 -0.001 of 1 1986­92 0.008 1997­2002 0.003 0.001 coefficient Expenditure per capita 0.5 1983­2002 0.002 1983­86 0.017** 1986­92 0.007 0 1992­97 -0.018 1975 1979 1983 1987 1991 1995 1999 2003 1997­2002 0.018 Regional GDP per capita Non-mining regional GDP per capita PCE Source: Authors' calculations. *** Significant at 1 percent. ** Significant at 5 percent. Source: Authors' calculations. * Significant at 10 percent. Note: All numbers are calculated at 1993 prices. 130 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Social indicators regional GDP per capita and HDI, albeit We then ask how the economic and social with much clustering close to the averages. indicators correlate with one another. Table Jakarta stands out, with the highest ranking 8.4 provides a summary picture. We include on both measures, while the Nusa Tengga- here a health indicator (infant mortality), ras and Papua are among the lowest. There an education indicator (average years of are several provinces with below-average schooling), and the percentage of the popu- income per capita but above-average HDI. lation below the poverty line. The first two The two major cases are North Sulawesi and are based on the Population Censuses of Yogyakarta, both with traditionally strong 1971 and 2000, while the poverty estimates educational achievement. There are no cases are available only from 1984. of above-average (non-mining) regional Two general points deserve emphasis. GDP per capita but below-average HDI. First, there are dramatic improvements in This suggests that the resource-rich prov- the social indicators: by 2000, infant mor- inces (with the possible exceptions of Papua tality was just a third of the 1971 rate, while and Aceh) have been reasonably successful average years of schooling had risen almost at translating the benefits of the resource threefold. Moreover, these improvements booms into improved social indicators. One have been experienced practically through- qualification that needs to be attached to out the country. Although the rankings have these conclusions is that all of the provincial not changed significantly, in all but one case HDIs prepared thus far include an income or infant mortality rates have at least halved, expenditure variable, typically with a weight and years of schooling have doubled. The of one-third, thus limiting their value as an one exception is Papua, for which the early independent check on economic and social data series are incomplete. For the shorter correlates. time series of the poverty estimates, also, Social conflict there is broad-based decline. Here too, Papua goes against the trend, partly owing Particularly since the fall of the Soeharto to weaknesses in the data, but also reflect- regime, several regions have experienced ing the unequal nature of Papua's develop- episodes of severe social conflicts that have ment. Aceh is the only other province where significantly interrupted their development poverty increased, owing to the effects of the progress. The most serious incidents have prolonged conflict. occurred inAceh,West and Central Kaliman- Second, coefficients of variation (CVs) tan, Central Sulawesi, Maluku, and Papua. are low, but there is no clear trend. The Only as an indicative exercise, we pres- health and education CVs are very low, ent a number of variables hypothesized to well below those of the regional accounts be likely explanations of interprovincial series. They reflect the fact that, as with variations in conflict. A number of these are intercountry comparisons, interprovincial interrelated, and therefore any quantitative social inequalities are lower than economic approach would need to deal with the prob- inequality. The poverty CV is higher, which lem of multicollinearity. is to be expected because it is generated from The first indicator is the volatility of pro- the consumption expenditure estimates. vincial growth rates, defined as the coeffi- There is a slight increase in the poverty and cient of variation through a certain period health CVs and a fall in the education CVs. of time (table 8.5). Its inclusion is based on These trends are to be expected and indicate, the premise that higher variations in growth in particular, the government's emphasis on rates will lead to heightened insecurity and universal mass primary and lower secondary possiblyconflict.Theseareshownincolumns education since the 1970s. 1­3 for each of the three series. As would be There are now several estimates of the expected, the resource-rich regions experi- human development index (HDI) for Indo- ence more volatile growth, with the CVs of nesian provinces (UNDP 2004).They are not Aceh, Riau, East Kalimantan, Maluku, and presented here, but they show the expected Papua at least double, and the CV of Papua positive relationship between non-mining four times, the national average. The high Economic geography of Indonesia: location, connectivity, and resources 131 figures for Aceh and Papua lend prima facie and religious diversity. There are cases of support to the hypothesis. an apparently strong association, such as in However, the direction of causality is Maluku and East Nusa Tenggara. Yet there unclear. For example, Maluku was peaceful are more examples where the converse and experienced fairly stable growth until applies. Aceh has one of the highest major- the onset of serious conflict in 1998. In other ity-religion shares and serious conflict. words, this was a case of conflict causing the North Sulawesi is at the opposite end of the volatility of growth, rather than the con- spectrum, with the highest religious diver- verse. A similar observation applies to some sity and little conflict. North Sumatra and extent in the case of Central Kalimantan. some of the Kalimantan provinces are reli- The second indicator is the share of nat- giously mixed but have low recorded conflict ural resources in regional GDP (table 8.1). (but note caveats). There are also instances This is a subnational variant of the "natu- of little religious diversity but considerable ral resource curse": a large natural resource conflict, such as West Nusa Tenggara. sector will result in a more volatile income Papua is a special case in this context. stream (that is, the first factor) and also pos- There are two main sources of spatial sibly exacerbate conflict over the allocation inequality, which together explain the per- of natural resource rents. In 2004 mining ception that the benefits of growth have generated more than one-third of regional been enjoyed primarily by immigrant com- GDP in three of the resource-rich provinces munities. The first is the urban-rural divide. and more than one-quarter in the fourth, Much of this was fueled by the growth of the Aceh.12 High shares are also evident in West provincial capital, Jayapura, the center of the NusaTenggara(of veryrecentorigins),South rapid expansion of the mainly non-Papuan Sumatra, and South Kalimantan. Severe and civil service and major development projects. protracted conflict has occurred in two of These growing centers also attracted many these provinces, Aceh and Papua, again migrants from other provinces in search of lending prima facie support to this hypoth- business opportunities. Poverty in the urban esis. Nevertheless, the other resource-rich areas was quite low in 2004, around 8 per- regions have been relatively peaceful, while cent. By contrast, in rural areas, where the serious conflict has occurred where mining majority of Papuans reside, poverty was still shares are low, for example, Maluku, Central around 50 percent. The second major source Kalimantan, and Southeast Sulawesi. Hence, is the huge Freeport mine, whose impact is the presence of mining enclaves per se is an confined mainly to Timika. insufficient explanation for conflict. Fourth, the percentage of the popula- A third variable relates to ethnic frag- tion born outside the province indicates the mentation, data that we include on the extent of settler arrivals. It too is suggestive grounds that greater ethnic diversity is of the possibility of conflict, as in-migrants alleged by some to hinder the development compete for jobs and access to land and of local cohesion and trust and to heighten sometimes introduce customs at variance the potential for conflict. We lack precise with local traditions (for example, con- estimates of ethnic diversity at the provin- cerning gender relations, diet, and so forth). cial level in Indonesia, but a good proxy for Obviously, this variable is highly correlated it is religious belief, especially as the latter with the share of the natural resource sector. has been a source of tension in some of As would be expected, a high presence of Indonesia's most serious conflicts, such as migrants is found in Jakarta, the resource- in Poso (Central Sulawesi) and Maluku. A rich regions, remote "frontier" regions, and convenient proxy for religious diversity is areas formerly designated by the central the percentage share of the largest religion government as transmigration sites. in each province, with the hypothesis being Here too the evidence for this variable is that the lower the share, the greater the pos- mixed. There are examples where conflict sibility of conflict. and in-migration are significantly corre- There does not appear to be a clear rela- lated, such as Papua, Southeast Sulawesi, tionship between the incidence of conflict Central Kalimantan, and Jakarta. Yet, there 132 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA are also cases of large migrant communi- and social progress have been remarkably ties generally living in harmony (some of even. There has been no significant change the Sumatran provinces and Yogyakarta), in the concentration of economic activity while some of the most serious conflict has across the major island groupings. As with occurred in regions with below-average all the economic variables, this conclusion in-migration, such as Aceh and Maluku. is somewhat sensitive to whether or not the Finally,it might be expected that intrapro- mining sector is included.Excluding mining, vincial inequality in income and expendi- Java's share has risen, mainly at the expense tures would predispose a province to conflict. of Sumatra. That is, all things being equal, high-inequal- More generally,economic activity has con- ity provinces are more likely to experience tinued to cluster around some key regional conflict. We include estimates of provincial economies. Java has remained dominant, expenditure inequality for 1984 (the first year along with Bali, Sumatra, and Kalimantan, they were available) and 2002 to examine as compared to the eastern region (although this proposition. Predictably, above-average Sulawesi has gone from below-average to inequality is found in the resource-rich prov- above-average growth over the two periods). inces, except Aceh. Papua particularly stands Moreover, Greater Jakarta has assumed ever out. There is also high inequality in the two greater prominence in the nation's key eco- most urbanized provinces of Java: Jakarta and nomic agglomeration. Yogyakarta. With the exception of Papua, all Nevertheless, the poorest regions, located the high-inequality provinces have been quite mainly in eastern Indonesia, have generally peaceful. By contrast, inequality is generally performed about as well as the national below average in areas of major conflict,such average. There is no case of a province with as Maluku and Central Sulawesi. Therefore, consistently poor performance for decades, inequality per se does not appear to be a in the sense of being well below the national major explanatory factor. average growth rate, let alone with pro- This discussion highlights the fact that tracted periods of negative growth. the magnitude and determinants of local Second, as a corollary, regional dispari- conflict are complex, interrelated, and not ties are either high and declining or moder- easily amenable to quantitative explanation. ate and stable, depending on which series is The quality of local leadership is a key factor used. The former conclusion is based on the and thought to be one of the reasons why with-mining series. However, these provide one of the most religiously diverse prov- a misleading indicator of local-level welfare inces, North Sulawesi, has been largely free and should be interpreted with caution. of conflict. In the case of Aceh, one of the The other two series--that is, non-mining most conflict-prone provinces, the conflict regional GDP and household consumption has been principally between the central expenditure per capita--suggest no signifi- government and the very strong local iden- cant change in inequality or catch-up dur- tity, which, when mismanaged, has spawned ing both the 1980s reforms and the crisis a separatist movement. It required a terrible periods. Over the entire period, there was natural disaster (the December 2004 tsu- no convergence in non-mining regional nami), presidential leadership, and a local GDP per capita, while household expendi- capacity to negotiate to reach the 2005 peace ture per capita showed weak convergence. settlement. Similarly, the protracted conflict It is also notable that the policy reform in Papua reflects its complex history and a period of 1984­96 produced an even record troubled record of central government and of provincial economic performance, as military intervention. compared to the mining boom, crisis, and post-crisis periods, when major exogenous Conclusions events had uneven subnational impacts. Our major conclusions include the Third, while there have been strong following. performers--notably, Bali, Jakarta, and First, there continues to be great diversity occasionally East Kalimantan and Riau-- ineconomicandsocialoutcomes,butgrowth the group of top performers has been quite Economic geography of Indonesia: location, connectivity, and resources 133 diverse, as to location, size, and socioeco- University; Budy P. Resosudarmo is a fellow nomic characteristics. In general, the better- in the Ardnt-Corden Division of Economics, performing regions are typically those that Research School of Pacific and Asian Studies at are the most connected to the global econ- the National University; and Yogi Vidyattama is a lecturer in the Faculty of Economics at the Uni- omy. In this respect, Jakarta stands out as a versity of Indonesia. The authors thank Chris special case, growing richer than the rest of Manning and seminar participants at the Austra- the country over time. lian National University and at the World Bank, Although two of the strongest performers Tokyo, for helpful comments on an earlier draft areresource-richregions,thereisnoclearnat- and Terry Hull for advice on the demographic ural resource story, in that the performance and social database. of this group of provinces has varied consid- 1. Thus, for example, West Java refers to erably. The impact of enclave-style develop- the existing provinces of West Java and Banten, ment has also varied among them, with the North Sulawesi to North Sulawesi and Goron- most challenging being the special case of talo, and so on. Papua. Moreover, it is evident that conflict 2. The revenues of all kabupaten and kota governments in the province have increased by at is particularly harmful to economic devel- least 300 percent since the 2001 decentralization. opment, as illustrated in the case of Maluku 3. In 2002, following the introduction of since 1997 and to a lesser extent Aceh. special autonomy measures, the budget of the Future research in this area might focus Papuan provincial government was three times on two areas. The first is an examination of that of 1999­2000 in nominal terms. the impact of decentralization on regional 4. In 1997­98, the economies of Jakarta and dynamics. This will need to be a longer-term West Java contracted by about 50 percent more project because, as illustrated by the experi- than the economy as a whole. This was explained ence in the Philippines and elsewhere, it will mainly by the effects in finance, construction, take at least a decade to discern impacts. and import-substituting manufacturing, all dis- Second, the fragmentation (pemekaran) of proportionately important in these two prov- inces (Akita and Alisjahbana 2002). administrative boundaries is greatly com- 5. But note that Bali's position has slipped plicating longitudinal analysis. This paper significantly since the 1990s, mainly due to the has consolidated the current 34 provinces downturn in international tourism following the back to 26, to facilitate comparisons over terrorist incidents. time. Even this process is a laborious one. It 6. After having one of the highest per capita is currently not possible to draw inferences incomes in the country, this province has slipped at the kabupaten level, the administrative more than most in this group. Part of the expla- unit to which authority and resources have nation is that it was one of the first oil-refining been decentralized, because the process of centers in the country, with Pertamina's Musi boundary changes has proceeded much fur- plant. However, this large sector of its economy ther. However, it may be possible to develop has grown slowly since the 1970s, and, unlike Riau, new growth engines have yet to emerge, such a database with the cooperation of apart from palm oil. Indonesia's Central Board of Statistics.With 7. The very high growth rates of small prov- a finer level of disaggregation, it would be inces like Southeast Sulawesi in the earlier period possible to examine the development of need to be interpreted with great caution. The regional clusters in more detail, because statistical infrastructure was still rudimentary, these invariably straddle provincial bound- and the transition from subsistence to a mon- aries. It may also be the case that our main etary economy may have inflated measured conclusions, of no major change in inter- growth rates. regional inequality and no major dropouts 8. For an economic survey of the province (apart from Maluku in recent years), would since the crisis, see Sondakh and Jones (2003), have to be modified. which extends their earlier work on this province in Hill (1989). 9. See also Jones and Hull (1997). Notes 10. The special case of Lampung deserves Hal Hill is a professor in the Ardnt-Corden Divi- note. It was traditionally designated as a major sion of Economics, Research School of Pacific transmigrant-recipient region and in 1971 had and Asian Studies at the Australian National by far the highest share of migrants outside 134 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Jakarta (Bakir and Humaidi 1989). However, its since 1970. Singapore: Oxford University slower growth, combined with the emergence of Press. other more attractive destinations and the lower Jones, Gavin W., and Terence H. Hull, eds. 1997. cost of movement, meant that by 2000 it had Indonesia Assessment: Population and Human slipped to seventh ranking in terms of the pro- Resources. Singapore: Institute of Southeast portion born outside the province. Asian Studies. 11. See also the work by Garcia Garcia and Soelistianingsih (1998). Keyfitz, Nathan. 1965."Indonesian Population 12. The high share of mining in West Nusa and the European Industrial Revolution." Tenggara is of recent origins and dates from the Asian Survey 5 (10): 503­14. establishment of the sometimes controversial Krugman, Paul. 1991. Geography and Trade. Newmont copper and gold mine on Sumbawa. Cambridge, MA: MIT Press. The share of mining in the province's GDP rose Manning, Chris, and Michael Rumbiak. 1989. from 4 percent in 1999 to 28 percent in 2000. "Irian Jaya: Economic Change, Migrants, and Indigenous Welfare." In Unity and Diversity: References Regional Economic Development in Indonesia Akita, Takahiro, and Armida S. Alisjah- since 1970, ed. Hal Hill, pp. 77­106. bana. 2002."Regional Income Inequality Singapore: Oxford University Press. in Indonesia and the Initial Impact of the Nashihin, Muhammad. 2007. Poverty Incidence Economic Crisis." Bulletin of Indonesian in Indonesia, 1987­2002: A Utility-Consistent Economic Studies 38 (2): 201­222. Approach Based on a New Survey of Regional Arndt, Heinz W. 1973."Regional Income Prices. PhD dissertation, Australian National Estimates." Bulletin of Indonesian Economic University, Canberra. Studies 9 (3): 87­102. Neary, J. Peter. 2001."Of Hype and Hyperbola: Bakir, Siti Z., and Mochtar Humaidi. 1989. Introducing the New Economic Geography." "Lampung: Spontaneous Transmigration." Journal of Economic Literature 39 (2): 536­61. In Unity and Diversity: Regional Economic Pangestu, Marie. 1989."East Kalimantan: Development in Indonesia since 1970, ed. Beyond the Timber and Oil Boom." In Hal Hill, pp. 349­62. Singapore: Oxford Unity and Diversity: Regional Economic University Press. Development in Indonesia since 1970, ed. Balisacan, Arsenio M., and Hal Hill, eds. 2007. Hal Hill, pp. 151­75. Singapore: Oxford The Dynamics of Regional Development: The University Press. Philippines in East Asia. Cheltenham: Edward Resosudarmo, Budy P., and Yogi Vidyattama. Elgar. 2007."The East Asian Experience: Indonesia." Barro, Robert J., and Xavier Sala-I-Martin. 1991. In The Dynamics of Regional Development: "Convergence across States and Regions." The Philippines in East Asia, ed. Arsenio M. Brookings Papers on Economic Activity 1: Balisacan and Hal Hill, pp. 123­55. Chelten- 107­58. ham: Edward Elgar. Castles, Lance. 1989."Jakarta: The Growing Sachs, Jeffrey D., and Andrew M. Warner. 2001. Centre." In Unity and Diversity: Regional "The Curse of Natural Resources." European Economic Development in Indonesia since Economic Review 45 (4­6): 827­38. 1970, ed. Hal Hill, pp. 233­53. Singapore: Sala-I-Martin, Xavier. 1996."The Classical Oxford University Press. Approach to Convergence Analysis." Economic Central Board of Statistics. Various years. Statis- Journal 106 (437): 1019­36. tical Year Book of Indonesia. Jakarta: Central Sondakh, Lucky, and Gavin Jones. 2003."An Board of Statistics. Economic Survey of Northern Sulawesi: Garcia Garcia, Jorge, and Lana Soelistianingsih. Turning Weaknesses into Strengths under 1998."Why Do Differences in Provincial Regional Autonomy." Bulletin of Indonesian Incomes Persist in Indonesia?" Bulletin of Economic Studies 39 (3): 273­302. Indonesian Economic Studies 34 (1): 95­120. UNDP (United Nations Development Pro- Hill, Hal. ed. 1989. Unity and Diversity: gramme). 2004. Indonesian Human Develop- Regional Economic Development in Indonesia ment Report 2004. Jakarta: UNDP. Spatial considerations on decentralization and economies of concentration in Indonesia Francisco Javier Arze del Granado 9 Urbanization and industrial agglomeration local governments with adequate incentives are two fundamental characteristics in the for development and to ensure a healthy transition of countries toward greater levels degree of equalization across local govern- of development. Both have to do with the ments by addressing vertical and horizontal concept of concentration, the former with imbalances. regard to the concentration of people and Despite the natural association among the latter with regard to the concentration the concepts of fiscal decentralization, of economic activity. While the concept of urbanization, and industrial agglomeration, c h a p t e r "urban" is broad and requires that a dis- there is scant understanding of how these trict's population also has access to a basket issues interrelate with one another and with of basic services, a necessary first condition other geographic dimensions. Two ques- is that the district scores above a benchmark tions are of particular importance: What is of population density. Concentration of the expected effect of fiscal decentralization production (or economic density) is driven on patterns of urbanization and industrial by agglomeration economies, which result concentration in developing countries? from a self-reinforcing interaction among What is the expected effect of deconcentra- (a) increasing returns at the plant level, tion on regional economic growth through leading producers to concentrate; (b) trans- its impact on the patterns of urban and portation costs, leading the concentration to industrial concentration? The second ques- occur close to large markets; and (c) factor tion is more important for policy making mobility, making large markets even larger and flows naturally from the first. as producers and labor relocate to them This study focuses on Indonesia, a large (Krugman 2000). Concentration of pro- middle-income country economically and duction and consumption generates benefits demographically concentrated on the island but also costs resulting from congestion and of Java, but experiencing strong decentral- increased interregional disparity. Govern- ization since 2001. It addresses these ques- ments intervene in these market-led levels tions from a spatial economics point of view, of urbanization and agglomeration by pro- by examining the ways in which the spatial moting deconcentration through legislation distribution of districts affects their eco- and regulations as well as through economic nomic development. In particular, regard- policy making in the fiscal and monetary ing urbanization, it investigates whether arenas. Fiscal decentralization has been, distance between a leading and a lagging perhaps, the most widespread and profound district has an effect on the migration of type of deconcentration of the government citizens toward the larger district. Second, structure in developing countries over the it explores whether public expenditures in a past two decades. Decentralization pro- district would generate further in-migration motes fiscal structures that aim to provide to leading regions (or lesser out-migration 135 136 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA from lagging regions). Finally, based on the The distinction between urban and rural role of the spatial characteristics examined, areas has declined in most regions due to a this study speculates on the likely results of large pattern of temporary, seasonal, and dispersing leading regions geographically. weekly commuting of rural residents to urban In regard to economic concentration, this or periurban regions, where they work in the study examines first whether the spatial dis- informal sector, transportation, and minor tribution of districts is increasingly driven trade. Moreover, small and medium cities are by agglomeration economies in industries simultaneously urban and rural,with farming at the two-digit level (where employment activities in close proximity to modern build- growth is used as the measure of growth).1 ings and housing. This pattern, which has Second, it examines whether a district's pub- been referred to as kotadesasi--kota (city) and lic expenditures on infrastructure or busi- desa (village)--in Indonesian, is a common ness development programs are correlated characteristic of urbanization in Indonesia with industrial growth and, if so, whether (Firman 1997; Firman, Kombaitan, and Pra- decentralization can lead to lower industrial dono 2007; Hugo 1975, 1997; McGee 1992) concentration. Third, it examines whether and in other Asian countries (Hugo 2003b; decentralizing public expenditures can pro- Lin 1994). mote development by improving efficiency Indonesia's urbanization rate has through intergovernmental competition. It increased significantly over the past decades, concludes by discussing the likely implica- from 14.6 percent in 1960 to 42.7 percent in tions of policies such as fiscal decentraliza- 2006 (see table 9.1). Urbanization in Indo- tion for regional economic growth and for nesia results largely from natural population policy making. growth: Indonesia has doubled in popula- The rest of the chapter proceeds as follows. tion since the 1960s (from 94 million in First, it presents an overview of the patterns 1960 to 218 million in 2005) and remains and trends of urbanization in Indonesia, the fourth most populous nation on earth examining the patterns and trends of natural (BPS 1960, 2005b).3 However, the annual urban population growth across "1 million exponential population growth rate has plus" cities, urban fringes, and peripheries, declined sharply from 2.34 percent during the determinants of interdistrict net migra- the period of 1971­80 to 1.61 percent dur- tion, and the magnitude of urban growth ing the period of 2000­05. The decrease in due to reclassification. This is followed by a the population growth rate is, at least par- discussion of the dynamics and geographic tially, a result of the increasing tendency to patterns of agglomeration economies that postpone marriage, growing awareness and tests the existence of urbanization and local- effective use of modern contraceptives, and ization economies and examines the pat- unsatisfactory record in reducing mortality. terns of spatial interdependence in regional Urban population growth reached its peak employment growth. It then examines the in the 1980s, accompanied by a significantly role of government policy making and inter- lower rate of growth of the rural population. vention by examining the impact of public While the absolute percentage of urban pop- expenditures at the subnational level on ulation increased in the 1990s, the growth industrial concentration and intergovern- rate of the urban population declined, as did mental competition for mobile factors of that of the rural population. production. A final section concludes. The pattern of rapid urbanization in Indonesia is apace with the average of other Decentralization of countries in the East Asia and Pacific region, but lower than the average of lower middle- expenditures,urban-to-rural income countries generally (a category in migration,and urban density in which it falls according to gross domestic congested regions product [GDP] per capita). Although the From a demographic perspective, urbaniza- direction of causality is difficult to deter- tion is typically measured by the urban popu- mine, urbanization is highly correlated lation's proportion of the total population.2 with a country's level of socioeconomic Spatial considerations on decentralization and economies of concentration in Indonesia 137 Table 9.1 Urban and rural population and population growth rate in Indonesia, 1960­2005 Population (thousands) Annual exponential rate of growtha Urbanization Year Urban Rural rate Urban Rural Indonesia tb 1960 -- -- 14.6 -- -- -- -- 1971 20,568 99,712 17.1 -- -- -- -- 1980 32,845 113,930 22.4 4.68 1.33 2.34 9 1990 55,433 123,811 30.9 5.23 0.83 2.00 10 1995 71,657 123,143 36.8 5.13 -0.11 1.66 5 2000 85,380 115,961 42.4 4.47 -0.68 1.20 9.66 2005 92,919 124,452 42.7 1.69 1.41 1.61 5 Source: Author's calculations based on BPS (2005a, 2005b). -- Not available. a. The exponential formula is r = ln(P0/P1) * (100/t). b. Intercensal period, t years. Rates for 2000 are from October 31, 1990, to July 1, 2000. The figures for population growth differ from official government projections for 2005, as they are based on actual data as opposed to projections. development. In this regard, a review of East factor--remained constant over the 1990­ Asian economies estimates that the elasticity 2000 period. World Bank (2003) estimates of economic growth to urbanization is about that 30­35 percent of urban growth in the 2.71 (Iimi 2005).Whether larger populations 1990s was due to reclassification of rural to (and larger markets) generate economic urban areas. Yet urban growth due to lat- activity or mobile capital is driven to larger eral extension increased further in 2000, as labor markets (and so lower wages), retain- 7 percent of Indonesia's more than 65,000 ing a region's population by preventing out- villages were recoded from rural to urban. migration seems to be desirable if the goal is As much as 48 percent of urban population to promote economic development. growth between 1999 and 2000 was due to The implications of rapid urbaniza- reclassification (BPS various years). tion for development are particular to the underlying driving factors. When urban Distance and factor mobility growth is caused by natural population The concentration of population in a few growth and rural-to-urban migration, it metropolitan areas has gone beyond the leads to higher levels of urban density. In capacity of current factor markets and contrast, urban growth caused by reclas- infrastructure facilities, and so it generates sification reflects a lateral extension of the congestion costs. Rural-to-urban migration urban limits due to changes in the labor can affect urban growth in positive or nega- market structure of previously rural villages tive ways, depending on the degree to which toward nonagricultural jobs or the origina- population flows are monitored and whether tion of new industrial areas. According to they are planned in ways that promote an ESCAP­UN (1993) and Firman (2004), adequate distribution of urban centers and natural population growth accounted for minimize congestion costs. Urbanization is only around 35 percent of urban popula- not necessarily accompanied by economic tion growth from 1980 to 1985, whereas the growth, and in fact concentration of urban- remaining 65 percent resulted from migra- ization (in the form of urban primacy) can tion and reclassification (lateral extension). generate negative growth (Ades and Glaeser Gardiner (1997) asserts that the proportion 1995; Williamson 1965). Natural popula- of urban growth due to natural population tion growth in rural areas transforms rural growth, reclassification of rural regions as regions into small urban centers or hinter- urban, and net rural-to-urban migration lands that do not share the main character- over the period of 1980­90 was 35, 30, and istics of large metropolitan regions and often 35, respectively. As Gardiner and Oey-Gar- digress to slums,characterized by the absence diner (1991) report, the number of villages of infrastructure and sanitation services,such that changed from rural desas to urban areas as electricity, tap water, sewerage, and hous- almost doubled from 1980 to 1990. This ing, and the presence of air and water pollu- distribution--roughly one-third for each tion, crime, and heavy traffic (Iimi 2005). 138 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA In 2005, 32 million Indonesians (14 growth--is explained by a significant propor- percent of the total population) lived out- tion of large districts' own population that is side their district of origin. The pattern of driven to the peripheries mainly by the costs lifetime net migration in Indonesia reveals of congestion. Results from an econometric that large areas in Java and Bali, the eastern estimation of the determinants of net migra- and southern regions in Sulawesi, and some tion across districts shed light on the dynamic landlocked districts in Sumatra lost the most between pull forces (urban gravity) from dis- population (see figure 9.1). Strong dynamics tricts with a higher level of urbanization and of population mobility are led by the attrac- push forces from areas with a lower level of tion that regions with higher levels of urban- urbanization (see the annex for descriptive ization exert on regions with lower levels of statistics and variable definitions). urbanization.4 Urban gravity, resulting from On average, one rural district located in opposing centrifugal and centripetal forces, a peripheral area (that is,one standard devia- decreases with distance. Centrifugal forces tion above the average distance to a higher-tier lead districts proximate to districts with a urban neighbor and one standard deviation higher level of urbanization to benefit from below the average level of urbanization) is their neighbor's economic growth, due to expected to have net out-migration that is as the deconcentration of population and much as 21 percent higher than the average employment. This usually reflects conges- district: 5 percent due to distance-sensitive tion in the leading region and the propensity pull and 16 percent due to push from its of firms to take advantage of the lower costs own urbanization. As expected, the effect of of labor and land in the suburbs. Centrip- distance is nonlinear. Districts close to the etal forces lead rural residences to migrate average level of urbanization and situated to urban centers in search of better access to within a given radius (125 kilometers) of a employment and services. district with a higher level of urbanization While population growth in large cities experience negative net migration. This cir- (1 million plus population) is lower than the cle of "urban gravity"is larger (with a radius national average, there is a strong pattern of of 200 kilometers) for districts with signifi- migration, mostly from rural districts to dis- cantly lower levels of urbanization (one tricts with higher levels of urbanization.What standard deviation below the average). This might seem to be a contradiction--high suggests that the neighbor's pull is reinforced levels of in-migration and slow population by an additional push from the district of Figure 9.1 Net migration of population in Indonesia, by districts, 2005 Population net migration 100,000 to 1,050,000 0 to 100,000 -45,000 to 0 -145,000 to -45,000 -450,000 to -145,000 Source: Author's calculations based on net migration (in-migration minus out-migration) data from the Supas (BPS 2005b). Spatial considerations on decentralization and economies of concentration in Indonesia 139 origin. On the contrary, districts within 125 highest level of urbanization is that of Medan to 600 kilometers of a larger neighbor tend and Jakarta metropolitan area, illustrated in to gain population. figure 9.3. While the results of the migration Figure 9.2 presents a simulation of model predict that cities closer to larger cit- expected net migration patterns at increas- ies lose population (attracted by their strong ing levels of distance between a district urban gravity),the results reverse at the high- and its nearest neighbor at a higher tier of est level (as shown by the negative sign on the urbanization. Urban-gravity pull decays as fringe interaction term in table 9.2). That is, distance between two districts increases until net migration reaches a positive maximum. Distances to the nearest higher-tier neigh- Figure 9.2 Simulation: Urban gravity at alternative distances and levels of urbanization bor beyond 600 kilometers, which could be 60 considered the limit after which a district falls into the category of "very remote area," 40 are associated again with negative values of net migration. Suburbanization determines 20 that, within the highest level of urbaniza- thousands) tion, fringes gain instead of lose population inmigration (in 0 due to congestion costs and high-quality amenities and housing in suburban enclaves. ­20 This is reflected in the negative value of the migration outmigration coefficient for an interaction term defined net ­40 as the product of the variable for distance to the nearest higher-tier urban neighbor and ­60 a dummy variable for fringe areas of "1 mil- 10 20 30 40 50 60 70 80 90 100 125 150 175 200 300 400 500 600 lion plus cities" (see table 9.2, column 3). kilometers to a higher-tier urban district One of the clearest examples of conges- Net migration at avg. urbanization Net migration at low urbanization tion costs experienced by districts at the Source: Author's calculations. Table 9.2 Determinants of districts' net migration dependent variable: net migration (in-migration minus out-migration), thousands Explanatory variable (1) (2) (3) Distance to higher-tier-urbanization neighbor (gravity pull) 0.153* 0.443*** 0.09 (3.04) (3.37) (1.75) Distance to higher-tier districts squared -0.001** (-2.32) Incremental distance to urban level four 0.07 (0.27) Incremental distance to urban level five 0.1 (1.64) Urbanization 103.35 -182.76** 102.9*** (4.36) (-2.17) (4.08) Urbanization square 293.15*** (3.37) Dummy "1 million plus" cities -46.6** (-2.45) Fringe * Distance to higher-tier-urbanization neighbor -4.1 e-3 (-0.07) Constant 47.80 -22.38 -36.5** (4.65) (-1.48) (-3.06) Number of observations 371 371 371 R2 0.07 0.11 0.10 Note: Numbers in parentheses are t-statistics. *** Significance at the 1 percent level. ** Significance at the 5 percent level. * Significance at the 10 percent level. 140 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 9.3 Out-migration from large metropolitan districts to the fringes in Indonesia Population net migration 100,000 to 1,050,000 0 to 100,000 -45,000 to 0 Medan City -145,000 to -45,000 -450,000 to -145,000 Central Jakarta Singapore Bandung City Source: Author's calculations based on data in the Supas (BPS 2005b). fringe areas around the district of Central to total nonmigrant population.9 Table 9.3 Jakarta, consisting of the remaining districts presents the results of three models aimed of province DKI Jakarta (East, West, North, at estimating the effects of public expen- and South Jakarta) and the districts of Bogor, ditures on net lifetime migration.10 These Tangerang, Depok, and Bekasi (abbreviated specifications include the following regres- usually as BOTADEBEK),experienced signif- sors: exogenous fiscal variables from the dis- icant positive net migration (gained popula- trict of origin, which measure the fiscal push tion) instead of losing population, while the (or analogous fiscal pull from the district of district of Central Jakarta itself experienced origin when its level of expenditure is high negative net migration (-330,000).5 Nega- enough to keep its population from leaving); tive net migration in Jakarta metropolitan one variable accounting for the urban-grav- area reflects a push out of villages due to ity pull from larger urban centers (defined as congestion, as the city attracts population the distance to the nearest district in a higher from throughout Indonesia.6 Similar cases of quintile of urbanization); and a group of congestion can be seen in the city of Medan exogenous geographic variables. The group (province of Sumatra Utara) and the district of geographic variables includes dummies for of Bandung (which is now the largest district, the main island,a dummy for isolated islands, with a population of 4.05 million) and in 8 and a variable for landlocked districts. Model out of 10 of the districts with a population 1 presents a linear regression,whereas models greater than 2 million.7 2 and 3 explore further the spatial interde- pendence of net migration patterns by means Migration and regional fiscal of a spatial lag of the dependent variable and structure spatial error estimations.11 Furthermore, An examination of per capita public expendi- all sectoral public expenditure variables are tures after controlling for the forces of urban highly correlated with one another, and so gravity suggests that public expenditures on they have not been used simultaneously in social services and infrastructure have an the same specification,but rather in indepen- effectonhouseholdmigration.8 Publicexpen- dent equations. ditures are expressed as the log of the ratio of Results from these estimations suggest a four-year average of nominal expenditures that per capita public expenditures on Spatial considerations on decentralization and economies of concentration in Indonesia 141 Table 9.3 Population dynamics: public expenditures and net migration in Indonesia dependent variable: net migration Ordinary least squares Spatial lag Spatial error model (1) (2) (3) Variable Education Health Infrastructure Education Health Infrastructure Education Health Infrastructure Distance to district of highest- 0.07 0.07 0.07 tier urbanization (gravity pull) (4.91) (4.97) ( 5.00) Log sectoral expendituresa 1.61 2.52 3.31 1.76 2.29 2.97 1.30 1.74 2.63 (2.13) (3.33) (3.31) (1.58) (1.98) (4.3) (1.2) (1.56) (3.87) Isolated -6.53 -7.01 -5.59 -7.39 -7.76 -6.26 -8.20 -8.39 -7.25 (-0.98) (-1.06) (-0.83) (-1.45) (-1.52) (-1.25) (-1.52) (-1.56) (-1.37) Landlocked -10.71 -10.80 -10.38 -11.71 -11.81 -11.49 -9.84 -9.94 -9.84 (-3.42) (-3.45) (-3.47) (-3.65) (-3.69) (-3.66) (-3.04) (-3.07) (-3.09) Island Sumatra 13.01 12.82 11.45 8.66 8.71 7.87 14.31 14.64 16.52 (3.30) (3.29) (2.89) (2.19) (2.21) (2.02) (1.4) (1.45) (1.76) Island Kalimantan 15.11 13.89 12.31 10.31 9.66 8.94 24.53 24.27 22.50 (3.17) (2.93) (2.81) (1.97) (1.84) (1.74) (1.87) (1.87) (1.86) Island Sulawesi -2.84 -3.28 -3.44 -1.24 -1.39 -1.32 -0.81 -0.46 0.91 (-0.74) (-0.85) (-0.91) (-0.26) (-0.3) (-0.29) (-0.06) (-0.03) (0.07) Island Nusa Tenggara and -7.62 -8.51 -9.42 -3.65 -4.29 -4.90 -4.54 -4.71 -4.18 Maluku (-1.85) (-2.04) (-2.24) (-0.66) (-0.78) (-0.9) (-0.37) (-0.39) (-0.37) Island Papua -0.83 -3.79 -6.73 18.37 16.39 14.63 78.71 76.90 72.02 (-0.10) (-0.48) (-0.85) (2.08) (1.83) (1.68) (3.01) (2.93) (2.79) Constant -19.32 -25.97 -35.23 -16.84 -19.27 -27.35 -11.20 -13.90 -25.45 (-2.09) (-3.33) (-3.35) (-1.24) (-1.59) (-3.49) (-0.72) (-0.96) (-2.34) (spatial-lag coefficient) 0.57 0.56 0.53 (4.99) (4.89) (4.53) (spatial autoregressive 0.77 0.76 0.74 parameter) (7.99) (7.8) (6.92) R2 and square correlationb 0.23 0.24 0.28 0.204 0.206 0.235 0.104 0.108 0.143 Number of observations 354 354 354 354 354 354 354 354 354 Source: Author's calculations. Note: Numbers in parentheses are t-statistics. a. Four-year average. b. Squared correlation is reported instead of R2 in models 2 and 3. All spatial lag and spatial error regression models were computed by maximum likelihood using STATA ml routine modules (see Pisati 2001). infrastructure and health services are posi- by as much as 55 percent from 2005 to 2007. tively correlated (independently) with net If those resources are channeled toward ser- migration at the district level, whereas vices and infrastructure, decentralization is they provide no evidence of an associa- likely to have a large effect on the pattern of tion between net migration and education urbanization by minimizing the push effect expenditures.12 Fiscal policy--in particu- and encouraging a more homogeneous pat- lar, public expenditures--could be used to tern of urban development. control the push of population from rural to urban areas and to promote faster urban Alternative policy interventions to development in small towns and medium- reduce urban concentration size cities. In particular, the effects of infra- An examination of population at an initial structure spending are larger--and its cor- point in time and subsequent growth at the relation is stronger--than those of health district level confirms the existence of a ten- expenditures. This is of particular impor- dency for population to converge (decon- tance in Indonesia, given what has been centrate), although at a very slow pace (see called a second "big bang" of fiscal decen- table 9.4).13 The change in the pattern of tralization in 2006. This substantial increase urban concentration can also be seen by cat- in subnational resources resulted from an egorizing cities by quintile in terms of their increase in national fiscal space (mainly population size and computing the per- because of a reduction in fuel subsidies), centage that each quintile represents of the increasing subnational revenues, on average, total population. The 20 percent of smallest 142 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA districts increased from only 2.8 percent lation on the island of Java and in Bali and of the population in 1983 to 11 percent in 20 percent on the island of Sumatra. The 2005. On the contrary, districts forming maximum-to-minimum ratio at the island the 20 percent of largest districts decreased and provincial level decreased slightly, sug- from 49 to 44 percent over the same period. gesting that regions with the lowest popu- An alternative natural aggregate level of lation have been catching up with regions observation is the six largest islands of the with the most population (converging). The country (see table 9.5). Population dis- same figure increased at the district level, at tribution in Indonesia, at the island level, least partially, as a result of the prolifera- remained almost constant over the last two tion of new districts, which have some of decades, with 60 percent of the total popu- the smallest populations. Overpopulation and the high degree of population concentration in inner Indonesia Table 9.4 Population convergence (as opposed to the outer islands) have been dependent variable: log of population widely regarded as among the most pressing Variable Coefficient problems facing the nation since the 1970s. Log population 1983 -0.107 (-6.68) As a result, a government resettlement pro- [-9.12] gram,denominated Transmigrasi,which was Constant 1.71 initiated as early as 1904, was strengthened (8.33) by the late 1970s and mid-1980s under the [11.13] New Order government (during the periods Source: Author's calculations. Note: Numbers in parentheses are t-statistics. Numbers in square of the Repelita: five-year economic develop- brackets are t-statistics corrected for spatial correlation (error ment plans III and IV; see table 9.6). These model). programs aimed to intervene in the patterns of population growth and to reduce concen- Table 9.5 Distribution of population in Indonesia, by district, province, and island, 1983 and 2005 tration by resettling the landless population from areas with high population density District level Provincial level Island level (thousand) (million) (million) to those with low population density. The Indicator 1983 2005 1983 2005 1983 2005 number of families relocated decreased Mean 531.7 496.7 4.7 6.9 24.9 36.8 markedly after 1980 due to a decline in avail- Standard deviation 464.5 571.8 7.6 10.0 34.0 49.7 able resources; at that time, the government Minimum 14.0 12.7 0.32 0.64 1.3 2.5 devised a program to encourage"voluntary" Maximum 291.9 4.102 31.1 39.1 91.3 134 migration of people who would receive Share maximum (percent) 1.9 1.9 20.2 17.7 61.3 60.6 none or only partial funding. As a result, the Ratio of maximum to minimum 201 323 95 65 70 53 Number of observations 281 444 31 32 6 6 second half of the 1980s was characterized Source: Computed based on information in the Susenas (BPS various years). by a large increase in the number of spon- taneous (swarkasa) transmigrants. These reached approximately 500,000 families, Table 9.6 Number of people relocated under Indonesia's transmigration program twice the number of sponsored families. In thousand all, between 1969 and 2000, approximately Number of Number of families Number of people 1 million families--or 6.2 million people-- Period target families actually moved actually moved were moved from Java-Bali to the outer Pre-Repelita period, 1950­69 -- 100 500 islands, in particular Sumatra. Repelita perioda I: 1969­74 39 37 174 Similar types of government-financed II: 1974­79 250 118 544 resettlement programs were also practiced III: 1979­84 500 535 2,470 in several other countries in Southeast Asia, IV: 1984­89 750 230 1,062 including Malaysia, the Philippines, Thai- V: 1989­94 550 110b -- land, and Vietnam (Leinbach 1989). It is VI: 1994­99 600 300 1,500 VII: 1999­2000 16 4 22 widely believed that the success of those pro- Total 2,705 1,024 6,271 grams, as well as that of Indonesia's Trans- Source: Sri Adhiati and Bobsien (2001). migrasi program, was, at best, mixed and -- Not available. controversial, because of problems includ- a. Rencana Pembangunan Lima Tahun: five-year economic development plan in Bahasa, Indonesia. b. Approximation: equal to one-fifth of the target number of families, as stated by Hugo (1997). ing inadequate income levels, improper Spatial considerations on decentralization and economies of concentration in Indonesia 143 site selection, poor matching of settlement experiencing the larger rate of growth. Sec- models to the specific sites, environmental ond, outer islands are growing faster than deterioration, difficulties of adjustment, Java and Bali. Third, within Java and Bali, land conflicts, and financing (Fearnside "1 million plus cities"are still growing faster 1997; Hugo 2003a). As highlighted by Hugo than the fringe and peripheries, whereas (1997), however, the transmigration pro- peripheral districts in the outer islands are gram was not the largest component of a growing significantly faster. The first and substantial spontaneous migration from second points suggest a deconcentration of Java to the outer islands. This is shown by population, as both categories in which pop- significant and increasing figures for overall ulation was initially lower--peripheries and lifetime migration out of Java: 2.35 million outer islands--are growing faster than those in 1980 and 2.71 million in 1990. In all, the in which population was initially higher. populations of Java and Bali are growing at The third point reveals that the dynamics of a slower rate than those of the outer islands. deconcentration are driven by growth in the In fact, the annual population growth rate in outer islands and not in Java and Bali, where Java and Bali over the period 2000­05 (1.47 the trend of population growth exhibits fur- percent) was almost half that in the outer ther concentration. islands (2.86; see table 9.7). A more disag- gregated categorization on the basis of the Agglomeration economies initial size of districts and their spatial loca- In Indonesia, economic activity is clearly tion can be set to distinguish among districts becoming more concentrated in large cities. with a population greater than 1 million (in In 2004, 66 districts (15 percent of all dis- 2000), fringe districts (defined as all neigh- tricts) with populations larger than 1 mil- boring districts of districts in the previous lion accounted for 51 percent of total GDP category), and districts in the periphery (excluding oil and gas). From this group of (defined as all districts that do not fall into districts, those classified as cities or kotas (13 either of the two previous categories). Three in total) contributed 26 percent, while "non- results stand out. First, peripheries and city districts"(53 in total) contributed 25 per- fringe districts are growing faster than "1 cent.14 Large districts generate about half of million plus cities," with the former group national GDP, but this does not necessarily reveal the existence of disparities in income per capita, as large districts also account for Table 9.7 Population and annual population growth about half of the population of the country in Indonesia, 2000 and 2005 (49 percent). It is important to recognize that Annual Population even largely populated districts in Indonesia growth Area 2000 2005 (percent) have both urban and rural areas. In particu- All areas lar, as much as 35 percent of employment Java and Bali 124.1 133.5 1.47 in large districts is agricultural. Disparities Outer islands 75.5 87.0 2.86 in productivity become more evident with Indonesia 203.5 220.6 1.99 a more detailed analysis of the patterns of "1 million plus" cities sectoral employment (see table 9.8). In 2004, Java and Bali 82.2 88.7 1.52 45 percent of Indonesians worked in agricul- Outer islands 11.0 11.4 0.81 Indonesia 93.2 100.0 1.43 ture, which accounted for only 17 percent of Fringe areas national GDP (net of oil and gas).On average, Java and Bali 27.3 29.1 1.33 residents of agricultural regions have lower Outer Islands 41.6 46.8 2.2 income per capita than persons working in Indonesia 68.8 76.0 1.98 nonagricultural jobs,whether located in large Periphery areas cities or elsewhere.Moreover,the agricultural Java and Bali 14.6 15.7 1.42 Outer islands 22.9 28.8 3.52 sector in Indonesia (together with wholesale Indonesia 37.5 44.5 3.41 and retail trade, hotels and restaurants, and Source: Author's calculations based on data from BPS (2000, construction) exhibits a low level of relative 2005a, 2005b). concentration, indicating that agricultural a. The populations of all districts in 2005 are aggregated at 2000 boundaries. activity is spread somewhat evenly across 144 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 9.8 Composition and concentration of employment, by sector Composition Average Index of relative concentration (employment) percentage Annual rate of regional Change, Sector Share, 1994 Share, 2004 of growth a GDPb 1994 2004 Rank 2004 1994­2004 1 Agriculture, forestry, and fishery 48.3 45.0 0.1 18.3 0.202 0.235 5 0.032 2 Mining and quarrying 1.1 0.8 -0.6 10.9 0.326 0.430 1 0.104 3 Manufacturing industry 10.5 12.7 1.2 24.6 0.297 0.312 4 0.015 4 Electricity, gas, and water 0.4 0.2 -2.4 1.4 0.372 0.395 3 0.023 5 Construction 3.9 4.5 1.0 5.9 0.233 0.225 6 -0.007 6 Wholesale or retail trade, restaurants, hotels 16.5 18.5 0.9 19.9 0.183 0.182 9 -0.002 7 Transport, storage, and communications 3.6 5.8 2.5 6.4 0.260 0.210 7 -0.049 8 Finance, insurance, real estate, and business 0.7 1.3 3.1 9.9 0.469 0.423 2 -0.046 9 Public services 14.0 11.5 -0.4 9.5 0.217 0.192 8 -0.025 Source: Employment figures are computed using data from the Susenas (BPS various years). a. Of the form: (log employment 1994 ­ log employment 2004) / 10. b. Excluding oil and gas manufacturing. regions as opposed to being concentrated.15 and concentration in further detail, the next Although there is widespread agreement that sections present an econometric analysis of concentration generates economies in some the (a) patterns of regional concentration industries, the large majority of these results by sectors, taking into account the strategic refer to industries within the manufacturing positioning of districts relative to neighbors sector. The manufacturing sector as a whole in higher orders of the city hierarchy and (b) accounts for only 25 percent of GDP in Indo- the magnitude and significance of agglom- nesia, which demonstrates the importance of eration economies. observing the spatial interaction across two- As outlined by the new economic geog- digit classification sectors, including services raphy literature (Fujita 1988; Fujita and and agriculture. Thisse 1996; Venables 1996), the decrease in The dynamics of concentration (lead- transportation costs over time has enabled ing to further divergence or convergence) closer links between suppliers and custom- can be examined roughly by looking at ers and made it easier to ship agricultural changes in the index of relative concentra- goods to industrial and urban centers. At tion. The nonservice sectors have become the extreme, if transportation costs and more concentrated, while the service sec- congestion costs16 are significantly low, tors have become less concentrated. This regional economies can be expected to is reflected in an increase in the sectors of converge to form unique urban centers of mining; agriculture; electricity, gas, and production that benefit from increasing water; and manufacturing and in a decrease returns, that ship in inputs and agricultural in all service sectors (sectors 6 to 9 in table goods, and that distribute products across 9.8). However, this indicator reflects an the nation. Several studies document how "average," which can be biased by outlier improved technologies have led to a pattern observations and may not statistically rep- of decreasing transportation and commu- resent the pattern of concentration in dis- nication costs over time (Cairncross 1997; tricts in Indonesia. Furthermore, the index Glaeser and Kohlhase 2004). Although the of relative concentration provides a view net effects of lower transportation costs, of concentration by sector, but it does not which lead to concentration, reduce the shed light on the geographic patterns of dis- costs of communication, and lead, in turn, tribution and does not take into account the to deconcentration, cannot be determined space in which each municipality is located a priori, an implicit corollary of both of (Ruiz Valenzuela, Moreno-Serrano, and these hypotheses is that interjurisdictional Vaya-Valcarce 2006). To examine the spa- distance would have a smaller effect on the tial dimension of economic development structure of local economies. Spatial considerations on decentralization and economies of concentration in Indonesia 145 Homogeneous growth nearby urban center has a positive effect on Even if it is not possible to fit hundreds of employment growth and that this effect is millions of people into a very few regions growing over time. due to congestion costs,concentrated growth Where decentralization provides addi- can be expected to spill over across an entire tional resources to lagging regions, as is region, except that growth spillovers fade off the case in Indonesia (World Bank 2007), with distance. Arze del Granado and Sumell small and medium-size cities are likely to (2008) examine employment growth over a experience accelerated growth, increas- 10-year period in Indonesia and its relation ing the scope for economic spillovers. The to the spatial location of districts at differ- appropriateness of the fiscal structure in a ent levels of the urban hierarchy.17 Employ- decentralized country depends essentially ment growth in lower-tiered regions can be on its geography. Hence the most direct expected to be statistically lower if the regions effects of decentralization policy should are subject to a"distance penalty"due to fewer be examined in terms of the potential ben- opportunities to commute, less access to efits of deconcentrating economic activity urban amenities and high-ordered services, across subnational territorial units. Decen- fewer opportunities to earn higher salaries, tralization entails the delegation of deci- less access to lower-price goods (achieved in sion-making authority to a large number larger districts through economies of scale), of subnational governments over resources and weaker trade links in general (Par- previously allocated by the federal govern- tridge and others 2008). In other words, by ment. This is often coupled with larger not accruing the benefits of agglomeration amounts of public spending per capita at economies, distant districts may experience the subnational level in targeted sectors such a loss in growth, or districts close to larger as education and health (Arze del Granado, ones may benefit from growth spillovers. In Martinez-Vazquez, and McNab 2005). A addition, the magnitude of the distance pen- direct consequence would be the more alty is likely to be positively associated with homogeneous development of new centers the distance between lower- and higher-tier of economic activity across the national ter- regions.18 The positive economic value of ritory as opposed to the complete central- distance to core centers of development ization in one unique mega urban center. In could include higher congestion costs in turn, this implies a decrease in the average the form of higher levels of crime, taxes, distance between lagging regions and the wages, land prices, traffic congestion, and nearest leading region, which, as mentioned environmental pollution (Glaeser 1997) in the previous section, increases the scope or by the benefits of less competition from for economic spillovers. large urban centers--the new economic Urbanization and diversity geography's agglomeration shadows (Krug- man 1991; Dobkins and Ioannides 2001). A wide array of empirical evidence sug- Arze del Granado and Sumell (2008) find gests that regional economic growth is that employment growth is larger in districts affected by agglomeration economies in geographically proximate to a higher-order the form of urbanization and localization urban center. In particular, the incremental economies. The former is usually proxied distance coefficients indicate that increas- either by the overall size of a local economy ing the distance from a district to the near- or by an indicator of sectoral diversity, est district of a higher tier by one standard whereas the latter reflects sectoral gains deviation decreases its expected employ- from concentration of a specific indus- ment growth by 0.49 percent.19 This is a sig- try. In a recent study of agglomeration nificant figure, considering that the mean economies in Indonesia over the period of annual employment growth of Indonesia's 1994­2004, Arze del Granado and Sumell districts is 1.1 percent. This result is consis- (2008) study the relationship of economic tent with the findings of Partridge and oth- structure and growth, controlling for spa- ers (2006) based on U.S. data at the county tial characteristics such as distance to the level, which show that closer proximity to a nearest urban center and employment 146 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA growth in the surrounding regions (to cor- growth. Combes (2000), in a study covering rect for spatial autocorrelation). This study 341 local areas in France from 1984­93, also concludes that lagging regions seem to be finds negative specialization effects in sec- catching up in terms of overall employ- tors in which employment is growing at the ment growth, led mainly by agriculture national level. This would also be the case in and mining. This is reflected in a negative Indonesia, where employment is growing in coefficient on districts' total employment 7 out of 10 main sectors (exceptions being and employment density in the initial year electricity, mining, and public services). of the period examined. Having controlled Combes attributes this result to a pattern in for sectoral effects--both localization and which sectors "first develop in a few places diversity--employment density reflects and then develop across space." Unlike the potential effect of the size of local mar- results previously found in the United kets, sometimes considered part of urban- States and Europe, Combes's findings can ization economies. It can also capture be explained partly by the fact that France negative externalities such as pollution has significantly lower levels of labor mobil- or high rents for land (see Combes 2000). ity. This is likely also the case in developing Because of transportation cost savings and countries and, in particular, in Indonesia, increasing returns, firms find it profitable which has a territory fragmented into more to locate in large input and output centers. than 13,000 islands, with more than 700 In addition, industries might benefit from living languages spoken across its regions knowledge spillovers across industries due (Gordon 2005). However, these results are to higher communication among people in based on a two-digit level of aggregation cities or due to supply-demand linkages. Yet in which all manufacturing industries are when markets are saturated with an exces- aggregated into one sector, and so they do sive number of suppliers, output prices not shed light on the patterns of industries tend to fall and input prices tend to rise. In within the infrastructure sector. addition, an urban area's capacity to absorb land-intensive industries, such as in agricul- Room for further ture and mining, is physically constrained. decentralization and Therefore, these land-intensive industries seem to be less prone to achieving the ben- policy implications efits of agglomeration economies.20 It is too early to examine the full impact of Arze del Granado and Sumell (2008) decentralization in Indonesia, but it is still find evidence of a pattern counter to that possible to infer some of its likely effects on expected from the presence of localiza- local economic development based on an tion economies. That is, the coefficient of analysis of two additional dimensions: (a) localization or specialization is statistically the extent to which decentralized expen- significant and negative in all sectors. The ditures generate new centers of industrial estimated localization elasticities range from activity and (b) the extent to which decen- -0.1 to -0.01 and are significantly larger tralized public expenditures can enhance for the service sector (from -0.1 to -0.04) efficiency by generating intergovernmen- than for the nonservice sectors, agriculture tal competition. Fiscal decentralization-- being the lowest (-0.01). These results sug- perhaps the most widespread and profound gest that employment in all sectors became type of transformation of the government distributed more equally across districts structure in developing countries over over the period of analysis, as opposed to the past two decades--has significantly a pattern of increased clustering of sectoral transformed Indonesia since 2001 (Alm, employment. Furthermore, these results are Martinez-Vazquez, and Indrawati 2004; consistent with the theoretical conclusions Bahl 1995; Fengler and Zaini 2006; Lewis of Jacobs (1969) and the empirical findings 2005). Public policy at the subnational level of Glaeser and others (1992), among oth- can affect the migration of firms and indi- ers, which suggest that urbanization, and viduals through the regulation of licensing, not localization, contributes to economic zoning, and the provision of various types Spatial considerations on decentralization and economies of concentration in Indonesia 147 of subsidies, tax exemptions, industrial lic expenditures should exhibit a systemic parks, and so forth. The regional allocation pattern of spatial interaction.23 Indonesia is of subnational expenditures across sectors among the most decentralized developing is just one of several factors that affect the countries in the world, with a ratio of sub- mobility of production factors in a country, national to total expenditures of 40 percent but an important one. The determinants of (World Bank 2007). Yet, in 2004, as much as firm location have been widely examined 48 percent of expenditures at the subnational in Indonesia (Deichman and others 2005; level were de facto nondiscretionary.24 The Kuncoro 1994; Kuncoro and Dowling current decentralized structure increased 2007; see also Kuncoro in chapter 10 of this significantly the overall level of expenditures volume), and this topic is not considered transferred from the central government that further here. Instead, this section examines could be used, in principle, to attract mobile the effects of public expenditures on indus- private investments. On the revenue side, trial concentration, defined as the percent- the ability of a district to compete is almost age of manufacturing in total regional GDP. nil, as most important taxes, such as prop- The theoretical justification for this type of erty and income taxes, are administered by empirical model is discussed in Chen, Jin, the central government and subsequently and Lu (2005) and Wen (2004). The main shared with provinces and districts, whereas variables of interest in this section are public the corporate income tax is still under the expenditures on infrastructure and on other complete jurisdiction of the central govern- sectors related to business development.21 ment. The lack of subnational tax autonomy There is no evidence that decentralization in Indonesia is reflected in the distribution of is generating new centers of industrial activ- revenue by source, with own revenue sources ity. In particular, decentralized expenditures accounting for as little as 8.8 percent of total over the period of 2001­04 are not statisti- revenues in 2005 (World Bank 2007).25 cally correlated with the level of industrial There is evidence that fiscal decentraliza- concentration in Indonesia.22 Results from a tion can increase interjurisdictional com- linear regression reveal a significant relation petition in developed countries, but there between the degree of industrial concentra- has been little research on these effects in tion and expenditures on infrastructure, yet the developing world. Arze del Granado, these results lose their significance once a Martinez-Vazquez, and Simatupang (2008) spatial lag is included in the model to correct examine the expenditure patterns of districts for spatial autocorrelation. This indicates in Indonesia, finding evidence of spillover the existence of spatial correlation in the effects in expenditures on administration, sense that other types of spatial characteris- but failing to find a similar effect in expen- tics might be leading to industrial clustering. ditures on other sectors.26 This is of particu- It is still too early to find an effect from local lar interest, as expenditures in the govern- public expenditures over the period exam- ment sector account for the largest share of ined, given that public investments may take subnational expenditures in Indonesia. The several years to generate results. estimated spatial elasticity for subnational An additional channel by which a decen- administrative services is almost twice as tralized government structure can enhance large as that for discretionary total expendi- economic development is by increasing the tures. Some sort of an imitation effect, and efficiency of public spending. Yet districts in not necessarily of the good kind, appears Indonesia seem to possess insufficient fis- to occur among Indonesia's local districts: cal autonomy to generate efficiency gains spending by neighboring districts on local from interregional competition. The cur- buildings, cars, and so on leads to imitation rent extent of competition can be examined by other districts. On the contrary, Arze del in the form of spatial interdependence in Granado, Martinez-Vazquez, and Simatu- discretionary public expenditures (largely pang (2008) find evidence of the existence expenditures other than personnel). If local of yardstick competition (spatial interaction governments compete for mobile labor on the quality of services). This is of par- and capital, the composition of their pub- ticular significance because accountability 148 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA mechanisms in decentralized developing In relation to the effects of government countries may be reinforced through the intervention on industrial concentration, presence of interjurisdictional competition the results of an empirical model on the in terms of local government performance. determinants of industrial concentration The constraints on the expenditures of do not support the existence of a significant subnational governments are now weakening relation between industrial concentration as the amount of intergovernmental transfers and public spending on infrastructure or from the central government is increasing in on "business development." This result is in what has been referred to as Indonesia's sec- line with studies finding that, while natural ond "big bang" (World Bank 2007). Hence advantages and production externalities greater autonomy on the expenditure side affect the location decisions of firms, the may be generating stronger intergovernmen- effects of government intervention on loca- tal competition across districts, but this will tion patterns, through investments on infra- only be shown when new budgetary data structure in lagging regions, are not effec- become available in the coming years. tive (Deichman and others 2005). Wells and Allen (2001) find that special tax incentives Conclusions and tax holidays aimed at developing local This chapter suggests that gravity forces industry are not likely to develop sustainable from districts in higher tiers of the urban growth, as these incentives attract mostly hierarchy pull population from nearby dis- footloose industries that leave soon after the tricts in lower ranks of the hierarchy. This policy is over. pattern reverses for districts at the highest Whether further concentration in Indo- levels of urbanization, which repel popu- nesia is desirable or not is a topic that lation toward their fringe areas due to the requires further debate. The benefits of costs of congestion. Higher urbanization deconcentration in reducing congestion rates do not imply greater urban concentra- costs and regional disparities should be tion. The level of urban concentration in weighed against the potential losses it could existing urban areas does not increase as a generate from agglomeration economies. result of the reclassification of regions from Results from the examination of employ- rural to urban, as this is a lateral extension ment growth across two-digit sectors of the urban boundaries. On the contrary, reveal that higher concentration of overall rural-to-urban household migration does economic activity--as measured by employ- lead to higher urban concentration. An ment density--does not generate produc- examination of rural-to-urban migration tive externalities in most sectors, including patterns in relation to urbanization reveals manufacturing (encompassing all industries that a district's own level of urbanization as a whole) and, in fact, is correlated with and level of public expenditures on social lower rates of growth in total employment.27 services and infrastructure determine its On the contrary, these results suggest that migration gravity. The higher these levels, economic growth is positively associated the lower the outflow of population toward with sectoral diversity (urbanization) and larger districts. If a district's own migration negatively associated with sectoral concen- gravity is sufficiently weak, residents are tration (localization) at the two-digit level pushed outward toward districts with bet- of aggregation. These results do not pertain ter services or employment opportunities. to the dynamics of growth in industries Fiscal decentralization could serve to cur- within the manufacturing sector, which tail rural-to-urban migration, whether this could be benefiting from economies of is desired or not, to the extent that decen- localization or urbanization. Likewise, these tralization increases per capita spending in results should not be interpreted as suggest- sectors such as infrastructure and health ser- ing that Indonesia has reached an intrinsic vices. Moreover, anecdotal evidence on the stage of excessive concentration leading to Transmigrasi program suggests that explicit congestion costs. On the contrary, it is more policy interventions that resettle population likely that Indonesia has not yet received can cause several secondary problems. the benefits that other more developed Spatial considerations on decentralization and economies of concentration in Indonesia 149 countries have accrued from agglomeration Wolfgang Fengler, task team leader, for his inputs economies and that further efforts should and guidance. Also beneficial were discussions be directed to improving the productive with Jorge Martinez-Vazquez, Renata Simatu- environment in ways conducive to achiev- pang, Sebastian Eckardt, Ignacio Navarro, Bam- bang Suharnoko, Neil McCulloch, Elif Yavuz, ing these benefits. Deconcentrating policies and Timothy John Bulman. such as fiscal decentralization, discussed 1. Most studies in the literature focus on the in relation to agglomeration economies, dynamics of sectors within the manufacturing are likely to enhance growth in nonmanu- sector, whereas this study explores the intersec- facturing sectors, to strengthen industries toral dynamics between two-digit sectors,includ- that have settled in leading regions outside ing agriculture, services, and manufacturing. Java, and to further develop some industries 2. The Indonesian Bureau of Statistics (BPS) across the country that do not exhibit strong defines a locality as "urban" if it satisfies three gains from agglomeration (see Deichman conditions: (a) a population density of 5,000 and others 2005; Kuncoro in chapter 10 of people or more per square kilometer, (b) 25 per- this volume; Kuncoro and Dowling 2007 cent or less of the households working in the agricultural sector, and (c) eight or more types for a review of agglomeration economies by of urban infrastructure and facilities. These industry in Indonesia). The potential gains facilities include a primary school or equiva- from developing nonmanufacturing sec- lent, a cinema, a hospital, a maternity hospital tors should not be taken lightly, particularly or mother-child hospital, a primary health care considering that nonmanufacturing sectors center or clinic, a road that can be used by three- account for 75 percent of Indonesian GDP. or four-wheel motorized vehicles, a telephone or Distance matters, and so does the homo- post office agency, a market with buildings or a geneous distribution of economic activity. shopping center, a bank, a factory, a restaurant, This chapter finds that employment growth public electricity, and a party-equipment rental is inversely related to the distance between service (Firman 1997). a district and its nearest higher-order urban 3. Data from Supas 2005 were comple- mented with data from the Aceh Census post- center. The more disperse leading regions tsunami (BPS 2005a, 2005b). (districts of higher urban hierarchies), the 4. Figures for net migration (in-migration lower the average distance from lower- to minus out-migration) and urbanization are higher-ranked districts and, in turn, the drawn from the Intercensal Population Survey higher the expected level of overall employ- (Supas) for 2005, which allows the identifica- ment growth. This suggests that fiscal decen- tion of important patterns of interdistrict rural- tralization, as far as it relates to the spatial urban migration, yet does not allow the study of distribution of growth and growth spillovers, intradistrict migration from rural to urban areas could generate more homogeneous levels of (BPS 2005b). Migration patterns are examined growth across the country. However, these in the economics literature by "gravity models" results do not lead to conclusions on whether that consider the characteristics of origin and destination districts as well the distance ("fric- more homogeneous levels of growth at the tion") between them (Sen and Smith 1995). subnational level translate into higher lev- 5.After a pronounced increase from 100,000 els of national economic growth, which is a in 1900 to approximately 9.1 million in 1995 topic worthy of further research. (Han and Basuki 2001), Jakarta's population decreased to 8.4 million in 2000, as recorded by Notes the census, but increased again to 8.82 in 2005 (BPS 2005b). Francisco Javier Arze del Granado is technical 6. According to the Supas, in 2005 only 68 assistance advisor with the International Mon- percent of the city's population was born in the etary Fund. This study was conducted within city (BPS 2005b). the Poverty Reduction and Economic Man- 7. These districts are Cianjur, Cirebon, agement Unit of the World Bank, Jakarta. The Garut, Jember, Malang, Medan, Sukabumi, and author would like to thank Yudha Permana for Tasikmalaya, with the district of Tangerang and his remarkable research assistance; Sukmawah city of Surabaya being the exceptions. Yuningsih and Fitria Fitrani for their inputs and 8. Expenditures on services such as educa- data; Albert J. Sumell, Yukon Huang, and Ales- tion and health, police, and parks and recreation sandro Magnoli Bocchi for their inputs; and 150 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA tend to improve the welfare of most individuals, 16. Congestion costs in the form of higher and so they can be expected to increase the like- wages and land prices tend to disperse manufac- lihood of a district being chosen as a migrant's turing activities to less-congested regions. destination. Infrastructure expenditures can 17. Districts are categorized into one of five be expected to be positively correlated with net tiers based on their degree of urbanization. See migration as the development of infrastructure the annex for details on the definition of the enhances the productivity of existing firms in urban-tier categories and also of the control several ways (for one, by decreasing transporta- variables used in this study. tion costs) and attracts additional firms to the 18. This refers to a negative-effect "penalty," region, which in turn increases employment as this is the sign found in their empirical analy- opportunities and wages. sis.A positive distance effect could be interpreted 9. This is because net migration is clearly as a"distance benefit"for more regions distant to correlated with total population in a district, and urban centers, as they benefit from lower compe- so any "per capita" measure would be endog- tition with their larger neighbors (Partridge and enous to the model. others 2006). 10. The 2005 Supas provides data on lifetime 19. The coefficient for distance against migration, which are used here to construct total employment as the dependent variable is the dependent variable. Lifetime migration is negative and statistically significant, indicat- defined by where individuals were born and ing that an additional kilometer of distance where they currently live. This is suboptimal, between a district and a higher-tier urban cen- as migration that occurred before the fiscal ter decreases the district's employment growth period being observed may distort the results. by -0.002 percent. Yet data on 5-year migration are only available 20. None of the remaining sectors has a sig- every 10 years, the next one being in 2010, and nificant elasticity for initial levels of employ- fiscal data at the subnational levels are avail- ment density, whereas all sectors experienced able only since 2001. positive intersectoral effects from diversity 11. This is important given that the units of (the coexistence of various sectors in a given observation are likely subject to spatial correla- locality benefiting from supply and demand tion, as migration is without doubt defined by linkages). geographic proximity. 21. Infrastructure expenditures are defined as 12. This is inconsistent with other studies, regional development, housing, and settlement which have found that in-migration is positively sector; water resources, irrigation, and trans- associated with education expenditures, but not portation sector; telecommunication subsector; with health expenditures (Clark and Hunter energy subsector of mining and energy sector 1998; Conway and Houtenville 1998). The com- (World Bank 2007). mon argument suggests that some segments 22. Reverse causality could lead to inconsis- of the population, such as the elderly, do not tent estimates of the coefficients. Yet this esti- benefit directly from education and that higher mation initially sought to find the mean of the expenditures on education may be associated dependent variable conditional on the poten- purely with higher local taxes. Yet the association tially endogenous variable, abstracting from the between taxes and expenditures is weak in Indo- sign of causality. The resulting coefficients of nesia, which raises a question regarding the true interest are not statistically significant, and thus reason for this relation. no further estimations with instruments were 13. This is revealed by a negative correla- deemed necessary. tion between initial population and population 23. Brueckner (2003) and Revelli (2006) dis- growth (or population convergence). tinguish among at least three types of govern- 14. All cities (regardless of their population) ment interaction: expenditure spillovers, tax account for 43 percent of GDP. competition, and yardstick competition. 15. This index varies between 0 and 1 and 24. To identify the presence of expenditure measures, for each sector, the aggregate differ- spillovers, it is necessary to assume that subna- ence between a district's share of total employ- tional governments have discretion regarding ment in that specific industry and the district's how to spend their resources. This assumption share of employment in total national employ- is not always met in developing decentralized ment. A value of 0 denotes the absence of countries. In the case of Indonesia, despite the regional concentration (for example, the share devolution of public spending in 2001 a large of a district's employment in a specific industry portion of the General Allocation Fund (Dana is the same as its share of employment in total Alokasi Umum) is used to cover the full amount national employment). Spatial considerations on decentralization and economies of concentration in Indonesia 151 of the district's civil service wage bill, while the Arze del Granado, Francisco J., Jorge Marti- central government has retained the authority nez-Vazquez, and Renata R. Simatupang. to manage the subnational civil service. In prac- 2008."Local Government Fiscal Competi- tice, district governments have circumvented tion in Developing Countries: The Case of this by hiring contractual employees to cover Indonesia." Urban Public Economics Review 8 additional needs in different sectors. Unfortu- (2008): 13­45. nately, the current budget classification does not Arze del Granado, Francisco J., and Albert J. allow the separation of those expenditures from Sumell. 2008."Growth and Spatial Concen- civil servants salaries. To test the predictions of tration: Indonesia." Paper submitted to the the expenditure spillover model, discretionary Journal of Regional Science. expenditures are separated from nondiscretion- ary expenditures at the subnational level. Non- Bahl, Roy. 1995."Worldwide Trends in Fiscal discretionary expenditures are defined as capital Decentralization." Policy Research Center expenditures plus routine expenditures other Working Paper, Georgia State University. than for personnel (goods and materials, opera- BPS (Biro Pusat Statistik, Indonesian Bureau tions and maintenance, other routine expendi- of Statistics). 1960. 1960 Population Census. tures, and others). Jakarta: BPS. 25. In turn, taxes on hotels and restaurants ------. 2000. 2000 Population Census. Jakarta: account for 75 percent of own-source revenues. BPS. 26. The lack of statistical significance of the coefficients for the remaining sectors is not sur- ------. 2005a. Aceh Population Census 2005. prising, as many of the studies conducted in Jakarta: BPS. other countries have failed to find a spatial inter- ------. 2005b. Intercensal Population Survey dependence on expenditure estimations based (Supas). Jakarta: BPS. on distance weight matrixes (Case, Rosen, and ------. Various years. National Socio-Economic Hines 1993). These authors do not find evidence Survey (Susenas). Jakarta: BPS. of the presence of tax competition. 27. The lack of evidence of production exter- Brueckner, Jan K. 2003."Strategic Interaction nalities in manufacturing suggests only that among Governments: An Overview of Empir- there is no evidence of production externalities ical Studies." International Regional Science at this level of aggregation. Studies of industries Review 26 (2): 175­88. within the manufacturing sector find effects of Cairncross, Frances. 1997. The Death of Dis- agglomeration economies in determined indus- tance: How the Communications Revolution tries (Deichman and others 2005; Kuncoro in Will Change Our Lives. Cambridge, MA: chapter 10 of this volume). Harvard Business School Press. References Case, Anne C., Harvey S. Rosen, and James R. Hines Jr. 1993."Budget Spillovers and Fiscal Ades, Alberto F., and Edward L. Glaeser. 1995. Policy Interdependence: Evidence from the "Trade and Circuses: Explaining Urban States." Journal of Public Economics 52 (3): Giants." Quarterly Journal of Economics 110 285­307. (1): 195­227. Chen, Zhao, Yu Jin, and Ming Lu. 2005."Eco- Alm, James, Jorge Martinez-Vazquez, and Sri nomic Opening and Industrial Agglom- Mulyani Indrawati, eds. 2004. Reforming Inter- eration in China." Industrial Organization governmental Fiscal Relations and Rebuilding 0511012, Econ WPA. Indonesia: The Big Bang Program and Its Eco- nomic Consequences. Studies in Fiscal Federal- Clark, David E., and William J. Hunter. 1998. ism and State-Local Finance, series ed. Wallace "The Impact of Economic Opportunity, E. Oates. London: Edward Elgar Publishing. Amenities, and Fiscal Factors on Age-Specific Migration Rates." Journal of Regional Science Arze del Granado, Francisco J., Jorge 32 (3): 349­65. Martinez-Vazquez, and Robert McNab. 2005. "Fiscal Decentralization and the Combes, P. P. 2000."Economic Structure and Functional Composition of Public Expendi- Local Growth: France, 1984­1993." Journal of tures." International Studies Program Work- Urban Economics 47 (3): 329­55. ing Paper 0501, Georgia State University, Conway, Karen Smith, and Andrew J. Houten- Andrew Young School of Policy Studies, ville. 1998. Elderly Migration and State Fis- International Studies Program. cal Policy: Evidence from the 1990 Census 152 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Migration Flows. Aging Studies Program Gardiner, Peter. 1997."Migration and Urbaniza- Paper 13. Syracuse University, Maxwell tion: A Discussion." In Indonesia Assessment: School of Citizenship and Public Affairs. Population and Human Resources, ed. G.W. Jones and T. H. Hull. Canberra: Australian Cushing, Brian J. 1989."Use and Misuse of the National University; Singapore: Institute of Allocation Rate in Models of Population Southeast Asian Studies. Migration." Annals of Regional Science 23 (1): 51­58. Gardiner, Peter, and M. Oey-Gardiner. 1991. "Pertumbatian dan Perluasan Kota di Indo- Deichman, Uwe, Kai Kaiser, Somik V. Lall, and nesia [Growth and Lateral Expansion of Zmarak Shalizi. 2005. Agglomerations, Trans- Cities in Indonesia]." Kompas, May 7. port, and Regional Development in Indonesia. Policy Research Paper 3477. Washington, DC: Glaeser, Edward L. 1997."Are Cities Dying?" World Bank. January. Journal of Economic Perspectives 12 (2): 139­60. Dobkins, Linda H., and Yannis M. Ioannides. 2001. "Spatial Interactions among Cities."Regional Glaeser, Edward L., Hedi D. Kallal, Jose A. Schei- Science and Urban Economics 31 (6): 701­31. nkman, and Andrei Shleifer. 1992."Growth in Cities." Journal of Political Economy 100 (6, ESCAP­UN (Economic and Social Commission centennial issue): 1126­52. for Asia and Pacific­United Nations). 1993. State of Urbanization in Asia and the Pacific Glaeser, Edward L., and Janet E. Kohlhase. 2004. 1993. New York: ESCAP­UN. "Cities, Regions, and the Decline of Trans- port Costs." Papers in Regional Science 83 (1): Fearnside, Philip M. 1997."Transmigration in 197­228. Indonesia: Lessons from Its Environmental and Social Impact." Environmental Manage- Gordon, Raymond G. Jr., ed. 2005. Ethnologue: ment 21 (4): 553­70. Languages of the World, 15th ed. Dallas: SIL International. Fengler, Wolfgang, and Bastian Zaini. 2006. "Increasing Subnational Government Han, Sun Sheng, and Ann Basuki. 2001."The Resources: Magnitude and Implications." Spatial Pattern of Land Values in Jakarta." Unpublished policy note, World Bank, Urban Studies 38 (10): 1841­57. Jakarta. Henderson, J. Vernon, Ari Kuncoro, and Matt Firman, Tommy. 1997."Patterns and Trends of Turner. 1995."Industrial Development in Urbanization." In Indonesia Assessment: Popu- Cities." Journal of Political Economy 103 (5): lation and Patterns, ed. G. W. Jones and T. H. 1067­90. Hull. Singapore: Institute of South East Asian Hugo, Graeme J. 1975. Population Mobility in Studies. West Java. Yogyakarta: Gadjah Mada Univer- ------. 2004."Demographic and Spatial Pat- sity Press. terns of Indonesia's Recent Urbanization." ------. 1997."Changing Patterns and Processes Population, Space, and Place 10: 421­34. of Population Mobility." In Indonesia Assess- Firman, Tommy, Benedictus Kombaitan, and ment: Population and Patterns, ed. G. W. Jones Pradono Pradono. 2007."The Dynamics of and T. H. Hull. Singapore: Institute of South Indonesia's Urbanization, 1980­2006." Urban East Asian Studies. Policy and Research 25 (4, December): 433­54. ------. 2003a. "Demographic Change and Fujita, Masahisa. 1988."A Monopolistic Com- Implications." In South East Asia Trans- petition Model of Spatial Agglomeration." formed: A Geography of Change, ed. Chia Regional Science and Urban Economics 18 Lien Sien. Singapore: Institute of South East (February): 87­124. Asian Studies. Fujita, Masahisa, Paul Krugman, and Tomoya ------. 2003b."Urbanization in Asia: An Over- Mori. 1999."On the Evolution of Hierarchical view." Paper prepared for the Conference on Urban Systems." European Economic Review African Migration in Comparative Perspec- 43 (2): 209­51. tive, Johannesburg, South Africa, June 4­7. Fujita, Masahisa, and Jacques-Franēois Thisse. Iimi, Atsushi. 2005."Urbanization and Devel- 1996."Economics of Agglomeration." Journal opment of Infrastructure in the East Asian of Japanese and International Economics 10 Region." JBICI [Japan Bank for International (4): 339­78. Cooperation] Review 10 (March): 88­109. Spatial considerations on decentralization and economies of concentration in Indonesia 153 Jacobs, Jane. 1969. The Economy of Cities. New Revelli, Federico. 2006."Spatial Interactions York: Vintage. among Governments." In Handbook of Fis- Krugman, Paul. 1991."Increasing Returns and cal Federalism, ed. Etisham Ahmad. Giorgio Economic Geography." Journal of Political Brosio: Edward Elgar. Economy 99 (3): 483­99. Ruiz Valenzuela, Jenifer, Rosina Moreno- ------. 2000."Where Is the New Economic Serrano, and Esther Vaya-Valcarce. 2006. Geography?" In The Oxford Handbook of Eco- "Has Concentration Evolved Similarly in nomic Geography, ed. Gordon L. Clark, Mary- Manufacturing and Services? A Sensitivity ann P. Feldmann, and Meric S. Gertler, pp. Analysis." Institut de Recerca en Economia 49­60. Oxford: Oxford University Press. Aplicada 2007/8, Universitat de Barcelona, Spain. Kuncoro, Ari. 1994. Industrial Location Choice in Indonesia. Ph. D. dissertation, Brown Uni- Sen, Ashish, and Tony E. Smith. 1995. Gravity versity. Models of Spatial Interaction Behavior. Berlin: Springer-Verlag. Kuncoro, Mudrajad, and J. Malcolm Dowl- ing. 2007."The Dynamics and Causes of Sri Adhiati, M. Adriana, and Armin Bobsien. Agglomeration." In Economic Integration and 2001. Indonesia's Transmigration Programme: Multinational Investment Behaviour, ed. Peter An Update. Report prepared for Down to Buckley. London: Edward Elgar. Earth, London. Leinbach, Thomas R. 1989. "The Transmigra- Venables, Anthony J. 1996."Equilibrium Loca- tion Programme in Indonesian National tion of Vertically Linked Industries." Interna- Development Strategy: Current Status and tional Economic Review 37 (2): 341­59. Future Requirements." Habitat International Wells, Louis T., and Nancy J. Allen. 2001. "Tax 13 (3): 81­93. Holidays to Attract FDI: Lessons from Two Lewis, Blane. 2005."Indonesian Local Gov- `Experiments.'" Unpublished mss., Harvard ernment Spending, Taxing, and Saving: An Business School, Cambridge, MA. Explanation of Pre- and Post-Decentraliza- Wen, Mei. 2004."Relocation and Agglomeration tion Fiscal Outcomes." Asian Economic of Chinese Industry." Journal of Development Journal 19 (3): 291­317. Economics 73 (1): 329­47. Lin, George C. S. 1994."Changing Theoretical Williamson, Jeffrey G. 1965."Regional Inequal- Perspectives on Urbanization in Asian Devel- ity and the Process of National Development." opment Countries." Third World Planning Economic Development and Cultural Change Review 16 (1): 1­23. 13 (4, June): 3­45. McGee, T. G. 1992."Indonesia: Towards a World Bank. 2003."Cities in Transition: Urban New Spatial Order." IUIDP Implementation Sector Review in an Era of Decentralization Support Project INS/89/014, Jakarta. in Indonesia." East Asia Working Paper Series, Partridge, Mark D., Dan S. Rickman, Kamar Dissemination Paper 7. World Bank, Urban Ali, and M. Rose Olfert. 2008."Employment Sector Development Unit, Infrastructure Growth in the American Urban Hierarchy: Department, East Asia and Pacific Region, Long Live Distance." B. E. Journal of Macro- Washington, DC. economics 8 (1): art. 10. ------. 2007. Spending for Development: Mak- Pisati, Maurizio. 2001."Tools for Spatial Data ing the Most of Indonesia's New Opportunities. Analysis." STATA Technical Bulletin Reprints Jakarta: World Bank. 10 (sg162): 277­98. Annex Definition of the variables and estimation results T his annex provides details regard- otherwise, based on population data from districts included in the analysis, 133 qualify ing the definition of variables and the Supas 2005 (BPS 2005b). as urban: 18 in category three, 4 in category econometric model specifications. Fringes areas is a dummy variable with two, and 74 in category one. Further details and descriptive statistics are a value of 1 for all districts adjacent to a "1 For rural districts, estimations include available from the author upon request. million plus" city and 0 otherwise. the distance to the center of the near- est urban district regardless of category Definition of the variables Estimation results and incremental distance to the center of the nearest higher-category district. For Results reported here are drawn from a min- Specialization or localization is defined urban districts, the distance to the nearest imum least squares regression. The variables as S=emps,d/emps, where emps,d is employ- urban district equals 0, but other values are and their sources are defined as follows. ment in sector s in districts d, and emps assigned to incremental distances of the Net migration is the number (thou- is total national employment in sector nearest higher-category district. All cate- sands) of lifetime in-migrants minus out- s. Employment density is E = empd /a, ,t0 gory-three urban districts have a value of migrants from the Intercensal Population where employment in district d in time t0 0 for both nearest urban district and incre- Survey (Supas) for 2005 (BPS 2005b). Net = 1994, and a is district d's area. This vari- mental distances. However, a category-one out-migration is used instead of an "alloca- able is a proxy for the size of local markets, urban district that is 20 kilometers away tion rate" (net migration as a percentage of which are quasi-proportional to the size from the nearest category-two district and the total number of out-migrants from the of the local economy. Diversity, defined as S 50 kilometers away from the nearest catego- place of origin), as recent studies prove that divs =1 / [empd /(empd-empd )]2 is ,d ,s ,s ry-three district will have a value of 0 to the using "allocation rates" as dependent vari- s=1 ss nearest urban district, an incremental value ables is valid only with a very narrow inter- the inverse of a Herfindahl index. This vari- of 20 to the nearest category-two district, pretation when origin variables are included able reaches a maximum when all sectors and an incremental value of 30 to the near- in the model (Cushing 1989). except the sector being studied have the est category-three district. As a final exam- Urbanization is the district's urban pop- same size in district d (see Combes 2000; ple, assume that a rural district is 40 kilome- ulation as a percentage of total population Henderson, Kuncoro, and Turner 1995). ters from the nearest category-one district, from the Supas (BPS 2005b). Nearest urban (distma) and incremental 80 kilometers from the nearest category-two Incremental distances are computed fol- distances (inchigherma) are in line with the district, and 60 kilometers from the nearest lowing a methodology similar to the one incremental distances described earlier, but category-three district. If the distance to the used by Partridge and others (2006). The the definitions are slightly different. Each nearest urban area equals 40, then the incre- variable minimum distance to urban center district is categorized either as "mainly" mental values to the nearest category-two is the distance from a district to a district urban or as rural, based on both the total and category-three districts would equal 0 at a higher tier of urbanization. The tiers population and the population density of and 20, respectively. are defined by quintiles relative to the per- the district. Urban districts are then catego- The specification is estimated as centage of urban population in the district. rized into tiers according to total popula- follows: The location of districts is not necessarily tion. That is, districts can be categorized as sequential, so a district in the top quintile either rural or as one of three categories of ys = +1D +2L +3E + 1Dist could be the closest district to one in the urban (with category three being the larg- ,d bottom quintile. In this case, all variables est). Specifically, a district is classified as + 2IncPop + 3G of incremental distance to levels three, four, urban if it has a population greater than + 4PopCat + us , (A.1) ,d and five would be 0. Yet if a district in the 100,000 and a population density greater bottom quintile is closest to a district in the than 100. The reason for using both popu- where ys,d is employment growth in district second quintile, the incremental distances lation and population density in categoriz- d and sector s, D is diversity, L is specializa- three, four, and five would record the "addi- ing districts is that several geographically tion or localization, E is employment den- tional" distance needed to reach a district large districts have a substantial, but sparse, sity, Dist is distance to nearest urban center, at those levels. For example, if a district in total population without a significant urban and IncPop is incremental population to category one is 50 kilometers from a cate- center. If the population is above 100,000 category-four and category-five districts, gory-two district and 110 kilometers from but below 400,000 with a population den- G is a matrix of geographic and ethnolin- a category-three district, the incremental sity above 100 or if the population is above guistic variables (including landlocked, distance to category three would be 60 kilo- 400,000 with a density below 100, it is a cat- island district, remote district, ethnolin- meters. Incremental distances are computed egory-one urban district; if the population guistic fractionalization, longitude, lati- based on geographic coordinates exported is between 400,000 and 700,000 with a den- tude, and regional dummies for Sumatra, from MapInfo. Incremental distance three sity above 100, it is a category-two district; Kalimantan, Sulawesi, Nusa Tenggara and was dropped because it did not have enough and if the population is above 700,000, it is Maluku, and Papua), PopCat2-PopCat5 are observations.1 a category-three district. The categories are dummies for population categories, and 1 million plus cities is a dummy variable based on 1994 population figures. All dis- u is the error term. is an autoregressive with a value of 1 for districts with popula- tricts with a population above 500,000 have parameter, which takes the form of a mixed tion greater than 1 million in 2005 and 0 a population density above 100. Of the 286 regressive spatial autoregressive process Spatial networks, incentives and the dynamics of village economies 155 corresponding to the following spatial are weights defined according to a predefined computed by maximum likelihood using regression model: Y = WY + X + . See criteria of neighborliness (district with a cen- STATA ml routine modules developed by M. Arze del Granado and Sumell (2008) for troid within a band of five digitizing units, Pisati (see Pisati 2001). The full set of results further details and a complete set of results. using spawmat in Stata software), yj are is available upon request. The specification is estimated as expenditures of district i's neighboring dis- follows: tricts, and E, I, S, and Geog are explanatory Note variables corresponding, respectively, to pub- 1. By definition, the incremental dis- Yi = w lic expenditures (infrastructure and expendi- tance to the third quintile (incremental dis- i,jyj +1E +2I ji tures in industry and business development tance three) is 0 for all districts in quintiles + 3S + 4Geog + ui , (A.2) sectors), relative income per capita, average four and five. In addition, for districts in ,t years of schooling, and a set of district i's geo- the second quintile, the minimum distance graphic characteristics, including the inverse to larger urban center variable is actually the where Yi is district's i share of manufactur- distance to the closest port, landlocked, iso- distance to the nearest third quintile, and so ing in total regional GDP (as a proxy for lated island, and main island dummies. values for the incremental distance to the regional industrial concentration), and Two more specifications (which include third quintile are also 0. 1 ­ 5 are parameters to be estimated, wi,j a spatial lag of the dependent variable) are Spatial agglomeration, firm productivity, and government policies in Indonesia: concentration and deconcentration in the manufacturing sector Ari Kuncoro 10 This chapter focuses on two of the three by building and improving roads in rural central themes of World Development Report areas. The goal is to decentralize jobs to 2009, namely, spatial concentration (density) small cities and thus ease the pressure on and industrial decentralization (distance). large cities. During the 1980s and 1990s, Spatial centralization of resources and spa- efforts to decentralize manufacturing firms tial concentration of economic activities in to outlying locations in Java were relatively a few of the largest metropolitan areas are successful mainly because localization forces issues facing many developing countries. were stronger than urbanization forces. The c h a p t e r Concentration will bring spatial disparities forces of decentralization pushed toward between leading and lagging regions, widen- deconcentration, while local agglomeration ing gaps in living standards and welfare, and and specialization encouraged industries to negative externalities associated with very reconcentrate in smaller, medium-size cities. large urban areas, such as congestion, crime, In the following sections, this chapter pres- and pollution. From the standpoint of eco- ents empirical evidence to explain why this nomic efficiency, of course, there are benefits was the case. from such concentration or agglomeration.1 The challenge is to minimize the unintended Evolution of the manufacturing negative effects. One solution is regional or industry's spatial configuration territorial development within countries, which means encouraging the development in Java of alternative centers in lagging regions. The concentration of economic activity in a Using series of firm-level data for Indo- few places is a common phenomenon.While nesian manufacturing, this chapter illus- the concentration of economic activity (and trates how the concentration took place as the concomitant economic efficiency) is the economy was becoming more developed itself desirable, the large spatial disparities and how public policies could mitigate the in welfare associated with this process are problem by mixing infrastructure develop- mostly unwelcome. Manufacturing activi- ment in lagging regions with private incen- ties in Indonesia offer good examples of tives to encourage industries to concentrate this process. In the Indonesian context, in smaller cities in lagging regions. The gov- manufacturing, especially its labor-intensive ernment has made it possible for manufac- branches, is instrumental in alleviating turing firms to locate in outlying locations poverty. It provides millions of people with 156 Spatial agglomeration, firm productivity, and government policies in Indonesia 157 more rewarding off-farm jobs and, at the 19.3 percent in 2003. In the meantime, its same time, relieves pressure on agricultural neighboring districts, BOTABEK, increased wages. But the concentration of manufac- their share to 23.9 percent. Evidently, some turing in a few places in Java also attracts deconcentration took place, albeit relatively people to cities, putting severe pressure on close to the old center. Surabaya showed overburdened urban infrastructure. some gains, while low-income regions suf- As in other countries, the Indonesian fered a loss of 6 percentage points. government has been eager to stem this tide Table 10.4 illustrates the role of the by improving infrastructure, particularly manufacturing sector in the local econ- roads, in the hinterland (Albala-Bertrand omy. The huge drop in the share of manu- and Mamatzakis 2004). The social capital facturing in GDP in Jakarta reflects more program, called INPRES, was created to than the movement from manufacturing improve infrastructure in the countryside. to the service sector. The soaring costs of By investing in road infrastructure, the land, tighter environmental regulations, government made it possible for manu- and worsening congestion have made it facturing firms to locate in outlying loca- increasingly uneconomical for manufac- tions. In effect, the government attempted turing to locate in Jakarta. The next-door to move jobs to rural cities to ease the pres- neighbors of BOTABEK are the most logi- sure on large cities. This chapter examines cal sites. Using money from the oil boom whether this policy was indeed successful in of the 1970s, the government has made a deconcentrating manufacturing out of old conscious effort to improve road infrastruc- central locations. ture in Java. Indeed, in contrast to Algeria, For the purpose of comparison, all dis- Ecuador, Nigeria, and Venezuela, Indonesia tricts in Java are divided into high income per is the only oil-rich country in which agri- capita versus low income per capita regions cultural output has expanded during an oil (see table 10.1).As part of the Dutch colonial boom (Gelb and others 1988).2 Although, legacy, Jakarta (the capital city) and Sura- from the infrastructure point of view, it baya (the capital of East Java province) were was possible to relocate farther away from among the few places where manufacturing Jakarta, the choice of BOTABEK--about firms were originally concentrated (see figure 60 kilometers from Jakarta--suggests that 10.1). Naturally, the supporting service sec- the national capital still acted as a magnet, tors, such as banking and trade, also agglom- drawing firms to locate in close proximity erated in those high-income regions. to it. Table 10.2 calculates the income gap Why did the deconcentration happen so between the national capital--Jakarta-- close to Jakarta rather than farther inland, and other cities and regions. Only Surabaya for example, in Bandung, West Java, which has been able to catch up with Jakarta, while is about 180 kilometers to the southeast of others either have maintained the same Jakarta? Infrastructure apparently was not gap (BOTABEK or Bogor, Tangerang, and the main reason. Indeed, road infrastruc- Bekasi) or have fallen behind (low-income ture in Java, which connects the industrial districts in lagging regions). agglomerations of Jakarta and Surabaya In terms of value added, the high con- centration of manufacturing in regions with low income per capita is rather deceiving, Table 10.1 Comparison of high-income per capita and low-income per capita regions in Java, because it is shared by about 83 (out of 97) select years, 1986­2003 GDP per capita in Rp million per year districts, which occupy about 90 percent of Region 1986 1995 2001 2003 the land area in Java (see table 10.3). Manu- facturing firms in low-income regions are High income per capita regions Jakarta 1.4 8.9 28.7 36.2 typically in food processing, mostly tradi- BOTABEK 0.4 3.1 9.3 10.5 tional food products. From 1986 to 2003, Lamongan, Gresik, Sidoarjo 0.5 2.6 6.9 8.7 Jakarta lost its dominance as creator of Surabaya 0.8 5.8 18.0 23.0 value added, as its share of manufacturing Low income per capita regions 0.5 1.9 4.8 6.1 gross domestic product (GDP) dropped to Source: Calculated from the regional income data of the Central Statistical Agency. 158 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 10.1 Java island and its vicinity Kalimantan Palembang Belitung Bandjarmasin Sumatera Bandar Lampung Java Sea D.K.I. JAKARTA JAKARTA JAWA TENGAH Serang Bandung Semarang Madura BANTEN Jawa Surabaya Bali JAWA BARAT Yogyakarta JAWA TIMUR Lombok D.I. YOGYAKARTA Denpasar National Capital Mataram Provincial Capitals Provincial Boundaries INDIAN OCEAN Source: www.lonelyplanet.com/java/. Table 10.2 Ratio of local GDP per capita to Jakarta's GDP in Java, select years, 1986­2003 licensing regime in Jakarta was one factor Region 1986 1995 2001 2003 drawing firms to central locations.4 High income per capita regions BOTABEK 0.32 0.35 0.32 0.29 Choice of firm location Lamongan, Gresik, Sidoarjo 0.38 0.30 0.24 0.24 This section examines firm-level decisions Surabaya 0.59 0.66 0.63 0.64 regarding location in the manufacturing Low income per capita regions 0.34 0.21 0.17 0.17 sector. Conceptually, a firm will choose a Source: Calculated from the regional income data of the Central Statistical Agency. location in which it believes it can earn the highest profit. The trend of spatial concentration or Table 10.3 Spatial concentration of manufacturing GDP in Java, select years, 1986­2003 deconcentration is presented in tables 10.6 percent and 10.7. Based on share of all firms as a Region 1986 1995 2001 2003 simple measure of concentration, it appears High income per capita regions that the trend among all industries between Jakarta 26.2 25.2 19.3 19.3 1990 and 2003 was toward deconcentra- BOTABEK 14.9 10.9 24.0 23.9 tion (table 10.6). Despite a period of slight Lamongan, Gresik, Sidoarjo 4.9 8.7 6.9 5.0 reconcentration between 1980 and 1990 as Surabaya 4.1 6.5 5.0 7.1 Low income per capita regions 50.0 48.9 44.7 43.7 the nation underwent liberalization, the Source: Calculated from the regional income data of the Central Statistical Agency. number of firms in Jakarta as a share of all firms dropped from 19 percent in 1980 to 11.5 percent in 2003. As expected, BOTA- to the hinterland, improved a lot between BEK increased its share from only 5 percent 1986 and 1990 (see table 10.5).3 The share in 1980 to 12.4 percent in 1990. Low-income of villages with paved roads in districts other regions, after experiencing a drop from 63.7 than Jakarta, including those in low-income percent in 1980 to 57.4 percent in 1990, regions, increased significantly during this regained much of the loss in 2003, returning period. In this respect, Henderson and almost to the level in 1980. So from the firm- Kuncoro (1996), using the Java sample, sug- level standpoint, the deconcentration was gest that, in 1986­90, centralization of the of firms moving to lower-income regions. Spatial agglomeration, firm productivity, and government policies in Indonesia 159 This would have been almost impossible if Table 10.4 Fraction of manufacturing value added in local GDP in Java, select years, 1986­2003 the central and local governments had not Region 1986 1995 2001 2003 improved infrastructure in Java, particularly High-income per capita regions roads (table 10.5). Jakarta 0.54 0.21 0.22 0.22 At the outset, hinterland locations had BOTABEK 0.26 0.49 0.59 0.58 difficulty attracting industries, despite gov- Lamongan, Gresik, Sidoarjo 0.30 0.33 0.35 0.33 ernment policies encouraging decentral- Surabaya 0.23 0.35 0.34 0.34 Low-income per capita regions 0.13 0.17 0.21 0.19 ization, such as the creation of industrial Source: Calculated from the regional income data of the Central Statistical Agency. zones in outlying regions. Many industrial zones remained largely empty until the mid-1990s. Later on, after the second wave Table 10.5 Fraction of villages with paved roads in Java, select years, 1986­2000 of economic liberalization in the mid-1990s, Region 1986 1990 2000 some deconcentration to areas farther from High-income per capita regions Jakarta did take place. Only after congestion Jakarta 0.94 0.99 0.96 and wage and price increases began to erode BOTABEK 0.31 0.61 0.72 their competitiveness did firms start to fill in Lamongan, Gresik, Sidoarjo 0.30 0.46 0.55 new locations adjacent to the original met- Surabaya 0.90 0.97 0.99 ropolitan areas of Jakarta, BOTABEK, and Low-income per capita regions 0.42 0.62 0.69 Greater Surabaya.5 Source: Calculated from the village potential data of the Central Statistical Agency. At the industry level, the picture is more interesting. In terms of number of firms, other industries, but they were less significant as the economy was liberalized in the mid- than in the nonmetallic industry. Although 1980s, textile firms began to concentrate less pronounced, the same picture can be in Jakarta and BOTABEK at the expense of observed for wood. Finally, there was little low-income regions,particularly the old cen- change in the concentration of food process- ter of textiles in Bandung, West Java. But in ing.TherewasmovementbetweenJakartaand 1995 this trend was reversed and continued BOTABEK, but little movement elsewhere. until well after the economic crisis of the late Table 10.7 presents the concentration of 1990s. For machinery, the deconcentration employment as a consequence of firm-level took place mainly from Jakarta to the neigh- choice of location. As expected, the general boring districts of BOTABEK and not much picture resembles the concentration of firms elsewhere. So, in effect, BOTABEK became a in table 10.6. For machinery and nonmetal- new center of agglomeration for machinery. lic minerals (including cement), where scale In the case of chemicals, low-income is important, the concentration of employ- regions enjoyed significant gains, increasing ment and any change associated with it is their share of firms from 30.4 percent in 1980 more pronounced than the concentration of to 45.7 percent in 2003. Tighter environmen- firms in Jakarta and BOTABEK. tal regulations in urban areas may continue to force firms to relocate in less-regulated The movement of population districts in low-income regions. However, Industry concentration is measured by a this does not necessarily mean the recon- normalized Hirschman-Herfindahl index. centration of industry, because these firms For each industry, the normalization con- are shared by 83 districts. But new indus- trols for changes in industry concentration trial agglomerations evidently are emerging brought about by changes in population in low-income regions, contributing to the concentration over time. For industry j at increasing share of firms in these regions. time t, the concentration is given by: The nonmetallic industry has experienced deconcentration, mainly to low-income dis- I tricts. The share of firms in lagging districts gj (t) = Eij (t ) - Pi (t)2 , (10.1) increased significantly, from 64.9 percent in = i =1 Ej P(t) 1980 to 83.5 percent in 2003, at the expense of Jakarta and BOTABEK. Districts in Gresik, where Ej(t) is national employment in indus- Lamongan, and Sidoarjo enjoyed gains in try j at time t, P(t) is national population 160 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 10.6 Concentration of manufacturing firms in Java, select years, 1980­2003 Region and year Food Textiles Wood Paper Chemicals Nonmetals Machinery All 1980 High-income regions Jakarta 9.0 13.6 24.1 37.5 38.8 14.9 37.3 19.0 BOTABEK 2.9 2.3 4.5 4.2 11.7 9.2 9.3 5.0 Lamongan, Gresik, Sidoarjo 8.5 5.4 3.0 2.2 4.6 4.1 5.7 6.0 Surabaya 5.5 1.8 2.5 6.4 14.4 6.8 13.9 6.3 All leading regions 26.0 23.1 34.2 50.3 69.6 35.1 66.2 36.3 All lagging regions 74.0 76.9 65.8 49.7 30.4 64.9 33.8 63.7 All regions 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 1990 High-income regions Jakarta 6.7 29.8 17.5 36.1 25.5 4.0 29.2 19.8 BOTABEK 5.2 8.7 20.0 12.5 23.9 11.5 22.8 12.4 Lamongan, Gresik, Sidoarjo 8.3 4.3 5.4 2.4 6.1 3.9 6.2 5.8 Surabaya 3.5 2.0 5.2 10.3 7.6 2.5 9.2 4.6 All leading regions 23.7 44.8 48.1 61.4 63.1 21.8 67.4 42.6 All lagging regions 76.3 55.2 51.9 38.6 36.9 78.2 32.6 57.4 All regions 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 1995 High-income regions Jakarta 5.7 22.2 9.6 30.6 20.4 2.5 19.4 14.7 BOTABEK 7.1 13.1 15.9 16.9 25.8 12.0 31.1 15.6 Lamongan, Gresik, Sidoarjo 7.1 6.5 7.3 4.4 8.9 3.7 8.3 6.8 Surabaya 3.0 1.9 3.7 9.4 6.2 1.2 6.8 3.7 All leading regions 22.8 43.7 36.5 61.4 61.3 19.3 65.6 40.8 All lagging regions 77.2 56.3 63.5 38.6 38.7 80.7 34.4 59.2 All regions 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 2003 High-income regions Jakarta 5.4 17.1 4.6 25.8 15.5 2.0 14.9 11.5 BOTABEK 6.5 10.8 10.8 16.5 24.1 8.0 34.3 14.5 Lamongan, Gresik, Sidoarjo 8.1 6.9 6.6 5.7 9.4 4.3 6.3 7.2 Surabaya 2.9 2.2 4.4 8.2 5.3 2.1 6.4 3.9 All leading regions 23.0 37.0 26.4 56.1 54.3 16.5 61.8 37.1 All lagging regions 77.0 63.0 73.6 43.9 45.7 83.5 38.2 62.9 All regions 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Source: Calculated from the Central Statistical Agency, Manufacturing Annual Survey (various years). at time t, and i is the total number of geo- 1990 and 1995, food and textiles became graphic units. The minimum value of gj(t) is more concentrated, while other industries 0; that is, when all locations' share of indus- showed the opposite trend.Wood,chemicals, trial employment is exactly the same as their and machinery were highly concentrated in share of population.6 the beginning of the period, but afterward The results are presented in tables 10.8 became significantly less concentrated. and 10.9. For simpler categorization, Jakarta For some industries, the 1998 economic is combined with BOTABEK to become crisis brought an abrupt change in the trend JABOTABEK, while Greater Surabaya com- of deconcentration. Wood, chemicals, non- bines the city of Surabaya itself with the metallic minerals, and machinery became industrial districts of Lamongan,Gresik,and more concentrated after the crisis. This Sidoarjo. The Hirschman-Herfindahl con- reconcentration, however, was not a result of centration index for all industries confirms the influx of new firms; rather it was caused that overall manufacturing industries were by the decline in the number of firms in more concentrated in 2003 than in 1990, outlying districts. This trend resumed in the which means that industrial concentration post-crisis period of 2001­03. For all indus- tends to deviate from where the population tries, the index increased from 3.1 percent resides. In the booming period between (chemicals) to 35.1 percent (machinery), Spatial agglomeration, firm productivity, and government policies in Indonesia 161 Table 10.7 Concentration of manufacturing employment in Java, select years, 1980­2003 Region and year Food Textiles Wood Paper Chemicals Nonmetals Machinery All 1980 High-income regions Jakarta 4.9 15.6 16.6 37.9 36.2 24.0 52.0 20.0 BOTABEK 1.3 9.8 8.2 8.4 15.0 13.3 9.2 7.3 Lamongan, Gresik, Sidoarjo 4.7 2.9 19.7 8.4 5.0 10.4 5.5 5.0 Surabaya 3.6 2.4 3.0 2.7 14.2 11.9 13.2 6.0 All high-income regions 14.5 30.7 47.5 57.5 70.4 59.6 79.9 38.3 All low-income regions 85.5 69.3 52.5 42.5 29.6 40.4 20.1 61.7 All regions 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 1990 High-income regions Jakarta 4.6 19.0 11.1 29.0 20.1 14.1 32.8 17.0 BOTABEK 3.9 21.7 24.3 17.5 28.2 25.6 22.6 18.7 Lamongan, Gresik, Sidoarjo 5.0 1.9 10.8 11.4 8.7 6.8 5.0 5.2 Surabaya 4.0 1.1 6.0 6.5 6.6 5.9 12.4 4.7 All high-income regions 17.4 43.6 52.1 64.4 63.6 52.4 72.8 45.5 All low-income regions 82.6 56.4 47.9 35.6 36.4 47.6 27.2 54.5 All regions 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 1995 High-income regions Jakarta 3.8 12.3 8.4 19.7 18.2 10.3 26.2 12.8 BOTABEK 0.4 1.1 42.0 9.7 1.0 17.4 4.4 1.1 Lamongan, Gresik, Sidoarjo 5.2 4.8 14.6 14.5 11.7 7.6 8.7 7.3 Surabaya 5.3 1.7 6.0 5.6 7.7 4.4 9.0 4.7 All high-income regions 20.3 47.7 49.3 58.2 64.1 53.5 77.7 48.1 All low-income regions 79.7 52.3 50.7 41.8 35.9 46.5 22.3 51.9 All regions 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 2003 High-income regions Jakarta 4.5 12.1 4.0 14.1 11.9 6.4 19.5 10.8 BOTABEK 6.0 19.5 15.7 15.3 23.8 26.1 37.7 19.7 Lamongan, Gresik, Sidoarjo 6.3 4.3 10.8 17.1 11.0 10.3 5.7 7.1 Surabaya 7.2 1.2 4.6 5.6 4.3 5.9 6.6 4.2 All high-income regions 23.9 37.1 35.1 52.1 51.0 48.6 69.4 41.8 All low-income regions 76.1 62.9 64.9 47.9 49.0 51.4 30.6 58.2 All regions 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Source: Calculated from the Central Statistical Agency, Manufacturing Annual Survey (various years). Table 10.8 District industrial concentration index in Java, select years, 1990­2003 Industry 1990 1995 1997 2001 2003 Food, beverages, tobacco 0.027 0.050 0.022 0.027 0.029 Textiles, garments, leather, footwear 0.015 0.019 0.009 0.013 0.016 Wood and wood products 0.038 0.026 0.028 0.033 0.033 Chemicals 0.042 0.029 0.028 0.032 0.033 Nonmetallic minerals 0.037 0.030 0.033 0.049 0.054 Machinery 0.061 0.051 0.046 0.077 0.104 All industries 0.019 0.017 0.017 0.026 0.023 Source: Calculated from the Central Statistical Agency, Manufacturing Annual Survey (various years). a significant increase in just over two years atmosphere. It is possible that outlying dis- (see table 10.9). This suggests that, for those tricts in Java experienced more firm death particular industries, the spatial distribution or exit than central locations. It is also of firms and population deviated from each possible that population or workers moved other. This can be attributed to the move- to "cheaper" districts adjacent to JABOTA- ment of new firms to central locations, but BEK and Greater Surabaya and spent more this is highly unlikely in the post-crisis time commuting to their workplace.Actually, 162 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 10.9 Change in the district industrial concentration index in Java, 1990­2003 percent Industry 1990­95 1995­97 1997­2001 2001­03 Food, beverages, and tobacco 85.2 -56.0 22.7 7.4 Textiles, garments, leather, footwear 26.7 -52.0 44.4 23.1 Wood and wood products -31.6 7.7 17.9 0.0 Chemicals -31.0 -3.4 14.3 3.1 Nonmetallic minerals -18.9 10.0 48.5 10.2 Machinery -16.4 -9.8 67.4 35.1 All industries -10.5 0.0 52.9 -11.5 Source: Calculated from the Central Statistical Agency, Manufacturing Annual Survey (various years). Table 10.10 Annual growth of stock of firms and respectively. Evidently, there was more firm labor employment in large and medium manufactur- death in the two central locations than in ing enterprises in Java, 1990­2003 remote districts. But these central locations JABOTABEK had a far larger stock of firms to begin with and Greater Period Surabaya Java and thus could sustain higher losses, while 1990­97 outlying areas had a meager initial stock of Stock of firms 4.0 4.6 firms, leaving them with almost nothing if Labor employment 8.4 7.2 only a very few firms disappeared. 1997­2001 Stock of firms 0.2 -0.9 Labor employment 1.8 1.7 Empirical methodology: 2001­03 externalities and firm Stock of firms -2.6 -2.4 Labor employment -1.4 -2.2 decentralization Source: Calculated from the Central Statistical Agency, For government policies to succeed, it is Manufacturing Annual Survey (various years). important to understand the behavior of firms with respect to choice of location, par- there was a slight decrease in the share of ticularly their behavior related to agglomer- population living in both central areas, from ation externalities (Henderson and Kuncoro 23.6 percent in 1995 (using intercensal data 1996; Mitra 1999). for 1995) to 22.9 percent in 2000 (using the Firm productivity is closely linked to 2000 population census). overall changes in employment and pro- What is interesting is that the concen- ductivity. Firms have the potential to cap- tration of all industries fell 11.5 percent ture efficiency gains from learning by doing between 2001 and 2003 (table 10.9). The as well as from increasing returns to scale category "all industries" includes paper, due to specialization and mechanization printing and publishing, and "other" cate- (Romer 1990). gories. Taking into account all industries in In the Indonesian context, one important all districts, the deviation between industrial question is which type of externalities is actu- and population concentration is not great, ally stronger. Whatever the form, externali- compared with the situation in which just a ties have important implications for urban single individual industry is considered. development. If externalities are in the form The puzzling question pertaining to the of localization--which in dynamic form post-crisis increase in industrial concentra- are often called Marshall-Arrow-Romer tion is resolved by figures presented in table (MAR) externalities--smaller cities are 10.10.In the pre-crisis era,between 1990 and more likely to specialize in just one industry 1997, each year the stock of firms grew 4 per- or in closely connected industries. Spatially, cent for JABOTABEK and Greater Surabaya this means that standardized manufacturing and 4.6 percent for the whole of Java. This activities tend to locate disproportionately implies that some deconcentration did take in smaller, specialized cities. Thus policies to place. But in the post-crisis era, these figures decentralize industries from their historical turned negative: -2.6 and -2.4 percent, agglomerations are more likely to succeed. Spatial agglomeration, firm productivity, and government policies in Indonesia 163 However, if the externalities happen to be and ownership status (foreign direct invest- urban in nature, an industry needs to find ment or domestic investment).9 a location in a large, diverse urban environ- Localization (MAR) externalities are ment. Such industries are more likely to be measured by total employment in the same found in large urban areas. industry in the same district. This measure A related question is whether externalities is meant to capture interaction among are mainly static or dynamic. If externalities firms within a district. Urbanization are dynamic, past industrial development externalities are measured by a diversity in cities will affect present productivity, index. For district i, for example, the index because, over time, a particular location of diversity is: accumulates a large body of knowledge. The J 2 paradox for industrial location is that firms gi (t) = s EEi ((tt)) - EE (t ) ij j , (10.3) become more static--tied to a particular (t) j =1 industrial agglomeration--and less willing to move to cities where historically a par- where E(t) is total national manufactur- ticular industry has never existed and where ing employment and Ej(t) is total national there is no built-up stock of knowledge. employment in industry j. Meanwhile, Ei This section examines whether the pattern and Eij are the corresponding local magni- of concentration or deconcentration across tudes. The measure of urbanization econo- s industries in Java is consistent with the esti- mies gi( t) has a minimum value of 0, where mated externalities.7 within a district, each industry's share of The approach is to relate firm productiv- local manufacturing employment is exactly ity as a function of local industry inputs and the same as its share nationally, so the dis- the external environment generating spill- trict is completely unspecialized because its overs (Henderson, Lee, and Lee 2001). The industrial composition merely copies that equation for assessing local externalities is of the nation. At the other end, the maxi- s based on the firm production function with mum value of gi(t) approaches 2 for a dis- constant returns to scale technology. In the trict completely specialized in one industry, intensive form, the firm technology is rep- while at the same time national employment resented by: is concentrated in another industry. The s higher is gi( t), the less diverse and the more xhij(t) = A[Shij(t)], f [khij(t)], (10.2) specialized is the district. where xhij(t) is real value added per worker In estimating equation 10.2, we use a in firm h, in city i, in industry j, and in time log-linear form of technology and assume t, and khij(t) is real capital per worker. The city, time, and individual fixed effects. We function f(.) represents firm technology also introduce firm characteristics such based on the original--extensive form-- as legal status, firm ownership, and age to production function F(.) = f(.)Lhij(t), where control for the shift in production function Lhij(t) is the number of workers.8 To obtain due to individual effects. The estimating the real value added, the nominal value equation is: added is divided by the wholesale price index at the appropriate three-digit indus- ln[xhij (t)] = j + jln[khij (t)] trial code for the relevant years. Firms' capi- + j ln[Eij (t )] tal is constructed from the estimated market s value of machinery and building. To convert + j g i (t ) + j (t ) this into real terms, the nominal values are + hhij + ij + hij . (10.4) deflated by the wholesale import price of machinery (including electrical machin- The equation is estimated at the level of ery). Shij represents the shift in the produc- industry j. Localization externalities are tion function, which also includes measures represented by the district's employment of spillover externalities, time and industry in the same industry, Eij. Urbanization dummies, and firms' characteristics that economies are represented by the gi index s supposedly affect productivity, such as legal in the linear form. 164 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA For industry j, the error term comprises So textiles are more likely to be found in four components; time fixed effects, j(t), more specialized, smaller cities. which can be used to infer productivity Chemicals encompass smaller subgroups gains; district fixed effects, µij; individual of industries such as basic and industrial firm fixed effects, hij; and contemporane- chemicals, petroleum refinery and prod- ous errors, hij, which are assumed to be ucts, and rubber and plastic products. The i.i.d. Exploiting the nature of the data, in results for chemicals are presented in table order to capture time fixed effects, we intro- 10.12. The coefficient of same-industry duce time dummies for the relevant year of employment is significant before and after manufacturing survey. District dummies are the economic crisis, but not in between. The introduced to capture district fixed effects. evidence for the dominance of localization District fixed effects capture time-invariant forces is quite strong, because the urbaniza- aspects, which are perhaps unique to that tion coefficient is never significant. particular location, such as resource endow- Table 10.13 reveals the results for non- ment, climate, urban layout, and internal metallic minerals, a group consisting of infrastructure.10 glass products, ceramics, clay, cement, and other nonmetallic minerals such as marble Results and granite. Unlike textiles and chemicals, Table 10.11 shows the results for textiles, a the evidence supporting localization is very category that includes textiles, garments, weak or nonexistent, and the coefficients leather, and footwear. The coefficient of the for the entire period are never significantly same-industry employment is positive and positive. The coefficients turn negative and highly significant, reflecting strong local- significant after the economic crisis, which ization economies. For the period before implies that firms go to where the presence the crisis, the coefficient for urbanization of the particular industry is weak. The coef- is positive but insignificant. In the crisis ficient for urbanization is also weak: none of period, the coefficient is positive and signifi- them is significant in three periods of analy- cant. It turns negative and significant in the sis. So the category of nonmetallic minerals post-crisis period, which means that firms exhibits no clear pattern either in localiza- located in more diverse environments have tion or in urbanization. higher productivity. Because the sign of the Table 10.14 shows the results for machin- coefficient of same-industry employment is ery. Unlike nonmetallic minerals, this indus- always consistent, the forces of localization try consists of more uniform products, rang- are stronger than the forces of urbanization. ing from metallic products, nonelectrical Table 10.11 Externality and productivity in Java: textiles, garments, leather, and footwear, 1990­2003 dependent variable: log value added per labor Explanatory variable 1990­95 1997­2000 2001­03 Log of capital per labor 0.21** 0.27** 0.15** (9.67) (10.34) (6.47) Log of same-industry employment (localization) 0.13** 0.03 0.19** (4.87) (0.33) (13.39) Index of districts' diversity (urbanization) 0.67 9.04** -0.70** (1.58) (3.67) (4.20) Year dummies Yes Yes Yes District dummies Yes Yes Yes Firm characteristics Yes Yes Yes Constant -0.35 -0.74 0.80** (1.21) (0.55) (4.03) Number of observations 18,807 7,768 7,539 R2 0.34 0.41 0.33 Source: Author's calculations. Note: Numbers in the parentheses are t-statistics with robust standard errors. ** Significant at 5 percent. * Significant at 10 percent. Spatial agglomeration, firm productivity, and government policies in Indonesia 165 Table 10.12 Externality and productivity in Java: chemicals, 1990­2003 dependent variable: log value added per labor Explanatory variable 1990­95 1997­2000 2001­03 Log of capital per labor 0.32** 0.22** 0.23** (19.11) (9.58) (10.68) Log of same-industry employment (localization) 0.22** 0.05 0.16** (3.05) (0.35) (2.01) Index of districts' diversity (urbanization) 1.17 0.05 -1.52 (1.48) (0.35) (1.18) Year dummies Yes Yes Yes District dummies Yes Yes Yes Firm characteristics Yes Yes Yes Constant -0.52 1.37 0.71 (1.46) (1.57) (1.31) Number of observations 8,642 5,144 3,464 R2 0.35 0.41 0.27 Source: Author's calculations. Note: Numbers in parentheses are t-statistics with robust standard errors. ** Significant at 5 percent. * Significant at 10 percent. Table 10.13 Externality and productivity in Java: nonmetallic minerals, 1990­2003 dependent variable: log value added per labor Explanatory variable 1990­95 1997­2000 2001­03 Log of capital per labor 0.28** 0.32** 0.21** (14.01) (5.93) (5.81) Log of same-industry employment (localization) -0.17* 0.11 -0.26** (1.67) (1.17) (2.19) Index of districts' diversity (urbanization) 0.19 1.20 2.26** (1.04) (0.69) (6.17) Year dummies Yes Yes Yes District dummies Yes Yes Yes Firm characteristics Yes Yes Yes Constant 1.07 -0.10 2.56 (1.36) (-0.09) (3.07) Number of observations 6,858 3,975 3,121 R2 0.54 0.43 0.72 Source: Author's calculations. Note: Numbers in parentheses are t-statistics with robust standard errors. ** Significant at 5 percent. * Significant at 10 percent. Table 10.14 Externality and productivity in Java: machinery, 1990­2003 dependent variable: log value added per labor Explanatory variable 1990­95 1997­2000 2001­03 Log of capital per labor 0.26** 0.29** 0.17** (17.10) (10.61) (9.89) Log of same-industry employment (localization) 0.10 0.31 0.24** (0.98) (1.27) (5.66) Index of districts' diversity (urbanization) -65.52* 0.13 156.76** (1.82) (0.07) (12.69) Year dummies Yes Yes Yes District dummies Yes Yes Yes Firm characteristics Yes Yes Yes Constant 1.07 -1.39 -0.36 (1.36) (-0.75) (1.08) Number of observations 8,151 3,996 3,675 R2 0.30 0.32 0.31 Source: Author's calculations. Note: Numbers in parentheses are t-statistics with robust standard errors. ** Significant at 5 percent. * Significant at 10 percent. 166 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA machinery, electrical machinery, transpor- static model. So, potentially, nonmetallic tation, and scientific equipment. The forces minerals and machinery may have a better of localization and urbanization are clearly chance than other industries of deconcetrat- nonexistent before and during the economic ing further to outlying locations. crisis. The coefficient of the same industry, though positive, is very weak statistically. Conclusions The same also applies to the urbanization As commonly observed in other countries, variable or the diversity index. Only later in in Indonesia as the economy was liberalized, the post-crisis years does the coefficient of economic activities tended to become more the industry, along with the diversity index, concentrated in a few places, which brought become positive, which means that the unintended negative externalities associated previous districts in which the industry is with agglomeration. This chapter illustrates found become even more specialized in the how public policies interacted with private production of machinery. incentives to mitigate this problem. After Previously it was said that if an indus- experiencing a period of concentration, try exhibits dynamic externalities, the past these policies were able to mitigate the con- industrial environment in that particular centration trend and to bring about distance location will affect the present-day produc- from the initial, historical agglomerations, tivity. Consequently, firms are reluctant to enabling industries to reconcentrate in move to locations with no prior history of smaller, less expensive cities, including those that particular industry. This makes it dif- in low-income or lagging regions in Java. ficult for industries to deconcentrate from Based on empirical exercises conducted primelocationstothehinterlands.Toaddress on Indonesia's four most important indus- this concern, we estimate the dynamic ver- tries, the chapter finds that this occurred sion of equation 10.3. For this, employment because the nature of externalities and in the industry and the diversity index are agglomerations favored industrial spill- replaced by their relevant past value with a overs--that is, localization was stronger lag of five years. The model is tested for the than urbanization effects. If externalities are 1990­95 period, when there was no huge in the form of localization, smaller cities are economic shock. The results are presented more likely to specialize in just one industry in table 10.15. The lag of the same-industry or in closely connected industries. However, employment is positive and significant for if the externalities happen to be urban in textiles and chemicals, while none of the nature, an industry will have to find a loca- coefficients of the diversity index or urban- tion in a diverse, large urban environment. ization is significant. The results mimic the The deconcentration process from Jakarta Table 10.15 Test of dynamic externalities in Java, 1990­95 Explanatory variable Textiles Chemicals Nonmetals Machinery Log of capital per labor 0.21** 0.32** 0.29** 0.27** (9.62) (19.25) (14.26) (16.97) Lag of log of same-industry 0.10** 0.10** 0.02 0.08 employment (4.11) (2.84) (0.42) (1.16) Lag of index of districts' diversity 0.06 0.71 -0.28 -1.13 (0.12) (0.93) (1.57) (0.83) Year dummies Yes Yes Yes Yes District dummies Yes Yes Yes Yes Firm characteristics Yes Yes Yes Yes Constant 0.28 -0.39 -0.18 0.36 (1.16) (1.08) (0.65) (0.54) Number of observations 18,636 8,480 6,818 7,980 R2 0.34 0.35 0.54 0.30 Source: Author's calculations. Note: Numbers in parentheses are t-statistics with robust standard errors. ** Significant at 5 percent. * Significant at 10 percent. Spatial agglomeration, firm productivity, and government policies in Indonesia 167 and BOTABEK is very apparent in non- important because it provides the basic metallic minerals and machinery industries, services that the local population needs but it is less evident in textiles and chemicals, and wants, reducing the incentives to since their externalities are dynamic--that migrate (permanently) to cities and is, firms are less willing to move to locations easing the pressures on urban areas. By without a prior history of the industry and locating in nonurban areas, firms can also hence no accumulated stock of knowledge. meet their need for workers with various Government policies were designed not skill levels and keep jobs in local areas. to interfere with private incentives (that is, firms will always look for the most efficient Notes sites in which to locate), but rather to com- plement them. Drawing from these lessons, Ari Kuncoro is a research associate at the Insti- the following measures could be imple- tute of Economic and Social Research, Univer- sity of Indonesia. mented in the future: 1. One important benefit of agglomeration · Improve our understanding of the behav- is that firms can learn from each other, creating ior of firms in different industries, espe- a synergy that collectively boosts their average cially regarding their choice of location. productivity. In this regard, there are two types of "positive" externalities. First is localization, · Improve infrastructure in lagging regions. in which firms learn about local inputs, output At the initial stage, the most effective pol- markets, and technological conditions in the icy is to improve or construct roads to cut same industry. Alternatively, firms learn from the costs and time of traveling between all firms in a city, where the diversity of local factory sites and markets or ports. industries enhances the environment for local Although, at the outset, the relocating information. This type of externality is called firm or new entrants may only be willing urbanization or, in the dynamic context, Jacobs to move within close proximity of the old externalities (Jacobs 1969). center, given enough time, they will grad- 2. Not only roads but also village infrastruc- ually fill up sites farther away, as the initial ture, such as village halls, schools, health centers, and markets, were constructed in rural areas. ones become more congested. 3. We use PODES (village potential) from · Decentralizethenationallicensingregime various years to construct road indicators. to local governments. This would allow 4. The influence was strongest for incorpo- firms, especially those that are the most rated firms that were most reliant on the central- reliant on the centralized bureaucracy for ized bureaucracy of financing, export licenses, financing, export-import licenses, and and other aspects of business. The economic other aspects of business, to locate closer liberalization in the mid-1980s gave firms bet- to the national capital. Responding to ter access to government and other centralized this, in 1999 the legislature passed the services; to take advantage of these opportuni- ties firms had to centralize in close proximity Decentralization Law, which was subse- to Jakarta because the bureaucratic process was quently enacted in 2001. Greater auton- centralized and communication was poor. omy is delegated to around 400 districts 5. For example, Krawang and Cikarang in in many areas, including in the fields of western Java, Kabupaten Semarang and Mage- public works, health, education, agricul- lang in the Semarang-Yogyakarta corridor in ture, industry, trade, and environment. central Java, and Pasuruan, Jombang, Mojokerto, It is true a lot of problems have emerged and Lamongan in eastern Java (see figure 1). since the launching of decentralization, But they did not locate exclusively in industrial such as the proliferation of new local zones. Batam Island was not taken into account taxes and local government corruption. because it had special regulations regarding However, these should be viewed as tran- industry exclusively for export. 6. This happens when there is total decon- sitory problems, as the economy is mov- centration of industries in which employment ing to a new equilibrium. in local industry is a fixed fraction of the local · Construct basic infrastructure, such as population in all cities. At the other end of the village halls, schools, health centers, spectrum, when gj(t) approaches its maximum and markets. Such infrastructure is very value of 2, an industry is totally concentrated in 168 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA one location, for example, at location k, such that the manufacturing survey, such as foreign direct Ekj(t)/Ej(t) = 1, when population share is minis- investment (FDI) versus other investment, the cule, or Pk(t)/P(t) 0, when population is con- share of capital equity ownership belonging to centrated in another location. various entities such as the central government, 7. We focus on four industries; textiles, local governments, private domestic investors, chemicals, nonmetallic minerals, and machinery. and foreign investors, and legal status. A study on manufacturing location in Indonesia looks at spatial concentration in Java,where about References 80 percent of manufacturing firms are located. Albala-Bertrand, José M., and Emmanuel C. Another consideration is that the location's Mamatzakis. 2004."The Impact of Public choice model is probably more applicable to Java Infrastructure on the Productivity of the because infrastructure, the quality of labor, and Chilean Economy." Review of Development other amenities located in the outer islands are Economics 8 (2): 266­78. not really comparable to those in Java. 8. The reason for choosing value added Central Statistical Agency. Various years. instead of output is that the former is less suscep- Manufacturing Annual Survey. Jakarta: tible to the extent of outsourcing and the use of Central Statistical Agency, Jakarta. same-industry intermediate inputs. At the aggre- Gelb,Alan, and others. 1988. Oil Windfalls: Blessing gate level,this could overstate the true net industry or Curse? New York: Oxford University Press. output with magnitudes that vary by the diversity Henderson, J. Vernon, and Ari Kuncoro. 1996. and size of the district, potentially biasing the esti- "Industrial Centralization in Indonesia." mate of true urbanization externalities. World Bank Economic Review 10 9. In the empirical formulation, agglomera- (3, September): 513­40. tion effects are modeled as external to the firm. Hence we can assume a constant return to scale Henderson, J. Vernon, Todd Lee, and Yung Joon (CRS) technology for firms so the output can be Lee. 2001."Scale Externalities in Korea." written in terms of output per labor or produc- Journal of Urban Economics 49 (3): 479­504. tivity. This can be easily modeled as the impact Jacobs, Jane. 1969. The Economics of Cities. New of agglomeration on firm productivity. There is York: Random House. no contradiction between the use of CRS with Mitra, Arup. 1999."Agglomeration Economies, increasing returns to economic scale: 100 firms as Manifested in Technical Efficiency at the can agglomerate in one location to create exter- Firm Level." Journal of Urban Economics nalities, which is obviously different than when 45 (3): 490­500. one CRS firm locates alone in a location with no externality, because nobody else is around. Romer, Paul. 1990."Increasing Returns and 10.To control for firm fixed effects,we employ a Long-Run Growth." Journal of Political several important firm characteristics available in Economy 94 (5): 1002­37. Spatial disparities and development policy in the Philippines Arsenio M. Balisacan, Hal Hill, and Sharon Faye Piza 114 Economic growth in the Philippines has view, for example, is that development efforts been quite anemic, barely exceeding the pop- have favored Luzon, particularly the national ulation growth rate, which has continued to capital region, Metro Manila, and discrimi- c h a p t e r expand rapidly at 2.3 percent a year for most nated against theVisayas and,especially,Min- of the past 25 years. It has quickened in the danao (see figure 11.1). Proponents of this present decade, but questions linger regard- view say that this development pattern has ing its sustainability. Even at the present pace led to substantial spatial differences in access c h a p t e r (per capita gross domestic product [GDP] to economic opportunities, in rates of pov- growth of 3­5 percent a year in 2004­07), erty reduction, and in the incidence of armed one can hardly argue that the Philippines has conflict. Indeed, economic activity has been come close to the growth trajectories of its highly uneven and concentrated particularly neighbors. It is thus not surprising that seri- in Metro Manila. Together with the two adja- ous students of Philippine development con- cent regions, Metro Manila produces about tend that shifting the economy to a higher 55 percent of the country's GDP. Socioeco- growth path--and keeping it there for the nomic indicators also vary significantly across long term--should be first and foremost on regions (and even across provinces within a the development agenda. region). The headcount poverty estimate for The country's similarly disappointing the two poorest regions is more than 10 times performance in poverty reduction simply that for the national capital. The Philippine mirrors its growth performance. This is not Human Development Report 2005 shows that unexpected. Every country that has chalked measures of deprivation,such as disparities in up significant achievements in poverty access to reliable water supply, electricity, and reduction and human development has also especially education, predict well the occur- done quite well in securing long-term eco- rence of armed encounters (HDN 2005). nomic growth. This correlation is not unex- To be sure, spatial economic disparities pected: economic growth is an essential con- need not be growth-reducing if these arise dition for the generation of resources needed from efficiencies associated with agglomera- to sustain investments in health, education, tion.Given scale economies and factor mobil- infrastructure, and good governance (law ity, as well as scarcity of investment funds, enforcement, regulation), among others. the spatial concentration of economic activi- Yet, more than a few observers of the Phil- ties leading to differential patterns of growth ippine economy contend that the poor per- across regions or areas of the country may in formance in economic growth and poverty fact be inevitable and even desirable from an reduction has to do partly with the large dis- overall economic growth perspective. How- parities in access to infrastructure and social ever, to prevent unreasonable spatial dispari- services across regions and island groups and ties in welfare during the development pro- between urban and rural areas. A widely held cess, the priority should be to improve the 169 170 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 11.1 Regions of the Philippines, 2007 PHILIPPINES Luzon Strait Regional Boundaries Provincial Boundaries Tuguegarao CAR CAGAYAN VALLEY ILOCOS Luzon San Fernando Baguio CENTRAL LUZON Philippine San Fernando CALABARZON Sea MANILA NATIONAL CAPITAL REGION Calamba BICOL Calapan Legaspi Mindoro Samar EASTERN VISAYAS Panay Tacloban Iloilo Cebu Leyte Cebu WESTERN MIMAROPA VISAYAS CENTRAL VISAYAS Bohol Palawan CARAGA Butuan NORTHERN Sulu Sea ZAMBOANGA MINDANAO PENINSULA Cagayan de Oro Pagadian Mindanao Cotabato Davao DAVAO Koronadal AUTONOMOUS REG. IN MUSLIM MINDANAO SOCCSKSARGEN MALAYSIA Celebes Sea market links between the leading and lagging The Philippines is ideally suited to a regions through greater factor mobility, par- study of regional dynamics and develop- ticularly labor mobility. Improving access to ment policy. With a population nearing 90 social services, particularly education and million people, the country is highly diverse health, in lagging regions should also be part in its geography, ecology, natural resource of the development agenda. endowments,economy,ethnicity,and culture. Spatial disparities and development policy in the Philippines 171 Comprising 7,100 islands, it is the second- Manila is by far the wealthiest region, largest archipelagic state in the world, after with a per capita income about 2.5 times Indonesia. It is estimated to have 110 ethnic the national average (see table 11.2). This is groups and 170 spoken languages. about double the income of the next richest This chapter provides an overview of region and 10 times that of the poorest. In spatial development dynamics in the Philip- fact, only two of the remaining 15 regions, pines in the past 25 years. Spatial develop- both distinctive in nature, have per capita ment is seen in the context of the country's 16 incomes above the national average.2 A regions and 77 provinces. Because the group- third group of regions may be regarded as ing of the country's provinces into regions is moderately well-off by national standards: based on considerations beyond economics, those whose per capita income is below the the spatial development story that emerges national average of P52,470 in 2003 but from the analysis of provincial data differs above the national average excluding Manila from that of regional data. Specifically, the of P38,600. They include a diverse group of chapter is organized as follows. It first dis- regions: the two adjacent to Manila, Cen- cusses the dimensions and patterns of spatial tral and Western Visayas, and Southern (regional, provincial, urban-rural) diversity. and Central Mindanao. A fourth group Diversity is seen in terms of economic per- comprises six poor regions: three in Luzon formance, economic and spatial attributes (Ilocos and Cagayan Valley in the north and (such as infrastructure development,agrarian Bicol in the south), Eastern Visayas, and structure, location), and various indicators of two regions in the western part of Min- social development outcomes (health status, danao (Caraga and Western Mindanao). literacy). The chapter then uses econometric Finally, the Autonomous Region of Muslim techniques to explain the differences in income growth and poverty reduction. The Table 11.1 Regional growth and structure in the Philippines, by region, 1975­2005 analysis makes use of an updated provincial Percent panel database covering 1985 and every three Region 1975­85 1985­95 1995­2005 1975­2005 years thereafter, which the authors have built over the years. Although the regions have Average growth of regional GDP (1985 prices) longer data, covering years before 1985, data Philippines 2.5 2.5 4.3 3.4 comparability over time is a major problem Luzon 2.6 2.8 4.3 3.6 owing to numerous changes in the grouping NCR 2.4 2.8 4.9 3.7 of provinces into regions. Finally, the chapter Central Luzon and South Tagalog 2.6 3.1 3.6 3.4 Other Luzon 3.0 2.3 4.3 3.4 highlights the policy lessons and implications Visayas 2.4 2.1 4.4 3.4 of the study for regional development and Central Visayas 2.7 2.6 5.1 3.9 poverty reduction. Other Visayas 2.3 1.7 3.9 3.1 Mindanao 2.2 1.7 3.8 2.8 Regional development patterns Share of national GDP Luzon 62.6 64.8 65.7 64.4 Manila dominates the Philippine economy, NCR 28.8 31.6 30.7 29.9 with the National Capital Region (NCR) Central Luzon and South Tagalog 23.3 23.2 24.7 24.3 generating a little more than one-third of the Other Luzon 10.5 10.0 10.3 10.1 country's GDP in recent years (see table 11.1).1 Visayas 16.7 16.3 16.3 16.3 With the two regions surrounding it-- Central Visayas 6.4 6.5 6.9 6.5 Other Visayas 10.3 9.8 9.4 9.8 Central Luzon and Southern Tagalog--this Mindanao 20.8 19.0 18.0 19.3 central zone produces about 55 percent of Share of total population the country's GDP. The island of Luzon, on Luzon 54.3 55.1 56.0 55.1 which they are located, contributes almost NCR 12.3 13.2 13.0 12.8 two-thirds of the national economy, making Central Luzon and South Tagalog 22.8 23.9 26.0 24.2 it by far the largest of the three major island Other Luzon 19.2 18.0 17.1 18.1 Visayas 23.2 21.4 20.3 21.7 groupings. Luzon's economy has also grown Central Visayas 7.9 7.5 7.5 7.6 marginally faster than the national economy Other Visayas 15.3 13.9 12.8 14.0 since the 1970s, resulting in a gradual rise in Mindanao 22.5 23.5 23.7 23.2 its national share. Source: Authors' calculations based on NSCB, National Income Accounts, regional link series for 1975­2003. 172 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 11.2 Key economic indicators in the Philippines, by region, 1988 and 2003 Percent Per capita regional GDP Composition of regional GDP (Philippines, 1988 =100) Agriculture Industry Services Regional GDP growth rates, Region 1988 2005 1988 2005 1988 2005 1988 2005 1988­2005 Philippines 100.0 126.5 23.2 14.5 34.8 31.7 41.9 53.7 3.6 NCR 232.6 318.7 0.0 0.0 45.5 31.2 54.5 68.8 4.0 CAR 98.7 159.8 20.4 9.7 56.4 59.6 23.2 30.7 4.8 Ilocos 50.6 68.9 42.5 34.9 15.9 14.9 41.6 50.1 3.7 Cagayan Valley 53.0 68.2 52.5 41.2 13.7 16.4 33.7 42.4 3.4 Central Luzon 94.0 99.3 22.6 18.6 39.3 36.2 38.0 45.2 3.1 Southern Tagalog 114.0 123.7 29.0 20.7 40.6 38.6 30.5 40.7 3.7 Bicol 42.7 59.1 41.3 19.7 18.6 24.0 40.1 56.3 3.2 Western Visayas 76.6 114.4 33.4 20.3 27.1 28.6 39.6 51.1 3.6 Central Visayas 86.5 120.5 14.5 8.8 36.4 31.7 49.1 59.5 4.1 Eastern Visayas 47.6 59.5 35.3 27.6 33.1 32.2 31.6 40.1 2.7 Western Mindanao 64.8 90.6 46.8 37.8 21.0 19.4 32.2 42.8 4.2 Northern Mindanao 104.4 132.2 48.0 27.2 20.4 32.9 31.6 40.0 5.2 Southern Mindanao 101.3 123.9 42.0 23.3 24.4 29.9 33.6 46.8 1.3 Central Mindanao 67.6 102.3 41.6 38.4 38.3 32.2 20.1 29.3 6.4 ARMM 70.4 30.6 57.2 51.0 13.8 10.9 29.0 38.1 -1.0 Caraga 88.2 59.7 33.3 33.5 36.2 27.1 30.5 39.4 -0.5 Sources: Authors' calculations based on NSCB, National Income Accounts (various years); NSO, Family Income and Expenditure Surveys (various years). Note: Average per capita GDP for the Philippines in 1988 was P 50,242 (in 2005 prices). Regional GDP shares and growth rates are averages for three adjoining years (that is, the 1988 figure is the average for 1987­89, while the 2005 figure is the average for 2004­06). Regions are defined consistently across years. Provincial income shares from the household survey data are used. Figure 11.2 Regional growth versus initial regional income including ARMM, grew more slowly than the national average of 3.6 percent. At the 5 Quadrant IV CAR Quadrant I other end of the range, the richest region, the NCR, grew at about the same pace as the 1985­2003 national average. Figure 11.2 investigates rate, the relationship between regional growth 3 rates and (initial year) average income. The C. Visayas growth W. Mindanao top right-hand quadrant (quadrant I) rep- W. Visayas N. Mindanao Bicol NCR GDP Cagayan resents regions with above-average growth 1 C. Mindanao and income; quadrant II represents regions E. Visayas S. Luzon Ilocos C. Luzon S. Mindanao with below-average growth and above-aver- regional Caraga age income; quadrant III represents regions capita with below-average growth and income; and ­1 per quadrant IV represents regions with above- ARMM average growth and below-average income. In general, the more heavily quadrants II average Quadrant III Quadrant II ­3 and IV are populated, the more likely are 8.3 8.8 9.3 9.8 10.3 log of per capita regional GDP, 1985 regional differentials to be narrowing. In fact, the majority of regions are in these two Source: Authors' calculations based on NSCB, National Income Accounts (various years); NSO, Family Income and quadrants: 2 in quadrant II and 8 in quad- Expenditure Survey (various years). Note: Regions are consistently defined across the period. rant IV, out of a total of 16. However, the clustering of regions close to the national Mindanao (ARMM) has to be grouped sep- average growth and the fact that the NCR arately owing to its extremely low income-- and ARMM are such outliers caution against less than half that of the poor grouping and drawing too robust a conclusion. less than one-quarter the national average. We formally test for the presence of con- The last column of table 11.2 shows aver- vergence by estimating a standard regional age annual growth of regional GDP between growth equation, to determine whether 1985 and 2005. Most of the poor regions, incomes are converging to the mean over Spatial disparities and development policy in the Philippines 173 time. The evidence is mixed, and the results table 11.2), poverty indicators vary con- are sensitive to the selection of administra- siderably across regions. However, Metro tive boundaries. That is, as shown below, Manila consistently has the lowest poverty, the provincial data indicate convergence, while Bicol, Western Mindanao, and the whereas the regional data do not.One plausi- Visayas, have the highest. In 2003 the pov- ble explanation for these mixed results is that erty incidence in Bicol and Western Mind- a number of administrative regions contain anao was roughly 10 times higher than in groups of provinces with a wide range of per Metro Manila. Some significant re-rankings capita income. We return to the provincial also occurred: ARMM became the poorest data later in the chapter to explore further region in 2003, after being the third-least- the determinants of local income growth. poor region (out of 16 regions) in 1988. Even more significant is the differential Social indicators evolution of poverty over time. In 2 regions, Table 11.3 shows indicators of poverty, Western Mindanao and ARMM, poverty was inequality, the human development index higher in 2003 than in 1988. This increase (HDI), life expectancy, and literacy by region also shows up in measures reflecting human and between two periods. These indicators development deprivation, particularly in generally correlate quite closely, although the areas of health and education (HDN there are some deviations. For instance, 2005). Toward the close of the 1990s, these regional mean income is highly correlated two regions, particularly ARMM, were at the with poverty incidence (the Spearman cor- center of violent confrontations between the relation coefficient is 0.78), the HDI (0.85), military and armed dissidents. and functional literacy (0.75). But it is weakly The Philippines is a high-inequality correlated with the Gini ratio (0.20) and pri- country compared with most of Asia, with mary enrollment (0.39).As expected,the cor- all but one of its regions (Central Luzon) relation between the HDI and poverty is also registering a Gini ratio of at least 40 in 2003. high (0.85), but not the correlation between Income inequality is particularly high in the HDI and the Gini ratio (0.05). most of theVisayas as well as in Mindanao-- As expected given the regions'very diverse ARMM being a notable exception--owing records of growth (see the last column in to the highly inequitable distribution of Table 11.3 Social indicators in the Philippines, by region, 1988 and 2003 Poverty Primary and Human Income Gini Life expectancy Adult functional secondary development Incidence Percent of total ratio at birth literacy rate enrollment rate index Region 1988 2003 1988 2003 1988 2003 1988 2003 1988 2003 1994 2003 1990 2003 Philippines 34.4 26.0 100.0 100.0 44.0 46.6 64.4 68.3 73.5 84.2 82.7 91.7 0.713 0.721 NCR 9.5 4.9 3.8 2.6 44.2 42.9 66.4 70.0 90.0 94.6 91.7 92.5 0.944 0.804 CAR 39.1 15.3 2.2 1.0 37.2 43.0 60.5 66.2 82.9 85.5 90.7 95.3 -- 0.648 Ilocos 25.5 16.9 4.3 3.4 38.1 41.3 65.4 69.5 71.9 88.4 90.2 91.6 0.592 0.649 Cagayan Valley 39.2 26.2 4.4 3.4 40.5 47.1 62.5 67.0 71.8 84.3 86.3 92.6 0.560 0.603 Central Luzon 15.3 13.6 4.4 5.7 39.6 37.7 67.2 70.9 82.1 86.8 87.8 91.0 0.695 0.654 Southern Tagalog 31.7 20.8 11.1 13.0 41.3 43.7 65.4 68.9 75.9 88.7 84.6 92.8 0.654 0.646 Bicol 60.9 45.7 12.8 10.7 41.1 49.7 63.0 68.6 67.5 79.8 84.1 90.6 0.488 0.538 Western Visayas 34.4 26.7 9.2 7.8 42.2 46.2 63.8 68.3 66.0 81.5 85.0 93.9 0.527 0.601 Central Visayas 55.2 36.6 12.0 10.4 44.5 47.3 66.1 70.7 68.2 81.6 80.7 90.6 0.528 0.592 Eastern Visayas 53.7 45.0 8.6 8.2 39.4 48.2 59.8 65.6 60.4 76.5 80.0 91.5 0.473 0.520 Western Mindanao 47.6 49.7 5.8 7.6 45.3 52.6 61.4 66.3 62.7 73.0 76.3 94.9 0.458 0.524 Northern Mindanao 44.9 29.8 4.9 4.1 48.8 47.9 62.4 68.6 75.5 82.6 72.2 90.6 0.531 0.610 Southern Mindanao 46.9 26.8 8.8 7.2 41.6 50.7 63.2 68.8 68.7 77.4 72.4 90.1 0.571 0.624 Central Mindanao 35.8 34.1 3.2 4.2 40.8 45.9 61.2 66.5 61.0 80.0 81.1 93.1 0.479 0.551 ARMM 23.4 63.4 2.0 7.2 34.3 40.6 52.0 54.2 55.2 65.9 57.6 81.0 -- 0.370 Caraga 30.1 36.9 2.5 3.7 37.8 44.9 60.2 64.8 75.2 80.5 76.2 93.2 -- 0.531 Sources: Authors' calculations based on NSO, Family and Income Expenditure Survey; NSO, Functional Literacy, Education, and Mass Media Survey; HDN (various years). -- Not available. 174 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA physical assets, particularly land, and the remaining 45 percent of the population is mix of economic activities (mining, planta- divided fairly evenly between the Visayas tion agriculture) in these regions. and Mindanao. Within Luzon, Manila and These high-inequality regions have, for the two surrounding regions dominate, with example, land Gini ratios of close to 60 per- a gradually rising share in the range 35­40 cent, while the comparable figures for most percent of the population. of Luzon (excluding Bicol) are close to, or These patterns reflect the interplay of below, 50 percent. As Balisacan (2003) has regional fertility differentials and migration. shown, it is the inequality within regions-- Historically, the major migration flows were not the inequality between regions--that into the national capital and its surrounds accounts for more than 80 percent of the and into the frontier regions,principally Min- national variation in household income. He danao (Pernia and others 1983). Since 1980, further shows that high-inequality regions the dominant migration stream has been into tend to have low steady-state growth rates the two regions surrounding Manila, espe- compared to their lower-inequality coun- cially Southern Tagalog (see table 11.4). Only terparts. Moreover, his analysis of the data two other regions have had (modest) net in- covering the past two decades indicates that migration: the region of CentralVisayas,with changes in poverty incidence (as well as its capital the relatively prosperous second in other measures of income poverty) are city of Cebu, and the resource-rich region of attributable largely to changes in overall per Northern Mindanao. Thus relative income capita income within regions, rather than to differentials, together with employment and changes in income or asset inequality within education opportunities,drive these patterns. or between regions. In spite of decentralization, and the disman- tling of the centralizing bias in favor of the Population, labor, and migration capital region, it remains the dominant des- Philippine demographics more or less reflect tination of migrant flows.3 In other words, economic patterns. A little more than half migration continues to be, de facto, a key of the nation's population lives in Luzon, instrument of regional adjustment, includ- whose share of population has been rising ing the well-known phenomenon of migra- gradually since the 1970s (table 11.1). The tion out of poverty. Regional labor markets Table 11.4 Population and intraregional migration in the Philippines, by region, 2000 Average annual growth Total Population density rate (percent) population (people per square Region (thousands) kilometer) 1980­90 1990­2000 Migration ratea Philippines 76,504 255 2.3 2.3 0 NCR 9,933 16,091 2.9 2.2 -22 CAR 1,365 70 2.3 1.8 -1 Ilocos 4,200 318 2.0 1.7 -1 Cagayan Valley 2,813 90 2.0 1.8 -5 Central Luzon 8,031 437 2.6 2.6 12 South Tagalog 11,794 239 3.0 3.6 26 Bicol 4,687 258 1.2 1.8 -10 Western Visayas 6,211 301 1.8 1.4 -6 Central Visayas 5,707 359 1.9 2.2 -- Eastern Visayas 3,610 155 0.9 1.7 -6 Western Mindanao 3,091 161 2.2 2.3 -9 Northern Mindanao 2,748 170 2.2 2.2 4 Southern Mindanao 5,189 183 3.0 2.6 -1 Central Mindanao 2,598 144 3.3 2.5 -9 ARMM 2,412 95 3.0 2.7 -9 Caraga 2,095 98 2.5 1.7 -6 Source: Authors' calculations based on NSO, Census of Population and Housing (1990, 2000). -- Not available. Note: Calculations are based on intracountry migration. a. Net migrants, defined as in-migrants less out-migrants, per 1,000 population in 2000. Spatial disparities and development policy in the Philippines 175 have also been liberalized gradually. The set- tion, land-line telecommunications, major ting of minimum wages has been decentral- trunk roads, international airports), which ized, and some regions, mainly poorer ones, in turn prescribe a role for government as are beginning to compete for employment regulator, though not necessarily as pro- by offering more flexible labor market regu- vider. Third, following a decentralization lations (Sicat 2003). program, there will be many players in the industry, including several tiers of govern- Infrastructure and integration ment, the state-owned providers, and some Infrastructure is the glue that unifies the foreign firms, as well as a number of regu- national economy, and it is in many respects latory agencies. There are therefore major the single most important instrument of coordination issues. regional policy. It enables people and goods Indicators of Philippine infrastructure to move quickly and efficiently around a generally follow per capita income rank- country. The composition of this infrastruc- ings, with the better-off regions having the ture also matters. For example, efficient con- capacity (and political influence) to fund nections to the global economy alongside the better-quality physical facilities. This is poorer provision of domestic networks--an illustrated in the standard indicators of road increasingly accurate characterization of the density, access to water, irrigation, electric- situation in the Philippines--will result in ity, and telephone density (see table 11.5). a series of internationally oriented enclaves Manila and its two surrounding regions of economic activity weakly integrated to clearly register above-average physical the hinterland. infrastructure indicators in most respects. Effective infrastructure provision Outside this central region, the picture is requires competent governance. First, many more variable. One notable feature is that infrastructure projects entail long gestation Mindanao does not emerge as a particularly periods and therefore require predictable infrastructure-deficient region by Philip- financing and policies. Second, a num- pine standards, reflecting the region's high- ber of sectors have "natural monopoly" priority status with both the government characteristics (for example, power genera- and the donor community. Table 11.5 Infrastructure indicators in the Philippines, by region, 1988 and 2004 or 2005 Road density (kilometers Access to potable water Access to electricity Telephone line density Irrigation serviced per square kilometer)a (percent of households) (percent of households) per 100 households (percent)b Region 1988 2005 1988 2004 1988 2004 1988 2005 1988 2005 Philippines 0.27 0.78 71.9 80.1 59.9 79.5 1.6 7.8 46.4 45.2 NCR 4.29 15.55 92.0 85.7 97.6 99.0 10.1 25.2 -- -- CAR 0.12 0.33 66.2 76.2 51.7 75.5 0.6 5.7 35.3 75.5 Ilocos 0.53 1.12 83.9 89.5 70.0 86.2 0.3 4.3 67.4 64.5 Cagayan Valley 0.14 0.43 80.2 87.9 61.3 78.4 0.1 1.2 54.3 42.8 Central Luzon 0.61 0.94 96.0 96.2 83.4 94.4 0.5 4.8 64.7 53.8 Southern Tagalog 0.28 0.62 78.1 84.5 63.8 86.1 0.4 8.4 48.8 49.6 Bicol 0.14 0.44 60.9 74.2 40.7 66.6 0.2 2.1 38.6 49.6 Western Visayas 0.35 0.77 54.4 73.4 43.5 72.6 0.6 6.2 59.4 39.4 Central Visayas 0.36 1.42 57.6 74.8 43.6 74.1 0.9 7.8 43.3 57.8 Eastern Visayas 0.37 0.75 60.9 79.5 33.2 68.4 0.1 3.7 45.4 59.4 Western Mindanao 0.10 1.11 40.8 59.7 43.4 54.5 0.3 1.0 58.5 48.4 Northern Mindanao 0.23 0.83 66.2 79.8 56.3 72.5 0.2 4.6 49.1 43.4 Southern Mindanao 0.12 0.32 73.6 69.9 52.1 70.9 0.5 5.5 41.0 36.3 Central Mindanao 0.12 0.56 69.7 74.3 46.6 66.8 0.1 2.8 34.6 28.3 ARMM 0.13 0.34 22.9 40.9 20.2 44.0 0.1 1.4 17.9 14.8 Caraga 0.15 0.36 77.7 79.7 61.1 69.2 0.1 5.1 33.0 24.7 Sources: Authors' calculations based on data from the Department of Public Works and Highways; NSO, Annual Poverty Indicators Survey (various years); NSO, Family Income and Expenditure Survey (various years); NSCB, Philippine Statistical Yearbook (various years). -- Not available. a. Road density is adjusted for quality (concrete equivalent). b. Irrigation serviced refers to the ratio of total irrigated area to potential irrigable area. 176 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA The Philippine infrastructure report card structure projects in many countries. But is deficient in key respects (Llanto 2007), large infrastructure investments appear to and this appears to be holding back the be unusually politicized in the Philippines, process of efficient regional economic inte- with several key projects over the past decade gration. In a recent Global Competitiveness remaining incomplete. Report, the country's infrastructure perfor- The picture varies considerably by subsec- mance was ranked 71 out of 131 countries tor. There have been some positive achieve- in 2007­08 (World Economic Forum 2007). ments. For example, domestic shipping, civil The country is underinvesting in infrastruc- aviation, and cellular telecommunications ture: its ratio of infrastructure investment to services were effectively deregulated during GDP is about half the East Asian average. the 1990s.5 Roads are perhaps the weakest There are three interrelated problems. link, and here coordination failures are seri- First, the country has chronic fiscal con- ous. The two major national agencies with straints, as a result of past fiscal crises and responsibilities for roads--the Department the continuing poor revenue performance of of Public Works and Highways and the Toll the national government. Fiscal constraints Regulatory Board--do not coordinate their have a particularly adverse effect on infra- activities effectively. There also appears structure, because capital works invariably to be a "missing middle" in the road net- are the first to be cut in budget-pruning exer- work. The national government assumes cises.4 Second, the overall regulatory frame- responsibility for the major trunk network. work lacks cohesion, coordination among Although local governments have limited national agencies and among the various infrastructure budgets,6 they are responsive tiers of government, and clear division of to local constituencies demanding farm- responsibilities. About 30 national agencies to-market roads. Reflecting the division of are involved in infrastructure decision mak- political power, secondary roads connecting ing. Third, national-level decision makers the national and local road networks suffer appear unable or unwilling to deliver the from continued neglect and constitute the long-term policy predictability and guaran- major weak link. tees that major private (and especially for- Is the Philippines becoming a more eign) providers require, resulting in potential spatially integrated economy over time? In suppliers factoring in very large risk premi- figure 11.3, we test for this by presenting ums. Corruption and political patronage estimates of coefficients of variation for are associated with the award of large infra- provincial prices during 1985­2003.7 Two sets of price indexes are shown, one consist- ing of a basket of food and nonfood items ("all commodities") and the other consist- Figure 11.3 Variation in provincial prices, 1985­2003 ing of food items only.8 Ideally, the spatial 19 comparison should involve only tradable goods. Arguably, food is highly tradable; hence the latter price index can be regarded 18 (%) as a reasonable measure for comparing the noitari regional prices of tradable goods. As figure va 17 11.3 shows, the two indexes exhibit the same oftneic all commodities pattern: the coefficients of variation tend to ffi rise in the second half of the 1980s through coe 16 early 2000s, suggesting that impulses for food domestic integration have been muted by widening regional price variations in recent 15 1985 1988 1991 1994 1997 2000 2003 years. This pattern is an outcome partly of evolving disparities in infrastructure devel- All Food opment and institutional arrangements and Source: Balisacan (2001), updated to 2003 using the consumer price indexes in NSCB, Philippine Statistical Yearbook partly of deregulatory reforms in transport (various years). Note: Figures pertain to coefficients of variation of cost-of-living indexes for a basket of goods and services. and related services. Spatial disparities and development policy in the Philippines 177 Determinants of local growth growth, together with a host of other factors, and poverty reduction influences poverty reduction. In his model, these other factors affect the speed of poverty In this section, we use subnational panel reduction either directly by changing the dis- data to explore the determinants of income tribution of a given economic pie (hereafter growth and poverty reduction. The units referred to as the redistribution channel) or of observation are provinces, which show indirectly by expanding the economic pie for remarkable diversity in terms of economic each person in society (hereafter referred to performance and poverty reduction. The as the growth channel). These factors can be units and variables are consistently defined, grouped into two types: both across space and over time. The his- torical and institutional contexts are largely · Initial economic and institutional condi- similar across these units (same legal system, tions (in or around 1988), which include same political administration).Moreover,the initialmeanprovincialpercapitaincome, major sources of heterogeneity--that is,tech- initial distribution of per capita income, nologies, tastes--are likely to be less severe initial stock of human capital, political for these data than for cross-country data. "dynasty" (as a proxy for political com- Hence, the estimation problems concerning petitiveness), and ethno-linguistic frag- cross-country data are likely to be less serious mentation and for the subnational panel data set. · Time-varying policy variables (difference The long-term relationship between Phil- during 1988­2003), which include the ippine poverty and income growth is evident simple adult literacy rate, agricultural in data on the country's 77 provinces. This is terms of trade (as a proxy for economic shown in figure 11.4, which plots the change incentives),accesstoinfrastructure(repre- in poverty incidence between 1985 and 2003 sented by electricity and good-quality and the corresponding percentage change road), and implementation of the Com- in real family income per capita, adjusted prehensive Agrarian Reform Program for provincial cost-of-living differences.9 (CARP). Clearly, as in cross-country data on growth The income growth regression is specified and poverty, the pace of poverty reduction at as in the standard Barro and Sala-i-Martin the provincial level is closely linked to local framework. The poverty reduction regres- economic performance. However, there are sion adds the income growth rate variable significant departures from the fitted line to the set of explanatory variables associ- (that is, provinces not conforming to the ated with the rate of poverty reduction. This "average pattern"), suggesting that factors amounts to estimating the income growth other than the local economic growth rate are influencing the evolution of poverty. One set of such factors may have to do Figure 11.4 Income growth and poverty reduction, Philippine provinces, 1985­2003 with the relatively large variation in access 60 to infrastructure and social services across regions, island groups, and provinces. As points) 40 noted earlier, a widely held view is that devel- opment efforts have favored Luzon and dis- 20 criminated against theVisayas and,especially, (percentage Mindanao. Proponents of this view say that 0 this development pattern has led to substan- ­20 tial spatial differences in access to economic incidence opportunities, in rates of poverty reduction, ­40 and in incidence of armed conflict. poverty in Adopting the growth framework devel- ­60 oped by Barro and Sala-i-Martin (2004), change Balisacan (2007) traces the quantitative sig- ­80 ­0.4 ­0.2 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 nificance of the channels by which income percent annual increase in per capita income, 1985­2003 178 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA and poverty reduction equations simultane- of convergence (2.2 percent), it would take ously using the three-stage least squares esti- 31 years to halve the gap between the initial mation technique. Only variables that are and the steady-state incomes. Compared to significant in the reduced-form estimates a similar estimate for China's 30 provinces of the growth and poverty reduction equa- and municipalities of 1 percent a year on tions are retained. The regression results are average during 1960­2000 (Song 2007), the summarized in table 11.6. The annex to this estimate for the Philippines is quite high. chapter shows the complete list of variables, Among the initial conditions, the level including descriptive statistics. of human capital stock (as proxied by the The magnitude of the coefficient estimate child mortality rate) is found to be statisti- for initial income implies that (conditional) cally significant at conventional levels. This convergence of provincial incomes occurs finding of a positive association between at a rate of 2.2 percent a year. There is thus growth performance and human capital is a growth premium for late starters; that is, consistent with most other studies on deter- provinces that have initially lower mean minants of income growth. The magnitude incomes tend to grow faster. The estimate of the coefficient, however, is comparatively is, however, much lower than the figure of 9 small. An increase of 10 percent in the mor- percent a year given by Balisacan and Fuwa tality rate relative to the mean for all prov- (2004).10 The present estimate is compa- inces (84.7 in 1988) would reduce the rate of rable to estimates of regional income con- provincial income growth by 0.2 percentage vergence for Europe, Japan, and the United point a year. Put differently, if the mortality States, which cluster around 2 percent a year rate in the province with the highest mortal- (Barro and Sala-i-Martin 2004). At this rate ity rate (Western Samar) were to fall to the average level for all provinces--that is, from Table 11.6 Determinants of local growth and poverty reduction in the Philippines 121.1 to 84.7 or by 30 percent (annex)--the income growth rate for that province would Explanatory variable Mean income growth Rate of poverty reductiona increase by 0.7 percentage points a year, all Mean income growth -1.30161** other things remaining equal. (-5.18) All the time-varying policy variables are Change in literacy 0.00066** -0.00077 (2.66) (-1.45) significant and have the expected signs. In Change in electricity 0.00031** conformity with theory and most cross- (2.81) country regressions, improvements in liter- Change in road density 0.04649** -0.07067** acy and access to infrastructure (electricity (2.41) (-1.95) and roads) have a positive effect on income Change in CARP 0.03211** 0.00748 (3.55) (0.38) growth. The magnitude of those effects, Change in agricultural terms of trade 0.01346** however, is surprisingly small. In the case (1.95) of literacy, even a 20 percent improvement Initial per capita income (log) -0.02106** in the overall provincial average increase of (-3.29) 3.8 percentage points a year (annex) would Initial mortality -0.00019* 0.00035* (-1.89) (1.86) see income growth increasing by only 0.05 Landlock 0.00754** 0.00615 percentage points. This limited gain from (2.29) (1.05) an improvement in the simple literacy rate Initial Gini ratio 0.00806** can be attributed to the relatively high rate (3.02) Initial Gini ratio squared -0.00012** for the provinces as a group (91.4 percent (-2.98) in 2003). This average, however, conceals Constant 0.06261 -0.01666 the large variation that exists across prov- R2 0.62850 0.64880 inces. For provinces that are well below the Sample size 71 71 national average, an improvement in the Source: Balisacan (2007). literacy rate to, say, the national average Note: The estimation procedure used is three-stage least squares regression. Figures in parentheses are t-ratios. Other variables included in the estimation but not significant in both the growth and poverty regressions are not shown. could have a major impact on local income a. The poverty measure used is headcount, defined as the proportion of the population deemed poor. The dependent variable is the average annual rate of headcount reduction between 1988 and 2003 so that a negative coefficient for a growth. For example, if the province with variable implies that the variable has a positive effect on poverty reduction. the lowest literacy rate in 2003 (Tawi-Tawi, ** Significant at 5 percent level. * Significant at 10 percent level. at 63.3 percent) were to achieve the average Spatial disparities and development policy in the Philippines 179 rate for all provinces (91.4 percent), the elasticity"of poverty reduction.This elasticity income growth rate of that province would clusters around 1.3: a 10 percent increase in increase by 1.8 percentage points a year, all the income growth rate increases the poverty other things remaining the same. reduction rate by roughly 13 percent. These Increments in land reform implementa- estimates are much lower than those reported tion (CARP) have a positive and significant for other developing countries. For example, effect on the mean income growth rate. A 25 using parameter estimates of inequality distri- percent increase in the pace of CARP imple- bution for each country, Cline (2004) obtains mentation (that is, an increase in the average growth elasticities of 2.9 for China, 3.0 for change for all provinces from 80 percent to Indonesia, and 3.5 for Thailand.11 Ravallion 100 percent, thereby effectively completing (2001) obtained a growth elasticity of 2.5 for implementation) would raise the income 47 developing countries, based on a bivariate growth rate by 0.6 percentage points a year. regression of the proportionate changes in This is a significant result considering that their poverty rate and mean income. A simi- land reform is often seen as a policy tool lar bivariate regression of the data used in this mainly for achieving noneconomic objec- chapter gives an elasticity of 1.5. Hence, by all tives. The result suggests that addressing these indications, the growth elasticity in the access to productive assets would improve Philippines has been quite muted by interna- efficiency, thereby raising the economy's tional standards. subsequent income growth rates, as argued Clearly, the very low income growth cogently by Bourguignon (2004). achieved in recent years is a key factor in The policy variables and the variables rep- the country's sluggish rate of poverty reduc- resenting initial conditions, except those per- tion. Still, even this modest level of income taining to human capital and infrastructure, growth could have delivered more pov- are found mainly to exert an indirect effect erty reduction than what would have been on poverty reduction through their effect on realized if the growth elasticity in the overall income growth. For infrastructure, Philippines had come close to that in neigh- particularly transport, and, to some extent, boring countries. initial human capital, both direct and indi- The finding that policy levers often iden- rect effects are operative and, taken together, tified as tools for achieving equity objec- have a positive impact on the pace of poverty tives--human capital and asset reform reduction. Particularly remarkable is the lack through CARP--have rather weak discern- of direct response of poverty to CARP. Con- ible direct effects on poverty reduction is sidering that the agrarian reform program is quite disturbing. Their effects are felt mostly touted as an equity tool, this result is not only indirectly through the income growth pro- surprising but also inconsistent with earlier cess. In other words, even programs suppos- findings. This is not to say that CARP has edly targeted at poverty, such as CARP, have no effect on the poor. It has, but its effect is been largely neutral from an income distri- mainly through the income growth channel. bution viewpoint. One interpretation of this Taken together, the regression results show result is that the implementation of such very limited direct effects of recent policies programs has been poorly targeted. Indeed, and institutions on the speed of poverty the country's record in administering direct reduction; their effects get transmitted indi- antipoverty programs, such as food, credit, rectly to poverty reduction, mainly through and housing subsidy programs, has been overall income growth. quite disappointing (Balisacan and Edillon Another interesting observation from the 2005). These programs have had high leak- above study, as well as other studies using the ages to the nonpoor, been administratively same provincial data (for example, Balisacan costly to implement, and encouraged unin- and Fuwa 2004),concerns the extent to which tended rent-seeking processes. poverty responds to overall income, after accounting for the influences of other fac- Conclusions tors noted above. This response can be aptly The very high spatial disparity in economic summarized by what is referred to as"growth performance and social development in 180 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA the Philippines is quite remarkable. This Spending priority should be accorded chapter has shown that, indeed, poverty as well to social services, especially health has a strongly spatial dimension, with some and education, in lagging regions. Revers- regions and provinces far more multidimen- ing the significant decline in education and sionally deprived than others. Some areas health spending in recent years is expected of the country have human development to unlock the potential of human capital outcomes comparable with those found as a "deep determinant" of income growth in more economically advanced countries; and poverty reduction. However, given the for example, Metro Manila's HDI for 2003 fiscal bind, the targeting of public spending is comparable with that of Thailand, and must be improved so that poorer individu- the province of Rizal's HDI is comparable als, especially in lagging areas, would receive with that of Ukraine. Sadly, many other proportionately more opportunities for pub- areas have outcomes comparable with those licly funded social services. Unfortunately, of the poorest countries of the world; for the country's record in administering direct example, the ARMM provinces have HDI antipoverty programs, such as agrarian scores comparable with those of Ghana, reform and food, credit, and housing sub- Myanmar, and Sudan. In recent years some sidy programs, has been quite disappointing. regions have done quite well in attaining These programs have had high leakages to high per capita income growth and poverty the nonpoor, been administratively costly reduction, but others have experienced falls to implement, and encouraged unintended in their average per capita income and an rent-seeking processes. Clearly, investing in increase in poverty. good governance has to be part of the over- The Philippines is significantly under- all reform agenda. investing in infrastructure, particularly in transport and electricity, owing to continual Notes fiscal crises and an unattractive commercial Arsenio Balisacan is professor of economics at climate for long-term private investors. This the University of the Philippines Diliman; Hal not only reduces overall efficiency (growth) Hill is the H. W. Arndt Professor of Southeast but also limits domestic mobility of fac- Economies at the Australian National University; tors, goods, and people, hindering the full and Sharon Faye Piza is a research fellow at the participation of lagging regions from the Asia-Pacific Policy Center. growth process in leading regions or urban 1. The fragmentation of administrative centers. The high cost of mobility, especially boundaries complicates the task of regional that of labor, creates spatial disparities in development analysis over time. For the pur- welfare levels. poses of consistency, throughout this chapter we use the 1997 classification, which groups the The government's allocation of scarce provinces into 16 regions, unless otherwise spec- infrastructure funds has had implications ified. Currently (that is, in 2007), the number of for regional development patterns. Follow- regions is 17, following the division of Region IV ing the dismantling of the old import sub- (Southern Tagalog) into 2 regions. stitution growth regime, the new driver of 2. These are the Cordillera Administrative spatial development patterns has been the Region (CAR) and Northern Mindanao. location decisions of export zones. In this 3. This region is also the major source of context, the Philippine government (and the estimated 8 million Filipinos residing abroad. donors) has been more inclined to invest Their remittances, estimated to be equivalent to in internationally oriented infrastructure almost 50 percent of merchandise exports, are (ports, harbors, and associated facilities) the third largest in the developing world (Bur- gess and Haksar 2005). than in domestic transport networks and 4. As a corollary, there is a tendency to corridors. The effect has been to reinforce rely on donor agencies to supply infrastructure, the internationally connected enclaves at the resulting in an investment strategy that is short expense of a denser set of domestic connec- term in orientation and poorly integrated. tions, a factor exacerbated by the regulatory 5. In the case of telecommunications, for barriers erected between firms inside and example, Salazar's (2006) comparative study outside the export zones. shows that the Philippines moved more quickly Spatial disparities and development policy in the Philippines 181 than several of its neighbors, particularly Barro, Robert J., and Xavier Sala-i-Martin. 2004. Malaysia. Economic Growth, 2d ed. Cambridge, MA: 6. Moreover, while the expenditure of local MIT Press. governments as a percentage of GDP has dou- Bourguignon, Franēois. 2004."The Poverty- bled since decentralization, their infrastructure Growth-Inequality Triangle." Paper presented budgets have not expanded commensurately. at the Indian Council for Research on Inter- 7. Available regional price indexes for the national Economic Relations, New Delhi. 1980s and beyond are not strictly comparable owing to the marked changes in the composi- Burgess, Robert, and Vikram Haksar. 2005. tion of regions over time. Moreover, the avail- "Migration and Foreign Remittances in able data do not capture price variation across the Philippines." IMF Working Paper regions, because each region has a price index WP/05/111, International Monetary Fund, value of 100 for the base year. Washington, DC. 8. Details of the construction of the price Cline, William R. 2004."Technical Correction to indexes are shown in Balisacan (2001). the First Printing." In Trade Policy and Global 9. Poverty estimates are those used in Bal- Poverty. Washington, DC: Institute of Inter- isacan (2007). These are not comparable with national Economics. official data released by the National Statistical HDN (Human Development Network). 2005. Coordination Board. Philippine Human Development Report: Peace, 10. Apart from the longer period covered by Human Security, and Human Development in the present study, Balisacan and Fuwa's results the Philippines. Manila: HDN. pertain to the convergence of per capita provin- cial mean expenditures, not incomes. Moreover, ------. Various years. Philippine Human Devel- the end year in Balisacan and Fuwa's study is 1997, opment Report. Manila: HDN. marking the start of the Asian financial crisis. Llanto, Gilberto M. 2007."Infrastructure 11. Cline's estimate for the Philippines is 2.2. and Regional Growth." In The Dynamics of While higher than the other estimates quoted Regional Development: The Philippines in East here, it is still low by Asian standards. Asia, ed. Arsenio M. Balisacan and Hal Hill, pp. 316­44. Cheltenham: Edward Elgar. References NSCB (National Statistical Coordination Balisacan, Arsenio M. 2001."Rural Develop- Board). Various years. National Income ment in the 21st Century: Monitoring and Accounts. Makati City: NSCB. Assessing Performance in Rural Poverty ------. Various years. Philippines Statistical Reduction." In The Philippine Economy: Alter- Yearbook. Makati City: NSCB. natives for the 21st Century, eds. Dante B. NSO (National Census and Statistics Office). Canlas and Shigeaki Fujisaki. Quezon City: Various years. Annual Poverty Indicators University of the Philippines Press. Survey. Manila: NSO. ------. 2003."Poverty and Inequality." In The ------. Various years (1990, 2000). Census of Philippine Economy: Development, Policies, Population and Housing. Manila: NSO. and Challenges, ed. Arsenio M. Balisacan and Hal Hill, pp. 311­41. New York: Oxford ------. Various years. Family Income and University Press. Expenditure Survey. Manila: NSO. ------. 2007."Local Growth and Poverty ------. Various years. Functional Literacy, Edu- Reduction." In The Dynamics of Regional cation, and Mass Media Survey. Manila: NSO. Development: The Philippines in East Asia, Pernia, Ernesto M., Cayetano W. Paderanga Jr., eds. Arsenio M. Balisacan and Hal Hill, pp. V. P. Hermoso, and associates. 1983. The Spa- 398­421. Cheltenham: Edward Elgar. tial and Urban Dimensions of Development in Balisacan, Arsenio M., and Rosemarie G. Edillon. the Philippines. Manila: Philippine Institute 2005."Poverty Targeting in the Philippines." for Development Studies. In Poverty Targeting in Asia, ed. John Weiss, Ravallion, Martin. 2001."Growth, Inequal- pp. 219­46. Cheltenham: Edward Elgar. ity, and Poverty: Looking beyond Averages." Balisacan, Arsenio M., and Nobuhiko Fuwa. World Development 29 (11): 1803­15. 2004."Going beyond Cross-country Averages: Salazar, Lorraine. 2006. Getting a Dial Tone: Growth, Inequality, and Poverty Reduction in Telecommunications Liberalization in Malay- the Philippines." World Development 32 (11): sia and the Philippines. Singapore: Institute of 1891­907. Southeast Asian Studies. 182 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Sicat, Gerardo P. 2003. Economics, New Edition. Philippines in East Asia, ed. Arsenio M. Bal- Vol. 3: Philippine Economic and Development isacan and Hal Hill, pp. 93­122. Cheltenham: Issues. Manila: Anvil. Edward Elgar. Song, Ligang. 2007."The East Asian Experi- World Economic Forum. 2007. The Global Com- ence: The People's Republic of China." In petitiveness Report 2007­2008. Geneva: World The Dynamics of Regional Development: The Economic Forum. Annex Determinants of growth and poverty reduction in the Philippines: descriptive statistics Standard Variable Description Mean deviation Minimum Maximum Income 1988 Log of per capita income, 1988 9.868 0.270 9.168 10.562 (Romblon) (Metro Manila) Income 2003 Log of per capita income, 2003 10.059 0.290 9.058 10.717 (Sulu) (Northern Vizcaya) Headcount 1988 Proportion of the population deemed poor, 1988 0.394 0.175 0.075 0.852 (Kalinga-Apayao) (Romblon) Headcount 2003 Proportion of the population deemed poor, 2003 0.321 0.176 0.044 0.884 (Northern Vizcaya) (Sulu) Average income growth rate Average annual growth rate of per capita income, 0.012 0.016 -0.030 0.049 1988­2003 (Maguindanao) (Batanes) Average headcount growth rate Average annual rate of change in poverty -0.008 0.032 -0.0568 0.115 incidence, 1988­2003 (Batanes) (Mindoro Occidental) Gini 1988 Expenditure Gini ratio, 1988 33.594 5.077 21.190 43.230 (Tawi-Tawi) (Iloilo) Gini squared 1988 1,153.988 339.961 449.016 1,868.833 (Tawi-Tawi) (Iloilo) Dynasty Proportion of provincial officials related by blood 0.140 0.246 0 1.000 or affinity Ethnic fragmentation 1988 Herfindahl index 0.579 0.190 0.287 0.884 (Catanduanes) (Palawan) Mortality Mortality rate per 1,000 children ages 0­5 years, 84.688 14.847 55.920 121.120 1988 (Pampanga) (Western Samar) Landlock Dummy variable (1 if a landlocked province, 0 0.203 0.405 0 1.000 otherwise) Change in literacy Change in simple literacy rate, 1988­2003 3.847 5.288 -8.960 16.0000 (Zamboanga del (Abra) Norte) Change in road density Change in (concrete-equivalent) road density, 0.123 0.286 -0.076 2.466 1988­2003 (Romblon) (Metro Manila) Change in electricity Change in share of households with electricity, 18.761 13.931 -11.800 67.380 1988­2003 (Agusan del Sur) (Batanes) Change in CARP Change in CARP accomplishment, 1988­2003 0.802 0.144 0.263 1.000 (Sulu) (Batanes/Squijor) Change in agricultural terms of Change in agricultural terms of trade, 1988­2003 -0.004 0.186 -0.310 0.460 trade (Northern Mindanao (CAR provinces) provinces) Source: Balisacan (2007). Note: The last two columns show the provinces with the lowest and highest scores, respectively. Spatial disparities in Thailand: does government policy aggravate or alleviate the problem? Nitinant Wisaweisuan 12 Thailand's economy, measured by gross After the crisis, the figure resumed its down- domestic product (GDP) at market prices in ward trend and ultimately stabilized at 9.6 2006, was about B 7.8 trillion and had been percent in 2007. c h a p t e r growing at a rate of 6 percent a year for more Despite the decline in absolute poverty, than 25 years.1 After experiencing a deep eco- the income gap between the richest and the nomic crisis in 1997, with a drastic decline poorest has been worsening. From 1990­ in real output of approximately 10 percent 2006, the ratio of the richest to the poorest c h a p t e r in 1998, Thailand took five years to recover income quintiles (Q5/Q1) increased from completely,2 achieving a post-crisis growth 13.3 to 15.9, indicating a wider income gap rate (1999­2005) of about 5 percent a year. between the rich and the poor.Yet that ratio The structure of the Thai economy began declined during two periods: 1992­98 and to change in the early 1980s, during which 2000­04. According to Siamwalla and Jitsu- the Thai government promoted industri- chon (2007), the earlier decline occurred as alization and shifted the policy emphasis a result of growth-promoting policies, while from import substitution to export promo- the later decline occurred as a result of pol- tion. Accordingly, the manufacturing sector icy packages implemented by the Thaksin took over a large area of agricultural land, cabinet, which sought to stimulate domes- which reduced the proportion of agriculture tic demand without offering incentives for in GDP from 23 percent in 1980 to 13 per- businesses to improve productivity. This cent in 2005 and increased the proportion explains why Thailand is facing a deterio- of industry from 27 to 38 percent. Services rating situation, as shown by the bounce in remain the largest sector,contributing half of 2006 of the Q5/Q1 ratio to 15.9, soon after the country's GDP in 2005 (see table 12.1). the packages were removed. Together with satisfactory economic The Thai economy is growing and devel- growth, per capita GDP has increased over oping satisfactorily, but there is a question time, with the only exception being dur- regarding whether these benefits are distrib- ing the economic crisis (see figure 12.1). In uted evenly to different areas of the country. 2005 Thailand's GDP per capita was about B This paper attempts to outline the existence 60,000 a year, which is equivalent to approx- and evolution of spatial disparities and their imately B 164 a day, a level just above the relationship with economic development minimum wage rate. and to delineate the factors that create such As this impressive economic growth was spatial differentials, including both market- taking place, the proportion of people living driven and government-directed influences. below the poverty line declined from 38 per- Special emphasis is placed on urban-rural as cent in 1990 to 17 percent in 1996 (see figure well as regional differences, with a particular 12.2). During the economic crisis, poverty attempt to determine whether the growth- increased, approaching 21 percent in 2000. promoting policies as well as public finances 184 Spatial disparities in Thailand: does government policy aggravate or alleviate the problem? 185 Table 12.1 Economic indicators in Thailand, 1980­2005 Composition of GDP ( percent) GDP (1988 prices, Growth (percent a Year baht billion) year) Agriculture Manufacturing Services 1980 913.7 n.a. 23 27 50 1985 1,191.3 5.4 18 27 55 1990 1,945.4 10.3 14 33 53 1995 2,941.7 8.6 12 36 52 2000 3,005.4 0.4 11 37 52 2005 3,851.3 5.1 13 38 49 Source: National Economic and Social Development Board (NESDB). n.a. Not applicable. have alleviated or aggravated spatial dispari- Figure 12.1 GDP per capita in Thailand, 1980­2005 ties in Thailand. 70,000 Growth and spatial disparities 60,000 A quick observation of the income dispar- 50,000 ity between urban and rural areas in 2004 (Baht) shows that the urban population in Thai- 40,000 land earned approximately 2.2 times what capita 30,000 those living in rural areas earned. The fig- per ure shows a satisfactory decline from 1994, GDP 20,000 when the disparity was 2.56 (see table 12.2). Intraregional differences in income between 10,000 the 20 percent richest and the 20 percent 0 poorest, however, can be as large as 1,000 in 19801981198219831984198519861987198819891990199119921993199419951996199719981999200020012002200320042005 Bangkok and as low as 3 in the northeast. year This suggests that richer areas can be subject Source: Bank of Thailand. to wider gaps in income distribution than poorer areas. Bangkok, the capital of Thailand, and its vicinities play a significant role in cre- Figure 12.2 Absolute poverty and income distribution in Thailand, 1990­2006 ating jobs and produce as much as half of 45 18 the country's GDP (see figure 12.3). Greater 15.9 Bangkok has the smallest proportion of peo- 40 38.2 14.9 16 14.5 ple defined as poor, while the northeastern 14.1 35 13.5 13.1 13.2 14 region has the highest (see figure 12.4). (%) 32.5 13.3 12 Greater Bangkok generates per capita 30 poor 12.1 income about 2.6 times that of the country 25 25.0 10 the 21.3 (times) average and about 8 times that of the north- of 20 18.8 8 eastern region, where the majority of the 15.6 Q5/Q1 poor are located. Yet income convergence is 15 17.0 6 11.3 9.6 evident in the central and eastern regions, proportion 10 4 which are catching up with Bangkok (see 5 2 figure 12.5). They had accelerating growth rates of 12.1 and 13.5 percent a year, respec- 0 0 1990 1992 1994 1996 1998 2000 2002 2004 2006 tively, in the early 1990s and kept growing at year an impressive rate afterward, finally outpac- Proportion of poor (%) Q5/Q1 ing the country's average rate of growth and becoming second, after Bangkok, in contrib- Source: National Economic and Social Development Board (NESDB) and Bank of Thailand. Note: Q1 and Q5 = the first and fifth quintile of income, indicating the proportion of population who earn the least 20 uting to Thailand's output (see table 12.3). percent and the highest 20 percent of income, respectively. 186 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 12.2 Urban-rural income gap in Thailand measured by per capita income, 1994­2004 Indicator 1994 1996 1998 2000 2002 2004 Urban income per capita Baht per month 3,868 5,220 5,657 5,772 6,394 6,885 Rate of change (percent) n.a. 34.96 8.36 2.02 10.78 7.68 Rural income per capita Baht per month 1,510 2,007 2,343 2,300 2,680 3,130 Rate of change (percent) n.a. 32.92 16.78 -1.85 16.53 16.80 Urban income times rural income 2.56 2.60 2.41 2.51 2.39 2.20 Thailand income per capita Baht per month 2,217 2,978 3,356 3,372 3,867 4,331 Rate of change (percent) n.a. 34.35 12.67 0.50 14.68 11.99 Source: Data from National Statistical Office, computed by the NESDB. n.a. Not applicable. Figure 12.3 Regional share of GDP in Thailand, 1990­2005 For the past 15 years, they have enjoyed an increase in per capita income, attaining a 100 level 2.17 times the country average in 2005. The central region, whose per capita 80 income was 88 percent of the country aver- age in 1990, now has per capita income 60 that is approximately 1.5 times the country average. While both regions are success- percent fully narrowing the gap with Bangkok, other 40 regions--western, southern, northern, and northeastern--are growing relatively slowly. 20 As a result, they are maintaining their low levels of income, which are 75, 72, 48, and 31 0 percent, respectively, of the country average. 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 year Explaining spatial disparities NE S E W N Central BKK Different levels of income and growth can be achieved partly by income-generating Source: National Economic and Social Development Board. activities that are inherently different in each region (see figure 12.6). In terms of economic structure, Bangkok's major activ- Figure 12.4 Proportion of the poor people in Thailand, by region, 1990­2004 ities are service related. Other regions, like 60 the central and eastern areas, have benefited from a variety of growth-promoting poli- 50 cies, including industrialization, globaliza- 40 38.18 tion, and the creation of export-promotion 32.5 zones, which are eligible to undertake a sig- 30 25.04 nificant structural transformation in eco- percentage 18.78 21.32 20 17.03 nomic activities, meaning essentially a shift 15.55 to manufacturing. Improved infrastructure 10 11.25 and a lot of incentives attract both local and 0 multinational firms, encouraging them to 1990 1992 1994 1996 1998 2000 2002 2004 year establish production sites and eventually allowing the area to enjoy agglomeration Bangkok and Vicinities Northern Northeastern economies as well as government support Central Southern Whole Kingdom in several forms. In the central region, the Source: NESDB. manufacturing sector constituted only one- Note: Numbers in the figure represent the national average. The western and eastern regions are included in the central region, following the classification of NESDB. tenth of economic activity in 1980, but this Spatial disparities in Thailand: does government policy aggravate or alleviate the problem? 187 Figure 12.5 GDP per capita in Thailand, by region, 1990­2005 300,000 250,000 Bangkok and Vicinities Northern 200,000 Northeastern 150,000 baht Central Western 100,000 Eastern 50,000 Southern 0 5 1990 1991 1992 1993 1994 199 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 year Source: National Economic and Social Development Board. Table 12.3 Per capita regional GDP in Thailand, 1990­2005 Per capita regional GDP Regional GDP growth rate (Thailand = 100) (percent a year) Region 1991­95 1996­2000 2001­05 1990­95 1995­2000 2001­05 Bangkok and vicinities 310.9 280.8 262.8 8.4 -2.0 4.5 Eastern 147.7 185.0 217.0 13.5 5.3 9.1 Central 87.6 120.4 149.6 12.1 12.5 6.5 Western 75.6 74.9 75.3 8.4 0.1 3.5 Southern 66.8 72.0 71.5 7.4 1.9 3.8 Northern 50.0 50.1 48.1 6.2 0.3 3.6 Northeastern 33.7 33.3 30.8 7.4 -1.2 3.7 Thailand GDP per capita (baht) 56,113.8 76,931.6 94,345.8 n.a. n.a. n.a. Growth rate (percent) n.a. n.a. n.a. 8.6 0.4 5.1 Sources: Data from NESDB and author's calculations. n.a. Not applicable. had increased to about 60 percent in 2005. relatively more relevant than manufactur- The pattern of structural shift in the eastern ing. Households generate income mainly region is similar to that in the central area: from agriculture and services based largely the manufacturing sector has replaced the on natural resources, including rubber plan- agriculture and service sectors. tations and tourism. Its ability to catch up As a result, both the central and eastern with other regions is generally weak. regions have become a magnet for labor from Several studies attribute the success of other low-income, lagging areas, including the catching-up process in Thailand to a the western, southern, northern, and north- number of growth policies that have pro- eastern regions. The northeastern region, in moted industrialization, globalization, and particular,has long housed the poorest of the urban-based development (see, for example, country, as reflected in its high poverty rates. Krongkaew 1996; Siamwalla and Jitsuchon Its regional GDP was only about 30 percent 2007). However, Ikemoto and Limskul of the country average.The southern region's (1987) and Tinakorn (1995), among many economic structure is quite distinct from that others, assess the impact of such growth- of other regions, with the agriculture sector promoting policies on income distribution 188 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 12.6 Composition of regional GDP in Thailand, by region and economic activity, 1981­2005 A. Thailand B. Bangkok and Vicinities 100 100 90 90 80 80 70 70 60 60 50 50 percent 40 percent 40 30 30 20 20 10 10 0 0 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 year year C. Central region D. Northeastern region 100 100 90 90 80 80 70 70 60 60 50 50 percent 40 percent 40 30 30 20 20 10 10 0 0 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 year year E. Eastern region F. Western region 100 100 90 90 80 80 70 70 60 60 50 50 percent 40 percent 40 30 30 20 20 10 10 0 0 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 year year G. Northern region H. Southern region 100 100 90 90 80 80 70 70 60 60 50 50 percent 40 percent 40 30 30 20 20 10 10 0 0 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 year year Agriculture Manufacturing Services Source: Author's calculation based on data from National Economic and Social Development Board. and conclude that they tend to deteriorate in Thailand, exhibited a continual upward equality. According to a survey by Siripra- trend between 1981 and 1997, rising from chai, Wisaweisuan, and Srisuchart (2004), 0.13 to 0.24 (see figure 12.7). This indicates industrialization policy did not lead to per- that the income gap widened even as eco- manent job creation and thus resulted in nomic growth accelerated. Thanks to the long-term inequality, and the export pro- economic crisis in 1997, Thailand's regional motion policy helped to reduce absolute Gini index fell sharply to 0.1560. In 2005 the poverty but widened income distribution. figure stayed at 0.1683, which was higher The regional Gini index,3 constructed than the figure in 1980. based on income earned by population However, an assessment of income dis- living in 76 provinces over seven regions tribution based on the Gini coefficient of Spatial disparities in Thailand: does government policy aggravate or alleviate the problem? 189 Figure 12.7 Gini coefficient and economic growth in Thailand, 1981­2005 0.25 15.0% 0.20 10.0% 0.15 0.10 5.0% 0.05 0.00 0.0% growth coefficient 6 7 0 2 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 199 199 1998 1999 200 2001 200 2003 2004 2005 GDP gini­0.05 year ­5.0% ­0.10 economic ­0.15 crisis ­10.0% ­0.20 ­0.25 ­15.0% Gini coefficient GDP growth Source: GDP growth from NESDB and GINI coefficient from author's calculation. consumption spending between 1988 and of total export values, and machinery, fuel, 2006 shows an improvement in equality (see and lubricants together share about 57 per- table 12.4). Yet the level of Gini was still as cent of total import values. high as 0.4148 in 2006, which, according to Israngkura (2000) employs a social Siamwalla and Jitsuchon (2007), is similar to accounting matrix to evaluate the impact of the level in Latin American countries such free trade areas on income distribution and as Argentina and Mexico. Gini coefficients concludes that export-led growth via global- at the regional level improved everywhere, ization and greater intensity of regionalism, except in the northern and northeastern particularly in the form of a free trade area, regions and in rural areas. These areas are results in deteriorating income distribution experiencing greater intraregional dispari- in Thailand. In particular, nonagricultural ties, with Gini coefficients of 0.4048, 0.3948, sectors earn triple the income earned by the and 0.3882, respectively, in 2006. agricultural sector (B 154,774 and B 57,010, As asserted earlier, globalization and respectively). More important, inequality is industrialization are engines of growth. higher in the nonagricultural sector, which Thailand has a close relationship with inter- has a Gini coefficient of 54.41, compared national economies in terms of both trade with the nonagricultural sector, which has a and investment, as indicated by an increase Gini coefficient of only 17.78. in the degree of openness: from 90 percent According to Israngkura (2000), the of GDP in 1995 to 149 percent of GDP in worst scenario would occur in the presence 2005 (see table 12.5). Thailand also enjoys of financial liberalization that brings about an influx of capital, which generates greater growth in the service sector. In this regard, employment and migration of labor into Wattanakuljarus (2007) studies the tour- the export sector. Although it is impossible ism sector to assess the impact of growth on to establish a concrete relationship between income distribution and finds that for every greater connectivity to the global market 10 percent increase in tourism, 3.72 percent and a lower Gini coefficient, in the manu- in additional income will accrue to nonagri- facturing sector, machinery and manufac- cultural labor and 2.53 percent in additional tured goods together share about 70 percent income will accrue to agricultural labor. 190 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 12.4 Gini coefficient of consumption spending in Thailand, by region, 1988­2006 Year Thailand Bangkok Central North Northeast South Urban Rural 1988 0.4387 0.3627 0.3767 0.3952 0.3877 0.3773 0.4058 0.3787 1990 0.4433 0.3845 0.3864 0.4111 0.3816 0.3611 0.4173 0.3856 1992 0.4500 0.3926 0.3717 0.3898 0.3960 0.3739 0.4230 0.3639 1994 0.4377 0.3641 0.3732 0.3983 0.3909 0.3983 0.4085 0.3811 1996 0.4313 0.3484 0.3597 0.3871 0.3785 0.3742 0.4104 0.3592 1998 0.4092 0.3261 0.3354 0.3583 0.3428 0.3612 0.3784 0.3489 2000 0.4283 0.3289 0.3657 0.3745 0.3517 0.3742 0.3951 0.3594 2002 0.4179 0.3650 0.3539 0.3819 0.3559 0.3661 0.3939 0.3552 2004 0.4255 0.3585 0.3647 0.4062 0.3760 0.3871 0.3956 0.3838 2006 0.4183 0.3584 0.3589 0.4048 0.3908 0.3740 0.3897 0.3882 Source: Community Economic Development and Income Distribution Office (CEDIO). Table 12.5 Openness and income distribution in Thailand, 1995­2005 GDP per capita Change in Trade value openness Year (percent of GDP) Ginia Baht Change (percent) (percent)b 1995 90.4 0.2191 70,474 n.a. n.a. 1996 84.8 0.2227 76,847 9.0 -6.2 1997 94.6 0.2299 78,093 1.6 11.6 1998 101.9 0.1560 75,594 -3.2 7.7 1999 104.0 0.1686 75,026 -0.8 2.1 2000 124.9 0.1697 79,098 5.4 20.1 2001 125.1 0.1732 81,915 3.6 0.2 2002 122.0 0.1693 86,322 5.4 -2.5 2003 124.3 0.1664 92,960 7.7 1.9 2004 136.4 0.1682 101,092 8.7 9.7 2005 148.8 0.1683 109,440 8.3 9.1 Source: Bank of Thailand. n.a. Not applicable. a. Based on the regional Gini index in Fu (2004). b. Degree of openness is measured by the ratio of trade value to GDP. In terms of capital account, Thailand is For several decades, the policy has been to a net recipient of foreign direct investment raise income in rural areas through mea- (FDI), with net flows amounting to B 383 sures such as price supports. This approach billion (see table 12.6). FDI grew rapidly, at has not succeeded in raising agricultural 71 percent a year, between 1985 and 1990, as income. As Siamwalla and Jitsuchon (2007) a result of the government's industrialization argue, agricultural development policy that policy and several export promotion mea- improves productivity and promotes com- sures. Thailand experienced only one period petitiveness is preferable because it helps of continuous decline in FDI--between 1990 to protect the Thai agricultural sector from and 1995, a period in which financial liberal- the adverse impacts of agricultural liberal- ization created an influx of short-term capi- ization in the years to come. tal and portfolio investment that replaced a Attempts to alleviate the large proportion of FDI. Since then, FDI has increased steadily, growing, on average, 22 problem percent a year. The government's approach to alleviating A great deal of FDI in the manufac- regional disparities is contained in the offi- turing sector has intensified the imbal- cial National Economic and Social Develop- ance between agricultural development ment Plans (see table 12.7). The first one was and industrialization, exacerbating the launched in 1963, with major emphasis on uneven distribution of income. The need investment in infrastructure throughout the to raise income in the agricultural sector country. A large proportion of the govern- has received far less government attention. ment budget was spent on the construction Spatial disparities in Thailand: does government policy aggravate or alleviate the problem? 191 Table 12.6 Net inflows of foreign direct investment Table 12.7 Major emphasis of national economic in Thailand, 1970­2006 and development plans in Thailand FDI Five-year growth Plan Period Major emphasis Year (baht billion) (percent a year) 1 1963­66 Investment in infrastructure 1975 1.7 14.4 2 1967­71 Sectoral development 1980 3.9 17.3 3 1972­76 Job promotion 1985 4.4 2.6 4 1977­81 Alleviating poverty and reducing 1990 64.7 71.2 income distribution 1995 50.0 -5.07 5 1982­86 Balanced development 2000 115.3 18.2 6 1987­91 Productivity-based development 2005 262.6 17.9 7 1992­96 Decentralization of public 2006 382.9 45.8 authorities 8 1997­2001 Focus on human resources Source: Bank of Thailand. 9 2002­06 A self-sufficient economy 10 2007­11 Green and happiness society Sources: NESDB, author's compilation. of roads, railways, and several facilities that support growth of the industrial sector; a 2004. The figure was especially high in 1994, negligible amount was spent on research when the gap was larger than 2,000 times. and development in the agricultural sector. The northeastern region is the poorest in the At the end of the first plan, rural-urban country and has relatively even distribution differences became evident, which led the of income, with the Q5/Q1 ratio being only government to attempt to reduce the income about two to three times. gap. The success of this effort is mixed. The The fourth National Economic and government succeeded in reducing the Social Development Plan (1977­81) gave income gap from 2.56 times in 1994 to 2.20 high priority to alleviating poverty and in 2004, but it was unable to address rural- reducing income inequalities. The issue urban differences in income. As shown in was again addressed in the seventh plan, table 12.2, monthly per capita income in which sought to decentralize development. urban areas increased from B 3,868 in 1994 However, the plan was poorly implemented, to B 6,885 in 2004, while that in rural areas especially in the regional context, as only a increased from B 1,510 in 1994 to B 3,130 limited amount of infrastructure was made in 2004. available to the regions and rural areas In urban areas in 2004, the top 20 per- (Dixon 1999). cent of income earners earned 80 times the The problems persist, owing in part earnings of the bottom 20 percent (see table to the uneven spatial distribution of the 12.8). This difference was even higher in public budget. The most recent figures for 1994, when the top 20 percent earned 110 2003­06 indicate that more than 50 per- times the earnings of the bottom 20 per- cent of the government budget was spent cent.4 In rural areas, the income gap has in Greater Bangkok (see figure 12.8). As remained narrow, with the top 20 percent of Greater Bangkok continues to grow, the income earners earning approximately four budget inequality deepens even further. In times the earnings of the bottom 20 percent 2006 the government budget spent within throughout the decade. Bangkok and its vicinities grew at 14.8 per- The disparity may be observed not only cent, followed by the northeastern and the across regions, but also within regions: the southern regions, which grew at 9.9 and richer the area, the wider the income gap. 1.9 percent, respectively. The northeastern Bangkok has the highest per capita income region received the second-highest propor- and the widest income gap.The northeastern tion of the budget, about 13 percent, but region has the lowest per capita income and the sum was still far lower than the amount the smallest income gap. Based on an analy- going to Bangkok and its vicinities. In fis- sis of the ratio of the richest to the poorest cal 2006 Bangkok received B 123,057 per income quintiles (Q5/Q1), the top 20 per- capita of government spending, while the cent income group in Bangkok earned 986 northeastern region received only B 8,448. times what the bottom 20 percent earned in The following regions receive a share of the 192 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 12.8 Ratio of Q5 to Q1 in Thailand, by region, 1994­2004 percent of income Region and quintile 1994 1996 1998 2000 2002 2004 Bangkok 2,263.94 2,890.00 1,991.56 4,499.84 1,409.69 985.95 Central 36.30 33.81 31.81 53.34 38.49 37.73 North 8.31 8.44 8.42 6.82 6.00 5.90 Northeast 3.15 3.14 3.16 3.13 3.49 2.92 South 11.18 10.45 9.61 12.14 10.52 11.91 Thailand 14.07 13.52 13.06 14.55 13.23 12.10 Urban 110.45 110.68 100.69 112.03 91.40 80.18 Rural 4.52 4.19 4.45 4.89 4.47 4.48 Source: NESDB. Note: Q1 = 20 percent lowest-income group; Q5 = 20 percent highest-income group. Figure 12.8 Allocation of the government budget in economic development was diverted to the Thailand, by region, 2003­06 northeastern region, suggesting that the government views infrastructure as a means 100 90 of promoting growth and decentralization. 80 The uneven distribution of public finance 70 results in the "poor" being poor not only in 60 terms of income but also in terms of access 50 to basic services required for improving their percent 40 quality of life,developing their skills,and rais- 30 ing their productivity.Table 12.9 presents the 20 spatial distribution of health care and educa- 10 0 tion resources by region,revealing once again 2003 2004 2005 2006 the concentration of productive resources in year Greater Bangkok. For example, 41 percent of South North Northeast West doctors, 26 percent of pharmacists, 30 per- East Central Bangkok and Vicinities cent of nurses, and 31 percent of dentists are Source: Comptroller-General's Department, Ministry of Finance. working in Bangkok and its vicinities. Gov- ernment spending per capita on health care was B 8,484 in Bangkok and merely B 763 in government budget, in descending order: the northeastern region. Even if the scarcity the north, the south, the east, the west, and of medical services does not necessarily lead the central regions. to low productivity, differences in life expec- The uneven distribution of government tancy do lead to differences in earning capac- budget across regions may be attributed to ity. It is not clear whether greater availability spatial disparities in Thailand, despite the of resources would lead to higher income. absence of empirical evidence that fiscal For example,per capita income in the central policies over the past five decades brought region is relatively high, but the proportion about a narrowing or a widening of spatial of health care resources is relatively small. disparities. Excluding Greater Bangkok, the This may be explained by the region's prox- northeastern region received the greatest imity to Bangkok and its vicinities. share of the public budget for water devel- Education also contributes to inequali- opment, education, and general support for ties.In general,higher education successfully local institutions (see figure 12.9), but its reduces the proportion of the population level of income remained low. The central living in poverty (see table 12.10). However, region received the greatest share of pub- Siamwalla and Jitsuchon (2007) argue that lic budget for infrastructure for economic higher education will not take people away development, and its per capita income rose, from poverty even in the medium to long approximating that of Bangkok. Recently, term,because the quality of education rather the public budget for infrastructure for than the proportion of the population who Spatial disparities in Thailand: does government policy aggravate or alleviate the problem? 193 Figure 12.9 Allocation of the government budget in Thailand, by activity and region (excluding Bangkok and its vicinities), 2001­06 A. Water development B. Education 40 50 35 40 30 25 30 20 20 percent 15 percent 10 10 5 0 0 2001 2002 2003 2004 2005 2006 2001 2002 2003 2004 2005 2006 year year C. Infrastructure for economic development D. General support for local institutions 50 50 40 40 30 30 20 20 percent percent 10 10 0 0 2001 2002 2003 2004 2005 2006 2001 2002 2003 2004 2005 2006 year year Northern Northeastern Central Southern Source: Ministry of Finance, Comptroller-General's Department. Table 12.9 Availability of health care resources in Thailand, by region percent Region Doctors Pharmacists Technical nurses Nurses Dentists Bangkok and vicinities 41 26 20 30 31 Central 4 5 7 5 4 North 15 19 17 18 19 Northeast 17 21 24 20 21 South 10 14 16 13 12 West 5 6 6 6 5 East 8 8 9 8 7 Thailand 100 100 100 100 100 Source: NESDB. Table 12.10 Proportion of the poor in Thailand, by education of the head of household, 1996­2004 Education 1996 1998 2000 2002 2004 None 28.8 29.6 32.7 28.9 24.2 Kindergarten (two years) -- -- -- -- 18.2 Junior primary school (three years) 17.3 20.2 23.2 16.8 12.3 Senior primary school (three years) 12.6 16.3 18.2 14.0 11.1 Junior high school (three years) 6.3 6.1 7.9 4.6 4.8 Senior high school (three years) 6.8 3.3 4.6 4.7 3.0 University level 2.6 1.2 7.3 1.3 0.5 Vocational school 1.1 1.4 1.0 2.5 1.9 Others -- -- 3.5 -- -- Percentage of Thailand's population 15.2 16.5 18.8 13.9 10.3 who are poor Source: Data from National Statistical Office, computed by NESDB. are educated is what raises productivity and national, not the regional, level. They find the ability to earn income. that more loans are allocated to the lower- In this regard, Tangkitwanit and Manus- income population at both the high school boonpoempoon (2007) evaluate the impact and the university levels, which successfully of the income-contingency loan for educa- distributes income to lower-income groups. tion (ICL)5 on income distribution at the However, when investigating the relative 194 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA effectiveness of the policy by education Gini index, which takes into account the relative level, a more satisfactory outcome may be importance of regions as follows: found at the high school than at the uni- 1 versity level. This is true largely because the GINI = n j k n yj - yk , ICL is a tool for facilitating education, not 2n2 j=1 k=1 investing in human capital, and the poor where n is the total population, is the average generally have limited access to a university- income equal to total income divided by total level education. population, j is the number of regions, k is the number of provinces, nj and nk are the popula- Conclusions tion in region j and province k, respectively; and The spatial disparities in Thailand that are yj and yk are income per capita in region j and outlined in this paper reveal a decline in province k, respectively. absolute poverty, as indicated by the pro- 4. These figures need to be interpreted with portion of the population who are poor, caution, as a number of areas could transform and the persistence of an income gap not themselves from rural to urban. only across but also within regions. For the 5. Commenced on January 16, 1996. As of 2005, the maximum annual loan varies by past 50 years, various policies have sought education level; that is, B 55,440 for high school; to tackle the problem of income distribu- B 62,500 for vocational study; B 70,240 for tion at the national level, but not at the higher vocational study; B 127,000 for sciences regional level. There is no clear evidence and health-related undergraduate studies; and that spatial disparities are less severe now B 100,000 for other undergraduate studies. than before; nevertheless, some of the regional policies that successfully raised References income in the area targeted also seem Dixon, Chris. 1999. The Thai Economy: Uneven to have created gaps among regions. A Development and Internationalisation. Rout- balance is needed between agricultural ledge: London and New York. development and industrialization. The Fu, Xiaolan. 2004. "Limited Linkages from allocation of government spending needs Growth Engines and Regional Disparities in to be assessed critically, because many China." Journal of Comparative Economics 32 provinces are allocated less than 1 percent (1): 148­64. of the total budget, which will not lead to Ikemoto,Yukio, and Kitti Limskul. 1987."Income a narrowing of regional disparities. Rather, Inequality and Regional Disparity in Thai- centralized policy is likely to remain inef- land." Developing Economics 25 (3): 249­69. fective. Further study is needed in areas Israngkura, Adis. 2000."The Distribution such as regional job creation and its rela- Impact of Globalisation and Policy Impli- tion to the income-generating process and cations." Paper submitted to Robert S. the linkage between government finance McNamara Fellowships, the World Bank, and the regional Gini index, to inform a Washington, DC. complete and comprehensive policy on Krongkaew, Medhi. 1985."Agricultural Devel- regional disparities in Thailand. opment, Rural Poverty, and Income Distribu- tion in Thailand." Developing Economics 23 Notes (4): 325­46. Nitinant Wisaweisuan is assistant professor of ------. 1996. Thailand Poverty Assessment economics at Thammasat University in Thailand. Update. Bangkok: Economic Research and 1. Thailand's currency is the baht. Training Centre. 2. "Recovery," according to Siamwalla and Siamwalla, Ammar, and others. 2006."An others (2006: 4), is defined as the year in which Economic Assessment on Impacts of Thak- economic growth started to outpace aver- sin Policies." Thailand Development and age growth during 1980­96, not including the Research Institute, Bangkok. year with economic shocks, and in which real national income bounced back to the same level Siamwalla, Ammar, and Somchai Jitsuchon. as the pre-crisis figure. 2007."Tackling Poverty: Liberalism, Popu- 3. This calculation is based on the alternative lism, or Welfare State." Paper presented at methodology of Fu (2004) for estimating the the 2007 Thailand Development Research Spatial disparities in Thailand: does government policy aggravate or alleviate the problem? 195 Institute Year-End Conference "How Can We Tinakorn, Pranee. 1995."Industrialisation Tackle Poverty? Liberal, Populism, or Wel- and Welfare: How Poverty and Income Are fare State Approaches," Chonburi, Thailand, Affected." In Thailand's Industrialisation and November 10­11. Its Consequence, ed. Medhi Krongkaew. New Siriprachai, Somboon, Nitinant Wisaweisuan, and York: St. Martin Press. Supachai Srisuchart. 2004."Macroeconomic Wattanakuljarus, Anan. 2007."Does the Growth Policies and Poverty in Thailand: A Survey." of the Tourism Industry Benefit the Low- Thammasat Economic Journal 22 (3): 94­142. Income Population?" Paper presented at Tangkitwanit, Somkiat, and Areeya Manus- the 2007 Thailand Development Research boonpoempoon. 2007."An Assessment of Institute Year-End Conference "How Can We Income-Contingency Loans for Education." Tackle Poverty? Liberal, Populism, or Wel- In Evaluating Public Policy by Economet- fare State Approaches," Chonburi, Thailand, ric Methods, ed. Somkiat Tangkitwanit, pp. November 10­11. 113­88. Bangkok: Thailand Development and Research Institute. SECTION III Reshaping economic geography in China Northeast Asia: China and the Republic of Korea Yukon Huang and Xubei Luo 13 Much has been made of China's impres- Geographic factors determined that sive growth as well as its distributional and regional development in China would be environmental consequences. But with 1.3 uneven long before the post-1980s reform billion people spread unevenly across 31 era. The potential for disparities to emerge, ecologically diverse provinces and munici- however, was less visible because of the palities, China is perhaps the best example low level of development after the postwar of how a country has reshaped its economic period of central planning (1950­80). To geography to reverse half a century of eco- encourage more balanced growth, invest- c h a p t e r nomic decline. This chapter provides an ment was directed to inland provinces and overview by drawing on chapters in this the northeast, rather than based on regional volume that deal in depth with particular comparative advantages. China's unique aspects and on other related studies. hukou residency restrictions also limited the potential for both rural to urban and China's historic and interprovincial migration and prevented geographic legacy cities from getting too large. As elaborated China's geography has been fundamental in by Yao in chapter 14, these location-specific shaping 5,000 years of social and economic industrial policies took place in the midst of history and has been a defining factor in altering periods of economic and fiscal cen- influencing China's pattern of population tralization and decentralization. These shifts settlement (see figure 13.1). Although by were motivated by the adverse consequences global standards its land mass is large, the of the Great Leap Forward and the Cultural population is concentrated in the more Revolution, which pushed senior policy fertile central plains and valleys of the Yang- makers either to tighten or to liberalize their tze andYellow rivers and along the urbanized grip on the local authorities. coastal areas.The western region,comprising By 1980, three decades after liberation, 12 provinces with large tracts of uninhabit- China was still a desperately poor agrarian able mountainous terrain and deserts, is iso- economy, although by international stan- lated, sparsely populated, and thus less suited dards it was an exceptionally egalitarian soci- to commercial activities. The central region, ety.With these facts in mind, Deng Xiaoping with 8 provinces, has large concentrations launched a series of reforms that, in open- of population along major river basins and ing the country to the outside world, fore- accounts for a substantial portion of agricul- shadowed the basic tenets underpinning the tural production. Finally, the coastal region, new economic geography and trade theories with 11 provinces, has traditionally been that were only just becoming prominent in China's industrial and commercial heartland the literature. During the ensuing reform and is historically linked to the outside world era, preferential policies were sequenced to through trade and labor migration. work with rather than against differences Reshaping economic geography in China 197 Figure 13.1 Regions and open cities in China 70° 80° 90° 100° 110° 120° 130° RUSSIAN FEDERATION KAZAKHSTAN H E I LO N G J I A N G Harbin MONGOLIA C E N T R A L 40° R E G I O N KYRGYZ Changchun Urumqi J I L I N REP. W E S T E R N TAJIKISTAN R E G I O N Shenyang LIAONING DEM. 40° AFGHANISTAN X I N J I A N G G O L PEOPLE'S Sea of O N Hohhot REP. OF I M BEIJING Japan N E KOREA Bo BEIJING Tianjin Hai 80° TIANJIN H E B E I Yinchuan Taiyuan Shijiazhuang REP. OF KOREA IA C O A S TA L X SHANXI Jinan Xining R E G I O N Lanzhou G Yellow IN SHANDONG Q I N G H A I Sea 0 300 Kilometers N JAPAN G A N S U W E S T E R N Xi'an Zhengzhou 0 100 200 300 Miles JIANGSU R E G I O N S H A A N X I H E N A N C E N T R A LHefei Nanjing X I Z A N G R E G I O N Shanghai 30° 30° G ANHUI SHANGHAI H U B E I S I C H U A N IN Hangzhou East Lhasa Chengdu Q Wuhan 90° G N ZHEJIANG China Chongqing CHO Sea Nanchang C O A S TA L CHINA Changsha R E G I O N HUNAN JIANGXI REGIONS AND OPEN CITIES F U J I A N Guiyang Fuzhou GUIZHOU Kunming 4 FIRST SPECIAL ECONOMIC PROVINCIAL CAPITALS TAIWAN ZONES (1980) Y U N N A N GUANGDONG G U A N G X I NATIONAL CAPITAL Guangzhou 14 COASTAL CITIES (1984) R PROVINCIAL BOUNDARIES A Nanning HONG KONG M 3 DELTAS (1985) MACAO 20° 20 N N REGIONAL BOUNDARIES AY VIETNAM LAO PROVINCE OF HAINAN (1988) INTERNATIONAL M PEOPLE'S BOUNDARIES Haikou 11 BORDER CITIES (1992) DEM. THAILAND REP. HAINAN PHILIPPINES 6 PORTS OFYANGTZE (1992) 100° 120° in natural endowments and comparative having dealt successfully with the three advantages; in the process, they fundamen- "Ds": realizing the benefits of agglomera- tally reshaped China's economic geography. tion economies through a higher density of The story of China during the past quar- economic activities in its major urban cen- ter century revolves around a series of prag- ters, overcoming the distance factor by mak- matic and at times trial-and-error reforms ing a concerted effort to improve regional that made use of the country's natural transportation and communications infra- advantages, unleashing the pent-up ener- structure, and eliminating internal and gies of the nation. These policies had three external divisions to promote national mar- distinct consequences: (1) they reshaped ket integration and China's participation in the spatial dimensions of development both the global economy. in terms of regional concentration and the This chapter begins by presenting a syn- dynamism of urban and rural areas; (2) opsis of how the government's reform poli- they broke the gridlock in the mobility of cies managed to reorient growth dynamics factors and goods both internally across to tap China's comparative advantages as provinces and between urban and rural influenced by its economic geography. It areas and externally between China and then explores more deeply several of the the rest of the world; but (3) they also set key spatial and policy factors that have in motion forces that increased disparities shaped these growth and distributional across regions and provinces and between trends: agglomeration benefits, fiscal poli- as well as within urban and rural areas. In cies, investment in transport services, and the process, China emerged as an exception- the migration cum urbanization process, ally competitive, growth-oriented economy, before assessing the distribution and equity 198 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA implications. Finally, this chapter explores the full benefits from the emergence of rural- the implications of these issues for future based"township-and-village enterprises"and policies and trends in inequality. the restructuring of state-owned enterprises in urban areas were still to be felt. Neverthe- Accelerating growth: less, these enterprise-related reforms spurred the movement from farm-based to off-farm coordinating structural, labor activities and sowed the seeds of rapid incentive,and fiscal policies urbanization. With the advent of economic reforms in the This urbanization process was integral to late 1970s, the past quarter century has seen the process of industrial agglomeration in a period of unprecedented growth, coupled China and influenced by the uniqueness of with equally sharp increases in income and the transition from a planned to a market regional disparities. Gross domestic product economy. The transition was supported ini- (GDP) growth averaged nearly 10 percent a tially by an agenda of dual-track reform--an year,liftinghalf abillionpeopleoutof poverty. "open-door policy" that strengthened the The record of rapid growth and increasing investment climate along the coastal prov- inequality, however, was marked by periods inces before going national and waves of of considerable variation (see figure 13.2). fiscal decentralization and recentralization The major factors underpinning this growth that initially favored more experimenta- have been well analyzed. Growth increased tion among certain coastal provinces but is sharply in the first half of the 1980s as house- now trying to deal with the disparities and holds became part of the agrarian economy, expenditure needs of a highly decentralized along with communes and state farms, as a intergovernmental fiscal system. result of the new "household responsibility As discussed by Chen and Lu in chap- system." By simply de-collectivizing produc- ter 15, the open-door policy began with tion and allowing farmers to sell their sur- the establishment of the first special eco- plus on the market, rural per capita incomes nomic zones (SEZs) in 1980 in four cities tripled during the 1978­84 period, contrib- in Guangdong and Fujian provinces (figure uting to a surge in the GDP growth rate to 13.1). This initiated a process that generated more than 15 percent by the mid-1980s. By urban-based agglomeration benefits but the second half of the 1980s, the momentum also triggered wider regional disparities. The from these reforms had petered out, while preferential status of SEZs was soon rolled Figure 13.2 GDP growth in China, by region, 1981­2006 18 16 14 12 10 percent8 6 4 2 0 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 year East Central West China Source: National Bureau of Statistics of China (various years). Reshaping economic geography in China 199 out to 14 other coastal cities in 1984, to the sion of the emergence during this period of three deltas in 1985, to Hainan in 1988, to the three major commercial regions centered Pudong in Shanghai in 1990, and then grad- on Guangzhou, Shanghai, and Beijing.) ually to 11 border cities and eventually to all Although growth moderated in the capitals of the inland provinces and autono- late 1990s due to the Asian financial crisis, mous regions during the 1990s. These SEZs an ambitious public investment program were the precursor to a more broad-based coupled with accession to the World Trade improvement in China's investment climate Organization (WTO) led to the rapid expan- and laid the foundations for a ratcheting up sion of exports, which has kept GDP growth of investment rates. above 10 percent in recent years. Further In the early 1990s, Deng Xiaoping's liberalization of the agrarian economy was famous southern tour gave a strong boost largely connected with WTO trade-related to the open-door policy and conferred reforms, which eliminated any remain- formal status on the "gradient develop- ing pricing biases against agriculture and ment model" (see Yao in chapter 14). This encouraged a shift to more profitable crops approach, which considered coastal, cen- in line with China's comparative advantages tral, and western regions as three ladders of (Huang and others 2007). This was subse- economic growth starting from the coast, quently reinforced by a whole-scale reduc- gave the coastal region large benefits as the tion in agricultural taxes and fees that lifted first movers of reforms. This was supported agrarian incomes and moderated regional by a strategy of "big inputs, big exports," disparities in the new millennium. which was centered in the coastal region and sought to stimulate processing trade Spatial factors and government and attract foreign direct investment (FDI). policies: growth and equity Foreign trade and investment provided the coast not only with capital but also with implications advanced technology and management, China's success in building a more competi- which enhanced productivity. In 1990, when tive economy over the past several decades only coastal provinces were open to foreign can be viewed from various perspectives, investors, more than 92 percent of FDI but overall its success can be interpreted as accrued to the coast. (Despite more region- the consequence of having initiated a major ally neutral incentive policies, by 2005, the agglomeration process that boosted the coastal region still accounted for more than productivity of Chinese enterprises (sup- 90 percent of total exports and imports and ported by policies that revived agricultural received 85 percent of a much larger volume productivity).As elaborated in chapter 16 by of FDI.) With the first-nature advantages of He and summarized here, in the aftermath geographic location supported by favorable of the open-door policies and selective fis- second-nature policy advantages, the coastal cal incentives, the triple forces of marketi- region took off economically. zation, globalization, and decentralization These industrial reforms were comple- jointly drove the spatial transformation of mented by additional agrarian reforms, the industrial sector. notably the overhaul of China's food-grain In 1980 the most agglomerated industries marketing system in the mid-1990s, which were state owned and capital intensive, with increased farm-gate prices and led to the majority of the dispersed industries another surge in rural incomes. Together being resource based. By allowing market these reforms pushed GDP growth rates well signals and globalization to play a more into the double digits during the first half of important role, economic reforms gave birth the 1990s. The growing divergence between to centripetal forces that encouraged com- growth of coastal and inland provinces wid- petitive industries, especially those with an ened, however, and only in the last several export orientation, to locate along the coast. years has there been some evidence of con- Coastal cities have not only first-nature vergence (figure 13.2). (In chapter 18,Yueng comparative advantages in factor endow- and Shen provide a comprehensive discus- ments but also second-nature advantages 200 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA as first-movers in attracting and retaining (Bhattasali, Li, and Martin 2004). Third, as domestic and foreign resources. labor markets became more flexible, wages Meanwhile,as decentralization of respon- began to reflect differences in quality, and sibilities encouraged provincial authorities skill premiums rose sharply. to protect their local industries, inefficien- As market integration deepened, provin- cies and duplicative industrial structures also cial industrial structures gradually became emerged. This was motivated in part by a more diversified, but by the late 1990s, the desire to protect local revenues and jobs, but agglomeration process began to take hold, over time it undermined the achievement of with the effect of nurturing more specializa- productivity gains for the overall economy tion. As He discusses in chapter 16, the rela- as a result of geographic and economic spe- tionship between regional specialization and cialization. As reforms deepened, however, per capita GDP is U shaped. Driven by mar- interactions between centripetal and cen- ket forces, both the very poor and very rich trifugal forces gradually reshaped the spatial regions are now more specialized, with the structure of activities. Differences increased more service-oriented or higher-technology between the productivity of agglomerated industries concentrated in the coastal urban competitive industries and that of dispersed areas. Moreover, as analyzed by Bai and resource-based, heavy industries. Lin in chapter 17, differences in returns to Gradually, a process of rapid urban- capital across provinces have decreased over ization and specialization evolved in the time although returns remain greater in the coastal areas. Several factors supported this coast than in the west. This lends support to process. First, national market integration the view that the agglomeration process has facilitated labor and capital mobility, while not become wasteful over time; rather, mar- cross-provincial commodity exchanges ket forces have been encouraging the more encouraged industrial development in efficient allocation of resources. line with locational advantages. This pro- Today, many agglomerated industries are cess allowed Chinese companies to exploit globalized, with more diversified ownership economies of scale (as discussed by Chen structures, while the dispersed industries and Lu in chapter 15 and by He in chap- either are oriented to the domestic market ter 16). Second, external market integra- or are strategic industries. Heavily pro- tion, as part of globalization, encouraged tected or state-controlled industries remain more dynamic activities to concentrate in relatively dispersed due to the strong local the coastal region, which, in turn, gener- pressures to retain profits (and associated ated agglomeration effects. As the coastal fiscal revenues) within provincial bound- cities became more linked to the global aries. Industries with less local protection economy, the benefits became obvious, as and government intervention typically exemplified by rapid employment creation, are more exposed to external competition. competitive pressures on enterprises to Eventually, as agglomeration effects have restructure, and a much improved domestic taken hold, these industries have become and external financial position. (Yeung and more regionally specialized as well as more Shen describe the contrasting experiences productive. in chapter 18.) This provided the basis for broad-based political support for WTO Role of fiscal policies membership and trade liberalization more These open-door policies were also sup- generally. The remarkable changes in the ported by shifting decentralization and degree of openness of the economy are evi- recentralization of fiscal policies. Although dent in the increase in the ratio of trade to far from meeting the needs of a modern GDP from 10 percent in 1978 to nearly 70 economy, these actions provided Guang- percent in 2006. Much of the trade liberal- dong and Fujian and then other coastal ization took place over the past decade, as provinces with stronger revenue incentives reflected in the decline in trade-weighted to experiment with reforms and thereby statutory tariffs from 40 percent in 1992 to improve their investment climate. But the an estimated 7 percent after WTO accession system also encouraged fiscal disparities Reshaping economic geography in China 201 to widen between the relatively better-off with the central government. The ratio of coastal provinces and the poorer inland central government revenues to total govern- areas. Only with the 1994 tax reforms was ment revenues declined from 55 percent in the fiscal system restructured to begin tack- 1980 to 31 percent in 1989 and to 22 percent ling distributional concerns. in 1993. Fiscal devolution, on the one hand, Before 1980, China's fiscal system was contributed to rapid economic growth by characterized by centralized revenue col- effectively enhancing the incentives of local lection and fiscal transfers, which created governments; on the other hand, it limited few, if any, incentives to develop the local the central government's ability to use tax economy. Under this system, subnational and expenditure policy instruments to nar- governments were highly dependent on the row regional fiscal disparities and support central government and had limited fis- the delivery of basic public services in poor cal autonomy (Ma and Norregaard 1998). localities. At this stage in China's transition From 1980 to 1993, China adopted the fiscal from a centrally managed to a market econ- contracting system, which decentralized tax omy, the priority was more to revive growth administration. Subnational governments, than to deal with distributional concerns. with relatively large discretionary powers to In the aftermath, the more advantageous grant tax privileges, had strong incentives revenue-sharing arrangements contributed to retain fiscal revenues and develop their to the rapid growth of the two provinces on localities by imposing the lowest taxes pos- the frontier of open-door policies, namely, sible on enterprises to compete with other Fujian and Guangdong. This was reinforced regions. As a result, the ratio of total gov- by incentives favoring the allocation of pub- ernment revenue to GDP declined, as did lic investment projects to provinces with a central government revenues relative to total greater financing capacity. As a result, the government revenues. The ratio of total gov- share of public capital expenditures going ernment revenues to GDP declined from 26 to the coastal provinces rose from about 50 percent in 1980 to 16 percent in 1989, and percent in the mid-1980s to nearly 65 per- to 12 percent in 1993 (see figure 13.3). The cent by the mid-1990s. ratio of total government expenditure to With the major tax reform of 1994, this GDP declined accordingly. discretion-based revenue-sharing system was The fiscal capacity of local governments replaced with a more rule-based fiscal assign- closely mirrored the availability of resources. ment system, allowing the central authori- To keep resources within their control, local ties to reassert themselves more actively governments avoided sharing their revenues and to use fiscal policy for redistribution Figure 13.3 Ratio of total government revenue and expenditure to GDP in China, 1980­2005 30 25 20 15 percent 10 5 0 1980 1985 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 year Total government revenue to GDP Total government expenditure to GDP Source: National Bureau of Statistics of China (various years). 202 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA (Dollar and Hofman 2008). The reform government revenues to total government package brought China's intergovernmental revenues rose sharply, more than doubling fiscal system much closer to international from about 22 percent in 1993 to 56 percent practice and paved the way for a turnaround in 1994 and hovering around 50 percent in in the ratio of government revenues to GDP recent years. The share of central govern- in the second half of the 1990s (figure 13.3). ment expenditures to total government To mitigate possible resistance from the expenditures remained at about 30 percent. provinces, the central government made The centralization of the fiscal system certain concessions, including tax rebates strengthened the central government's in favor of the richer provinces, and only capacity to redistribute in favor of poorer gradually changed the tax-sharing arrange- inland provinces. After a decade of decline ments between the central and subnational under the fiscal contracting system, the share governments. of total fixed-asset investment that went to Before the reforms, the share of central the inland region versus the coastal region government expenditures was roughly in increased gradually from the mid-1990s line with its share of revenues (see figure onward (see figure 13.5). 13.4). After the reform, the share of central The ratio of local government expendi- tures to regional GDP rose over time under the tax-sharing system, especially in the Figure 13.4 Ratio of central government revenue and expenditure to total government revenue inland region in the late 1990s to early 2000s, and expenditure in China, 1980­2005 when the"Go West"policy was implemented 60 (see figure 13.6). By 2005, the ratio of fiscal revenues in the eastern, central, and western 50 parts of China was 60:23:17, while the ratio 40 of their expenditures was 46:29:25, which suggests that overall the fiscal system has 30 percent had some redistributive effects (Lou 2008). 20 Figure 13.6 shows that total central trans- fers range widely across provinces in aggre- 10 gate terms as well as in per capita terms, but 0 there appears to be some preference in favor 1985 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 of inland provinces.1 Generally, the larger year the size of the provincial economy, the Central government revenue to total government revenue larger the central transfer; however, in per Central government expenditure to total government expenditure capita terms, the level of transfer is relatively Source: National Bureau of Statistics of China (various years). greater, the poorer the province, especially if it is inland (see figure 13.7). At the provincial level, a higher GDP per Figure 13.5 Share of total investment in fixed assets in coastal and inland areas of China, 1982­2007 capita is associated with a lower ratio of central transfers to subnational government 70 expenditures (see figure 13.8). Central trans- 60 fers play a more important role in inland 50 provinces, especially in the more remote and 40 poorer ones, as their subnational adminis- trative units are more dependent on inter- percent30 governmental support. In 2004 the ratio of 20 central transfers to subnational government 10 expenditures was, on average, about 40 per- 0 cent for coastal provinces, but almost 70 1982 percent for inland provinces. This suggests 1984 1986 1988 1990 1992 1992 1996 1998 2000 2002 2004 2006 year that central transfers have helped to reduce Coast Inland disparities in fiscal capacity across regions Source: National Bureau of Statistics (various years). (see also chapter 15 by Chen and Lu). Reshaping economic geography in China 203 Figure 13.6 Ratio of total provincial government expenditure to regional GDP in coastal and inland regions of China, 1999­2006 18 16 14 12 percent 10 8 6 1999 2000 2001 2002 2003 2004 2005 2006 year Coast Inland China Source: National Bureau of Statistics (various years). However, fiscal disparities across regions, spend the least on public health. In 2003 the as measured by per capita social expendi- maternal mortality rate was 73 per 10,000 tures, continue to be large. Wealthier prov- live births in the poorest fifth of the popula- inces, as measured by GDP per capita, tend tion covered by China's maternal and child to have a higher level of local government health surveillance system and 17 per 10,000 expenditures per capita. This positive rela- live births in the richest fifth (Wagstaff and tionship has remained virtually unchanged Lindelow 2008). over the years. In 2005 GDP per capita of Although revenue collections have been the richest province, Zhejiang, was about 5.5 gradually recentralized, expenditure assign- times that of the poorest province, Guizhou; ments in China continue to be exception- and subnational government expenditures ally decentralized. The mismatch between per capita of the former were about 1.85 expenditures and revenue assignments has times those of the latter. Disparities in public led to major financing gaps, as responsibili- spending contribute to disparities in social ties cascade from the center to the provinces outcomes across regions and between rural and then down through several layers to the and urban areas.2 Although the differences local level. Subnational governments have a in subnational government expenditures wide array of economic responsibilities in on education and health care across regions addition to the delivery of social services, have been decreasing in recent years,per cap- including basic health care, education, and ita expenditures in the coastal region are still social security schemes. Such a mismatch more than 1.5 times those of inland regions of risk pooling and resource redistribu- (see figure 13.9). For example, in 2005 the tion leads not only to deadweight efficiency national average per capita expenditure on loss but also to uneven provision of public public health was Y 78; major cities like Bei- services. As fiscal transfers from the upper jing spend several multiples of the average. levels are often inadequate, local govern- But worth noting is that some of the poor- ments largely rely on their own devices to est provinces like Tibet and Qinghai spend finance and deliver public services. Due to well above the average, largely because of the tighter fiscal constraints, many local govern- higher costs of serving a sparsely distributed ments, especially those in rural inland areas, population. Furthermore, budgets are set at which often suffer from significant resource the county level and are based largely on shortfalls, have to provide more limited pub- county government revenues. Local govern- lic services, while charging higher user fees. ments in the poorest parts of China, which In sum, although revenues have grown face the toughest public health challenges, substantially in recent years and the fiscal 204 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 13.7 Central transfers and GDP per capita in China at the provincial level, 2004 60 inland provinces Guangdong Sichuan yuan) 50 Henan Hunan Liaoning (billion Helongjiang Jiangsu Hebei Yunnan Shandong 40 Hubei Inner Mongolia (2004) Jilin Anhui Zhejiang Shaanxi Guangxi 30 Guizhou Xinjiang transfers Jiangxi Gansu Shanxi Chongqing central 20 Fujian coastal provinces total Tibet Qinghai 10 Ningxia Hainan 0 500 1000 1500 provincial GDP (2004) (billion yuan) coastal provinces inland provinces 5000 Tibet yuan 4000 inland provinces (2004) 3000 transfers Qinghai central 2000 coastal provinces Ningxia Inner Mongolia total Xinjiang Gansu Jilin Helongjiang Shaanxi Hainan Liaoning 1000 Yunnan Shanxi Guizhou Jiangxi Chongqing Guangdong Zhejiang Hunan Hebei Fujian Jiangsu per-capita Guangxi Sichuan Hubei Anhui Henan Shandong 0 5000 10000 15000 20000 25000 per-capita GDP (2004) yuan coastal provinces inland provinces Source: Shen (2008). system has now taken on a redistributive Overcoming the distance factor: role, the impact on the provision of more expanding transport services equitable access to public services is still China's experience illustrates the importance modest, in part because of the way expen- of spatial factors in development. There is a diture assignments are pushed down to the significant correlation between the distance local level without the provision of com- of inland provinces to the major commer- mensurate funding and in part because of cial centers along the coast and the level of the structure of revenue sharing. These con- economic development. This is illustrated sequences are more significant in the poorer by the growth rates of per capita income, inland provinces and partially explain why which decline with distance from the east urban-rural disparities are greater there coast. As seen in figure 13.10 (panel A), relative to the coastal areas. Reshaping economic geography in China 205 Figure 13.8 GDP per capita and ratio of central transfers to subnational government expenditure at the provincial level 2004 1 Tibet Qinghai inland provinces Ningxia to 0.8 Gansu expenditure Jilin Shaanxi Inner Mongolia Guizhou transfers Jiangxi Hunan Xinjiang Helongjiang Xunnan Sichuan Hubei 0.6 Hainan Guangxi Chongqing Anhui coastal provinces Henan Shanxi central government Hebei Liaoning of ratio 0.4 Fujian Shandong Jiangsu Zhejiang subnational Guangdong 0.2 5000 10000 15000 20000 25000 per-capita GDP (2004) yuan Coastal provinces Inland provinces Sources: National Bureau of Statistics (various years); authors' calculations. Figure 13.9 Subnational government expenditures per capita on education and health care in coastal and inland provinces of China, 1999­2006 600 185 on (2004) 180 500 175 capita 400 170 (percent) expenditure per 165 ratio care 300 160 inland health 200 155 government to and 150 100 145 coastal subnational 0 140 education 1999 2000 2001 2002 2003 2004 2005 2006 year Coast Inland China Coastal to Inland ratio Source: National Bureau of Statistics (various years). in the early years of the reform period provinces were to the coast, the faster they (1979­87), growth was less dependent on grew, with the effect illustrated by a more location, as measured by the relatively flat steeply sloped line in figure 13.10 (panel A). line showing the relationship between GDP Given the nature of the growth process in per capita growth rates and the "adjusted the reform period, a typical coastal province distance"3 of a province from the coast. This benefited from the compound effect of a is because the "balanced growth" strategy reduction in the adjusted distance and the in the earlier years partially offset any loca- presence of market-oriented policies that tional advantages, with the consequence capture the benefits of location. While a that growth rates tended to equalize across typical inland province benefited from the provinces, but at a lower level. With the same reduction in adjusted distance, the dis- onset of reforms and globalization dur- advantages of remote location in the reform ing the 1988­95 period, locational advan- era had a negative effect on its growth poten- tages became more important; the closer tial. However, the adverse impact of distance 206 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 13.10 Relationship between growth in provincial GDP per capita and adjusted distance from the major coastal commercial areas in China, 1979­2003 A. 1979­1995 B. 1988­2003 0.1 0.1 per per rate rate GDP GDP 0.075 0.075 growth growth annual annual capita capita average average 0.05 0.05 7.5 8 8.5 9 9.5 10 10.5 7.5 8 8.5 9 9.5 10 10.5 average adjusted distance in logarithmic form average adjusted distance in logarithmic form 1979­1987 1988­1995 1996­2003 1988­1995 Source: Authors' calculations based on National Bureau of Statistics (various years). on growth was mitigated over time as China early 1990s, emphasis was placed on devel- expanded its transport services. With the opment of the coastal provinces and on massive transport investment programs upgrading of logistics services to improve initiated in the 1990s, the distance factor connectivity with the outside world. After became less of a barrier during the third 2000, increasing priority was given to infra- period from 1996 onward (figure 13.10, structure investments in the western and panel B), making it possible for the growth central regions, as exemplified by a 45 per- rates of the inner provinces to move a bit cent expansion in highway mileage from closer to those along the coast, as shown in 1999 to 2004, compared with a 30 percent the line becoming slightly flatter. This may expansion in the eastern region (Li and Xu be one of the reasons why the gap in growth 2007). Over the past decade, completion of rates between the inland and coastal prov- the 44,000-kilometer national highway sys- inces has narrowed in recent years. tem along with improvements in 400,000 Part of China's success story is therefore kilometers of local and township roads con- due to sustained improvements in trans- nected almost all of China's major regional portation and communication networks, centers.4 This was supplemented by huge which mitigated some of the disadvantages investments in the railway network, ports, associated with distance. In the process, and inland waterways. All together, China this allowed competitive forces to reshape has been spending more than 5 percent of interprovincial industrial structures more GDP on transport investments annually, in line with evolving comparative advan- amounting to more than US$100 billion tages. These investments also contributed in 2006 (of which about a third was on the to globalization pressures by reducing the expressway network). This is roughly double costs of transport and logistics as a per- the amount spent in comparable countries. centage of the final price of traded goods, As seen in figure 13.11, highway expan- which helped to link production to both sion was initially concentrated in the three domestic and global markets. Internal divi- core economic centers in the eastern region sions in the form of provincial boundaries (Pearl River delta, Shanghai, and Beijing- have traditionally inhibited the integration Tianjin), spreading inward over time, follow- of national markets, as evidenced by abnor- ing the need to move finished products and mally large differences in product prices people. These patterns reinforced the fiscal and wages across provinces before the mid- and investment-related policy reforms,which 1990s. Due largely to massive investments in favored growth along the coast and helped to transport infrastructure in the 1990s, inter- promote agglomeration effects arising from nal transport and logistics costs have fallen unification of the national market. significantly, and interprovincial prices and The pattern of development for the rail- wage rates have begun to converge. In the way sector has been different, driven largely Reshaping economic geography in China 207 Figure 13.11 Density of highways in China, 1978, 1999, and 2003 kilometers per 10,000 square kilometers Density of Highways (km/10,000 km2) 1978 1999 2003 > 5,000 3,000­5,000 1,000­3,000 < 1,000 by the need to move bulk commodities like Improved connectivity has increased coal from resource-intensive regions in the efficiency and facilitated greater industrial northeast (see figure 13.12). Thus railway specialization through agglomeration econ- development was initially concentrated in omies. Reduced transport costs between the northeast, spreading first into the cen- inland cities and coastal mega cities and sea- tral provinces and then to the coastal areas. ports--and from there to overseas markets-- Investment in railways has lagged invest- have promoted development of inland firms ment in the highway system. In particular, in two ways: through lower costs of inputs connections between the north and south delivered by inland factories and higher net for passenger and processed goods have revenue from sales to external markets. The not kept up with demand. Consequently, result is greater access to both national and the rail system is overloaded, carrying 25 external markets, creating more competition percent of the world's railway traffic on just and eroding existing local monopolies. Both 6 percent of its track length, but capacity skilled and semiskilled labor is increasingly is expanding rapidly in line with an ambi- able to move from the interior to the coast tious investment program laid out for the and back in line with shifting economic coming decade. activity, resulting in greater economies of Evaluations of the rates of return on scale and rewards to innovation. highway projects over the past decade show These transport investments are part of that the majority cluster in the 15­25 per- a broader strategy to reduce logistics costs cent range. There is now enough evidence to driven, in part, by the competitive pressures support the view that transport investments of globalization and the key role that China have been fundamental in linking regional now plays in regional supply chains. Fac- markets. Only in the last few years has it been tors that influence logistics costs range from possible to drive by expressway from Beijing customs procedures, transport infrastruc- to Hong Kong (China) and from Shanghai ture, security, and regulations. China has to Xinjiang. Within the eastern provinces, made impressive progress over the past few transporters now have optional routes,which decades. According to rankings on the logis- are important for long-distance transport of tics performance index (World Bank 2007b), high-value freight, thus allowing distributors China ranks 30 globally among 150 coun- to hold smaller inventories and respond more tries, but 1 among lower-middle-income quickly to changing market tastes. Wal-Mart countries. However, there is still potential now has a single major distribution center in for improvement, as measured by estimates Guangdong for supplying stores throughout that logistics costs amount to 18 percent of China, a practice that would not have been GDP in China compared with 10 percent in possible a decade ago. the United States ("China's Infrastructure 208 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 13.12 Density of railways in China, 1978, 1999, and 2003 kilometers per 1,000 square kilometers Density of Railways (km/10,000 km2) 1978 1999 2003 > 200 100­200 50­100 < 50 Splurge: Rushing by Road, Rail, and Air," employment opportunities to be suffi- Economist, February 16, 2008). ciently attractive and durable, is still absent in China. In the absence of more formal Toward a more urbanized economy land use markets in rural areas, the equally and implications for labor mobility contentious and politically sensitive issue of China's urbanization process accelerated granting residency rights to migrant labor when the introduction of township-and- in the cities is difficult to resolve. However, village enterprises drew workers out of as addressed in many studies, perhaps the farm production and facilitated migration. most effective instrument for dealing with Over time, this spawned a large popula- rural-urban disparities would be to reform tion of migrant labor, which now totals an the hukou system and give migrant work- estimated 140 million, heavily concentrated ers better access to social services and equal in the major commercial centers along the employment rights. But understandably, coast: Pearl River delta (Guangzhou and elements of the system have helped China Shenzhen), Yangtze River delta (Shanghai), to avoid some of the worst features of urban and Bohai Bay area (Beijing-Tianjin). See slums typical of many Asian cities. Thus the figure 13.13. However, rural migrants often issue is really about how to manage better can only find jobs not wanted by urban resi- rather than halt rural-urban migration to dents, characterized by long working hours, moderate social tensions while providing poor pay, and inferior working conditions. more equitable access to employment and As reforms deepened, administrative con- social services. straints on labor mobility were progressively A quarter century after the reforms alleviated in the mid-1980s.Although access began, China's urbanization rate, including to housing, health services, and education migrant labor, has more than doubled and for children remained unequal for migrants is now about 50 percent. The urbanization without urban residency status (hukou), rate of Guangdong province now exceeds they became less prohibitive over time. 60 percent compared with about 35 percent For many migrants economic security in most western provinces. At the current is typically linked to their rural hukou in stage of development in metropolitan areas, their home province, which, in the absence positive agglomeration effects dominate of formal land markets, provides the right negative congestion effects, although China's to use agricultural land. A fully function- cities face severe environmental challenges ing land market, which would allow exist- and urban transport systems need to be ing landowners to sell or lease use rights to improved. Contrary to popular perceptions, others and migrate to the city if they found in relation to its population and land mass, Reshaping economic geography in China 209 Figure 13.13 Interprovincial migration flows to the coast in China, 1995­2000 CHINA INTER-PROVINCIAL MIGRATION FLOWS TO THE COAST 1995­2000 MIGRATION TOTALS: MIGRATION FROM WESTERN REGION TO 2,000,000 COASTAL REGION H E I L O N G J I A N G MIGRATION FROM 1,000,000 CENTRAL REGION TO C E N T R A L COASTAL REGION 500,000 R E G I O N MIGRATION WITHIN THE 250,000 COASTAL REGION 100,000 J I L I N W E S T E R N R E G I O N LIAONING Sea of G OL O N BEIJING Japan I M N E Beijing Bo Hai TIANJIN H E B E I C O A S T A L IA X R E G I O N G SHANXI Yellow Q I N G H A I IN SHANDONG N Sea G A N S U H E N A N S H A A N X I JIANGSU C E N T R A L W E S T E R N R E G I O N ANHUI R E G I O N SHANGHAI H U B E I East S I C H U A N China ZHEJIANG Sea CHONGQING JIANGXI C O A S T A L HUNAN R E G I O N F U J I A N GUIZHOU TAIWAN Y U N N A N G U A N G X I GUANGDONG HONG KONG MACAO NATIONAL CAPITAL PROVINCIAL BOUNDARIES REGIONAL BOUNDARIES HAINAN INTERNATIONAL BOUNDARIES China's major cities are too small rather than format of city development, has unnecessar- too large. Building new "secondary" towns ily increased infrastructure costs by creating on the edge of existing cities may be effective, less densely utilized enclaves and increasing but only if there is a strong demographic and the costs of urban transport and other social economic rationale for doing so. Fragmen- services. More efficient urban planning that tation in large cities (agricultural or vacant would infill "leapfrogged" areas will be an land within the contiguously built-up city), important issue as urban population growth resulting from a typical multiple-ring spatial continues to accelerate. 210 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Dealing with emerging disparities: production increased by around 4 percent spatial and welfare aspects a year; by international standards this is quite impressive. Yields for the major food Income inequality in China, as measured by grains, for example, are already quite high various indicators, has risen rapidly in the in China compared with other countries; past quarter century: between coastal and in fact, they are similar to or higher than inland regions, within provinces, as well as those in the United States and Japan and across and within rural and urban areas. much higher than in Argentina, Canada, The dynamics of spatial divergence across and Thailand (Yusuf and Nabeshima 2008). subnational areas have taken the form of Given technological constraints and the cost a "race to the top." Disparities are not the implications of increasing crop yields in a result of stagnant income growth among land- and water-scarce economy, it is hard to certain segments of society or regions but see how agricultural growth could have been rather the consequence of unusually high rapid enough to prevent rising urban-rural and sustained growth in coastal and urban income ratios,because urban income growth areas. As the biggest gains have gone to the rates have, at times, approached double dig- leading commercial centers, the income gap its. Thus experience suggests that inequality between the coastal and inland provinces as would be more effectively addressed by pro- well as between urban and rural areas has moting off-farm income opportunities in steadily widened. rural areas. This is evidenced by the higher As one indicator, the Gini coefficient has share of off-farm income in the coastal areas risen from around 30 to 45 over the past 25 compared with the inner provinces and the years (see figure 13.14). Although regional extent to which this has lowered urban-rural disparity is widely considered as the key income disparities in the former. determinant in China, the rural-urban The challenge in reducing inequality, divide is the more important factor in shap- however, is that, as impressive as growth ing overall inequality. Changes in the Gini in rural incomes has been, it is still much are closely associated with changes in the lower than growth in urban incomes. As a urban-to-rural income ratio and the coast- result, trend lines for indicators of disparity al-to-inland per capita GDP ratio. either level off or reverse during periods of Shifts in the urban-to-rural income ratio sharply rising rural incomes, notably in the as well as the Gini are largely explained by early 1980s with the household responsibil- performance of the rural economy. Over ity reforms, in the mid-1990s with agrarian the past two and half decades, agricultural Figure 13.14 Income inequality in China, 1978­2004 3.5 50 45 40 ratio 3 35 capita 30 per 2.5 25 Gini 20 (GDP) 15 2 Income 10 5 1.5 0 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 19891990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Urban to rural income ratio Coastal to inland GDP per capita ratio Gini (unadjusted) Source: Authors' calculations based on data from Ravallion and Chen (2004). Reshaping economic geography in China 211 marketing reforms, and more recently with employment and flat remuneration system the reduction in rural-based taxes and the among urban enterprises and the compre- increase in commodity prices. For example, hensive social welfare role provided by the the urban-to-rural income ratio fell from communal farming system in rural areas. 2:5 in 1980 to 1:8 by the mid-1980s and When marketization and structural transfor- then rose to 2:8 by the mid-1990s. It then mation deepened, individual circumstances fell again to 2:5 in the late 1990s, before ris- gradually played a more important role in ing sharply and leveling off at the current determining income, including the high pre- ratio of 3:1 (World Bank 2007b, forthcom- mium accorded to education.6 This suggests ing; also see chapter 14 by Yao). that increasing income inequality is, to some Regional factors do matter, however, extent, a consequence of the stage of China's because the larger urban-rural differences development: the growth process unleashed in the western provinces can be seen as competitive pressures and created incentives structural: in those areas, ecological con- for investment in skills enhancement. ditions militate against higher agricultural There is also a regional dimension to the productivity and lower urbanization rates. evolution of inequality. Among the coastal Moreover, more isolated settlement patterns provinces, the poorest groups in rural areas raise the costs of providing public services have experienced phenomenal growth in to rural inhabitants. Per capita GDP in the their incomes, while among inland prov- coastal region is now more than twice that in inces rural households have not experienced the inland region. As documented in many as significant an increase. Meanwhile, the studies, the coastal provinces have smaller incomes of the wealthier groups in urban urban-rural income gaps than the inland areas--both the coastal and inland regions-- provinces, with differences within rural and have been rising the most rapidly (see figure urban areas in the poorer regions being par- 13.15). Overall, in both urban and rural ticularly pronounced (World Bank 2007a). areas, the growth rate of income has been In 2006 the ratio of urban to rural per capita higher in coastal than in inland provinces. income for the three richest provinces was The divergence in growth rates between about 2:5 times, while the ratio for the three coastal and inland provinces peaked between poorest provinces was about 4:5.A large part the mid-1980s and 1990s, but there seems to of the inequality between regions is associ- be some convergence in recent years, espe- ated with the differences between their rural cially in the western region, as the growth areas and is related to the uneven degree of rates of many of the poorer provinces urbanization across provinces. Thus equal- increased more rapidly (see table 13.1). This izing mean incomes between rural areas and is also noted by Yao in chapter 14 and by Li urban areas would have a larger impact on and Xu (2007). These encouraging trends reducing overall inequality than equalizing may be due to recent regional policies, but mean incomes across regions (World Bank they are also an affirmation that globaliza- forthcoming). Both rural income levels and tion forces, embodied by WTO accession, urbanization rates are lower in inland prov- have not worsened the rural-urban divide. inces than in coastal provinces. In 2005 rural Accompanied by labor market reforms, the per capita incomes were 70 percent higher forces of globalization actually narrowed and urbanization rates were 65 percent the differences between international and higher in the coastal region than in the west- domestic market prices for agricultural ern region. As urban incomes are, on aver- products and eliminated domestic policy age, two to three times rural incomes, the distortions between farm prices and mar- higher income levels in coastal areas are due ket prices. In doing so, the terms of trade largely to their more urbanized labor force. improved in favor of agricultural products, Within provinces, inequality within rural and rural-urban inequality declined (Huang and urban areas has accounted for a larger and others 2007). share of total inequality over time.5 Before It might be too early,however,to attribute the reform era, the extent of urban and rural this convergence to the impact of various inequality was contained by the guaranteed regional development policies--"Go West" 212 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 13.15 Income growth in rural and urban areas of China, 1989­2004 Rural Urban 300 300 (%) (%) 200 200 household household increase 100 100 capita increase capita per 0 0 income per real income ­100 real ­100 0 25 50 75 100 0 25 50 75 100 percentile of the population (ranked by real percentile of the population (ranked by real per capita household income) per capita household income) Inland province Coastal province Inland province Coastal province Source: Luo and Zhu (2008). in 1999, "Revive the Northeast" in 2003, Table 13.1 GDP growth rates of central and western and "Central China Rising" in 2005--that regions as a percentage of the coastal region in China, 1980­2006 targeted rebalancing regional growth (see coastal = 100 chapter 14 by Yao). By remedying market Year Central Western failures and some of the earlier biases in 1980­85 94 91 spatial policies, these programs probably did 1986­90 72 80 help to address some of the issues under- 1991­95 75 63 pinning regional disparities. In focusing on 1996­2000 87 82 ecological restoration and infrastructure 2001­06 87 92 improvements in the west, restructuring Source: National Bureau of Statistics (various years). static industrial and institutional systems in the northeast, and developing intermodal between rural and urban areas are large. As connectivity facilities in the central region, of 2000, 2.5 percent of the urban popula- these efforts can be an effective means to tion between 15 and 64 years of age received develop the comparative geographic and no education, while the proportion in rural economic advantages of the various regions. areas was more than three times as high, at However, arguments for a more aggressive 8.7 percent. Child and maternal mortality investment-led strategy to raise agricultural rates are twice as high in rural areas as in productivity to levels comparable to those in cities. Moreover, social disparities between the industrial and services sectors, and thus urban and rural areas appear to be greater to moderate rural-urban disparities, should in the poorer provinces, particularly in the be pursued cautiously, given differences in west. For example, the urban-rural dispar- regional endowments and considerations of ity in life expectancy is less than 3.5 years in cost-effectiveness. the eastern region, but more than 8 years Judged by social indicators, China's per- in the western region. Over the past decade, formance has been favorable, with achieve- given the increased attention to regional ments exceeding what would be predicted differences, especially in the poorer western in relation to income levels. China's human provinces, interregional gaps at the primary development index (HDI) has risen contin- educational levels have narrowed. However, uously over the past quarter century to 0.78, urban-rural disparities in health conditions placing the country 81 in 2005 among 177 may have widened since the late 1990s due countries (UNDP 2007). China's primary to the persistence of urban-rural income school net enrollment rate of 97 percent inequality and slow development of rural and life expectancy of 72.5 years are higher health care insurance systems. This issue today than the average of lower-middle- has drawn increased attention from policy income countries. makers, with indications that more support However, as discussed earlier, dispari- will be brought to bear in the near future. ties in social indicators among regions and This reinforces the point made earlier: the Reshaping economic geography in China 213 persistence of such disparities in welfare the needs of the population. Protectionist indicators illustrates how much further the provincial regulations that discourage inter- distributional aspects of fiscal policies need regional movement of goods, finance, and to go to moderate trends. services and encourage duplicative industrial structures have lessened as national markets Looking to the future have become more unified, but many inef- Recent trends suggest a gradual convergence ficiencies persist. Regionally differentiated in growth rates between the coastal prov- policies should continue to recognize that the inces and the interior. But the advantages of priority for the western region is defined by location will likely persist even if narrowed, its fragile ecological conditions and the need with agglomeration effects continuing to to strengthen its base of human capital to favor the larger and more globalized urban prepare for the voluntary migration of labor areas along the east coast.What, then, should to better employment opportunities. For the be the course of future policies, given public northeast, the priority is to encourage more pressures to deal with increasing disparities? aggressive restructuring of enterprises, create Both theory and experience indicate that supportive social protection systems, and tap government initiatives should not try to the region's natural agriculture-based advan- "balance the location of productive capacity" tages. For the central region, the priority is across regions.However,a strategy to"moder- to strengthen intermodal transport links and ate differences in economic welfare" between logistics services, as commercial activities the coastal and inner provinces and between inevitably shift inward to serve major popu- rural and urban areas would involve a three- lation centers and growth becomes more prong approach that builds on China's domestically driven. past success by (1) strengthening the distri- In the future, however, China's evolv- butional aspects of the fiscal system so that ing economic structure is likely to favor a regional and rural-urban differences in access more balanced pattern of investment in to social services are reduced and allocations transport infrastructure. Investment in one of investment projects are less constrained by province will increasingly have spillover the financing capacity at the poorer subpro- effects on other provinces (see chapter 17 vincial levels; (2) eliminating jurisdictional by Bai and Lin; Luo 2005). In addition, as barriers that inhibit mobility of labor, finan- China's growth becomes driven relatively cial resources,and goods,while strengthening more by growth in domestic consumption infrastructure and logistics links so that the and services and less by external demand, regions and rural-urban areas are better con- highway investment in the central provinces nected; and (3) encouraging complementary will exert both a push and a pull effect on regional development policies that recognize adjacent western and eastern provinces, and build on the uniqueness of geographic and this may produce the best overall net and inherited economic differences rather economic effect (Luo 2004). In addition, than working against them. while a relatively complete transport net- The government's policies have been gen- work has developed along the coastal areas, erally consistent with this agenda, although many of the western and some of the central progress on some aspects could be acceler- provinces lack high-grade highway connec- ated and others refined given political pres- tions to each other and to the major eastern sures (see chapter 15 by Chen and Lu and cities. Over the coming decades, the regional chapter 14 byYao).The distributional impact pattern of investment priorities is likely to of the fiscal system in transferring resources involve a combination of the following: from richer to poorer provinces and between (1) eastern provinces: expand capacity where urban and rural areas should be further the volume of traffic warrants; (2) central enhanced, especially since revenue growth provinces: complete network connectivity; has been so buoyant in recent years.Although and (3) western provinces: improve acces- the bias favoring infrastructure needs in the sibility through a balanced investment coastal provinces has been reversed, interior program of expressways and lower classes regions remain less well served in relation to of roads (World Bank 2007a). 214 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Positive externalities within and across in China's hukou system, which are closely sectors have further intensified the concen- linked to policies governing rural and urban tration of activities. As the most developed land use rights. No other measure is likely coastal metropolitan areas move up the lad- to have as significant an impact as liberal- der into higher value-added services and izing internal migration from rural to urban high-technology industries, dispersion of areas and from disadvantaged inland regions labor-intensive activities remains limited to the more dynamic commercial centers within the coastal region (Catin, Luo, and along the coast as well as the newly emerging Van Huffel 2005). Only a few industries inland commercial centers such as Chongq- have relocated to adjacent inland prov- ing and Wuhan. inces, and some are attractive candidates, The government's policies have been if there are no major agglomeration econo- generally consistent with this agenda, plac- mies in being along the coast and if access ing renewed emphasis on improving welfare to domestic consumer centers and lower in rural areas and dealing with the environ- cost structures become more important. mental concerns from rapid urbanization. Thus reducing local protective barriers and Taken together, such initiatives will promote encouraging more regional integration will the agglomeration benefits from higher help to strengthen the spillover effects from density, help to minimize the distance fac- the coast to the rest of the country, which tor that impedes more efficient location of are necessary to facilitate continued gains economic activity, and take down the divi- in productivity. sions that limit factor mobility and discour- The disadvantages of being located in the age sustainable development outcomes. As remote interior may not always be a prob- discussed below, over time, location-specific lem, and, ironically, globalization may well disparities, as measured by social welfare be a significant factor in expanding trade to indicators, will become less pronounced. Central Asia and Europe through the old silk route via the relatively poorer western prov- Likely trends in inequality inces of Gansu and Xinjiang. Similarly, there The key question facing policy makers is is a noticeable increase in trade flows going whether China can maintain rapid growth into Southeast Asia via Guangxi and Yun- in a way that moderates existing inequalities. nan due in part to the promotional efforts This is all the more important given the con- to integrate the Greater Mekong region (see cern of the senior leadership for fostering a chapter 6 by Rigg and Wittayapak). In fact, more"harmonious society,"which is a major growth in exports along these border areas is objective of the current Five-Year Plan. Thus expanding more rapidly than anywhere else, how will the above policies affect distribu- albeit from a relatively low base. tion of income across income groups and Pressures will also be high to address space? The links between growth and equity widening urban-rural disparities. Improv- are complex. Sustained growth is likely to ing connectivity and encouraging spillover uplift the incomes of everyone--urban and effects that provide more wage-based rural rural, rich and poor--as has been the case employment is the most attractive option. thus far. But as impressive as this progress While designing policy reforms for rural has been, it does not necessarily lead to a and urban areas is often seen as separate more equitable distribution of income. As exercises, sustainable institutional solutions noted, over the past quarter century, the may emerge from thinking about the prob- Gini has been closely associated with move- lem in a more spatially neutral framework. ments in the ratio of urban-to-rural incomes For example, social welfare schemes that and coastal-to-inland GDP per capita. Small segregate rural and urban populations tend differences in annual growth rates can lead to accentuate differences and hamper finan- to wide gaps over time. Take urban and cial sustainability. rural income growth as an example. A 1 Perhaps the major challenge, however, is percent difference in the annual growth rate managing the pace and nature of reforms (4 percent for rural and 5 percent for urban) Reshaping economic geography in China 215 compounded over 24 years will result in an areas reaches a critical threshold, centrip- increase in the urban-to-rural income ratio etal and centrifugal forces may work in from 2.5 times to 3.2 times. If one ignores favor of the emerging new economic cen- the fluctuations over time, that is roughly ters and eventually lead to a narrowing of the change in the urban-to-rural income inland-coastal differences. ratio during the period of 1980­2004. · Current policies that emphasize a more In the future, inequality will very likely balanced approach in building a "more continue to rise. The reason, as observed harmonious society" may redistribute by Kuznets in his seminal work, National resources in favor of more social services Income and Its Composition (Kuznets 1941), for less-developed areas and poorer seg- is that, as a country develops and the popu- ments of the population. Implementa- lation moves from lower-productivity agri- tion of appropriate fiscal and regional culture to higher-productivity urban cen- policies will influence the emergence of ters, this leads to a lengthy period of rising new economic centers, but in ways con- overall inequality as the share of the higher- sistent with shifting comparative advan- paid urban workers increases relative to the tages and market forces. less well-off rural population (Bourguignon · Migration will become a powerful force 2008). But over time, as the rural popula- in shaping development and influenc- tion diminishes with migration, inequality ing distribution. Movement of labor, begins to decline as the vast majority of the typically from rural to urban and inland population becomes engaged in urban- to coastal areas, contributes to over- based activities. all income growth. On the one hand, it At this stage in China's development, for provides coastal and urban areas with a the foreseeable future growth in productiv- flexible labor supply at reasonable costs; ity will remain higher in the industrial and on the other hand, it augments the avail- service sectors than in agriculture due to ability of land and other resources for specialization and agglomeration effects. those staying in rural farming areas, China's GDP growth rates, however, will while generating higher incomes for likely decline somewhat from recent levels migrants (Zhu and Luo 2008). Although to a more sustainable 8­9 percent over the the impact of migration on lowering coming decade or two. The recent shift in the urban-to-rural income ratio may be the terms of trade between the industrial uncertain--because this depends on the and primary sectors will reverse itself at combined effects on migrants and non- some point or become a neutral factor in migrants in sending areas and residents affecting intersectoral incomes. As a conse- in receiving areas--migration is likely quence, the urban-to-rural income ratio will to reduce coastal-inland inequality by feed into rising overall inequality as mea- accelerating urbanization and increasing sured by the Gini.With appropriate policies, overall productivity. however, a plausible scenario could begin to reverse this trend: How long will it take before the forces that moderate income inequality begin to make a · As a large continental economy (similar difference? As the percentage of the popu- to the United States), domestic demand lation working in more productive jobs in will eventually play a more dominant role urban areas becomes large enough, at some than trade as China develops. Industries, point the pattern of income divergence and starting probably with those that are more inequality will have peaked and the Gini will "footloose" and geared to the domestic begin trending downward. These processes market, may choose to locate more inland take decades and perhaps even a genera- to optimize their cost-profit structure and tion to become significant if the experiences better respond to consumers in the major of developed countries are to be taken as a population centers in the central prov- guide. We project that with good policies-- inces. Once the economic mass in inland freer internal migration, more redistributive 216 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 13.16 Projected household income inequality in China, 1981­2021 Household income inequality projections (Gini coefficient) 55 50 45 40 35 30 25 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017 2019 2021 Gini (good policy) Gini (poor policy) Source: National Bureau of Statistics (various years) and authors' calculations. fiscal policies, and continued infrastructure ity is normal, but in China's case, the disparities investments to improve connectivity--the are unusually large. Gini will continue to rise over the next dec- 3.Adjusted distance is defined as the"traveling ade, peaking at around 48, but then begin distance" between the province in question and to decline before 2020.7 With less effective the economic centers along the coast, adjusted by policies, the Gini will continue to increase to the level of development of the transport network. Transport investments "shorten" the economic well over 50 and may then level off, but it will distance between the two provinces by reducing not begin to decline even by 2020 (see figure transportation costs. See Luo (2001, 2004). 13.16). Either way it is important to realize 4. For a comprehensive discussion of trans- that growth and spatial factors have a very port investment and its impact on development slow impact on distributional outcomes. in China, see World Bank (2007a), which is the This does not mean that social and economic source of much of this discussion. progress is not being made. With appropri- 5.Rural inequality continues to be higher than ate policies, all segments of society are likely urban inequality, although the rate of increase in to better off in the coming years. How China urban inequality is becoming more significant. handles this complex set of issues will have 6. A recent study based on household survey profound implications for shaping the loca- data of eight Chinese provinces in 1989­2004 (Luo and Zhu 2008) suggests that the most tion of future activity and for determining important factor explaining overall inequal- its impact on social disparities and growth. ity is the differential returns to schooling and sector of employment; the increase in returns Notes to education explains two-thirds of changes in Yukon Huang is former World Bank director for household income in urban areas and one-sixth China, and Xubei Luo is economist, East Asia in rural areas. and Pacific Region, Poverty Reduction and Eco- 7. See He and Kuijs (2007), who show that, nomic Management Unit of the World Bank. with a more labor-intensive, service-led pattern 1. The municipalities of Beijing, Tianjin, and of urban growth, urban-rural inequality will Shanghai are not included in this analysis given decline in the decades ahead as more migration their special fiscal situations. allows productivity of those left behind in agri- 2. In almost all countries, per capita social culture to increase. expenditures are higher for richer areas than for poorer and in urban areas than in rural. These References can be explained not only by differences in the Bhattasali, Deepak, Shantong Li, and Will availability of local financing but also by differ- Martin, eds. 2004. China and the WTO. ences in price indexes. Thus a degree of inequal- Washington, DC: World Bank. Reshaping economic geography in China 217 Bourguignon, Franēois. 2008."Growth, ------. 2005."Growth Spillover Effects and Inequality, and Fiscal Policy from a Regional Development Patterns: The Case of Historical Perspective: Are There Lessons for Chinese Provinces." Policy Research Working China?" In Public Finance in China: Reform Paper 3652, World Bank, Washington, DC. and Growth for a Harmonious Society, eds. Luo, Xubei, and Nong Zhu. 2008."Rising Jiwei Lou and Shulin Wang. Washington, DC: Income Inequality in China: A Race to the World Bank. Top." Paper presented at the Economists' Catin, Maurice, Xubei Luo, and Christophe Van Forum 2008, World Bank. Huffel. 2005."Openness, Industrialization, Ma, Jun, and Jan Norregaard. 1998."China's and Geographic Concentration of Activities Fiscal Decentralization." Unpublished mss. in China." Policy Research Working Paper International Monetary Fund, October. 3706, World Bank, Washington, DC. National Bureau of Statistics of China. Vari- Dollar, David, and Bert Hofman. 2008."Inter- ous years. China Statistical Yearbook. Beijing: governmental Fiscal Reforms, Expenditure China Statistical Publishing House. Assignment, and Governance." In Public Finance in China: Reform and Growth for a Ravallion, Martin, and Shaohua Chen. 2004. Harmonious Society, eds. Jiwei Lou and Shulin "China's (Uneven) Progress against Poverty." Wang. Washington, DC: World Bank. Policy Research Working Paper 3408, World Bank, Washington, DC. He, Jianwu, and Louis Kuijs. 2007. Rebalancing China's Economy: Modeling a Policy Package. Shen, Chunli. 2008."Fiscal Inequality in China: China Research Paper 7. Washington, DC: Measurement and Decomposition." In Three World Bank, September. Essays on Public Finance, Ph.D. dissertation, University of Maryland. Huang, Jikun,Yu Liu, Will Martin, and Scott Rozelle. 2007."Agricultural Trade Reform and UNDP (United Nations Development Pro- Rural Prosperity: Lessons from China."Unpub- gramme). 2007. China Human Development lished mss. World Bank, Washington, DC. Report. New York: UNDP. Kuznets, Simon. 1941. National Income and Its Wagstaff, Adam, and Magnus Lindelow. 2008. Composition, 1919­1938. Cambridge, MA: "Health Reform in Rural China: Challenges National Bureau of Economic Research. and Options." In Public Finance in China: Reform and Growth for a Harmonious Society, Li, Shantong, and Zhaoyuan Xu. 2007. The eds. Jiwei Lou and Shulin Wang. Washington, Trend in Regional Income Disparity in China. DC: World Bank. Unpublished mss. Asian Development Bank Institute. World Bank. Forthcoming. China Poverty Assess- ment. Washington, DC: World Bank. Lou, Jiwei. 2008."The Reform of Intergovern- mental Fiscal Relations in China: Lessons ------. 2007a. China Transport Sector Review. Learned." In Public Finance in China: Reform Washington, DC: World Bank. and Growth for a Harmonious Society, eds. ------. 2007b. Connecting to Compete: Trade Jiwei Lou and Shulin Wang. Washington, Logistics in the Global Economy. Washington, DC: World Bank. DC: World Bank. Luo, Xubei. 2001."La mesure de la distance Yusuf, Shahid, and Kaoru Nabeshima. 2008. dans le modčle de gravité: Une application "Optimizing Urban Development." In China au commerce des provinces chinoises avec le Urbanizes, eds. Shahid Yusuf and Tony Saich. Japon." Région et Développement 13: 163­80. Washington, DC: World Bank. ------. 2004."The Role of Infrastructure Zhu, Nong, and Xubei Luo. 2008."The Impact Investment Location on China's Western of Remittances on Rural Poverty and Inequal- Development." Policy Research Working ity in China." Policy Research Working Paper Paper 3345, World Bank, Washington, DC. 4637, World Bank, Washington, DC. The political economy of government policies toward regional inequality in China Yang Yao 14 With its vast territory and rich geographic, rising steadily. Public pressures to reverse economic, and social diversity as well as fast- the trend have been building since the mid- changing institutional setting, China offers a 1990s and have led to two significant recent valuable case study on economic geography. policies,"GoWest"(xibu da kaifa) and"Reviv- In particular, China's regional patterns of ing the Northeast" (zhenxing dongbei), both economic development in the last half cen- aiming to achieve a more balanced model of tury have been closely tied to government regional development. c h a p t e r policies. This paper examines how govern- After reviewing some key facts about ment policies have shaped and reshaped regional disparities, this paper seeks to China's patterns of regional development answer the following questions: Why did and how these policies have been linked with the Chinese government choose the uneven growth as well as other considerations. development model in the 1980s? To what China was an unevenly developed coun- extent can the resulting policies be rational- try when the People's Republic was estab- ized on the grounds of economic geogra- lished in 1949. Industries were concentrated phy? To what extent were they successful? in the Yangtze River delta, the northeast, and What factors--geography or government a limited number of industrial pockets in preferential policies--are more capable of inland provinces. The planning era drasti- explaining the superb record of growth of the cally changed this picture by deliberately set- coastal provinces? What factors, in addition ting up new factories in inland provinces and to geography and the uneven development moving existing factories from the east coast model, exacerbated China's regional dispar- and the northeast to inland provinces. After ities? What are the prospects of the recent reform and the "open-door" policy were government policies aiming to achieve more introduced in the late 1970s, this model of balanced development? Are there any alter- balanced development was abandoned and native ways to reduce regional disparities? replaced with one of uneven development. And, if yes, what are they? These are difficult Exemplifying this model was the opening up questions to answer. This paper does not of the coastal region through the creation of seek to conclude the debate; rather, it tries "special economic zones" and "coastal open to provide a basis for further discussion. cities."Concurrent with this opening up was deep fiscal decentralization, which provided Regional disparities in China local governments with incentives to pur- Many studies have examined China's sue economic growth. To a large extent, this regional disparities. This section gathers key model integrated the Chinese economy into second-hand statistics from the literature the world market and drove China's rapid to provide a broad picture of regional dis- economic growth. Its downside is equally parities since the founding of the People's obvious: China's regional disparity has been Republic of China. The political economy of government policies toward regional inequality in China 219 Convergence and divergence in gence began to stabilize after this policy was historical perspective implemented? This is an important question to which the paper returns later. Démurger and others (2002) provide a Consistent with this pattern of comprehensive study of China's regional income inequality in the period of 1952­98. -convergence,-convergence,orthe(uncon- ditional) convergence of growth rates, had Figure 14.1 summarizes their results of -convergence among Chinese provinces distinct features in the three periods of time. Figure 14.3,also adopted from Démurger and based on the coefficient of variation (CV) of others (2002), presents evidence for 1952­78. per capita GDP. The figure shows two sets of A strong trend of divergence clearly existed results,one with the municipalities of Beijing, in that period; that is, provinces with higher Shanghai,and Tianjin and one without them. Because these three municipalities have much Figure 14.1 Divergence of income among Chinese provinces, 1952­98 higher income than the other provinces,their inclusion results in much higher CVs. Both 0.8 sets of results, however, follow the same pat- 0.7 tern. Three periods of increased divergence are evident, all characterized by decentraliza- 0.6 tion. The first one was in the Great Leap For- Including ward period, the second was in the Cultural 0.5 municipalities Revolution period, and the third started in 0.4 the early 1980s and continues today. There were also three notable periods of declin- 0.3 ing divergence, all of which were associated Excluding 0.2 municipalities with centralization or measures to correct the decentralization that preceded them. The 0.1 first period (1952­56) was China's First Five- 0 Year Plan period, in which the central gov- 1952 1956 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 ernment controlled most of the government Source: Démurger and others (2002). investment. The second period (1960­65) Note: Hainan and Tibet are excluded; Chongqing is added to Sichuan. Per capita GDP is measured at constant featured a great famine and subsequent eco- 1995 prices. The coefficient of variation (the vertical axis) is used to measure the degree of income convergence (-convergence). nomic adjustments and recentralization. The third period (1976­82) followed the Cultural Figure 14.2 Divergence of income among Chinese provinces, 1999­2006 Revolution and witnessed many measures to reverse the decline of the national economy. 0.8 Before the third period of increasing 0.7 divergence began in the early 1980s, there was no clear trend of divergence and, except 0.6 for the Great Leap Forward period, the CVs 0.5 were relatively low when the three big cities are excluded from the calculation. In con- 0.4 trast, there has been a steady trend of diver- 0.3 gence since the early 1980s. Figure 14.2 extends the two series of data 0.2 to the period of 1999­2006. An interest- 0.1 ing finding is that the trend of divergence stopped for both series, and the one with all 0 the provinces even exhibited a weak trend of 1999 2000 2001 2002 2003 2004 2005 2005 convergence, indicating that the three large All provinces Exclude Beijing, Shanghai, Tianjin municipalities were growing more slowly than other provinces in this time period. Sources: National Bureau of Statistics of China, China Statistical Yearbook (2000­06); National Bureau of Statistics of The Go West policy was initiated in 1999. China, China Statistical Abstract (2007). Note: All the mainland provinces are included. Per capita GDP is measured in current prices. The coefficient of Was it a coincidence that income diver- variation (the vertical axis) is used to measure the degree of income convergence (-convergence). 220 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 14.3 Divergence of growth rates among Chinese provinces, 1952­78 initial per capita gross domestic product (GDP) tended to grow faster throughout the 4% planning period. This result was obtained even when the Chinese government delib- 1953­78 3% erately tried to spread industry evenly in the rate, country. There were two significant waves 2% of allocation and reallocation of industry to growth the inland areas. One was in the First Five- GDP Year Plan period, when many of the major 1% projects--notably the 156 projects aided by annual the Soviet Union--were allocated to inland 0% provinces. The other was the 1960s, when capita new investment was concentrated in the per ­1% so-called third front,and many factories were 5.5 6.0 6.5 7.0 7.5 8.0 reallocated from coastal to inland provinces.1 logarithm of 1952 per capita GDP (1995 constant prices) The evidence in figure 14.3 shows that high- Source: Démurger and others (2002). income provinces, mostly those on the east Note: Hainan and Tibet are excluded; Chongqing is added to Sichuan. coast and in the northeast,continued to grow faster than the rest of the country, despite Figure 14.4 Divergence of growth rates among Chinese provinces, 1978­98 these two large-scale government efforts, 14% indicating the strength of economic and geographic forces. 12% 1979­98 Figures 14.4 and 14.5 present evidence for the periods of 1978­98 and 1999­2006, rate, 10% respectively. Although neither period shows 8% growth a sign of convergence, the trend of diver- gence is not as strong as shown in the data GDP 6% for 1952­78. It is even less evident in the more recent period of 1999­2006, a result annual 4% that is consistent with the stabilized and capita 2% even declining income disparities shown for per this period in figure 2. 0% 6.0 6.5 7.0 7.5 8.0 8.5 9.0 If we group the Chinese provinces into logarithm of 1978 per capita GDP (1995 constant prices) the three conventionally defined regions-- coastal, central, and western--we can even Source: Démurger and others (2002). Note: Hainan and Tibet are excluded, and Chongqing is added to Sichuan. see growth rates in the central and western regions catching up with growth rates in the Figure 14.5 Divergence of growth rates among Chinese provinces, 1999­2006 coastal region. Table 14.1 shows the data. In the period of 1978­98, the average growth 16% rate of the central and western regions was 14% 81 and 77 percent, respectively, of that of rate, 12% the coastal region. However, in 1999­2006, 10% the central region caught up with the coastal growth (percent) 8% region, and the western region substantially GDP narrowed its gap to only 7.7 percent of the 6% coastal region's growth rate. In recent years, capita 1999­2006 4% there have been complaints that the central per 2% region has been neglected by the central 0% government and is sinking into a valley in 7.0 7.5 8.0 8.5 9.0 9.5 10.0 10.5 11.0 China's economic landscape. In 2006 these logarithm of 1998 per capita GDP complaints led the government to call for Sources: Author's calculations based on data in National Bureau of Statistics of China, China Statistical Yearbook development of the central region. However, (2000­06); National Bureau of Statistics of China, China Statistical Abstract (2007). Note: All the mainland provinces are included. Per capita GDP is measured in current prices. the data provided in table 14.1 show that the The political economy of government policies toward regional inequality in China 221 central region has not been sinking; instead, Table 14.1 Convergence of growth rates among the three regions in China, 1978­2006 it has accelerated its catch-up with the coastal Year and indicator Coastal Central Western region and has widened its distance from the 1978­98 western region (the growth rate of the west- Average growth rate 9.4 7.6 7.2 ern region was 95 percent of the growth rate Central and western as a percent of coastal n.a. 81.2 77.0 of the central region in 1978­98, but fell to 1999­2006 91 percent in 1999­2006). Average growth rate 10.1 10.3 9.3 Central and western as a percent of coastal n.a. 101.8 92.3 This analysis is simple but shows that Chi- Sources: Data for 1978­98 come from Démurger and others (2002); data for 1999­2006 come from National Bureau na's regional disparities have passed through of Statistics of China, China Statistical Yearbook (2000­06); National Bureau of Statistics of China, China Statistical a dynamic and complicated process.The bot- Abstract (2007). n.a. Not applicable. tom line is that the divergence of growth rates was not as serious in the reform era as in the planning era. The planning era had smaller not a new phenomenon; it was already 2.78 degrees of income disparities only because times when China embarked on its reform the level of income was low to begin with. and open-door policy in 1978. Except for a The continuing divergence of income growth brief decline to 1.8 times in the early half rates had already begun to set the stage for of the 1980s due to institutional reforms the divergence of the level of income, which implemented in the countryside and another finally showed up in the reform period. The period of smaller declines in the early 1990s diverging growth rates in the planning era due to higher agricultural prices, the gap has were not likely to be a consequence of gov- been growing for the past 30 some years. ernment policies, though. Rather, the estab- The most significant regional regularity lishment of the People's Republic provided about the urban-rural divide is that higher- a relatively stable economic environment income provinces have lower urban-rural (with little political turmoil), which enabled income gaps. Figure 14.6 shows the relation- the advanced regions to unleash their growth ship between per capita GDP and the urban- potential (Perkins 2005). Economic reform rural divide for Chinese provinces in 1999 and the open-door policy, viewed from and 2006. Beijing, Shanghai, and Tianjin are a historical perspective, are not the likely excluded because their urbanization rates causes of the growing regional disparities in are very high. There clearly exists an inverse the last quarter century. The signs of both relationship between per capita GDP and the -convergence and (unconditional) the urban-rural income gap. This should -convergence, especially among China's be obvious even to a casual observer travel- three regions since 1999, are encouraging. ing on Chinese highways. In coastal regions, The question is the extent to which positive modern houses dot the densely populated signs can be attributed to government poli- countryside along any highway; in western cies, especially the Go West campaign, which regions, a 20-minute drive out of any major aims to reduce regional disparities. It is also city encounters sheer poverty. possible that the Chinese economy was not Table 14.2 provides a sharp contrast ready to converge until the late 1990s.Empir- between the three richest provinces (Guang- ical research finds that there exists a hurdle dong, Jiangsu, and Zhejiang) and the three for convergence to happen both internation- poorest provinces (Gansu, Guizhou, and ally (Durlauf and Johnson 1995) and domes- Yunnan) for 2006. Per capita GDP of the tically (Peng, Wang, and Wu 2007). Perhaps three richest provinces was 3.62 times that China was only able to overcome that hurdle of the three poorest provinces. However, per in the late 1990s. capita urban disposable income of the three richest provinces was only 1.72 times that Regional or urban-rural divide? of the three poorest. In contrast, per capita The urban-rural divide has been much more rural net income of the three richest prov- serious than the regional divide in China. In inces was 2.86 times that of the three poor- 2006 per capita urban disposable income est. The urban-rural income gap was 2.65 was 3.14 times per capita rural net income, times for the three richest provinces, but the highest in the world. This large gap is 4.41 times for the three poorest provinces. 222 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 14.6 Per capita GDP and urban-rural divide in China, 1999 and 2006 A. 1999 B. 2006 5.0 5.0 4.5 4.5 4.0 4.0 income/ 3.5 income/ 3.5 3.0 3.0 income income 2.5 2.5 net 2.0 net 2.0 disposable 1.5 disposable 1.5 rural rural 1.0 1.0 Urban 0.5 Urban 0.5 0 0 7.5 8.0 8.5 9.0 9.5 8.5 9.0 9.5 10.0 10.5 Logarithm of 1999 per-capita GDP Logarithm of 2006 per-capita GDP Sources: National Bureau of Statistics of China, China Statistical Yearbook (2000­06); National Bureau of Statistics of China, China Statistical Abstract (2007). Note: Beijing, Shanghai, and Tianjin are excluded; figure of 1999 does not include Tibet. Per capita GDP is in current prices. Table 14.2 Comparison of the three richest and the three poorest provinces in China, 2006 1.68 times, a drop of 6.8 percent, and the gap Grouping Urban Rural Urban-rural between the coastal and the western region would decline from the actual 2.22 times to Richest three provinces 16,121.67 6,076.00 2.65 Poorest three provinces 9,369.33 2,123.33 4.41 1.91 times, a drop of 13.8 percent. Ratio of richest to poorest 1.72 2.86 0.60 A more precise estimation is provided by Source: National Bureau of Statistics of China, China Statistical Abstract (2007). Gajwani, Kanbur, and Zhang (2006), who Note: Urban income is per capita disposable income; rural income is per capita net income. Both are in current prices. show that the urban-rural divide has played The three riches provinces are Guangdong, Jiangsu, and Zhejiang, and the three poorest provinces are Gansu, Guizhou, and Yunnan. a much larger role than the regional divide in determining China's interprovincial Table 14.3 Urban-rural and regional divides in China, 2005 inequality. Using data from table 14.1, fig- Indicator Coastal Central Western ure 14.7 shows the change in interprovincial inequality measured by the Gini coefficient Actual Average income (yuan) 9,907.92 5,504.70 4,463.77 and the general entropy (GE) index for the Central and western as a percent of coastal n.a. 1.80 2.22 period 1952­2004. The two series parallel Urban income (yuan) 12,884.09 9,207.20 8,597.80 each other. Following a decline between the Rural income (yuan) 5,123.36 2,971.80 2,278.40 mid-1970s and early 1980s, both the Gini Ratio of urban to rural income 2.56 3.11 3.84 and the GE increased dramatically dur- Simulated Average income (yuan) 9,907.92 5,907.58 5,177.07 ing the last quarter century. However, the Central and western as a percent of coastal n.a. 1.68 1.91 urban-rural divide by and large has been Source: National Bureau of Statistics of China, China Statistical Yearbook (2006). the dominant factor in determining inter- n.a. Not applicable. provincial inequality (see figure 14.8). The Note: Average income is the composite income of urban per capita disposable income and rural per capita net income using the urbanization ratio as the weight. contribution of the coastal-inland divide was minimal before the reform started but Table 14.3 presents a rough estimation has increased substantially since then, except for the share of the urban-rural divide in the for a brief decline in the early 1990s. In the regional divide using data for 2005. It pres- meantime, the contribution of the urban- ents two sets of data: one actual and the other rural divide decreased. However, its contri- simulated. The simulation assumes that the bution remained at 72 percent in 2004, while central and western regions had the coastal the contribution of the coastal-inland divide region's urban-rural income ratio and recal- was only 11 percent. culates rural income leaving urban income In summary, the larger urban-rural unchanged. Using the ratio of urban popula- divide in inland provinces is an important, if tion as the weight, this yields the simulated not a decisive, factor in the regional divide. average income for the central and western This is not to deny the significance of the regions.Table 14.3 shows that the income gap regional divide; rather, the real gap exists between the coastal and the central region between the coast and the interior--that is, would decline from the actual 1.8 times to between the countryside of the two regions. The political economy of government policies toward regional inequality in China 223 Government policies to raise rural income Figure 14.7 Interprovincial inequality in China, 1952­2004 in inland provinces are likely to have a large 40 impact on lowering regional inequality. 35 30 The path to the uneven 25 development model 20 The model of uneven development was 15 adopted after 1978 in a conscious pursuit of 10 economic growth. This started with estab- 5 lishment of the "growth consensus," which was based on the painful lessons learned in 0 the Chinese encounters with the Western 1952 1955 1958 1961 1964 1967 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 year powers, the socialist world's lagging behind GE Gini the capitalist world,and the destructive forces of the political movements in China's his- Source: Gajwani, Kanbur, and Zhang (2006: table 1). tory. The uneven development model was a deliberate choice based on market principles, Figure 14.8 Contribution of urban-rural divide and regional divide to interprovincial inequality, notably those implied by economic geogra- 1952­2003 phy.Central to the model is the priority given 100 14 to the coast, characterized by several waves of 90 preferential policies in the 1980s.This section 12 80 reviews the formation of and rationale for the 70 10 uneven development model and the various 60 8 waves of preferential treatment given to the 50 coast. The emphasis is on the alignment of percent 6 40 percent government policies with economic geogra- 30 4 phy and the tradeoff between the opening of 20 2 the coast and the need to generate stable gov- 10 ernment revenues. 0 0 8 4 1952195419561958196019621964196619681970197219741976197 1980198219841986198819901992199 19961998200020022004 year The growth consensus Urban-rural (left axis) Coastal-inland (right axis) Since the Opium War, several generations of Chinese leaders have sought to build a Source: Gajwani, Kanbur, and Zhang (2006: table 1). strong China. The painful lesson learned from China's encounters with the West- firmly established and maintained through- ern powers in the 1800s was that, without out these years. economic power, China would be vulner- This consensus has its roots not only in able to pressures. Unfortunately, economic history but also in contemporary thought. growth was interrupted by civil wars and Socialism centered on state ownership was the Japanese invasion. The establishment of once thought to be the key to higher rates the People's Republic gave China a chance of economic growth, but the competition to concentrate on economic growth, but between the socialist and the capitalist the dream was shattered again and again by worlds provided decisive evidence that this political movements, one wave higher than version of socialism could not outperform the other. Pragmatic leadership was restored capitalism (Nee and Lian 1994). The fast with the ascension of Deng Xiaoping in the growth of the four East Asian Tigers was late 1970s, giving China a chance to pursue a particularly painful fact for the Chinese its century-long dream of economic pros- leadership to swallow. Every piece of evidence perity. Fortunately, China did not miss this showed that China had to introduce some ele- chance and was able to maintain an aver- ments of capitalism if it hoped to catch up age growth rate of 9.7 percent for the next economically with the developed nations. 30 some years. The growth consensus was However, the introduction of capitalism 224 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA threatened the orthodox ideology of the the central region and then to the western Communist Party.To win support within the region (Wang and Hu 1999). This model party, Deng Xiaoping initiated a nationwide was formally adopted by the government's debate at the end of 1978. The debate was Seventh Five-Year Plan, which started in not directed to the question of whether to 1986 (State Council 1986). In the late 1980s, have more markets; rather it was framed as a more radical theory named "big inputs, a philosophical discussion about the sources big exports" (dajin dachu) was put forward, of truth. This proved to be a wise political which called for China's coastal region to strategy, as it posed the question as a classic participate in the world market via process- Marxist issue, preempting the objection of ing trade. Those theories provided a strong party conservatives to the debate. The con- case for the government to concentrate its clusion of the debate was that there could be investment in the coastal region. When put only one source of truth, which was practice, into practice, they were reinforced by fis- opening a door for experiments and change. cal decentralization, which gave provinces These changes would ultimately transform unprecedented fiscal autonomy. The result, China. The leadership at the time was con- as we all know, was rapid economic growth vinced that embracing the market was the coupled with large regional disparities. But only way for China to grow economically. before discussing the consequences, let us To preempt the conservatives' rebuff, Deng step back and discuss the economic and Xiaoping set the tone with the following political rationale behind the uneven devel- simple words: "Do not debate." opment model. Today, however, many in China are con- The economic rationale has a lot to do cerned that the growth consensus has been with the geographic advantages of the coastal overdone. Indeed, inequality, social justice, region: access to international markets, his- and environmental problems are mount- torical traits, cultural proxy to overseas Chi- ing issues in today's China, and the growth nese communities, and the concentration of model deserves a reexamination. However, large cities. it is worth keeping in mind that the growth In terms of the access to international consensus propelled the country on its markets, China's coastal region has the remarkable path. advantage not only of transportation, but also of close proximity to Hong Kong, a Economic and political rationales dynamic and free-trade city second only to for the uneven development model Tokyo in East Asia. For a long time, Hong The uneven development model was a nat- Kong has been an important window for ural result of China's decision to enter the Chinese exports. Guangdong is the largest market economy. The first steps entailed exporter in mainland China, accounting for creating special economic zones (SEZs) in 30 percent of China's total exports, and 60 Shantao, Shenzhen, Xiamen, and Zhuhai, percent of its exports are routed through all located on the southern coast. The SEZs Hong Kong (Yang 2006). This means that were designated "to experiment with the 18 percent of the mainland's exports are development of an outward-looking, mar- routed through Hong Kong. In accordance, ket-oriented economic system and to serve 60­70 percent of Hong Kong's GDP is tied the country as a `window' and a `base' along to exports from the mainland (Yao and oth- these lines"(Ge 1999: 49).That is,the uneven ers 2006). Since one of the major aims of development model sought to experiment the uneven development model is to experi- with a market-based system from the very ment with a market-oriented system, learn- start. Reinforcing this idea was the gradient ing from Hong Kong became a convenient theory (tidu lilun) put forward in the early route toward that goal. The influence of 1980s, which distinguished China's three Hong Kong was the most evident in the ini- major regions (coastal, central, and western) tial stage of Shenzhen's astonishing transfor- as three ladders of economic growth.Accord- mation from a fishing village to a major city ing to this theory, growth should start in the with a large population of several million. coastal region and gradually expand first to Deng Xiaoping's idea of creating an SEZ in The political economy of government policies toward regional inequality in China 225 Shenzhen was intended to open a window The three large municipalities--Beijing, for Hong Kong's influence to slip gradu- Shanghai, and Tianjin--are all located on ally into the mainland. For the first quarter the eastern coast; Guangzhou, the fourth- century of existence, Shenzhen lived up to largest city in China today, is located on Deng's expectations and was a champion for the southern coast; and Shenyang, the fifth- economic reform, new ideas, and new forms largest city, is close to the sea in the northeast. of governance, although its advantages have In fact, the Chinese economy is heavily con- diminished in recent years. centrated in three regions centered on those History matters a lot in the divergence of big cities: the Pearl River delta, the Yangtze the Chinese provinces. The Chinese civiliza- River delta, and the Bohai Bay area formed tion started in the middle reach of the Yel- by Beijing, Shenyang, Tangshan, and Tianjin. low River in the Loess plateau and gradually The political rationale was the need to moved east to the lower reach of the river achieve a balance between reform experi- in the Song dynasty. The invasion of the ments and a stable flow of tax revenue. northern tribes forced the Song dynasty to Reform experiments were politically risky, move its capital to Hangzhou. The move as failure risked a backlash from the conser- was decisive, as it enabled China to move vatives. To make sure that the experiments its economic center from the north to the were successful, the central government more fertile south. The encounters with implemented a set of fiscal policies. On the Western powers in the 1800s added the one hand, it gave experimenting prov- another advantage to the southern and east- inces preferential policies, which granted ern coasts: access to international markets them more flexibility and helped them to through the treaty ports connecting China attract FDI and other investments. On the to the outside world.2 The most significant other hand, through fiscal contracting, it development was Shanghai's emergence as provided them with strong fiscal incen- the most dynamic city in the Far East. Since tives and gave the central government tight the early 1900s, Shanghai has been China's control over the nonexperimenting prov- economic powerhouse. inces. The next section is devoted to a dis- The cultural ties with overseas Chinese cussion of the preferential policies. This communities have been a valuable asset for section provides a review of the fiscal con- development of the coast. The early wave of tracting system implemented in the 1980s. overseas Chinese came from a few regions The fiscal relationship between the central in Fujian and Guangdong provinces. They and local governments was not settled until brought back a large amount of investment the 1994 tax reform. Several rounds of cen- in the early reform era. Indeed, until the late tralization and decentralization occurred 1990s, half of China's foreign direct invest- during the planning period. The fiscal con- ment (FDI) came from overseas Chinese. tracting system was inspired by the model By the late 1990s, several other sources of implemented in the decentralization period investment became significant: the invest- of 1959­67 (Wei 2000). Contracting was ment brought back by new emigrants from inspired by the success of the rural reform, Zhejiang province, the investment brought which was famous for its village-household by Singaporean businesses to Jiangsu prov- contracting of land. The central government ince, and the investment brought by Tai- negotiated different contracts with individ- wanese businesses to areas around Shang- ual provinces, and no two contracts were hai. Jiangsu's becoming the second-largest identical. Generally, there were five types of exporter in China had a lot to do with this contracts (Wei 2000): new wave of FDI. Finally, the concentration of large cities · The first type of contract was offered to enabled the coastal region to achieve econo- the two provinces on the frontier of the mies of agglomeration. Empirical research open-door policy, namely, Fujian and finds evidence in China to support the claims Guangdong. They had to hand in a fixed of the new economic geography (for exam- amount of revenue to the central gov- ple, Chen and Wang 2007; Lu and Tao 2007). ernment and were allowed to keep the 226 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA rest for their own budget. Later, Fujian this was stabilized at 46.5 percent in 1988. received a fixed amount of transfers from Jiangsu's retention rates were more stable, the central government each year. starting with 39 percent in 1980 and ending · The second type of contract was offered with 41 percent in 1988 (Wei 2000). to Jiangsu, which shared revenue with The case of Shanghai is especially inter- the central government according to a esting.In 1984 Shanghai contributed 5.6 per- formula that was fixed for four years. cent to the national GDP and 9.6 percent to · The third type of contract was offered to the gross value of national industrial output, 15 provinces, which received a fixed base while its population was barely more than payment and then shared a fixed per- 1 percent of the national total (Ge 1999). centage of revenue growth with the cen- Because of its significance in the national tral government. economy, Shanghai had the worst contract with the central government. Its retention · The fourth, and most favorable, type rate was merely 8.6 percent in 1980 and was of contract was for eight minority and only raised to 26 percent in 1985. Starting in border provinces, which received a fixed 1988, Shanghai began to pay a fixed amount amount of transfers from the central of Y 10.5 billion to the central government government. each year. For comparison, Guangdong's · The fifth type of contract was offered to highest fixed payment to the central gov- the three large municipalities, Beijing, ernment was Y 14.1 billion. Although it was Shanghai, and Tianjin. These three cit- made one of the 14 coastal open cities in ies retained the smallest proportion of 1984, Shanghai still got the most unfavor- revenue according to a formula that was able deal with the central government. The adjusted every year. contrast between Shanghai and Guangdong These five types of contracts can be shows clearly that the central government regrouped into two broader types, one with was seeking to pursue reform while ensur- fixed payments to or subsidies from the cen- ing a continuous flow of tax revenues. tral government and one with a marginal Roland and Verdier (2003) believe that sharing mechanism. Almost all of the west- one of the keys to China's successful tran- ern provinces had fixed-payment contracts, sition is the adoption of a "reform at the with each receiving a fixed amount of sub- margin" approach, by which the state sec- sidies from the central government. Fujian tor was left intact to provide revenues to the and Guangdong were the only two coastal government, while the private sector was provinces that had this type of contract. All allowed to expand. The same logic applies of the other provinces or cities had sharing to China's uneven regional development contracts. According to the tenancy theory, strategy, which allowed a few frontrunner fixed-payment contracts provide relatively provinces to experiment and grow quickly, strong incentives to the contractor, while while holding back other provinces for the sharing contracts provide relatively weak sake of tax revenues. incentives. The political goal here was for Preferential government policies the two leading reform provinces--Fujian and Guangdong--to have a strong incentive and economic geography to move forward, for the western provinces The uneven development model was car- to balance their budgets (most of them had ried out by a series of preferential policies deficits), and for the rest of the provinces to toward the coast. This started as a shift of contribute to the central budget in a pro- central government investment away from gressive way. inland provinces and toward coastal prov- The three large municipalities and Jiangsu inces through direct allocations and fiscal were taxed the heaviest. Beijing's retention arrangements. This was followed by creation rate was only 28.1 percent in 1980, although of the SEZs and coastal open cities, which this was raised to 50 percent by 1988, where institutionalized a set of preferential poli- it remained until the 1994 tax reform. Tian- cies toward the coastal region. Finally, tax jin started with 30.6 percent in 1980, and breaks for FDI benefited the coast, which The political economy of government policies toward regional inequality in China 227 was the largest recipient of FDI. However, ered almost the entire share it had lost in the it is not clear whether government policies previous period. As a result, the distribution or geography played a more important role resembled that of the period of 1979­91. in encouraging growth in the coastal region. Concurrent with the decline in equality In the 1990s, preferential government poli- in central government investment in the cies were extended to inland provinces. This 1980s was the decline in fiscal sharing among gives us a chance to disentangle government the three regions. This is clearly shown in policies from geography in determining table 14.5, which presents the trends in fis- regional variations in economic growth. cal transfers between 1953 and 2005. The coastal and central regions' net contribution Investment shifts and reduced to the central government budget increased interregional fiscal sharing until 1976, but declined thereafter, until the The planning era witnessed the central gov- 1993 tax reform reversed the trend. The ernment's intentional shift of investment to reform significantly increased fiscal sharing inland provinces. In the reform period, there among the three regions. By the period of was a shift back to the coast and a decline in 1999­2005, the contribution of the coastal interregional fiscal transfers. region had recovered to close to the level of Changes in investment since the 1980s the period of 1976­80. Notably, the net gain are shown in table 14.4, which compares of the central region became positive and four periods, 1953­78, 1979­91, 1993­98, reached almost the same level as that of the and 1999­2005, by the share of the coastal, western region. central, and western regions in the cen- The coastal region's dominance since tral government's investment. While it the 1994 tax reform was more subtle. Fis- was already the largest recipient of central cal decentralization in the 1980s was so deep government investment in the period of that the central government's share of tax 1953­78, the coastal region became even revenues declined sharply from 25 percent more favorable in the initial reform period in 1981 to 14 percent in 1994 (Wang and of 1979­91, receiving more than half of total Hu 1999: fig. 6.7). The reform was intended central government investment, an increase to reverse this decline by strengthening the of 14 percentage points over the earlier central government's position. In the mean- period. In contrast, the shares of both the time, the government promised that poorer central and western regions declined signifi- provinces would get more transfers through cantly, the central by 5.74 percentage points a predetermined formula based on provincial and the western by 8.26 percentage points. income and government revenues. While In terms of absolute value, the contrast is the first objective was obtained (currently, even starker. The investment received by the coastal region increased more than 200 Table 14.4 Central government capital investment, 1953­2005 percent, but that of the central and western Indicator Coastal Central Western regions increased only 89 and 55 percent, 1953­78 respectively. In 1979­91, the coastal region's Investment (billion yuan) 223.39 192.21 149.64 share was 1.9 and 2.9 times that of the cen- Percent of total 39.52 34.01 26.47 tral and western regions, respectively. The 1979­91 coastal region clearly dominated central Investment (billion yuan) 686.28 362.42 233.63 government investment in the 1980s. Percent of total 53.52 28.26 18.22 Change over last period 14.00 ­5.74 -8.26 The share of the central and western 1993­98 regions increased slightly in the period Investment (billion yuan) 1,003.6 690.2 464.0 of 1993­98, while the share of the coastal Percent of total 46.51 31.99 21.50 region declined. However, the trends Change over last period ­7.01 3.72 3.28 were reversed in the subsequent period of 1999­2005 1999­2005. The share of the central region Investment (billion yuan) 4,696.7 2,255.1 1,920.4 Percent of total 52.94 25.42 21.65 declined substantially, while the share of the Change over last period 6.43 -6.57 0.13 western region remained almost unchanged. Sources: Data for 1953­91 are from Li (1995: 89). Quoted in Wang and Hu (1999: 176). Data for 1993­2005 are from In the meantime, the coastal region recov- National Bureau of Statistics of China, China Statistical Yearbook (1994­2006). There are no data for 1992. 228 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 14.5 Net transfers received from the central government, 1953­2005a province and became the fifth SEZ. In Time period Coastal Central Western 1992 Shanghai's Pudong became the sixth SEZ. 1953­57 ­9.4 -1.4 7.6 1958­62 -15.6 -1.6 8.6 · Coastal open cities. In 1984, 14 coastal cit- 1963­65 -15.7 -4.2 9.1 ies were designated as coastal open cities, 1966­70 -20.2 -6.2 20.9 whose purpose was mainly to attract FDI 1971­75 -21.2 -4.1 15.2 through the establishment of economic 1976­80 -20.5 -2.9 10.8 1981­85 -12.8 -1.2 12.2 and technological development zones 1986­92 -4.0 -1.0 11.1 (ETDZs). 1994­98 -10.9 1.5 7.1 · Coastal economic open zones. Between 1999­2005 -17.8 10.1 11.4 1985 and 1988, five coastal economic Sources: Data before 1994 come from Wei (2000: table 4.9). Other data come from National Bureau of Statistics of China, China Statistical Yearbook (1995­2006); National Bureau of Statistics of China, Finance Yearbook of China (1995­2006). open zones were created along the coast, a. Percentage of net transfers received from the central government in total central government budget. Net transfers covering all the major economic centers equal transfers from the central government (including tax returns) minus taxes handed over from local governments to the center. there. · Customs-free zones. These were created between 1990 and 1993 along the coast. central government tax revenues are 60 In the 1990s economic opening spread to percent of the national total), the second other regions through the creation of new objective largely fell through. Instead, cen- open economic zones. These included major tral government transfers have been relying cities along the Yangtze River, border eco- largely on project financing. Provinces that nomic cooperation zones, capital cities of need money from the central government inland provinces and autonomous regions, must submit a proposal to the National ETDZs outside the coastal open cities, and Development and Reform Commission bonded areas (see Démurger and others (NDRC) and the Ministry of Finance, 2002). By the mid-1990s, opening finally among other central government agencies. spread to almost every corner of the coun- Once a project is approved, NDRC and the try, and"zone fever"led to the establishment Ministry of Finance require the province to of numerous ETDZs and high-tech zones match the central government's contribu- throughout the country. However, the num- tion. This matching rule creates a field tilted ber of ETDZs approved by the central gov- in favor of coastal provinces, which are ernment was limited. Even by the end of the rich, and against inland provinces, which 1990s, the distribution of special zones was are much poorer. As a result, more central tilted toward the coastal region (Démurger government revenues flow back to coastal and others 2002). provinces than to inland provinces. These special zones received substantial Government preferential policies preferential policy treatment,as summarized in table 14.6. These policies involved three Starting in 1980, China began to offer pref- types of preferential treatment: tax breaks, erential policies to a few cities through vari- more freedom to approve FDI, and a larger ous initiatives. In the 1980s, most of these retention rate of foreign exchange earnings.3 initiatives only covered cities in the coastal They provided substantial benefits to firms region. After Deng Xiaoping's visit to the operating in the zones. This raises the ques- south in 1992, inland provinces began to tion of whether the rapid development of receive the benefits of opening up through the coastal region has been a result of eco- the establishment of economic and techno- nomic forces including economic geogra- logical development zones. The initiatives phy or a result of preferential government undertaken since 1980 include the follow- policies that arose because the zones were ing (Wang and Hu 1999): concentrated in the coastal region (Wang · Special economic zones. Four SEZs were and Hu 1999). The next section turns to created in Shantou, Shenzhen, Xiamen, this question and tries to evaluate the roles and Zhuhai in 1980. In 1988 Hainan played by government policies and geogra- Island was separated from Guangdong phy in different periods since 1978. The political economy of government policies toward regional inequality in China 229 Table 14.6 Preferential policies offered to different zones Economic and technological Coastal economic Customs-free Policy Special economic zones Central open cities development zones open zones zones National income taxa 15 percent, exempted in first 24 percent 15 percent 24 percent three years Local income tax Reduced or exempted Reduced or exempted Reduced or exempted Reduced or exempted ICT on exportsb Exempted Exempted Exempted Exempted Exempted ICT and custom duties on imports Exempted ICT and custom duties on FDI's Exempted Exempted Exempted Exempted Exempted imported equipment Right to approve FDI Much greater Greater Much greater Greater Much greater Right to retain foreign exchange earningsc 100 percent 50 percent Source: Wang and Hu (1999: table 7.2). a. Standard rates were 30 percent for joint ventures and 20­40 percent for foreign-owned companies. b. ICT stands for industrial and commercial tax. It was replaced by value added tax after the 1994 tax reform. c. The standard ratio was 25 percent. Policy versus geography For geography, Démurger and her coau- Démurger and others (2002) provide thors create a variable, Pop100cr, measur- a comprehensive study of the factors ing a province's ease in getting access to determining China's uneven economic the sea. To be precise, it is "the proportion development for the period of 1978­98. of the population distribution of a prov- In particular, they examine the relative ince in 1994 within 100 km [kilometers] of importance of preferential policies and the coastline or ocean-navigable river[s], geography in determining variations in excluding the coastline above the winter growth across provinces. For that, they extent of sea ice and the rivers that flow to construct two variables, one measuring this coastline" (Démurger and others 2002: policy preferences and the other measuring 21). The correlation coefficient between geographic advantages. For the first, they Pop100cr and the average value of Policy in create a preferential policy index, called 1978­98 is 0.54. Therefore, the two variables Policy, based on the number of designated have enough variations for us to disentangle open zones in a province and the extent of the effects of geography and policy. preferential treatment they get. They then Using Pop100cr and the average scores of assign different weights to different zones Policy for different periods and controlling according to the following rule: initial per capita GDP, Démurger and her coauthors estimate separate growth equa- Weight=3: SEZ or Shanghai Pudong tions for three periods: 1979­84, 1985­91, New Area; and 1992­98. The results are presented in the first three rows of table 14.7. The last Weight=2: economic and technologi- row of the table presents results for the cal development zone or border eco- period 1999­2006. The Pop100cr variable is nomic cooperation zone; the same as in the first three regressions; Pol- Weight=1: coastal open city, coastal eco- icy uses its values in 1998. Several relevant nomic open zone, open coastal belts, results emerge from the table. major city, bonded area, or capital First,thereisnoevidencefor-convergence city of an inland province or autono- among Chinese provinces.The coefficient for mous region; and initial GDP is not statistically significant in Weight=0: no open zone. any of the four periods. Second, the role played by government If a province (such as Guangdong) has policy has changed throughout the years. one or more SEZs, it gets a score of 3 for its In the period 1985­91, the coefficient of preferential policy index. That is, the value Policy was significantly higher than in 1979­ of the variable Policy does not increase when 84. However, it became insignificant in a province has more than one zone. 1992­98 and 1999­2006.That is,preferential 230 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 14.7 Policy versus geography: Regression results preferential government policies to inland Period Initial GDP Pop100cr Policy R2 provinces in the 1990s. In 1991, the year before Deng Xiaoping's visit to the south, the 1979­84 -1.23 1.51 0.56 0.28 (1.47) (2.31) (2.88) average score of Policy for the coastal region 1985­91 -0.34 -0.64 1.19 0.38 was 2.18, but the average score for the other (0.29) (0.73) (2.67) two regions was 0.10. By 1998, however, the 1992­98 -0.60 4.27 0.99 0.71 score had a small increase for the coastal (0.57) (7.14) (1.35) region, reaching 2.36, but a large increase for 1999­2006 0.52 0.51 -0.58 0.06 the other two regions, reaching 1.50. (0.78) (0.66) (1.10) Source: The results for the three periods between 1979 and 1998 are from Démurger and others (2002: table 12); the results for 1999­2006 are calculated by the author based on data from National Bureau of Statistics of China, China Statistical Yearbook (2000­06); National Bureau of Statistics of China, China Statistical Abstract (2007). Recent government initiatives to Note: The dependent variable is average growth rate (percent) of per capita GDP in each period. Regressions for 1979­98 are based on data for 30 provinces (Chongqing is excluded); regression for 1999­2006 is based on data for all reduce regional disparities 31 provinces. Initial GDP is the logarithm of per capita GDP of the year immediately before each period started. Policy takes the average values of each period in the three regressions for 1979­98 and takes the 1998 values in the regres- Beginning in the late 1990s, regional sion for 1999­2006. A constant is added in each regression. Numbers in parentheses are t-statistics for the estimates. inequality caught the attention of the cen- tral government. This was partly related government policies were important in the to the 1994 tax reform. One major conse- 1980s, but not since the 1990s. quence of the reform was to increase the Third, the geography variable Pop100cr is central government's share of government statistically significant in the two expansion- revenues. Less-developed provinces felt the ary periods of 1979­84 and 1992­98, but not pressure more than developed provinces in the less expansionary periods of 1985­91 because their budgets were smaller. The and 1999­2006.4 The coastal region appar- amount of formula-based revenue trans- ently has experienced more volatility than fers did not increase to a level sufficient to the other two regions: in expansionary peri- counterbalance the inequality created. In the ods, it moved ahead of the other two regions; meantime, project-based revenue transfers in recessionary periods, it contracted more worked against less-developed provinces. than the other two regions. Less-developed provinces felt that they had Fourth, the predictive power of the been abandoned by the central government. growth regression declines significantly for The Go West policy, which was adopted in the period 1999­2006, with its R2 decreas- 1999, was an effort to respond to this senti- ing to only 0.06. Neither geography nor ment.At the same time, the northeast, one of government policy played a significant China's old powerhouses, also experienced a role in this period. This result is consistent sharp decline because of economic restruc- with the data presented in table 14.1, which turing. Industries in the northeast were over- show that the growth rates of the central whelmingly state owned, resource based, and and western regions were catching up with lacking new investment and research and the growth rate of the coastal regions in the development (R&D). In the new era of pri- period of 1999­2006. Some of this can be vate economy and globalization, these indus- attributed to convergence, and some can be tries lost their competitiveness and began an attributed to the diminishing role of gov- inevitable decline. A once glorious region ernment policies, but neither is significant became China's backwater of stagnation and enough to dominate. a source of social unrest. To revitalize the In summary, we have the following two industrial bases in the northeast, the cen- major conclusions. First, geography played tral government initiated the Reviving the a diverging role only in periods when the Northeast program in 2003. economy was on an expansionary track. This The central government set up a special is a piece of evidence for the coastal region's office for each of the two programs, com- deeper integration into the world economy. mitted considerable amounts of financial Second, preferential government policies resources, and offered them favorable poli- played a significantly diverging role in the cies. The positive responses from the cen- 1980s but have since lost strength. This has tral government, however, encouraged the been caused partly by the dispersion of central provinces to request preferential The political economy of government policies toward regional inequality in China 231 treatment. A theory of "the falling central" 28 million hectares of land's natural ecosys- was developed, which believes that, while the tem were restored. New trees were planted east is forging ahead as a result of geographic on some 7.65 million hectares of wasteland. and policy advantages and the west and the Under the Land for Grass program, 19.33 northeast are getting preferential policies million hectares of grassland were restored. and money from the central government, The Land for Forest and Land for Grass the central provinces (Anhui, Henan, Hubei, programs displaced a large number of Hunan, Jiangxi, and Shanxi) do not get any- farmers from agriculture. The central gov- thing from the central government and thus ernment started various complementary are falling behind. The central provinces programs to help displaced farmers find have used this argument to request support alternative employment. For the transitional from the central government. In April 2006 period, displaced farmers were eligible for a the central government, in its tenth directive subsidy of Y 20 for each mu (one-fifteenth of 2006, pledged to give the central prov- of a hectare) of land converted back to for- inces more support and to set up an office est. In 2007, the subsidy was raised to Y 105 in the NDRC to lead the efforts. in the south and Y 70 in the north (State This section reviews these three initia- Council 2007). tives and comments on their merits and The Go West program also allocates shortcomings. In particular, it seeks to find money to support social development in the political economy logic behind the cen- the western region. Investment in educa- tral government's regional policies. tion in this region amounted to Y 15 billion, and investment in public health services Go West reached Y 8 billion in the six-year period of The main purpose of the Go West program 2000­05.5 is to support infrastructural construction In addition to financial resources, the and environmental protection. Between central government gives the western region 2000 and 2005, 70 main construction proj- a variety of preferential policies to attract ects were started, and the total amount of FDI and domestic and foreign companies to investment reached Y 1 trillion. More than construct infrastructure and environmen- one-third of the funds raised by long-term tally friendly businesses. They include:6 government bonds for construction were · For domestic and foreign companies directed toward the western region during engaged in industries promoted by the this period, and the percentage exceeded 40 central government, the rate of corpo- percent from 2002 to 2005. About 220,000 rate income tax is 15 percent for a desig- kilometers of new roads were built in the nated period of time. region in the six years of 2000­05, among · In minority autonomous regions, cor- which 6,853 kilometers were highways. In porate income tax can be reduced or addition, 5,000 kilometers of railways were exempted after getting approval from the built, and 10 airports were under construc- provincial government. tion. Among these projects, some, such as the Qinghai-Tibet Railway, West-East Power · For new companies in transportation, Transmission Project, and West-East Natural power supply, water conservation, postal Gas Pipeline Project, have become national services,and radio and television services, landmarks. corporate income tax can be exempted or Environmental conservation is an impor- reduced for 3 years. tant part of the Go West program. The · In old revolutionary bases, minority "Land for Forest" and "Land for Grass" pro- regions, remote border areas, and pov- grams were introduced to restore the eco- erty regions designated by the central system. Under the Land for Forest program, government, corporate income tax can 5.26 million hectares of cultivated land be exempted or reduced for 3 years, sub- were converted to forest. In addition, about ject to government approval. 16 million hectares of land suffering from · For agricultural cash products covered by water and soil losses were controlled, and the Land for Forest and Land for Grass 232 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA programs, the agricultural cash crop tax first needs to be evaluated against the primary is exempted for 10 years.7 purpose--economic growth--it was meant · Road construction is exempted from to serve. In this regard, we do not have ade- land occupation tax subject to approval quate data to provide a conclusive answer. by the provincial government. Reviving the Northeast · For companies engaged in government- The Reviving the Northeast program seeks promoted industries, imported equip- to revive the industrial potential of key cit- ment for self use are exempted from ies and facilitate a smooth economic transi- tariffs and import value added tax. tion. The decline of the northeast as one of Table 14.8 compares the western region China's powerhouses was a result of several and the country in some key economic and concurrent factors. social indicators for the period of 2000­05. The first was the declining efficiency Road construction and telecommunica- of state ownership. The northeast lagged tions users grew faster in the western region behind the rest of China in privatizing its than in the nation as a whole, reflecting state-owned enterprises (Garnaut and oth- the emphasis of the Go West program on ers 2005). This, in turn, was a result of its infrastructural construction. In addition, relative superiority before the mid-1990s. the Land for Forest and the Land for Grass The performance of the region's state- programs also paid off because the western owned enterprises in the 1980s and early region lost arable land and gained forest 1990s was relatively good, and people work- more quickly than the nation as a whole. In ing for them enjoyed relatively high income. the same period of time, population in the However, the emergence of the private western region declined 0.6 percent. This economy in the southern and eastern parts decline probably was a result of large-scale of China since the mid-1990s has posed a migration from the west to the east. Lastly, serious challenge to the state-owned enter- the western region's growth rates for other prises, resulting in declining profitability. indicators were comparable to those of the In the eastern and southern parts of the nation as a whole. country, privatization was relatively easy The Go West program seems to have ful- because a viable private sector was ready to filled its primary goals of providing infra- absorb the redundant workers released from structure, conserving the environment, and state-owned enterprises. In the northeast, improving social development in the west- however, the task was much harder because ern region. While the last two achievements there were fewer employment opportunities should be applauded in their own right, the outside the state sector. Therefore, having a Table 14.8 Comparison of the western region and the country, 2000­05a 2000 2005 Growth rateb Indicator Western National Western National Western National Population (million) 362 1,267 360 1,306 [­0.6] [3.1] Arable land (million hectares) 4.846 12.824 4.503 12.208 [­7.1] [­4.8] GDP (billion yuan) 1,665.5 9,720.9 3,349.3 19,778.9 11.3 11.9 Per capita GDP (yuan) 4,624 7,766 9,180 15,386 14.7 14.7 Railroads (kilometers) 22,109 58,656 27,594 75,438 [24.8] [28.6] Paved roads (kilometers) 553,874 1,402,698 780,339 1,930,543 [40.9] [37.6] Highways (kilometers) 3,677 16,314 10,530 41,005 23.4 20.2 Number of airports 58 121 66 142 [13.8] [17.4] Landline phone subscribers (million) 2.623 14.483 7.030 35.045 21.8 19.3 Mobile phone subscribers (million) 1.382 8.453 8.012 39.341 42.1 36.0 Middle school students (million) 1.834 7.369 2.354 8.581 5.1 3.1 Hospital beds (thousand) 830 3,177 877 3,351 [5.7] [5.5] Source: Statistical yearbooks of various provinces. a. Figures for "western" are aggregated on all the provinces in the western region. For comparison purposes, figures for "national" are aggregated on all the provinces in the country. Growth rates are based on comparable prices. Other financial figures are based on current prices. b. Most figures are annual growth rates, but those in square brackets are accumulative growth rates in 2000­05. The political economy of government policies toward regional inequality in China 233 strong state economy became a curse for the of central government money have been northeast. invested in the program. According to the The second factor was related to the Office of the Leading Group for Reviving northeast's industrial structure. Many cit- the Northeast Old Industrial Bases (2005), ies in the region were dependent on a sin- in 2004 alone the central government: gle resource such as coal or crude oil. The · Pledged Y 108.9 billion of government "resource curse" had real bite. Most cities bonds for 297 projects in industrial were not prepared for the depletion of these structural adjustments and revitaliza- resources. As a result, many cities suffered tion of old factories. By the end of 2004, from massive unemployment in the mid- Y 880 million had been disbursed; 1990s. For example, Fuxin, a coal mining · Provided Y 560 million to support the city, had an unemployment rate of 40 per- commercialization of key high-tech cent in the early 2000s (Garnaut and others projects; 2005). Another deficiency of the industrial structure was that the economy was depen- · Pledged Y 3.43 billion of government dent on heavy manufacturing. This did not bonds for projects related to agriculture, appear to be a problem when economic plan- forestry, and water conservation; ning was in place and orders were secured, · Gave Y 5.31 billion to support the aboli- but it became increasingly problematic tion of agricultural taxes; when the market became the primary tool · Arranged Y 2.02 billion for key road con- of resource allocation. The dependence on struction projects and Y 2.2 billion for heavy industry slowed the process of priva- rural road construction; tization in the northeast. The development · Subsidized Y 1.82 billion to convert the in the east and south, especially the industri- old pension scheme to a new pension alization of their rural areas, was closely tied scheme; to their advantages in light industry. Private · Provided Y 2.75 billion for the settlement firms are mostly small and lack capital in of redundant workers in state-owned their early years. They are more suitable to enterprises; producing consumer goods than intermedi- · Invested Y 4.05 billion in the rehabilita- ate inputs. In the eastern and southern parts tion of 15 coal mining areas; and of the country, such firms benefited from technological spillovers from the existing · Gave Y 1.3 billion to settle redundant state sector (Lin and Yao 2001). The domi- workers in the petroleum industry. nance of heavy industry in the northeast The heavy investment has paid off in prevented this kind of spillover from hap- some respects, but it has not reversed the pening, retarding the process of privatiza- decline of the northeast in the national econ- tion. In addition, state-owned enterprises omy. An official report of the Office of the in heavy industry are difficult to privatize Leading Group for Reviving the Northeast because they are much larger in both capital Old Bases (2007) finds that the share of the stock and employment. three northeastern provinces (Helongjiang, The third factor was related to the inferior Jilin, and Liaoning) in national GDP was technology in the northeast. This was actu- 9.6, 9.3, 8.7, and 8.6 percent in 2003, 2004, ally tied to the first two factors. Because the 2005, and 2006, respectively. Their gap with profitability of the state-owned enterprises Guangdong grew wider in these four years. was declining and the industrial structure In 2003 the sum of the three provinces' GDP was inadequate, technological upgrading in was only 80.3 percent of Guangdong's GDP. the northeast has been very slow. The share declined further to 77.1, 76.6, and Notwithstanding these weaknesses, the 58.3 in 2004, 2005, and 2006, respectively. northeast has rich stocks of human capital The same report finds that the economy and technological know-how that, if used in the northeast is still dominated by state properly, could serve as the base for its ownership. In 2006 state-owned and state- revival. The Reviving the Northeast program controlled enterprises contributed 35.7 per- acknowledges this potential. Large amounts cent of the total industrial value added in the 234 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA country, but 53, 63, and 86 percent in Liaon- promotion of the construction of the Social- ing, Jilin, and Helongjiang, respectively. ist New Countryside, optimization and In addition, the industrial structure has upgrading of industrial structures, enhance- not changed. The total profit of enterprises ment of transportation advantages, promo- with an annual sales volume larger than Y tion of the development of urban clusters 5 million was Y 191.1 billion for the three and county economies, efforts to deepen provinces. This is a remarkable increase of the opening up, efforts to accelerate social 22.5 percent over the year before, but still 8.5 development, promotion of sustainable percentage points shy of the national rate. development, and strengthening of the lead- Most important, 74.7 percent of the profit ership. Although the details of the program was contributed by central government­ have not been worked out, this program has owned petroleum and natural gas compa- a broader mandate than the other two pro- nies. Indeed, except for a limited number of grams, placing more weight on continuous resource-based sectors such as petroleum, development of the central region. natural gas, steel, transportation equipment, power generation, and utility supply, other The political logic of regional sectors were barely making money. The oil development programs refinement and nuclear power sector, the The economic rationale for the three largest sector in the northeast, lost Y 14.5 regional development programs varies, but billion and Y 19.3 billion in 2005 and 2006, the political case for all of them is strong. respectively. The Go West program has lofty goals, but Finally, the banking system is siphoning is a relatively modest program. It does not money out of the region, while the central aim to narrow the gap between the west and government is pumping money into it. the east; rather, it aims to restore the region's Since 2004, more than Y 10 million have ecological balance, preserve the environ- been diverted by the banking system out of ment, and build better infrastructure. This the region each year, and the total amount is a much-needed program because China's diverted during this period reached Y 858.1 western region is environmentally fragile billion by the end of 2006. The banking sys- and experiencing serious ecological chal- tem diverts money out of the region because lenges. It was named da-kai-fa--big devel- investments in other parts of the country opment--for the sake of political viability. can bring better returns than investments in The true aims of the program are considered the northeast. Thus the question of whether to be too conservative to receive support direct government investments can do bet- in the western provinces and too limited ter than private investments warrants seri- to soothe the dissenting voices opposed to ous consideration. enlarging the gaps between east and west. In comparison, the Reviving the North- Central Rising program east program is socially driven more than The huge amounts of central money economically and environmentally moti- pledged to the west and the northeast were vated. While it is right to help the northeast the envy of other provinces. The plea made in transforming its ailing industrial struc- by the central region was especially appeal- ture, the program tends to ignore the real ing politically.Although there is no evidence impediments to growth in the northeast, that the central region was failing (see table which are rigid mind-set, poor incentive 14.1 for counterevidence), there is a strong structure, and overwhelming state domi- political case for the central government to nance of the economy. There is a danger provide support to provinces in the region. that the influx of money from the central In April 2006, the Central Committee of the government will disguise the consequences CCP and the State Council (2006) issued a of these impediments. To catch up, the joint directive to launch the zhongbu jueqi, northeast needs to adapt the reforms taken or Central Rising program. This directive by the south and east coast. The central gov- indicates nine areas for improvement: artic- ernment should facilitate the transition in ulation of overall objectives and principles, the northeast instead of trying to keep ailing The political economy of government policies toward regional inequality in China 235 enterprises afloat. However, caution is major barrier preventing people from mov- needed in drawing attention to the weak- ing from the countryside to the city, from nesses that have prevented the northeast one village to another, and from one city from catching up with the coastal regions to another. The fast growth of the coastal because doing so puts the blame on local region since the early 1990s, however, has governments. The Reviving the Northeast drawn a large number of migrants from the program is a benefit the central government central and western regions.These migrants, gives the northeast to win local support. now estimated at 140 million, have become Likewise, the Central Rising program is a an indispensable part of the growth of the response to regional pressures. The central coastal region. Yet it has taken more than government has already pledged money to 10 years, and sometimes political heat, for some of the key areas for improvement. For the government to acknowledge migrants' example, Y 200 billion has been allocated to rights to move freely about the country. the Socialist New Countryside for each year The hukou system is still the major obstacle of the Eleventh Five-Year Plan spanning preventing migrants from settling freely in from 2006 to 2010.While the spread of pref- a place of their choosing. While most of the erential treatment to regions other than the restrictions, such as no movement across coast may narrow the gap between the coast county or city borders, and the benefits, and other regions, the gaps among the other such as free housing and food subsidies, regions are likely to remain and even expand. associated with hukou have been reduced, There is a "race to the bottom" effect in the hukou still holds significant implications in competition for central government money, two important areas (Bhide and Yao 2007; in which every province compares itself with Yao 2001). the worst-performing province and is ready The first area is providing more political to ask for more central government support. representation to migrants in the recipient The effects of preferential treatment then city. The right to political representation cancel each other out and may fail to rebal- is important because it instills a sense of ance regional development. responsibility--not necessarily account- ability--on the part of local government Alternative ways to address officials to include migrants in their calcu- regional disparities lations. Without hukou, migrants are often The previous section shows that other regarded as "outsiders" who only stay for considerations are as prominent as eco- short periods of time and whose welfare nomic considerations in reducing regional can rightly be ignored. For example, local inequality in the three major government urban residents can be qualified for dibao, programs, although the Go West program the low-income maintenance program, but is more practical than the other two. This migrants cannot. Local governments need section discusses some alternative ways to worry about the employment of local to address regional disparities. It does not residents, but not about the employment of provide an exhaustive list of the alterna- migrants. tives; rather, it concentrates on three areas The second area is the education of where policy changes could bring significant migrant children. It used to be very expen- results. These three areas are migration and sive for migrants to send their children to urbanization policy, fiscal policy, and invest- local public schools, and migrants opted ment policy. to set up their own schools. However, the schools faced constant harassment from Migration policy local governments because they were not China's migration policy has long been licensed. Starting in 2003, the situation criticized for its lack of efficiency as well began to improve. Fees charged on migrant as its violation of people's basic economic students were lowered, and many migrant rights. Since it was instituted soon after the schools were licensed by the local govern- great famine of 1959­62, the hukou, or resi- ments. However, migrant students can- dent registration, system has served as the not take the college entrance exams in the 236 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA recipient province because each province heavily on project-based transfers, which has a quota for college admission. This invariantly favor rich provinces. As a result, severely limits a migrant family's future investments by the central government in the city. Migrants with children in high often exacerbate regional inequality rather school have to return to their home prov- than narrow it. This seemingly unwise ince if they want their children to receive a result has a sound political economy basis. higher education. The formula-based scheme does not allow The restrictions put on migration are the central government to provide selec- closely linked with China's urbanization pol- tive favors to certain provinces or regions, icy.While large urbanized areas are emerging whereas the project-based scheme gives the along the east coast, the government is still central government considerable discretion averse to population concentration. Local to exchange favors with select provinces. To governments of large cities all have popula- the extent that they allocate money via proj- tion caps. For example, Beijing's population ects, the programs that aim to narrow the is capped at 19 million by 2020,a target many regional gaps have the same logic. experts believe will be exceeded. At the provincial level, Zhang (2005) Yet migration may be one of the most demonstrates forcefully that the current fis- effective ways to neutralize regional dis- cal arrangements hurt poor provinces.As he parities. In a simulation study, Whalley and puts it in the summary of his paper, Zhang (2004) find that eliminating hukou Regions initially endowed with a broader could lower regional income inequality. The nonfarm tax base do not need to rely heav- intuition is a straightforward application of ily on new and existing firms to finance pub- the law of one price: free movement of labor lic goods provision, which creates a healthy tends to equalize the wage rates in different investment environment in support of non- regions. Although in reality the effect may farm sector growth. In contrast, local gov- not be as strong as the simulation predicts, ernments in regions where agriculture is the allowing people of inland provinces to settle major economic activity spend the majority permanently in the coastal region would of their resources on their own operating relieve some of the environmental stress costs, leaving little for public investment. widely observed in inland provinces. Because of the relatively high transaction The coastal provinces will certainly need costs associated with collecting taxes from the agricultural sector, local governments to confront the problems caused by popula- tend to levy the existing nonfarm sector tion concentration. But precisely because of heavily, thereby greatly inhibiting its growth. concentration, many of those problems are As a result, regional differences in economic easier to solve than when they happen in a structures and fiscal dependent burdens may dispersed area. For example, waste treatment translate into widening gaps in equality. is cheaper when the quantities are larger. Other issues can also be handled with proper Table 14.9 shows that inland provinces urban design and government policies. Cities tax their farmers more heavily than coastal like Tokyo and NewYork provide good exam- provinces, although the overall tax burden ples of big-city management for China. is about the same level in all three regions. Since inland provinces have higher shares of Fiscal policy agricultural GDP, the consequences of this China's fiscal system provides incentives to skewed distribution of the tax burden can- local governments that exacerbate regional not be underestimated. disparities. This happens through two In combination with inland provinces' mechanisms at the central and local levels, heavier tax burdens, extending the central respectively. government's preferential tax policies to At the central level, there is no properly inland provinces would likely do more harm designed and enforced revenue transfer than good. These policies almost always scheme. There is a formula-based transfer promise to cut the corporate income tax. For scheme, but the money allocated to it is lim- existing firms, the corporate income tax is a ited. Instead, the central government relies 100 percent local tax; for new firms, it is split The political economy of government policies toward regional inequality in China 237 evenly between the central and provincial Table 14.9 Composite tax rates of inland and coastal regions governments. As provincial government Per capita GDP Agricultural tax rate Overall tax burden officials like to say, "The central government Year and region (current yuan) (percent) (percent) invites the guests, but provinces end up pay- 1994 ing the bill." This is fine with rich provinces National 3,849 2.14 10.55 but becomes a burden for poor provinces that Coastal region 5,262 2.16 10.53 Inland region 2,773 2.78 9.58 are financially constrained.Western provinces 2000 often opt not to extend preferential tax poli- National 7,077 2.33 12.62 cies to enterprises,which invites the inevitable Coastal region 10,578 1.61 11.40 complaints from businesses (CPDF 2004). Inland region 5,670 2.73 9.91 A better approach to addressing regional 2005 disparities is to strengthen formula-based National 14,002 0.25 16.67 Coastal region 21,426 0.16 14.81 revenue transfers. Although this ties the Inland region 11,070 0.34 12.62 central government's hands, it also creates Sources: National Bureau of Statistics of China, China Statistical Yearbook (1995, 2001, 2006); National Bureau of tangible benefits. One benefit is that it allows Statistics of China, Finance Yearbook of China (1995, 2001, 2006). the central government to address regional Note: Agricultural tax rate = total agricultural tax / agricultural GDP; overall tax rate = total tax / total GDP. disparities in a more equitable and less dis- tortionary way. It is more equitable because Table 14.10 Marginal contribution of public investments to regional income inequality poorer provinces automatically get more Type of GDP and investment Coastal Central Western transfers from the central government; it is Agricultural GDP less distorting because it does not introduce Roads 0.004 0.003 -0.005 selective incentives to the economy. Another Education 0.137 0.086 -0.221 benefit is that it preempts local governments' Electricity 0.022 0.014 -0.033 Telephones 0.043 0.027 -0.068 demands for special favors, which makes the Irrigation 0.127 0.080 -0.204 central government's life easier. Agricultural R&D 0.018 0.011 -0.027 However, the central government may Rural nonagricultural GDP not want to adopt this approach precisely Roads 0.033 0.002 -0.036 because it ties its hands. As in other cases, Education 0.251 0.018 -0.268 political will is needed. Electricity 0.064 0.004 -0.068 Telephones 0.129 0.009 -0.138 Investment policy Total rural GDP Roads 0.018 0.009 -0.028 Even though the central government insists Education 0.185 0.093 -0.277 on project-based fiscal transfers, there are Electricity 0.041 0.021 -0.062 better ways to conduct projects. Zhang Telephones 0.084 0.042 -0.125 Irrigation 0.052 0.026 -0.078 and Fan (2000) study the contributions of Agricultural R&D 0.007 0.003 -0.010 different public investments to reducing Source: Zhang and Fan (2000: table 6). regional income inequality for the period of Note: Inequality is measured by logarithmic variance of each type of GDP among the provinces. Figures are percent- 1978­95.8 Table 14.10, adopted from table age changes of inequality as a result of a 1 percent increase in individual types of public investment. 14.6, reports their estimates for the marginal contribution of different public investments thy that investments in education have the to regional inequality for the western, cen- largest effect on reducing inequality in the tral, and eastern regions. All the investments western region, but they also have the largest in the coastal and central regions increase effect on increasing inequality in the coastal inequality, although the contribution of and central regions. To the extent that edu- investments in the central region is minimal. cation in the western region is inferior to All the investments in the western region education in both the coastal and central reduce inequality. Investments in educa- regions, investment in the western region to tion and irrigation have very large effects on ensure more equitable access to education is agricultural GDP; investments in education the most important way to reduce regional and telephones have very large effects on income inequality. rural nonagricultural GDP; and investments Another interesting finding in table in education and telephones have very large 14.10 is that roads have a very mild impact effects on total rural GDP. It is notewor- on both reducing and increasing regional 238 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA inequality. A more detailed study is pro- Notes vided by Fan and Chan-Kang (2005), who Yang Yao is a professor and deputy director of the find that high-quality roads do not help to China Center for Economic Research at Peking reduce poverty, but that low-quality and University. rural roads have a significant effect. This 1. Construction of the third front started result makes sense because poor people when China split from the Soviet Union and have few chances to use highways, and more the Vietnam War began to escalate. The Chinese rely on local roads for their production and leadership sensed that a war was looming and commercial activities. believed that the country should be prepared for it. The country was then divided into three stra- The results provided by Fan and Chan- tegic fronts: the coast and border area was the Kang (2005), Fan, Zhang, and Zhang (2002), first front; the hinterland was the third front; and and Zhang and Fan (2000) may not be between them was the second front (Wei 2000). conclusive, but at least they show the merits 2. In an interesting paper, Banerjee, Duflo, of identifying better ways of investing pub- and Qian (2005) find that the distance to the lic resources to reduce regional inequality. nearest treaty port still has predictive power for a Chinese officials tend to conduct large-scale county's contemporary growth rates. projects, such as building highways, and give 3. Before 1994,China had a dual-track system lower priority to long-term investments, for its exchange rate regime. The central bank set such as in education. The aforementioned an official exchange rate, and the Shanghai for- studies have shown that a more balanced eign exchange swap market generated a market exchange rate. Domestic firms had to buy for- approach is needed. eign currencies if they imported inputs beyond Conclusions the government plan. Therefore, a higher reten- tion rate saved firms' money that could be used This paper has reviewed regional income to import inputs. inequality in China and discussed the politi- 4. The period 1985­91 included 1989 and the cal economy behind China's uneven develop- slowdown years that followed; the period 1999­ ment model and its recent programs aiming 2006 included years of major deflation. to reduce regional inequality. These reforms 5. If not otherwise indicated, data used in have several facets. First, the uneven develop- this subsection all come from the Office of the ment model seeks to explore the geographic Leading Group for Western Region Develop- advantages offered by China's coastal region. ment (2006). 6. Seehttp://www.developwest.gov.cn/content. Second,theGoWestprogramaimstopreserve asp?filename=txt/200707318 for details. the environment and improve economic and 7. This policy lost its effect after the central social infrastructure in the western region, government abolished all kinds of agricultural but the Reviving the Northeast and the Cen- taxes in 2006. tral Rising programs are driven as much by 8. Fan, Zhang, and Zhang (2002) provide non-economic as by economic consider- another study for a longer period of time (1952­ ations. Third, there are often more efficient 97) and obtain similar results. approaches to addressing regional inequal- ity, but they are less likely to be implemented References precisely because government decisions are Banerjee, Abhijit, Esther Duflo, and Nancy Qian. often driven by factors other than efficiency 2005."Railway and Growth: Evidence from considerations. China." Paper presented at the "Inequality The driving force behind Chinese cen- and Economic Performance" workshop, Mac- tral-regional politics is the "selective favor Arthur Foundation, International Research exchanges" by which the central govern- Network. ment retains large discretionary power in Bhide, Amita, and Yang Yao. 2007."The State distributing government revenues and the at Work: A Comparative Study of the State's provincial governments bargain for favors Responses to the Educational Needs of Infor- and give regional support in return. But mal Settlers in Indian and Chinese Cities." such decisions are not limited to China, India-China Institute, New School, New York. although they may take a different form in Central Committee of the CCP (Communist other administrative systems. Party of China) and the State Council. 2006. The political economy of government policies toward regional inequality in China 239 "Several Suggestions for Promoting the Rise Agglomeration in China." China Economic of the Central Region." Directive no. 10, 2006. Quarterly 6 (3): 801­16. Central Committee of the CCP, Beijing. National Bureau of Statistics of China. 2007. Chen, Liangwen, and Wei Wang. 2007. China Statistical Abstract. Beijing: China "Variations in Labor Productivity and the Statistical Publishing House. Intensity of Agglomeration." Working Paper, ------. Various years (1994­2006). China School of Government, Peking University. Statistical Yearbook. Beijing: China Statistical CPDF (China Project Development Facility). Publishing House. 2004. Small and Medium Enterprise Mapping ------. Various years (1995­2006). Finance of Sichuan, China. Chengdu: CPDF. Yearbook of China. Beijing: China Statistical Démurger, Sylvie, Jeffrey D. Sachs, Wing Thye Publishing House. Woo, Shuming Bao, Gene Chang, and Andrew Nee, Victor, and Peng Lian. 1994."Sleeping with Mellinger. 2002."Geography, Economic Policy, the Enemy: A Dynamic Model of Declining and Regional Development in China." NBER Political Commitment in State Socialism." Working Paper 8897, National Bureau of Eco- Theory and Society 23 (2): 253­96. nomic Research, Cambridge, MA. Office of the Leading Group for Reviving the Durlauf, Steven, and Peter Johnson. 1995. Northeast Old Industrial Bases. 2005."A "Multiple Regimes and Cross-Country Summary of 2004 and Key Tasks for 2005 Growth Behavior." Journal of Applied [in Chinese]." (http://chinaneast.xinhuanet. Econometrics 10 (4): 365­84. com/2005-06/13/content_4427545.htm.) Fan, Shenggen, and Connie Chan-Kang. 2005. ------. 2007."An Appraisal of the First Three Road Development, Economic Growth, and Years of the `Reviving the Northeast' Program Poverty Reduction in China. Research Report [in Chinese]." (http://chinaneast.xinhuanet. 138. Washington, DC: International Food com/2007-05/23/content_10091405.htm.) Policy Research Institute, June. Office of the Leading Group for Western Region Fan, Shenggen, Linxiu Zhang, and Xiaobo Development. 2006."The Unusual Five Years: Zhang. 2002. Growth, Inequality, and Poverty Accomplishments of the `Go West' Program in Rural China: The Role of Public Invest- in the First Five Years [in Chinese]." (http:// ments. Research Report 125. Washington, www.chinawest.gov.cn/web/NewsInfo. DC: International Food Policy Research asp?NewsId=28366.) Institute, July. Peng, Fangping, Shaoping Wang, and Qiang Gajwani, Kiran, Ravi Kanbur, and Xiaobo Wu. 2007."A Study of China's Economic Zhang. 2006. Comparing the Evolution of Convergence." China Economic Quarterly 6 Spatial Inequality in China and India: A (4): 1041­52. 50-Year Perspective. DSGD Discussion Paper Perkins, Dwight. 2005. "Chinese Reform in a 44. Washington, DC: International Food Pol- Historical Perspective." In The China Miracle icy Research Institute. in Perspectives [in Chinese], ed. Justin Lin Garnaut, Ross, Ligang Song, Stoyan Tenev, and and Yang Yao. Beijing: Peking University Yang Yao. 2005. Ownership Transformation in Press. China. Washington, DC: World Bank. Roland, Gerard, and Thierry Verdier. 2003. Ge, Wei. 1999. Special Economic Zones and the "Law Enforcement and Transition." European Economic Transition in China. Singapore: Economic Review 47 (4): 669­85. World Scientific. State Council. 1986. The Seventh Five-Year Plan Li, Boxi. 1995. Regional Policy and Coordinated for Economic and Social Development, People's Development. Beijing: China Caizheng Jingji Republic of China. Beijing: Renmin Press. Press. ------. 2007. Directives for Improving the Lin, Justin, and Yang Yao. 2001."Chinese Rural "Land for Forest" Policy [in Chinese]. State Industrialization in the Context of the East Council Directive no. 25, 2007. (http:// Asian Miracle." In Rethinking the East Asian www.chinawest.gov.cn/web/NewsInfo. Miracle, ed. Joseph Stiglitz and Shahid Yusuf, asp?NewsId=37705.) chap. 4. New York: Oxford University Press. Wang, Shaoguang, and Angang Hu. 1999. The Lu, Jiangyong, and Zhigang Tao. 2007."A Political Economy of Uneven Development: The Study of the Determinants of Industrial Case of China. New York: M. E. Sharpe. 240 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Wei, Yehua D. 2000. Regional Development in Yao, Yang, Yinzhao Gao, Ye Zhang, and Rudai China: States, Globalization, and Inequality. Yang. 2006."A Study of the Economic Com- London: Routledge. plementarities between the Pearl River Delta Whalley, John, and Shunming Zhang. 2004. and Hong Kong [in Chinese]." Working "Inequality Change in China and (Hukou) Paper, China Center for Economic Research, Labour Mobility Restrictions." NBER Peking University. Working Paper 10683, National Bureau of Zhang, Xiaobo. 2005. Fiscal Decentraliza- Economic Research, Cambridge, MA, August. tion and Political Centralization in China: Yang, Rudai. 2006."Adjustments to Sino-U.S. Implications for Regional Inequality. Trade Volumes by Accounting for Exports DSGD Discussion Paper 21. Washington, Rerouted through Hong Kong [in Chinese]." DC: International Food Policy Research Working Paper, China Center for Economic Institute, July. Research, Peking University. Zhang, Xiaobo, and Shenggen Fan. 2000. Public Yao, Yang. 2001."Social Exclusion and Eco- Investment and Regional Inequality in Rural nomic Discrimination: The Life of Migrants China. EPTD Discussion Paper 71. Washing- in Coastal Regions." Strategy and Manage- ton, DC: International Food Policy Research ment 3 (2001): 32­42. Institute, December. Is China sacrificing growth when balancing interregional and urban-rural development? Zhao Chen and Ming Lu 15 In the past 30 years of reform and opening balanced development and sustainable eco- up, China has enjoyed unprecedented eco- nomic growth, we have to introduce a spatial nomic growth in the context of globaliza- perspective and take into account the role tion, industrialization, and urbanization. of the spatial agglomeration of economic When China became fully integrated into activities. This chapter uses panel data at the the global economy, this populous country provincial and city levels to describe indus- joined the global production system, bring- trial agglomeration in China. Generally ing not only vast, cheap, and high-quality speaking, cities achieve increasingly strong c h a p t e r labor but also great market demand. China's industrial agglomeration effects and scale integration has reshaped the global and, economies. The chapter finds that industrial especially, the Asian industrial map. At the agglomeration in China is indeed accompa- same time, China's manufacturing indus- nied by the widening of interregional and try has become concentrated along the east urban-rural gaps, which may exert a nega- coast and urban areas. Industrial agglom- tive influence on economic growth and eration has been beneficial for China's eco- social harmony. To achieve balanced devel- nomic growth, but it also has given rise to opment while maintaining sustainable eco- interregional and urban-rural disparities. nomic growth is a great challenge facing the Because income disparities are disadvan- Chinese government. tageous for sustainable economic growth, In the second part of this chapter, we China has been trying to balance inter- present the trend of industrial agglomera- regional and urban-rural development by tion, use panel data at the provincial level means of fiscal transfers. Despite the wis- to elaborate the interrelationships among dom of the strategy, China's government industrial agglomeration, urbanization, is unduly dependent on fiscal transfers to and globalization, and use panel data at achieve interregional and urban-rural bal- the city level to show the positive effects anced development. Fiscal transfers alone of scale on industrial growth.1 The third cannot keep interregional and urban-rural part shows the changes in interregional gaps from expanding further. To achieve and urban-rural gaps as well as the nega- this, the Chinese government will have to tive influences of income disparity on eco- adjust its policy measures to integrate the nomic and social development. The fourth interregional market, promote the agglom- part discusses the adjustment process of eration of labor resources to coastal and interregional and urban-rural develop- urban areas, and direct more investment to ment policy from a historical perspective human capital and infrastructure in lagging and examines adjustments in fiscal trans- and rural areas. fers of the central government and their In China, when considering the relation- influence on regional development. The ship between interregional, urban-rural fifth section concludes. 241 242 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Industrial agglomeration and proportion--below4percent--ofnational city development industrial output. Traditional economic growth theory does · Three provinces in northeastern China not stress the importance of space, whereas played an important role, with Liaoning new economic geography models consider ranking top in national industrial share. the economies of scale brought about by the Gansu and Shaanxi, which are two west- spatial agglomeration of economic activi- ern and inland areas of China, accounted ties to be an important driver of economic for more than 2 percent of national growth. As economic activities and popula- industry, surpassing that of some middle tion are increasingly concentrated in large and eastern provinces. cities, economic development will secure · Beijing, Shanghai, and Tianjin covered scale economies in at least three areas. The very small areas, but their industrial first is sharing. Producers can acquire exten- shares were larger than those of many sive supplies of inputs from a wider scope other provinces (Chen, Jin, and Lu 2006). of suppliers, which could reduce average Figure 15.1 divides China's provinces into production costs as the scale of produc- coastal and inland provinces and depicts the tion increases. At the same time, the sharing share of each province in national industrial of inputs encourages suppliers to provide gross domestic product (GDP) from 1987 to highly specialized products and services to 2005. Comparing the data for 2005 with the meet the demand. The second is match- data for 1987 shows significant industrial ing. In a large market, enterprises are more agglomeration in China. To be more spe- likely to employ workers with special skills, cific, during this time period: helping companies to meet specific market Industrialsharesof coastalareasincreased demands. Meanwhile, having many poten- · remarkably, with the top four provinc- tial employers and employees in the same es--Guangdong, Shandong, Jiangsu, and location makes it easier to match them Zhejiang--accounting for 12.27, 11.20, efficiently. The third is learning. Spatial 10.93, and 7.43 percent, respectively, in agglomeration can accelerate the spillover of 2005. knowledge and make it easier for employees and entrepreneurs to learn from one another · The industrial status of three northeast- (Gill and Kharas 2007). ern provinces decreased noticeably, with Before China's reform and opening up the industrial share of Liaoning, Hei- in 1978, many of China's industries were longjiang, and Jilin decreasing to 4.09, located in inland areas as a result of a strat- 3.16, and 1.60 percent, respectively. egy of balancing regional economic devel- · The industrial share of western prov- opment with military strategy. Since 1978, inces decreased on the whole. market forces have dominated economic · The industrial share of the four autono- development and industrial layout, with mous municipalities decreased distinctly, agglomeration occurring toward the east with that of Shanghai decreasing to 4.83 coast, especially the Yangtze River delta percent, Tianjin to 2.21 percent, Beijing to and the Pearl River delta. This provides an 2.0 percent, and Chongqin to 1.2 percent. excellent case in which to examine the inter- Industrial agglomeration showed other play of economic development, industrial patterns as well. In 1987 only 2 provinces agglomeration, and city development. (Jiangsu and Liaoning) had more than 8 Industrial agglomeration in percent and 13 had less than 2 percent of globalization and industrialization industrial shares. In 2005 3 had more than 8 percent, and 14 had less than 2 percent. Until the early years of reform and opening During the same period of time, 11 prov- up in 1978, China's industrial layout was inces increased their industrial share, includ- geographically scattered: ing 7 coastal provinces, while 18 provinces · A few coastal and middle prov- decreased their industrial share, including inces accounted for a relatively small 13 inland provinces. In the coastal areas, all Is China sacrificing growth when balancing interregional and urban-rural development? 243 except the 3 autonomous municipalities of Figure 15.1 Industrial share in China, by province, 1987, 1995, and 2005 Beijing, Shanghai, and Tianjin and the prov- A. Inland provinces inces of Guangxi and Liaoning increased Qinghai Ningxia their industrial share. Provinces lost indus- Gansu trial share for various reasons: 3 autono- Guizhou mous municipalities were entering the Xinjiang post-industrialization period, Liaoning was Chongqing part of a regional decline in northeastern Yunnan China, and Guangxi lost industrial shares Jilin to its neighboring province Guangdong. Jiangxi Although Guangxi is a coastal province, its Inner Mongolia economy is relatively backward and thus an Shaanxi object of China's "Go West" policy. Anhui Changes in industrial share were accom- Shanxi panied by the movement of labor, espe- Hunan cially redundant labor in rural areas, to the Hubei Heilongjiang southeastern coastal areas. As indicated by Sichuan an analysis of census data, labor flowed to Henan the provinces that increased their industrial share (see Ding, Liu, and Cheng 2005).2 Hainan B. Coastal provinces Guangxi Therefore, changes in industrial share Beijing reflect the trend of industrial agglomeration Tianjin rather than differences in industrial growth Fujian across regions. Lu and Tao (2006) use data Liaoning at the industry level to calculate the Ellison- Shanghai Glaeser index between 1998 and 2003. They Hebei find that regional industrial agglomeration Zhejiang in China was still rising during this period. Jiangsu Inland cities are far from coastal lines of Shandong transportation, which increases the costs of Guangdong transportation, but labor costs are relatively 0 2 4 6 8 10 12 14 low. Therefore, products suitable for inland percent production include staples, such as coal, that 1987 1995 2005 are carried by train or ship, or products with Source: National Bureau of Statistics of China (1999, 2006). high added value, such as computer chips, Note: Beijing is counted as a coastal area. Provinces are sorted in ascending order in coastal and inland areas, that are transported by plane (Gill and Kha- respectively, according their industrial shares in 2005. ras 2007). Is industrial growth in these areas related to globalization and urbanization? To answer Globalization is mainly the opening of this question, we use provincial panel data commodity markets and capital markets. and find that both economic opening up and Both opening of capital markets as mea- urbanization enhance industrial agglomera- sured by per capita foreign direct investment tion, while economic opening up is related (FDI) and opening of commodity markets to geographic location and initial degree of as measured by dependence on interna- openness.At the same time, we also find that tional trade indicate that coastal areas have a larger market size, improved transportation higher degree of openness than inland areas and telecommunications infrastructure, and and that globalization is the most important weaker government intervention in a prov- factor contributing to interregional income ince are beneficial for industrial agglomera- disparities (Wan, Lu, and Chen 2007). Does tion (Chen, Jin, and Lu 2006). Figures 15.2 economic opening have something to do and 15.3 depict the relationship between the with industrial growth? To answer this ques- degree of globalization and urbanization tion, we compare data for 1987, 1994, 2000, and the level of industrial development. and 2005. In figure 15.2, the horizontal axis 244 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 15.2 Globalization and industrial growth in China, 1987­2005 Graphs by year 1987 1994 60 60 40 40 20 20 (%) 0 0 rate 0 50 100 150 0 50 100 150 growth 2000 2005 60 60 industrial 40 40 20 20 0 0 0 50 100 150 0 50 100 150 international trade dependence ratioa Industrial Growth Rate Quadratic Line Source: National Bureau of Statistics of China (1999, 2005, 2006). Note: Weighted by GDP per capita. The equations of the fitted lines are y = -0.0223x2 + 0.9852x + 11.007 (R2 = 0.0998) for 1987, y = 0.0012x2 - 0.1168x + 18.588 (R² = 0.0913) for 1994, y = -0.0007x2 + 0.1178x + 10.799 (R² = 0.0576) for 2000, and y = 0.0003x2 - 0.0245x + 21.142 (R² = 0.0104) for 2005. a. Ratio of import and export volume to GDP. represents the proportion of import and 2005. In figure 15.3, the horizontal axis rep- export volume in GDP--that is, the inter- resents the proportion of urban dwellers in a national trade dependence ratio--of a spe- specific province, the vertical axis represents cific province, the vertical axis represents the industrial growth rate of this province, the industrial growth rate of the province, and the width of the circles represents per and the width of the circles represents per capita GDP. We find that provinces with a capita GDP. By comparison, we find that higher urbanization ratio experienced more the relationship between economic open- rapid industrial growth. In 2005 provinces ness and industrial growth followed an with a higher urbanization ratio and more inverse-U curve in 1987, 1994, and 2000, rapid industrial growth also had higher per with most provinces located in the left half capita GDP. of the curve. This indicates that the phase of openness is beneficial for industrial growth. Industrial agglomeration and scale In these three years, the trade-related sec- economy in cities tor squeezed industrial growth when the Urbanization supports industrial growth degree of openness was too high. However, mainly due to economies of scale. To elabo- the curve for 2005 indicates that the rela- rate the relationship between scale effect in tionship between economic openness and city development and industrial agglomera- industrial growth sloped to the northeast. tion, we use data at the city level to examine In other words, the higher economic open- the relationship between per capita GDP and ness is, the stronger is the effect of economic industrial share. The higher per capita GDP openness on industrial growth. is,the larger is the local market,and the more The agglomeration effect is manifested beneficial this is for industrial agglomera- mainly in the effects of urbanization on tion. However, when per capita GDP exceeds industrial growth. To elaborate this point, a certain level, the share of service industry is we compare data for 1987, 1994, 2000, and higher, and the city's economy may enter the Is China sacrificing growth when balancing interregional and urban-rural development? 245 Figure 15.3 Urbanization and industrial growth in China, 1987­2005 Graphs by year 1987 1994 60 60 40 40 20 20 (%) 0 0 rate 0 50 100 0 50 100 growth 2000 2005 60 60 industrial 40 40 20 20 0 0 0 50 100 0 50 100 urban population ratio (%) Industrial growth rate Fitted line Source: National Bureau of Statistics of China (1999, 2005, 2006). Note: Weighted by per capita GDP. In the data, the urbanization ratio in Shanghai in 2005 (point located in the most right position) was even lower than it was in 2000. This is because of a substantial increase in the number of rural population. The equations of the fitted lines are y = 0.1028x + 13.039 (R2 = 0.0878) for 1987, y = 0.1286x + 24.543 (R² = 0.0915) for 1994, y = 0.0546x + 10.126 (R² = 0.0407) for 2000, and y = 0.0884x + 17.262 (R² = 0.0134) for 2005. post-industrialization phase, which means figure 15.4, the horizontal axis represents that industrial shares may decrease instead. per capita GDP in a city, while the vertical A typical example is Shanghai, which has axis represents the industrial share of this always been the country's top city in share city in the value of national industrial pro- of industry; however, Shanghai's industrial duction. We present the data for 1991, 1995, share first ascended and then descended as 2000, and 2005 in a scatter diagram and add the city entered the post-industrialization quadratic lines. Observing and comparing era. In 1997 the value of its service industry the figures for these four years, we find that in total production surpassed 50 percent for (a) on the whole, the higher per capita GDP the first time. In 2006 its service industry is in a city, the higher is its industrial share provided more than half of employment (although we add quadratic lines, very few opportunities for the first time.3 sample cities appear on the right half of the In figure 15.4, the panel data we use inverse-U curve), and (b) with the elapse only refer to the cities; counties are not of time, the turning point of the quadratic included. Industrial share of a city means curve moves to the right. For 2000 and 2005, the proportion of industrial value that this the quadratic curves are substantially closer city accounts for in the sum of all sample to the upward trend line. That is to say, with cities. To eliminate the influences of some the elapse of time, economic development outlier points, we do not include either cit- and market volume measured by per capita ies with industrial growth above 100 per- GDP became more and more beneficial for cent or below -50 percent or autonomous further industrial agglomeration. Moreover, municipalities. Because Shenzhen has a in this period of 15 years, the goodness of fit high proportion of floating population, of the lines increased from 0.2552 to 0.3678, its per capita GDP, calculated according to 0.4628, and 0.4435. In other words, per population with local household registra- capita GDP in a city has stronger explana- tion (hukou) status, is extraordinarily high tory powers for industrial agglomeration in and thus is not included in our analysis. In recent years than in earlier years. 246 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 15.4 Per capita GDP and industrial shares of cities in China, 1991­2005 Graphs by year 60 1991 1995 60 40 40 20 20 (%) 0 0 shares 0 50,000 100,000 150,000 0 50,000 100,000 150,000 2000 2005 60 60 industrial 40 40 20 20 0 0 0 50,000 100,000 150,000 0 50,000 100,000 150,000 per capita GDP (Yuan) Industrial shares Quadratic line Source: National Bureau of Statistics of China, China City Statistical Yearbook (various years). Note: The equations of the fitted lines are y = -3E-08x2 + 0.0015x - 0.542 (R² = 0.2552) for 1991, y = -2E-09x2 + 0.0005x ­ 0.007 (R² = 0.3678) for 1995, y = -2E-09x2 + 0.0005x - 2.3832 (R² = 0.4628) for 2000, and y = -2E-09x2 + 0.0004x - 3.6737 (R² = 0.4435) for 2005. If scale economies are beneficial for (AuandHenderson2006a,2006b).Although industrial agglomeration and growth, then, a congestion effect will appear in the process in a city with a given area, enhancing a city's of city expansion, it could be alleviated with population density should improve labor improved transportation, environment, and productivity, as shown in figure 15.5.4 More security, which, in turn, would promote city notably, except for 1995, an inverse-U curve development on a larger scale. appears for the relationship between popu- Due to the obstacles to labor flow and lation density and per capita GDP, and most interregional market segmentation, indus- cities are found in the left half, which means trial agglomeration in China has been far that, to maximize per capita GDP, cities slower than possible and, indeed, necessary. could have their own optimal scale. In other Compared with Western countries, indus- words, during this period a great number trial agglomeration in China is still rather of cities in China were suffering produc- low (Lu and Tao 2006). Differences in scale tivity losses as a result of their small scale. among Chinese cities are much smaller than Econometric analysis indicates that 51 to 62 differences among cities in other countries percent of the cities in China have unduly (Fujita and others 2004). Lack of spatial small scale. In typical cities, losses caused agglomeration of population results in the by small scale account for 17 percent of the inefficient use of land; this is especially evi- average output of employees, and cities in dent in China. Since the mid-1990s, 338 big which losses reach 25 to 70 percent of the cities around the country have expanded average output account for a fourth of all their downtown area from 16,000 square the sample cities. As the services industry kilometers to 25,000 square kilometers, with continues to grow, diversifying the service an increase rate of 60 percent. In the same input in production chains will become period, population in the downtown area of increasingly important. At the same time, these same cities increased from 0.27 billion the scale effects of economic agglomera- to 0.3 billion, including migrant workers, tion on economic growth will also become with a growth rate of only 10 percent. The more and more important, and the optimal expansion rate of area is six times that of the scale of cities will become bigger and bigger population (Yan and Jiang 2007). Inefficient Is China sacrificing growth when balancing interregional and urban-rural development? 247 Figure 15.5 Population density and per capita GDP of cities in China, 1991­2005 Graphs by year 1991 1995 8 8 6 6 4 4 2 2 0 0 GDP 0 2,000 4,000 6,000 0 2,000 4,000 6,000 capita 2000 2005 8 8 per 6 6 4 4 2 2 0 0 0 2,000 4,000 6,000 0 2,000 4,000 6,000 population density (People per Sq-km) Per capita GDP Quadratic line Source: National Bureau of Statistics of China, China City Statistical Yearbook (various years). Note: The vertical axes are relative measures--namely, GDP per capita divided by the mean of the year. The equations of the fitted lines are y = -2E-09x2 + 0.0001x + 0.8461 (R² = 0.0226) for 1991, y = -7E-10x2 + 0.0002x + 0.8157 (R² = 0.0453) for 1995, y = -3E-08x2 + 0.0003x + 0.7212 (R² = 0.0581) for 2000, and y = -7E-08x2 + 0.0004x + 0.7406 (R² = 0.043) for 2005. use of land makes it difficult to improve the Growth at the cost of interregional quality of life and to increase the amount of and urban-rural inequality resources per capita in lagging areas. Worse Numerous studies have indicated that inter- still, if the population in inland areas cannot regional and urban-rural income disparities agglomerate toward coastal areas, migrants are the two major contributors to national will move toward cities in inland areas. And income disparities (Kanbur and Zhang 1999; once an unreasonable pattern of land use is Khan and Riskin 1998; Li 2003; World Bank formed, it will be very difficult to change. 1997; Yang 1999; Yao and Zhu 1998; Zhao and others 1999), while detailed analysis of China's urban-rural and income disparities shows that interregional interregional development: disparity itself has something to do with China's vast urban-rural income disparities is there a tradeoff between (Hussain, Lanjouw, and Stern 1994; Kan- efficiency and equality? bur and Zhang 1999; Tsui 1993). The recent To understand the relationship between decomposition of inequality shows that 70 China's economic growth and income to 80 percent of interregional income dis- inequality, a spatial perspective is absolutely parity is contributed by urban-rural income necessary. Correspondingly, if policy aims disparity (Wan 2007). to balance urban-rural and interregional Figure 15.6 indicates that urban-rural development, it is necessary to pay attention income disparity narrowed in the early to the positive effect of spatial agglomera- 1980s mainly because rural reform raised tion on economic growth. Nevertheless, it is the income of rural dwellers. Thereafter, very important to balance interregional and this ratio widened, before dropping again urban-rural development with sustainable after 1994, when the government raised the economic growth. In this section, we discuss purchase price of agricultural products. But the relationship between income disparities since 1997, urban-rural income disparity and economic growth. has been growing.Although the government 248 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 15.6 Urban-rural and interregional income disparities in China, 1978­2005 · Industrial agglomeration toward coastal 3.5 0.25 areas, especially the Yangtze River delta, Pearl River delta, and Bohai Bay area ratio 3.0 0.23 income (Chen and others 2007; Lu and Chen 2006; Wen 2004). income 2.5 0.21 capita per A report by the World Bank (1996) capita argues that, although disparities are increas- per 2.0 0.19 ing because of policy bias in trade and provincial 1.5 0.17 investment, the fundamental reason is that of comparative advantages of different areas urban-rural Gini 1.0 0.15 in China were inhibited before the reform and favorable policies granted to coastal 1978197919801981198219831984198519861987198819891990199119921993199419951996199719981999200020012002200320042005 year areas after the reform have brought inter- Urban-rural per capita income ratio regional comparative advantages into play. Gini of provincial per capita income In the process of opening up, the inflow of foreign capital and the development of an Sources: For urban-rural per capita income ratio, data for 1980 to 1987 are from Ravallion and Chen (2007); the Gini export-oriented economy began to create of provincial per capita income is calculated according to National Bureau of Statistics of China, China Statistical Yearbook (various years). Urban and rural per capita income is deflated by urban and rural consumer price index. interregional comparative advantages. Our Interregional inequality is calculated as Gini coefficient of per capita income in every province, and original data are from National Bureau of Statistics of China (2005) and National Bureau of Statistics of China, China Statistical Yearbook decomposition of interregional income (various years). We first deflate per capita urban disposable income and rural per capita net income by urban and rural disparities shows that per capita FDI and CPI, respectively. Then we conduct weighted average income with the prorportion of agricultural population as the weight, which yields the per capita income of every province. trade dependence have jointly become the most important among nine contributors to interregional income disparities (Wan, Lu, and Chen 2007). has made great efforts to treat the problems Of all the elements affecting interregional affecting agriculture, peasants, and rural disparities in the literature,the most essential areas--for example, the agricultural tax is the difference in geography and policies, exemption and subsidy for grain-planting while others are more likely manifestations peasants--the gap is still expanding. of disparities in interregional development. Figure 15.6 also indicates that the inter- Geographic advantages and preferential pol- regional gap is generally expanding and that icies are the major reasons for the current changes in interregional income disparity pattern of industrial agglomeration. are very similar to changes in urban-rural income disparity. In the existing research, Income disparities and interregional disparity in China arose for sustainable growth the following reasons: Spatial agglomeration and regional com- · parative advantages enhance efficiency but Preferential policies in which coastal also exacerbate interregional and urban- areas benefited from deregulation, which rural income disparities. However, income promoted economic openness and mar- inequality itself might be detrimental to ketization, improving their ability to social harmony and economic growth. Most compete in global markets and to absorb existing studies find that widening income more FDI (Démurger and others 2002; disparities will have a negative influence on Wan, Lu, and Chen 2007); economic growth by reducing the accumula- · Unbalanced development of private tion of physical and human capital. In recent economies and township-and-village years, some literature using data from China enterprises (Rozzelle 1994; Wan 1998; has studied the influence of income dispari- Wan, Lu, and Chen 2007); ties on economic growth. Ravallion (1998) · Fiscal transfers in favor of eastern areas uses survey data for rural areas in China (Ma and Yu 2003; Raiser 1998); and finds that inequality of wealth has a · Differences in infrastructure in different negative effect on the growth of consump- areas (Démurger 2001); tion per capita at both the family and the Is China sacrificing growth when balancing interregional and urban-rural development? 249 village levels. Benjamin, Brandt, and Giles accumulation of productive capital (see, for (2004) use panel data at the village level in example, Benhabib and Rustichini 1996). China and find no evidence that income dis- The first two mechanisms could hardly be parities block economic growth, but they do tested empirically, while the third mecha- find, in the long run, a negative relationship nism may find some indirect evidence.Figure between them. In our own study based on 15.7 shows the urban-rural income dispari- provincial panel data, we introduce the poly- ties and the number of infringement cases in nomial inverse lag (PIL) framework, which China. Both trends are very similar. allows us to measure the impacts of inequal- In China, greater interregional income ity on investment, education, and ultimately disparity is accompanied by market seg- on growth at precisely defined time lags. mentation and local protectionism, which Combining PIL with simultaneous systems is harmful for sustainable economic growth. of equations, we analyze the relationship Before 1978, China made many industrial between inequality and growth in post-re- investments in the inland areas, in effect form China, finding that this relationship is promoting interregional balanced devel- nonlinear and negative, irrespective of time opment. During the reform period, fiscal horizon (Wan, Lu, and Chen 2006). transfers from the central government have The ratio of urban-rural income per been invested in economically richer areas capita, a proxy of inequality at the provincial to promote preferential development of the level, has an effect on investment, education, coast. Meanwhile, local governments on var- and economic growth.A ratio of urban-rural ious levels have secured the power to make income that is one unit higher will have a local economic policies. Because lagged areas negative cumulative influence on investment, receive fewer fiscal transfers from the central proxied by the ratio of investment to GDP, government, these local governments have but a positive cumulative influence on educa- sought to protect disadvantaged enterprises tion, proxied by per capita schooling. Higher in the short term. Lagging areas invest in urban-ruralincomedisparitieswillhavealast- newly emerging industries and then protect ing negative influence on economic growth. their products to strengthen the bargaining Because the negative effects of inequality on power of local governments to negotiate fis- investment dominate the positive effects of cal transfers from the central government. inequality on education, and physical capi- Although the strategic actions of lagging tal accumulation remains the major driver areas could benefit the local area, they result of China's economic growth, it makes sense in numerous duplicative constructions and that the influence of income disparities on loss of efficiency, which are disadvantageous economic growth will be negative. for interregional specialization, economic Theoretically, income disparities can affect investment in many ways. First, due Figure 15.7 Urban-rural income disparities and infringement cases in China, 1981­2004 to imperfections in the credit market, higher 3.1 350 income disparities will constrain poor peo- ple by tightening credit lending and lower- 2.9 300 ing their investment in physical and human 2.7 250 capital (for example, Fishman and Simhon 2.5 disparity 200 2002; Galor and Zeira 1993). Second, in a 2.3 infringement democratic society, greater income gaps will 150 2.1 make more people support higher taxation 100 urban-rural 1.9 for redistribution, while it will have a nega- property 50 tive influence on economic growth (Alesina 1.7 and Rodrik 1994; Bénabou 1996; Persson and 1.5 0 Tabellini 1994). Third, higher income dis- 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 parity will also result in social and political year unrest, constrain the investment environ- Urban-rural per capita income ratio Property infringement cases per 100,000 people ment, and direct more resources to the pro- tection of property rights, thus reducing the Sources: National Bureau of Statistics of China, China Statistical Yearbook (2005); authors' calculations. 250 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA agglomeration, and sustainable growth (Lu government established SEZs in Shantou, and Chen 2006). In our empirical study con- Shenzhen, Xiamen, and Zhuhai in 1980 and cerning market segmentation, we find that in Hainan in 1988. SEZs were given greater government intervention--proxied by the powers of economic management and were government consumption ratio and share allowed to establish joint venture, coopera- of employment in state-owned enterprises tive, and wholly foreign-funded enterprises. lagged one year--increases the interprovin- Enterprises in SEZs enjoyed managerial cial market segmentation index. Although autonomy and preferential taxation rates.5 China's domestic commodity market has In 1984 the central government decided to been integrating gradually since the mid- open 14 coastal cities, granting preferen- 1990s, government intervention as measured tial treatment to foreigners who invest in by the ratio of local government expenditure China and bring advanced technologies and to local GDP has risen, which is disadvanta- expanding the ability of these cities to pur- geous for market integration, scale economy, sue foreign business activities. In 1985 the and sustainable economic growth in China Yangtze River delta, Pearl River delta, and a (Chen and others 2007). triangular area in south Fujian comprising Quanzhou, Xiamen, and Zhangzhou began Interregional and urban-rural to enjoy the status of coastal economic open economic development: policy areas; Shanghai began to enjoy the status adjustment and fiscal transfer of a coastal economic open area and an open city. In 1988 coastal open areas were Obviously, the Chinese government has expanded to 153 cities and counties in 7 recognized the need to adjust its regional eastern provinces and municipalities and in economic development policies. Deng Guangxi. The first 14 national economic and Xiaoping, the general architect of China's technological development zones (ETDZs) reform, mentioned in a speech dated 1986, established between 1984 and 1988 were all "We allowed first prosperity of some areas located on the east coast. The effect of the and some people just to better achieve opening policies adopted in the 1980s was common prosperity, and we need to pre- to widen the regional disparities between vent polarization. This is called socialism." coastal and inland areas. (Deng 1993: 195). In the early years of During the 1990s, the central govern- reform, China adopted economic opening ment began to seek balanced interregional policies intended to support the economic development. In 1992, 15 additional ports development of coastal areas, and these and 26 additional counties were declared policies widened interregional gaps while as "opening," which brought the number promoting the preferential development of open ports and cities or counties to 167 of coastal areas. However, since the end of and 825, respectively. In this period, open- the twentieth century, the government has ing policies began to reach inland areas. In sought to balance interregional and urban- March 1992, border economic cooperation rural development, as symbolized by a series areas were set up in 4 cities in the northeast- of regional development strategies, includ- ern provinces and Inner Mongolia, mak- ing the "Go West" policy, "Revitalizing the ing a national total of 14 border economic Northeast," and "Central Rising" programs, cooperation areas that year, most of which as well as by recent policies on agriculture, were located in middle and western areas peasants, and rural areas. (see table 15.1). In August 1992, the govern- ment announced its intention to declare as The history of regional and urban- coastal open cities 5 cities along the Yangtze rural development policies River, the capital cities of the 4 border prov- The policies of economic opening were inces, and the capital cities of 11 provinces first adopted in the coastal areas that had in inland areas. Concrete policies included geographic advantages, taking the form expanding the powers of open cities to con- of special economic zones (SEZs) or eco- duct foreign cooperation, introduce for- nomic development zones. The Chinese eign advanced technologies and managerial Is China sacrificing growth when balancing interregional and urban-rural development? 251 Table 15.1 Historical development of opening areas Period and type Total for of opening Year East Middle West period 1978­88 Special economic zone 1980 4 cities None None 5 1988 1 province Coastal open city 1984 14 cities None None 14 National economic and technological 1984 10 cities None None 14 development zone 1985 1 city 1986 3 development zones in Shanghai Coastal economic open area 1985­88 7 provinces, 2 autonomous municipalities None 1 province 10 1988­98 Capital city, city along Yangtze River 1992 1 city 11 cities 1 city 23 National border economic 1992 1 city 4 cities 9 cities 14 cooperation areas National economic and technological 1992 4 cities, 2 development zones None None 18 development zone in Fujian 1993 3 cities, 2 development zones in Guangdong 4 cities 1 city 1994 1 city None 1 city National high-tech industry 1988 1 development zone in Beijing None None 53 development zone 1991 10 cities, 1 development zone 5 cities, 1 5 cities in Shandong, Fujian, Guangdong, development respectively zone in Hubei 1992 11 cities, 1 development zone in Shanghai 7 cities 6 cities, 1 development zone in Inner Mongolia 1996 1 city None None 1997 None None 1 development zone in Shaanxi National industrial park 1989­94 1 park in Jiangsu, Shanghai, Zhejiang, Fujian, None None 5 Hainan, respectively National tax-free zone 1990­2000 10 cities, 1 park in Shanghai, Tianjin, Fujian, None None 13 respectively 1998­2008 National economic and 2000 None 4 cities 7 cities 17 technological development zone 2001 None 1 city 3 cities 2002 1 city None 1 city National high-tech industry 2007 1 city None None 1 development zone Source: http://www.cadz.org.cn/. Note: The east area includes 11 provinces or cities: Beijing, Fujian, Guangdong, Hainan, Hebei, Jiangsu, Liaoning, Shandong, Shanghai, Tianjin, and Zhejiang; the middle area includes 8 provinces: Anhui, Heilongjiang, Henan, Hubei, Hunan, Jiangxi, Jilin, and Shanxi; the west area includes 12 provinces or cities: Chongqing, Gansu, Guangxi, Guizhou, Inner Mongolia, Ningxia, Qinghai, Shaanxi, Sichuan, Tibet, Xinjiang, and Yunnan. Inner Mongolia and Guangxi are included in the west, because they are objects of the Go West policy. practices, grant preferential treatment to the cultivation of talented people, and open foreign-invested enterprises, and allow the up the economy. Thereafter, government creation of ETDZs when conditions per- investment grew each year, so that, by 2005, mitted. Some of the national ETDZs and the central government had invested Y 460 national high-tech industry development billion for construction projects in the west- zones established between 1992 and 1994 ern areas and distributed fiscal transfers were located in the middle and western and subsidies totaling more than Y 500 bil- areas. lion; one-third of national bonds for long- In recent years, the government has term construction were used in the western adopted various strategies to address the areas. The country established 60 key proj- interregional development gap. In 2000 the ects with an investment of Y 850 billion, in government adopted the Go West policy, which investment funded by national bonds which sought to speed up the construc- amounted to more than Y 270 billion. In tion of infrastructure in the western areas, the same period, western areas absorbed strengthen ecological and environmen- more than US$9 billion in FDI; together tal protection and construction, actively with loans provided by international adjust the industrial structure, accelerate organizations and foreign governments, 252 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA western areas absorbed US$15 million in tax and measures for collecting the tax on foreign capital. More than 10,000 enter- special agricultural products, and reform- prises entered western areas to invest and ing the collection and use of village funds. In develop their businesses, investing more 2004 the government declared its intention than Y 300 billion.6 Government also to lower the agricultural tax over the course invested in the construction of rural infra- of five years, canceling it altogether in 2006. structure and social affairs in the western In 2004 the government stopped subsidiz- areas. ing peasants indirectly by subsidizing the In 2003 government put forward the state-owned food supply and distribution Reviving the Northeast policy; deepened enterprises and began subsidizing peasants the reform of economic systems; promoted directly, which helped to stabilize the price the upgrading of industrial structure; accel- of grain and enhanced peasants' income. erated regional cooperation; sped up the Meanwhile, in education and medical transformation of resource-exhausted cit- care, government began to adopt a prefer- ies; strengthened ecological construction ential policy step by step. In 2006 govern- and environmental protection; quickened ment declared its intention to waive all of development in education, public health, the study and logistic fees for compulsory culture, and sports; and granted preferential education in the rural areas within two years policies in taxation, national fiscal invest- and pledged to provide poor students with ment, and introduction of foreign capital in free textbooks and to subsidize living costs the three northeastern provinces and some for those in boarding school. This policy was areas of Inner Mongolia. expanded gradually from the western areas In 2004 government explicitly launched to the middle and eastern areas. Finally, in the Central Rising policy, promulgating its 2004 the government experimented with a guiding documents in 2006. Moreover, the new type of medical cooperation system in national ETDZs and high-tech industry an attempt to reduce the burden of health development zones established since 2000 care for peasants. have favored the west (table 15.1). Finally, China's current household regis- With respect to urban and rural poli- tration system and the regionally segmented cies, China used to artificially depress social security system, together with poorly the price of agricultural products and to defined property rights of land, which limit restrict the interregional flow of labor the ability to trade land freely in the market, from rural to urban areas. After the reform, have become major obstacles to the inter- the household responsibility system was regional flow of labor. implemented in rural areas, the central- ized planning of the pricing and sale of Adjustment in the direction of agricultural produce was reformed, and fiscal transfers the purchase price of agricultural produce China's attempt to adjust interregional was enhanced. and urban-rural policies is clearly embod- Between 2004 and 2007, the government ied in the system of fiscal transfers. Figure stressed the problems affecting agriculture, 15.8 presents the change in the proportion rural areas, and peasants. As agricultural of eastern, middle, western, and the three produce came to be priced by the market, northeastern provinces in net fiscal trans- the government sought to narrow the urban- fers from the central government.7 After rural income gap. In 2000 government implementation of the Go West policy, experimented with reforms of the agricul- western areas accounted for a growing share tural tax and fees in Anhui province, extend- of central fiscal transfers between 2000 and ing these reforms to 16 provinces, cities, and 2002. autonomous regions in 2002. The reform Because provinces in these four major included canceling some administrative or areas are at different stages of development, institutional fees and governmental funds, figure 15.9 shows the relationship between reducing and then canceling all compulsory the area's share of net central fiscal transfers work, adjusting the policy of agricultural and its per capita GDP. Until 1998, wealthy Is China sacrificing growth when balancing interregional and urban-rural development? 253 provinces received more fiscal transfers from Figure 15.8 Share of net fiscal transfers in China from the central government, by geographic the central government than poor provinces, region, 1998­2004 but this relationship disappeared in 1999. 60% Since 2000, central government transfers have been directed to poor provinces. 50% The share of agriculture-related expendi- tures in total fiscal expenditure also changed, 40% as depicted in figure 15.10.8 Agricultural 30% expenditures increased remarkably in 2004. share This change appeared in 2002, as shown by 20% the fitted trend line, but was interrupted in 2003,perhaps as a result of the appearance of 10% SARS (severe acute respiratory syndrome). 0% Did more fiscal transfers bring higher 1998 1999 2000 2001 2002 2003 2004 growth rates? Figure 15.11 depicts the year relationship between fiscal transfers and East Middle West Northeast economic growth. In the figure, the horizon- tal axis represents the difference between the Sources: National Bureau of Statistics of China, China Finance Yearbook (various years); authors' calculations. Note: Data for Chongqing are missing for 2000. The share of eastern areas climbed abruptly in 2003 because Guang- provincial share of fiscal transfers in a par- dong obtained 38.71 percent of the central fiscal transfers in that year. Guangxi and Inner Mongolia are included in the western areas. ticular year and the average share in all years; the vertical axis represents the difference between the growth rate in a province in a Figure 15.9 Level of economic development and central fiscal transfers, 1998­2004 particular year and the average growth rate 1998 1999 2000 in all years. We de-mean the data to elimi- 10 10 10 nate the influence of time-invariant fixed effects of each province. Panel A of figure 5 5 5 15.11 depicts the relationship between cen- 0 0 0 tral fiscal transfers and economic growth in a given year. Because central fiscal transfers ­5 ­5 ­5 may be both the reason for and the result of 0 2 4 6 0 2 4 6 0 2 4 6 economic growth, panel A may not reflect 2001 2002 2004 10 10 10 how fiscal transfers affect economic growth. To alleviate the influence of two-way causal- 5 5 5 ity, in panel B of figure 15.11, the vertical 0 0 0 axis represents the de-meaned growth rate of the following year. Figure 15.11 shows that ­5 ­5 ­5 higher shares of central fiscal transfers did 0 2 4 6 0 2 4 6 0 2 4 6 not bring higher economic growth, at least per capita GDP in the short run. In other words, there is no Provincial share in central fiscal transfer Fitted values evidence to indicate that central fiscal trans- fers enhance development in any way other Sources: National Bureau of Statistics of China, China Finance Yearbook (various years); authors' calculations. than through income redistribution. For the moment, the market forces that drive inter- ment. Therefore, China offers a good case regional inequality may dominate govern- for studying industrial agglomeration and mental efforts to equalize regional income. regional economic development in the context of globalization, urbanization, and industrialization. China's industry is experi- Conclusions and policy encing agglomeration,with industry becom- implications ing highly concentrated in the coastal areas, The starting point of China's reform and especially the Yangtze River delta, the Pearl opening up was an economy dominated River delta, and the Bohai Bay area. Indus- for many years by a planned economic sys- trial agglomeration has boosted economic tem and interregionally balanced develop- growth but also exacerbated interregional 254 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 15.10 Changes in the share of agricultural expenditures, 1998­2004 economic growth. The Chinese government has realized the importance of balancing 8% 7.54% 7.26% urban-rural and interregional development 7% 6.77% 6.62% and has begun to invest more in lagging 6.23% 6.43% inland and rural areas by means of fiscal 6% 5.85% transfers. However, there is no evidence that 5% fiscal transfers by the central government have promoted economic growth, perhaps 4% share because market forces have dominated the 3% development-balancing function of fiscal transfers or because fiscal transfers have only 2% been able to alleviate income gaps, not boost 1% economic development. Therefore, interre- gional and urban-rural income gaps are still 0% growing. How to balance interregional and 1998 1999 2000 2001 2002 2003 2004 year urban-rural development is a challenging problem. Sources: National Bureau of Statistics of China, China Finance Yearbook (various years); authors' calculations. To sustain economic growth while bal- Figure 15.11 Central fiscal transfers and economic growth ancing interregional and urban-rural devel- opment, the government should consider A. Growth rate (percent) B. Growth rate the following year (percent) adjusting its policies in the following ways: 10 10 · Stop market segmentation and reform 8 8 the household registration and land (%) 6 6 property systems in rural areas to pro- year mote interregional and rural-to-urban 4 4 (%) next labor migration, improve urbanization, 2 2 the and enhance industrial agglomeration, rate of 0 especially toward the Yangtze River delta, ­2 ­1 00 1 2 ­2 ­1 0 1 2 rate Pearl River delta, and Bohai Bay area. growth ­2 ­2 Reasonable labor migration is beneficial ­4 growth ­4 for taking advantage of scale economies and improving the amount of resources ­6 ­6 per capita in lagging and rural areas; it ­8 ­8 also helps to narrow interregional and share in fiscal transfer (%) share in fiscal transfer (%) urban-rural income gaps. · Note: Data are missing for Chongqing in 2000, which means that the fiscal transfer data for this year are not Promote the interregional and urban- comparable with those of other years. Shares of Beijing in 2002 and Guangdong in 2003 are outliers and thus are excluded from the two figures. rural evenness of social services, rather than simply making fiscal transfers and urban-rural income disparities. Mean- or investing directly in industries in while, two other forces have kept industrial which inland areas have no comparative agglomeration from advancing further: advantages. market segmentation by local govern- · Emphasize investment in human capital ments and the household registration and and infrastructure in lagging and rural land property systems in rural areas, which areas. When labor cannot move freely, restrict the flow of labor. The size of cities in such investment would help to create China typically is controlled, and differences the conditions for long-term economic in their scale are small, stunting economic growth and enhance economic agglom- development in the long run. eration toward coastal areas. With the development of industrial · Address urban-rural income gaps, espe- agglomeration, interregional and urban- cially in inland areas. This could substan- rural income gaps are expanding, and this tially narrow interregional inequality, could have a negative impact on sustainable while narrowing urban-rural gaps. Is China sacrificing growth when balancing interregional and urban-rural development? 255 Notes expenditures for forestry, water conservation, and weather in 2003 and 2004. Zhao Chen is a professor with the Institute for Industrial Development and the China Center for Economic Studies at Fudan University. Ming Lu is References a professor in the Department of Economics, the Alesina, Alberto, and Dani Rodrik. 1994. Employment and Social Security Research Center, "Distributive Politics and Economic Growth." and the China Center for Economic Studies at Quarterly Journal of Economics 109 (2): Fudan University.The authors would like to thank 465­90. Indermit Gill and Yukon Huang of the World Bank for their helpful discussions and Shiqin Au, Chun-Chung, and J. Vernon Henderson. Jiang, Xiaofeng Liu, Zheng Xu, Jingmin Chen, and 2006a."Are Chinese Cities Too Small?" Hong Gao for their research assistance. Review of Economic Studies 73 (3): 549­76. 1. If not specifically mentioned or noted, ------. 2006b."How Migration Restrictions provincial panel data in this chapter come from Limit Agglomeration and Productivity in National Bureau of Statistics of China (1999, China." Journal of Development Economics 2005) and National Bureau of Statistics of China, 80 (2): 350­88. China Statistical Yearbook (2006). City panel data Beijing Bureau of Statistics. 1999. 50 Years of come from Beijing Bureau of Statistics (1999) Cities in New China. Beijing: BBS. and National Bureau of Statistics of China, China City Statistical Yearbook (various years). Bénabou, Roland. 1996."Inequality and 2. In the period of the National People's Con- Growth." NBER Macroeconomics Annual 1996, gress and the National Political Consultative Con- eds. Ben Bernanke and Julio Rotemberg, pp. gress of 2007, the number of migrant workers was 11­76. Cambridge, MA: MIT Press. widely quoted as amounting to 150 million. Benhabib, Jess, and Aldo Rustichini. 1996. 3. Data are from the Shanghai Statistical "Social Conflict and Growth." Journal of Bureau. Economic Growth 1 (1): 129­46. 4. The figure excludes the autonomous Benjamin, Dwayne, Loren Brandt, and John municipalities: Shenzhen for its abnormal per Giles. 2004."The Dynamics of Inequality capita GDP and Putian, in Fujian province, for and Growth in Rural China: Does Higher its abnormal population density. Inequality Impede Growth?" Working Paper, 5. China has frequently adjusted the pref- University of Toronto. erential policies for special economic zones (SEZs). The Income Tax Law of People's Repub- Chen, Min, Qihan Gui, Ming Lu, and Zhao lic of China for enterprises, enacted on January Chen. 2007."Economic Opening and Domes- 1, 2008, unifies the tax system between foreign- tic Market Integration." In China: Linking invested enterprises and domestically funded Markets for Growth, eds. Ross Garnaut and enterprises, which is expected to end the last Ligang Song, pp. 369­93. Canberra: Asia preferential policy for SEZs. Pacific Press. 6. Data are from the State Department's Chen, Zhao, Yu Jin, and Ming Lu. 2006."Eco- Office of the Development of the West, http:// nomic Opening and Industrial Agglomera- www.chinawest.gov.cn. tion in China." In Asian Regional Economic 7. The net central transfer is central-to-local Integration from the Point of View of Spatial subsidy less local-to-central contribution. Tradi- Economics, eds. Masahisa Fujita and Akifumi tionally, China was divided into east, middle, and Kuchiki, pp. 97­132. Joint Research Program west areas. The reason for separating the three Series 138. Chiba, Japan: Institute of Develop- northeastern provinces is to check the influence ing Economies. of the Reviving the Northeast policy. Among the three northeastern provinces, Liaoning is usually Démurger, Sylvie. 2001."Infrastructure Devel- included in the east, while Heilongjiang and Jilin opment and Economic Growth: An Expla- are included in the middle. nation for Regional Disparities in China?" 8. In the budget and final accounts of every Journal of Comparative Economics 29 (1): province, three categories of expenditures are 95­117. related to agriculture: production expenditures Démurger, Sylvie, Jeffrey D. Sachs, Wing Thye supporting rural areas; agricultural comprehen- Woo, Shuming Bao, Gene Chang, and sive development expenditures and institutional Andrew Mellinger. 2002."Geography, Eco- fees for agriculture, forestry, water conserva- nomic Policy, and Regional Development in tion, and weather before 2002; and agricultural China." Asian Economic Papers 1 (1): 146­97. 256 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Deng, Xiaoping. 1993. Selected Works of Deng National Bureau of Statistics of China. 1999. Xiaoping [in Chinese]. Vol. 3. Beijing: People's Comprehensive Statistical Data and Materials Publishing House. for 50 Years of New China. Beijing: China Sta- Ding, Jinhong, Zhenyu Liu, and Danming tistical Publishing House. Cheng. 2005."Regional Difference and ------. 2005. Comprehensive Statistical Flow Direction of Migration in China [in Data and Materials for 55 Years of New Chinese]." Geography Journal [Dili Xuebao] China. Beijing: China Statistical Publishing 60 (1): 106­14. House. Fishman, Arthur, and Avi Simhon. 2002."The ------. Various years. China City Statistical Division of Labor, Inequality, and Growth." Yearbook. Beijing: China Statistical Publish- Journal of Economic Growth 7 (2): 117­36. ing House. Fujita, Masahisa, J. Vernon Henderson, Yoshit- ------. Various years. China Finance Yearbook. sugu Kanemoto, and Tomoya Mori. 2004. Beijing: China Statistical Publishing House. "Spatial Distribution of Economic Activities ------. Various years (2005­06). China in Japan and China." In Handbook of Urban Statistical Yearbook. Beijing: China Statistical and Regional Economics, vol. 4, eds. J. Vernon Publishing House. Henderson and Jacques-Franēois Thisse, pp. 2911­77. Amsterdam: North-Holland. Persson, Torsten, and Guido Tabellini. 1994."Is Inequality Harmful for Growth? Theory and Galor, Oded, and Joseph Zeira. 1993."Income Evidence." American Economic Review 84 (3): Distribution and Macroeconomics." Review of 600­21. Economic Studies 60 (1): 35­52. Gill, Indermit, and Homi Kharas. 2007. An East Raiser, Martin. 1998."Subsidising Inequality: Asian Renaissance: Ideas for Economic Growth. Economic Reforms, Fiscal Transfers, Washington, DC: World Bank. and Convergence across Chinese Provinces." Journal of Development Studies 34 (3): 1­26. Hussain, Arthur, Peter Lanjouw, and Nicholas Stern. 1994."Income Inequalities in China: Ravallion, Martin. 1998."Does Aggregation Evidence from Household Survey Data." Hide the Harmful Effects of Inequality on World Development 22 (12): 1947­57. Growth?" Economics Letters 61 (1): 73­77. Kanbur, Ravi, and Xiaobo Zhang. 1999."Which Ravallion, Martin, and Shaohua Chen. 2007. Regional Inequality? The Evolution of Rural- "China's (Uneven) Progress against Poverty." Urban and Inland-Coastal Inequality in Journal of Development Economics 82 (1): China from 1983 to 1995." Journal of Com- 1­42. parative Economics 27 (4): 686­701. Rozelle, Scott. 1994."Rural Industrialisation Khan, Azizur R., and Carl Riskin. 1998."Income and Increasing Inequality: Emerging Patterns Inequality in China: Composition, Distribu- in China's Reforming Economy." Journal of tion, and Growth of Household Income, 1988­ Comparative Economics 19 (3): 362­91. 1995." China Quarterly 157 (June): 221­53. Tsui, Kai-yuen. 1993."Decomposition of Chi- Li, Shi. 2003."A Survey and Perspective on Chi- na's Regional Inequality." Journal of Compara- na's Income Distribution [in Chinese]." China tive Economics 17 (3): 600­27. Economic Quarterly [Jingjixue Jikan] Wan, Guanghua. 1998."An Empirical Study 2 (2): 379­403. on the Change in Interregional Income Lu, Jiangyong, and Zhigang Tao. 2006."Regional Disparity of Rural Residents [in Chinese]." Industrial Agglomeration and International Economic Research Journal [Jingji Yanjiu] Comparison [in Chinese]." Economic Research 5: 36­41. Journal [Jingji Yanjiu] 3: 103­14. ------. 2007."Understanding Regional Poverty Lu, Ming, and Zhao Chen. 2006. Market Integra- and Inequality Trends in China: Methodolog- tion and Industrial Agglomeration in China's ical Issues and Empirical Finding." Review of Regional Economic Development [in Chinese]. Income and Wealth 53 (1): 25­34. Shanghai: Shanghai People's Press. Wan, Guanghua, Ming Lu, and Zhao Chen. Ma, Shuanyou, and Hongxia Yu. 2003."Fiscal 2006."The Inequality-Growth Nexus in the Transfer and Convergence of Regional Econ- Short and Long Run: Empirical Evidence omy [in Chinese]." Economic Research Journal from China." Journal of Comparative Econom- [Jingji Yanjiu] 3: 26­33. ics 34 (4): 654­67. Is China sacrificing growth when balancing interregional and urban-rural development? 257 ------. 2007."Globalization and Regional Newspaper of Social Science [Shehui Kexue Income Inequality: Empirical Evidence from Bao], August 9. within China." Review of Income and Wealth Yang, Dennis Tao. 1999."Urban-Biased Poli- 53 (1): 35­59. cies and Rising Income Inequality in China." Wen, Mei. 2004."Relocation and Agglomeration American Economic Review Papers and Pro- of Chinese Industry." Journal of Development ceedings 89 (May): 306­10. Economics 73 (1): 329­47. Yao, Shujie, and Liwei Zhu. 1998. "Understand- World Bank. 1996. World Development Report ing Income Inequality in China: A Multi- 1996: From Plan to Market. New York: Oxford Angle Perspective." Economics of Planning 31 University Press. (2­3): 133­50. ------. 1997. Sharing Rising Incomes: Dispari- Zhao, Renwei, and others, eds. 1999. A Research ties in China. Washington, DC: World Bank. on China's Income Distribution: Income Distri- bution in Economic Reform and Development Yan, Wei, and Nan Jiang. 2007."Land Saving Is [in Chinese]. Beijing: China Finance and Eco- Necessary during Urbanization [in Chinese]." nomics Press. Industrial agglomeration and economic performance in transitional China Canfei He 16 According to mainstream economic theo- Sectors and regions that have undergone ries, comparative advantages and agglom- economic liberalization may be prone to eration economies are responsible for the the formation of agglomeration economies geographic clustering of industries (Ellison and would benefit significantly from indus- and Glaeser 1997). The forces driving indus- trial agglomeration. This chapter empiri- trial agglomeration in China might differ, cally investigates the evolution and extent however, because China has experienced two of industrial agglomeration in China. To fundamental changes since the late 1970s: the provide empirical justification for the geo- c h a p t e r transition from a command economy to a graphic clustering of Chinese industries, market-driven economy and the transforma- it also explores the relationship between tion from a closed economy to an open econ- industrial agglomeration and labor produc- omy. Such a process of economic transition tivity across and within industries. has been conceptualized as a triple process of decentralization, marketization, and global- Theoretical understanding ization, which has had a profound impact on of industrial agglomeration spatial development in China (He, Wei, and in China Xie 2008; Wei 2000). Industrial restructuring This section presents the theories that seek in China, like the process of economic tran- to understand economic marketization, glo- sition, has been a gradual, partial, spatially, balization,and decentralization as they apply and structurally uneven process. Chinese to industrial agglomeration. The following enterprises have been gradually exposed to section has detailed discussion regarding market forces. Liberalization of investment how the triple process of economic transi- and trade has opened the Chinese economy tion affects industrial agglomeration. to global competition. Marketization and globalization have encouraged plants to Economic marketization locate in areas with comparative advantages China's economic reform seeks to build and to exploit agglomeration economies. a market-oriented economy and to allow Meanwhile, decentralization has granted market forces to allocate resources. In the local governments more authority and command economy, governments distrib- responsibilities to develop local economies, uted resources. Literally, there were no well- resulting in fierce interregional competi- functioning markets. Economic geography tion and giving rise to local protectionism was heavily shaped by socialist ideology and and a rational imitation strategy in pursuing national defense considerations. Industrial economic development. As a consequence, location failed to match comparative advan- China is experiencing both significant cen- tages, because governments located indus- tripetal forces as well as centrifugal forces of tries based on shifting social, political, and industrial agglomeration. military considerations. The self-enclosed Industrial agglomeration and economic performance in transitional China 259 nature of the economic system ruled out agglomeration. As a consequence, indus- any horizontal economic flows or indus- tries and regions that have experienced eco- trial specialization (Zhao and Zhang 1999). nomic liberalization will be favorable to the As the economic transition has proceeded, formation of industrial agglomeration. As market forces have been progressively the economic transition proceeds, Chinese introduced, and limits on factor mobility industries, especially those driven by market and commodity exchanges have been grad- forces, will be increasingly agglomerated in ually lifted. Nonstate capital and private a few regions. firms have been playing an important role in local development. Economic globalization In the context of market economies, China has participated in economic global- neoclassical trade models, new trade mod- ization by trading with other economies and els, and new economic geography models by attracting foreign investment. Trade lib- are the theoretical inquiries underpinning eralization broadens the scope of industrial industrial location (Brülhart 1998). In the specialization along the lines of comparative neoclassical world, industrial agglomeration advantages and enhances the importance of is driven by exogenous endowments such as accessibility to international markets. Trade technology, labor, and natural resources. liberalization also provides trading firms with Industries are heavily agglomerated in loca- incentives to exploit scale economies and tions with matched comparative advan- heighten international competitiveness. In tages. The reduction of trade barriers makes China, labor migration resulting from relax- regions specialize in their production based ation of the household registration system on comparative advantages (Kim 1995). has allowed trading establishments to cluster In new trade models, internal economies along the coastal region and to benefit from of scale provide regions with incentives to the best use of cheap resources (He, Wei, and specialize even in the absence of differences Xie 2008). Moreover, Chinese trading enter- in resource endowments and encourage prises take full advantage of backward and firms to agglomerate their production in a forward business linkages through a deeper few locations. Economic activities concen- division of labor, significantly cutting pro- trate to realize scale economies, locating in duction and transaction costs (Wang 2001). large consumer markets to minimize trans- Fujita and Hu (2001) find that increases in portation costs (Krugman 1980). Regional exports have reinforced industrial agglom- integration allows underlying geographic eration in China, concentrating industries in advantages to play a greater role, encour- the coastal region, which is close to interna- aging industrial agglomeration. In the new tional markets,has locational advantages,and economic geography models, geographic enjoys the first-mover advantage of economic concentration is driven by the interaction globalization. However, as Krugman and of transportation costs, internal scale econo- Elizondo (1996) argue, trade liberalization mies, and labor mobility (Fujita and Thisse may significantly weaken the role of localized 1996; Krugman 1991). Demand linkages industrial linkages in promoting industrial represent incentives for producers to locate agglomeration because trading establish- close to buyers, whereas cost linkages gener- ments rely on external linkages. In China, ate incentives for consumers to locate close many assemblers focus on labor-intensive to suppliers (Venables 1996). functions, relying heavily on imported mate- Beyond the agglomeration economies rials and intermediate goods. The localized from industrial linkages, other sources of industrial linkages may not play the expected external economies are also driving plants role in industrial agglomeration. to cluster. Marshall (1898) pointed to the Foreign direct investment (FDI) has sig- pooling of markets for specialized skilled nificantly shaped the spatial structure of Chi- labor, the development of subsidiary trade nese industries. First, foreign establishments and suppliers of intermediate inputs, and are disproportionately agglomerated in the the easy flow of information and ideas coastal region and in strong industrial bases among firms as forces driving industrial (He 2002, 2003, 2006; Head and Ries 1996). 260 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Second, foreign enterprises create demand the key industries chosen by almost all of the for locally produced intermediate inputs and coastal provinces and many central prov- improve the efficiency of the whole sector, inces. The industrial duplication certainly making domestic producers more profitable discouraged the geographic concentration (Markusen and Venables 1999). Component of Chinese industries. To support the devel- sourcing in China is an important consider- opment of key industries, local governments ation for foreign firms because of local con- have taken serious measures to concentrate tent requirements (Belderbos and Carree them in development zones. 2002; Head and Ries 1996). Third, major Third, fiscal decentralization has trig- multinational corporations often bring a large gered fierce interprovincial competition number of suppliers to the host economy, for economic and political performance, facilitating industrial agglomeration. resulting in a rational imitation strategy of industrial policies (He and Zhu 2007). Economic decentralization Competition among provinces provides Unlike governments in the developed market incentives to replace poorly chosen strate- economies, central and local governments gies with strategies that appear to succeed in China are still rather powerful and influ- elsewhere. The economic-oriented evalua- ential in economic development. Economic tion system for local officials and a judicious transition in China has resulted in consider- combination of local autonomy, fiscal incen- able administrative decentralization from the tives, and hard budget constraints have cre- central to local governments.As a result,local ated a framework leading local governments governments now have the primary responsi- to follow the leaders in industrial develop- bility and authority for local economic devel- ment. Thun (2004) observes that decentral- opment (Qian and Weingast 1997). ization leads local governments to converge During the period of economic reform, on successful development policies through the government's intervention in economic a process of rational imitation. Rational development may have discouraged the imitation of successful industrial policies geographic concentration of Chinese indus- encourages local governments to attract tries, especially at the provincial level. First, duplicate industries that they believe could the central and local governments have rapidly improve local revenues or promote established a large number of economic and local economic growth. technological development zones (ETDZs) Finally, fiscal decentralization instinc- and high-tech industrial development zones tively and explicitly has emphasized autar- to attract domestic and foreign investments. chic development because the localities have There are more than 50 development zones had to self-finance their own budgets and across the major Chinese cities. The tar- their own development (Zhou 2000). Fiscal geted industries in those zones and parks decentralization has created conditions that are fairly similar, including electronics, encourage regionalism: disappearance of medicine, equipment, and other high- the traditional umbrella, unfairness to the tech industries. The centrally and locally poor regions, territorial segmentation and administered development zones promote confrontation, central-local vertical con- the geographic agglomeration of advanced frontation, and failure of spatial programs industries but discourage the industrial of specialization (Zhao and Zhang 1999). localization of labor-intensive industries. At the macroeconomic level, fiscal decen- Second, local governments intend to tralization has provided local governments duplicate industries highlighted in the with incentives to protect local industries, national Five-Year Plans and in the national significantly contributing to economic and industry-specific development plans. For revenue growth (Lee 1998; Young 2000). instance, the Tenth Five-Year Plan stressed As control over factor allocations has loos- the development of food processing,machin- ened, local governments have sought to ery and equipment, automobile, and high- capture these rents by developing high- tech industries such as electronics, biological margin industries. Continued reform and engineering, and medicine. These are also growing interregional competition among Industrial agglomeration and economic performance in transitional China 261 duplicative industries threaten the profit- Industrial agglomeration ability of these industrial structures, lead- and industrial specialization ing local governments to impose a variety in China of interregional barriers to trade (Young Following Wen (2004), this chapter applies 2000). The Development Research Center the widely used Gini coefficient to quantify of the State Council (DRCSC 2004) ranks industrial agglomeration of Chinese indus- the highly protected industries as follows: tries. The closer the distribution of industry tobacco, food, medical and pharmaceutical i is to a uniform distribution, the smaller the products, construction, agriculture, bever- index is. If an industry is equally distributed ages, real estate, power, gas and water pro- across all regions, the index will be equal duction, post and telecommunications, and to 0. An index close to 1 suggests that an machinery equipment. The least-protected industry is entirely concentrated in a region. industry is nonmetal mineral products, fol- Data on Chinese manufacturing industries lowed by cultural education and sporting are collected from various issues of China goods, chemical fibers, ferrous metal smelt- Industry Economy Statistical Yearbook and ing, petroleum refining and coking, rubber Annual Report of Chinese Industrial Statis- and plastic products, electric machin- tics and from the China Economic Census ery and equipment, instruments, meters Yearbook 2004. and office machinery, nonferrous metal smelting and pressing, and leather prod- ucts. Therefore, economic transition has Industrial agglomeration of fragmented domestic markets, distorted Chinese manufacturing industries regional production away from the pat- To investigate the overall temporal trend of terns of comparative advantage, and dis- industrial agglomeration in China, I have couraged the geographic concentration of computed the yearly weighted average of Chinese industries. the Gini coefficient of employment, value Overall, the spatial restructuring of Chi- added, and gross output during the period nese industries is the result of interactions of 1980­2004. of centripetal forces and centrifugal forces As shown in figure 16.1, Chinese manu- of industrial agglomeration. On the one facturing industries have undergone signifi- hand, there are driving forces for indus- cant spatial transformations since the early trial agglomeration as marketization and 1980s and have been more geographically globalization proceed. Highly liberalized agglomerated during the period of eco- and globalized industries tend to concen- nomic transition. The weighted 1980 and trate in regions with matched comparative 2004 Gini coefficients of gross output are and locational advantages and to become 0.51 and 0.64, while the weighted Gini coef- increasingly agglomerated in the coastal ficients of employment are 0.41 and 0.58, provinces. Geographic clustering of indus- respectively. However, industrial output tries generates substantial cost savings and experienced spatial dispersal in the 1980s leads to higher labor productivity. On the and has become increasingly concentrated other hand, the economic transition has since the early 1990s.Industrial employment created a market that preserves federalism, was much less agglomerated than industrial which provides local governments with output in the 1980s but has become more strong incentives to protect local industries concentrated since the early 1980s. Under from external competition and to imitate the influence of socialist ideology and egali- successful industrial policies. Local protec- tarian ideas during Mao's era, China's indus- tionism and rational imitation run counter trialization policy favored the traditional to the geographic agglomeration of Chi- industrial bases, tilted toward new indus- nese industries. The geographic dispersion trial cities in the interior, and sought to of industries loses scale economies and achieve full employment. Consequently, at makes it difficult for related enterprises the beginning of China's economic reform, to exploit external economies, leading to Beijing, Gansu, Heilongjiang, Hubei, Jilin, lower labor productivity. Liaoning, Shanghai, Sichuan, and Tianjin 262 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 16.1 Geographic concentration of manufacturing industries in China, 1980­2004 including Fujian, Guangdong, Jiangsu, Shandong, Shanghai, and Zhejiang. For 0.65 most industries, the share of industrial out- put of the top four provinces increased sub- stantially. Industries such as leather and fur 0.60 products, cultural education and sporting goods, chemical fiber, telecommunications 0.55 and electronic equipment, instruments, and meters concentrated more than 70 percent of coefficient industrial output in the top four provinces. 0.50 Gini The share of industrial output of the top four provinces ranged from less than 30 to 77 percent, indicating that industries 0.45 differed substantially in the extent of geo- graphic agglomeration. Table 16.1 presents the Gini coefficient of industrial output 0.40 for all two-digit manufacturing industries 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 year in select years. The concentration indexes Concentration of employment Concentration of value-added indeed differ significantly across industries. Concentration of gross output The Gini coefficient in 1980 ranged from 0.37 to 0.73, and it ranged from 0.46 to 0.83 Sources: Author's calculations based on data in SSB (2006); SSB, Annual Report of Chinese Industrial Statistics in 2004. In 1980 the most-agglomerated (various years); and SSB, China Industry Economy Statistical Yearbook (various years). industries included chemical fiber, cultural education and sporting goods, petroleum were the major industrial bases of China.The refining and coking, telecommunications top four provinces produced less than 50 per- and electronic equipment, ferrous metal cent of industrial output in most industries. smelting, and pressing, while the least con- With the shift of government policies centrated industries were food processing toward a market-driven economy and the and manufacturing, furniture making, opening up of the coast, Fujian, Guangdong, printing and copying, nonmetal mineral Jiangsu, Shandong, Shanghai, and Zhejiang products, leather and fur products, and have gradually taken the lead in attracting paper making and paper products. The investment, labor, advanced technology, most-agglomerated industries were capi- and firms. The rapid growth of the coastal tal intensive and had strong internal scale region and the relative decline of old indus- economies, while the dispersed industries trial bases and interior provinces brought a were resource-based or resource-processing decrease in the Gini coefficients of industrial industries. By 2004, cultural education output. Labor migration from the interior and sporting goods, telecommunications provinces to the coastal region stimulated and electronic equipment, and chemi- industrial employment to agglomerate in cal fiber were much more agglomerated the coast. and remained as the most-agglomerated Further economic reform has made the industries. The agglomeration of garments coastal region an engine of remarkable eco- and other fiber, leather and fur products, nomic growth in China, resulting in strong textiles, instruments and meters, furniture polarization effects. Gini coefficients of making, electrical machinery and equip- industrial output and employment have ment, plastic products, and metal mineral been growing since the early 1990s (figure products also gained momentum, with Gini 16.1). The Gini coefficient of gross indus- coefficients greater than 0.70. Many highly trial output increased from 0.50 in 1990 agglomerated industries in 2004 were also to 0.64 in 2004, and the Gini coefficient of highly globalized, exporting substantially industrial employment grew from 0.44 to or using a large amount of FDI. The most 0.58. Industrial output in most industries dispersed industries included nonferrous was concentrated in the coastal provinces, metal smelting and pressing, medical and Industrial agglomeration and economic performance in transitional China 263 Table 16.1 Gini coefficient of manufacturing industries in China in select years, 1980­2004 1980 1990 2004 Code Gini Rank Gini Rank Gini Rank Food processing and manufacturing 0.37 26 0.41 25 0.55 21 Beverage manufacturing 0.43 20 0.44 22 0.49 24 Tobacco processing 0.49 11 0.50 12 0.57 19 Textiles 0.58 6 0.58 5 0.75 7 Garments and other fiber 0.47 18 0.55 7 0.77 4 Leather and fur products 0.42 22 0.53 9 0.77 5 Timber processing 0.49 12 0.49 14 0.61 16 Furniture making 0.39 25 0.49 15 0.73 8 Paper making and paper products 0.42 21 0.42 23 0.68 13 Printing and copying 0.40 24 0.41 24 0.64 15 Cultural education and sporting goods 0.72 2 0.71 1 0.83 1 Petroleum refining and coking 0.70 3 0.59 4 0.54 23 Chemical materials and products 0.46 19 0.44 20 0.57 17 Medical and pharmaceutical products 0.47 17 0.45 18 0.47 25 Chemical fiber 0.73 1 0.63 2 0.79 3 Rubber products 0.49 13 0.45 19 0.69 12 Plastic products 0.53 9 0.55 6 0.72 10 Nonmetal mineral products 0.41 23 0.44 21 0.57 18 Ferrous metal smelting and pressing 0.58 5 0.51 11 0.54 22 Nonferrous metal smelting and pressing 0.54 8 0.46 26 Metal mineral products 0.48 14 0.50 13 0.71 11 General- and specific-purpose machinery 0.47 16 0.47 17 0.64 14 Transportation equipment 0.48 15 0.49 16 0.55 20 Electrical machinery and equipment 0.52 10 0.54 8 0.73 9 Telecommunications and electronic equipment 0.61 4 0.60 3 0.81 2 Instruments and meters 0.57 7 0.52 10 0.76 6 pharmaceutical products, beverage manu- telecommunications and electronic equip- facturing, petroleum refining and coking, ment and cultural education and sporting ferrous metal smelting and pressing, food goods were greater than 60 percent. In processing and manufacturing, transporta- the beginning of economic transition, the tion equipment manufacturing, and tobacco spatial pattern of the select industries was processing,among which petroleum refining rather diverse, with some agglomerated and and coking and ferrous metal smelting and others dispersed, because their provincial pressing were the most-concentrated indus- distributions were determined by central tries in 1980. The dispersed industries were and local governments. As the open-door either resource based or domestic market policy was successfully implemented, the oriented. Some were strongly favored and international market and foreign capital protected by local governments, because drove the globalized industries to the coastal they are strategic and profitable, such as provinces, particularly Fujian, Guangdong, tobacco processing, transportation equip- Jiangsu, Shandong, Shanghai, and Zhejiang ment, beverage manufacturing, and medical (see figure 16.3). Since the early 1990s, the and pharmaceutical products. globalized industries have been increasingly I now compare the temporal trends of agglomerated and more concentrated than industrial agglomeration for different types the domestic market­oriented industries, of industries to shed light on some influ- with Gini coefficients greater than 0.70. ential factors of industrial agglomeration. Figure 16.4 presents the pattern of Figure 16.2 shows the temporal change in industrial agglomeration of select domestic geographic agglomeration of select global- market­oriented industries, which exported ized industries. In 2004 all of these indus- less than 10 percent of their gross output tries exported more than 55 percent of their and had little foreign capital. In 1980 there gross output and had more than 45 percent were no significant differences in the spa- of foreign capital. The two percentages in tial pattern of globalized and less-globalized 264 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 16.2 Geographic agglomeration of industrial output of globalized industries in China, industries. For example, during the past two 1980­2004 decades, the Gini coefficients for medical and pharmaceutical products and beverages 0.85 were smaller than 0.50. Unlike the trend of 0.80 aggregate industries, favored and protected industries have not experienced significant 0.75 increases in industrial agglomeration since 0.70 the early 1990s. The tobacco industry expe- rienced significant concentration before the 0.65 mid-1990s, but it started to disperse in 1994, 0.60 when the central government introduced coefficient the new tax-sharing system, which further Gini0.55 hardened the local fiscal budgets (Zhou 0.50 2000). The machinery industry, character- ized by strong scale economies and strong 0.45 industrial linkages, was agglomerated dur- 0.40 ing the 1990s but remains at a relatively low level of agglomeration compared with the 0.35 least-protected industries. 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 year The location of least-favored and Electronics and telecommunication equipment Furniture making -protected industries was driven by market Instruments and meters Garments and other fiber products forces, and they have experienced a sig- Cultural education and sports goods Leather and fur products nificant polarizing process since the 1990s. Cultural education and sporting goods had industries. However, spatial restructuring of a Gini coefficient of 0.81 in 2004, with the domestic market­oriented industries was less top four provinces of Guangdong, Jiangsu, significant, and some became slightly dis- Shandong, and Zhejiang contributing 77 persed as globalized industries became more percent of gross industrial output. The Gini agglomerated. They were also less agglomer- coefficient of chemical fiber in 2004 was 0.79, ated than the globalized industries, with Gini with the top four provinces of Jiangsu, Shan- coefficients less than 0.60,because they served dong, Shanghai, and Zhejiang responsible local markets and valued the accessibility of for 75 percent of gross industrial output. As the domestic market.As shown in figure 16.5, expected, the spatial distribution of favored substantial shares of industrial output of the and protected industries was much more least-globalized industries were located in the dispersed, with significant presence in the central provinces. central and coastal provinces, while the least- During the economic transition,industrial favored industries were heavily agglomerated agglomeration in China has been associated along the coast (see figures 16.8 and 16.9). with how governments treat the industries. Figure 16.6 presents the spatial pattern of Causes of industrial agglomeration industries favored or protected by local gov- in China ernments, including food, beverage, tobacco, Chinese manufacturing industries have medical and pharmaceutical products, experienced a U-shape spatial restructuring and machinery and transportation equip- process, which is consistent with the tempo- ment. Figure 16.7 shows the spatial pattern ral pattern of interregional income inequal- of industries not particularly protected or ity. I argue that the spatial shift of Chinese favored by local governments, including manufacturing industries toward the coastal chemical fibers, rubber and plastic products, region has led to widening interregional cultural education and sporting goods, elec- inequality. The accelerating agglomera- trical machinery and equipment,and instru- tion of Chinese industries since the 1990s ments and meters. Overall, protected and has been the result of economic transition. favored industries were more geographically Specifically, the triple process of marketiza- dispersed than less-favored and -protected tion, globalization, and decentralization has Industrial agglomeration and economic performance in transitional China 265 Figure 16.3 Provincial distribution of industrial output of highly globalized industries in China Globalized Industries Electronics and Telecommunication Equipment Instruments and Meters Garments and Other Fiber Products Leather and Fur Products driven the spatial reorganization of Chinese Industries dependent on agricultural and industries. mining resource inputs are significantly First, the accelerating marketization pro- dispersed, indicating that industries based cess since 1992 has lifted the limits on fac- on immobile resources follow the resources tor mobility and commodity exchanges and (He, Wei, and Xie 2008). Marketization also stimulated labor and capital mobility and allows Chinese enterprises to exploit scale interregional trade, providing incentives economies. For example, Bai and others for Chinese enterprises to follow the line (2004) find that industries with large aver- of comparative and locational advantages. age firm size were significantly more con- A large number of workers migrated to the centrated during 1985­97, suggesting that coastal provinces of Fujian, Guangdong, internal scale economies drive industrial Jiangsu, Shandong, Shanghai, and Zhejiang, agglomeration in China. In addition, strong promoting industrial agglomeration in the interindustrial linkages are also positively coastal region. Meanwhile, interprovincial associated with industrial agglomeration, trade of commodities also stimulated Chi- indicating the importance of pecuniary nese provinces to specialize in production externalities in industrial agglomeration based on local resources. For instance, Fan (He, Wei, and Xie 2008). In a word, mar- and Scott (2003) relate the concentration ket forces have been a fundamental force in indexes of Chinese manufacturing indus- reorganizing China's economic geography tries to capital per labor and find that labor- by allowing comparative advantages and intensive industries are more agglomerated. agglomeration economies to drive industrial 266 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 16.4 Geographic agglomeration of domestic market­oriented industries in China, and fiscal revenues. Industries character- 1980­2004 ized by high tax rates and profit margins 0.85 are favored and protected by local govern- ments, resulting in low levels of industrial 0.80 agglomeration. Studies find less geographic 0.75 concentration in two-digit industries where the past tax rates, profit margins, and share 0.70 of state capital are high (Bai and others 2004; He, Wei, and Xie 2008). The negative 0.65 relationship between industrial agglomera- 0.60 tion and profit margins and tax rates also coefficient holds for three-digit manufacturing indus- Gini0.55 tries (see figure 16.11). The Gini coefficients 0.50 of three-digit industries are significant and negatively related to the ratio of income tax 0.45 and value added tax to sales revenues, with 0.40 a Pearson's correlation coefficient of -0.45. The Pearson's correlation coefficient for 0.35 the ratio of gross profits to sales revenues 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 year is -0.35. Due to interregional competition, local governments often duplicate profitable Petroleum refining and coking Ferrous metal smelting and pressing Chemical materials and products Nonferrous metal smelting and pressing and strategic industries that are successful Nonmetal mineral products in other provinces. In addition, local gov- ernments have strong incentives to protect state-owned enterprises under their admin- location and stimulate competitive indus- istrations, which are their base of political tries to agglomerate. power and their source of private benefits Second, China has successfully inte- as well as fiscal revenues (Bai and others grated its economy with the international 2004). Industries with a high share of state- market in recent decades, and global forces owned capital are also less agglomerated have been critical to reshaping China's (see figure 16.12). The correlation coeffi- industrial geography. Two-digit industries cient between industrial Gini coefficients with more exports and foreign capital are and share of state-owned capital is -0.37 more agglomerated and concentrated in and significant. These results provide indi- the coastal region. Figure 16.10 provides rect evidence to support the finding that additional evidence to support the argu- local governments act to discourage indus- ment that economic globalization leads to trial agglomeration of Chinese industries more industrial agglomeration in China. at the provincial level. Panel A presents the relationship between Other channels also lead to industrial the share of exports in industrial gross out- dispersion in China. On the one hand, local put and the Gini coefficient of three-digit governments duplicate industries high- manufacturing industries in 2004; panel B lighted in the national Five-Year Plans. On shows the relationship between the ratio of the other hand, local governments also imi- foreign capital in total capital and the Gini tate each other when choosing key indus- coefficient of three-digit manufacturing tries. He and Zhu (2007) find that provinces industries. The Pearson's correlation coef- with similar economic and political status ficients are 0.71 and 0.55, respectively, sug- and with common borders are more likely gesting that more globalized industries are to imitate each other and to converge in the more agglomerated. industrial structure. Machinery equipment, Third, as discussed theoretically, decen- medicine, food, automobiles, construction tralization results in local protectionism and materials, chemicals and petrochemicals, an imitation strategy in industrial policies, and high-tech industries are listed as key which is to promote local economic growth industries in most provinces in both of the Industrial agglomeration and economic performance in transitional China 267 Figure 16.5 Provincial distribution of industrial output of domestic market­oriented industries in China in 2004 Resource-Based and Domestic Market­Oriented Petroleum refining and coking Nonmetal Mineral Products Ferrous Metal Smelting and Pressing Nonferrous Metal Smelting and Pressing Five-Year Plans. Many local governments spatial disparities in China are widening as have granted financial and policy supports industries shift to the coastal provinces. for the development of key industries, lead- ing to less industrial agglomeration. As Spatial boundaries of industrial discussed, the select industries are fairly agglomeration in China dispersed, with Gini coefficients smaller To investigate the spatial boundaries of than the weighted average Gini coefficient industrial agglomeration in China, I apply of 0.64. the global Moran's I to uncover the spatial In summary, Chinese industries have autocorrelation of the geographic distri- become more geographically agglomerated bution of Chinese industries. The global as the economic transition proceeds. The Moran's I can be defined as follows: agglomeration forces of industries domi- nate the dispersion forces. Marketization n allows comparative advantages and scale w ij(xi - x)(xj - x) i j economies to play their roles in driving I = (2) industrial agglomeration, and globalization w ij (x i - x )2 allows underlying geographic advantages to i j i play a greater role. Market and global forces have constantly driven Chinese industries where n represents the number of prov- to agglomerate in the coastal provinces, and inces, wij = 1 if provinces i and j share a 268 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 16.6 Geographic agglomeration of favored and protected industries by local and there may be spillover effects across governments in China, 1980­2004 provinces. 0.85 I have computed the value of Moran's I for all two-digit manufacturing industries in 0.80 select years. First, some highly agglomerated 0.75 industries have insignificant values of Moran's I, indicating that geographic agglomerations 0.70 of these industries are confined within pro- 0.65 vincialboundariesandthatnospillovereffects occur across provinces.The industries include 0.60 coefficient cultural education and sporting goods, tele- Gini 0.55 communications and electronic equipment, instruments and meters, furniture making, 0.50 and printing and copying. Related companies 0.45 in these industries often seek suppliers from highly localized sources, limiting their spill- 0.40 over effects within a province. 0.35 Second, some agglomerated indus- 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 tries have significant and positive values of year Moran's I, suggesting that the geographic Food processing and manufacturing Medical and pharmaceutical products agglomeration of these industries is beyond Beverage manufacturing Machinery manufacturing the provincial boundaries and that there Tobacco processing Transportation equipment are significant spillover effects across prov- inces. The industries include chemical fiber, Figure 16.7 Geographic agglomeration of less-protected and -favored industries in China, paper making and products, garments and 1980­2004 other fiber, leather and fur products, textiles, 0.85 rubber and plastic products, metal mineral products, general- and specific-purpose 0.80 machinery, electrical machinery and equip- 0.75 ment, and timber processing. Most of these industries are labor intensive and market 0.70 driven,facilitating the spillover effects beyond provincial boundaries. Business linkages 0.65 in industries such as general- and specific- 0.60 purpose machinery, electrical machinery coefficient and equipment, metal mineral products, and Gini0.55 rubber and plastic products could easily go 0.50 beyond the provincial boundaries. Third, a couple of less-agglomerated 0.45 industries have positive and significant val- 0.40 ues of Moran's I. These industries include food processing and manufacturing, chemi- 0.35 cal materials and products, transportation 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 year equipment, and nonmetal mineral prod- Chemical fiber Electrical machinery and equipment ucts. Business linkages beyond provincial Rubber products Cultural, education and sports goods boundaries in these industries are the reason Plastic products Instruments and meters for spillover effects. For instance, Shanghai has provided substantial auto components common border, otherwise wij = 0, and xi and parts to the auto-related industries in and xj are the provincial share of an indus- Anhui, Jiangsu, and Zhejiang. Development try. A positive and significant value of of transportation equipment in Beijing has Moran's I indicates that industrial agglom- also stimulated the industry of auto compo- eration is beyond the provincial boundary nents and parts in Hebei (Thun 2006). Industrial agglomeration and economic performance in transitional China 269 Figure 16.8 Provincial distribution of favored and protected industries by local governments in China Protected and Favorite Industries Food Processing and Manufacturing Beverage Manufacturing Tobacco Processing Medical and Pharmaceutical Products Finally, protected and favored industries ship between industrial agglomeration and are spatially dispersed and experience no industrial specialization. As panel A shows, spillover effects across provincial boundar- the industrial structure within Chinese prov- ies. The industries include beverage manu- inces became more diversified in the 1980s, facturing, tobacco processing, medical and as Chinese industries became geographically pharmaceutical products, and petroleum dispersed, indicating an overall pattern of refining and coking. Provincial govern- convergence in provincial industrial struc- ments are able to exercise local protection- ture. Rapid development of labor-intensive ism because Chinese provinces have large and light industries was undoubtedly asso- markets, sufficient authority, and a favor- ciated with the decreasing provincial spe- able combination of resources with which cialization before the mid-1990s. Regional to develop most industries. Local protec- decentralization was apparent in the early tionism thereby prevents spillover effects stage of reforms (Wei 2000), providing beyond provincial boundaries. local governments with power to imple- ment protectionist policies. Interregional Industrial specialization of Chinese competition and provincial protectionism provinces are at least partially responsible for the con- Overall, Chinese industries have been vergence of provincial industrial structure increasingly agglomerated, which has (Young 2000). Beginning in the late 1990s, changed the industrial structure of Chinese the increasing agglomeration of Chinese provinces. Figure 16.13 shows the relation- industries has caused the gradual increase in 270 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 16.9 Provincial distribution of least-favored or -protected industries in China Least-Protected Industries Chemical Fiber Rubber Products Plastic Products Electrical Machinery and Equipment industrial specialization, as shown in panel B There are substantial provincial varia- of figure 16.13. First, marketization and glo- tions in industrial specialization in China. balization have forced manufacturing firms In 1980 the specialization coefficient ranged to specialize in production and to compete from 0.77 in Xizang to 0.47 in Guangdong. more effectively in the markets based on All provinces except Guizhou, Xizang, and geographic agglomeration and comparative Yunnan were less specialized in the 1980s. advantages. The interior provinces, in par- The three western provinces were specialized ticular, have specialized further in resource- due to the development of resource-based oriented production. Second, the increasing industries, including tobacco processing, specialization is also related to the heavy chemical materials and products,nonferrous industrialization strategy initiated by local metal smelting and pressing, and nonmetal governments in some coastal provinces such mineral products. The coastal provinces sig- as Fujian,Guangdong,Jiangsu,and Zhejiang. nificantly diversified their industrial struc- With the increasing importance of machin- tures as industries were driven to the coastal ery, chemical materials and products, petro- region by global and market forces. As most leum refining and coking, transportation Chinese industries increasingly agglomer- equipment, and ferrous metal smelting and ated in the coastal provinces, the central and pressing, some coastal provinces have gradu- western provinces became more specialized ally become more specialized. Therefore, the while the coastal provinces became more recent industrial agglomeration has resulted diversified. Recently, the inland provinces in industrial specialization in China. have become even more specialized, while Industrial agglomeration and economic performance in transitional China 271 the coastal provinces have begun to special- Figure 16.10 Relations between exports (panel A) and foreign capital (panel B) and industrial ize. Due to marketization and globalization, agglomerations in China in 2004 the inland provinces are able to fully exploit A. Exports their comparative advantages based on 1.00 natural resources, leading to a higher level of industrial specialization. For instance, 0.90 in 2004, the top four industries in inland provinces included typical resource-based 0.80 or resource-processing industries such as ferrous metal smelting and pressing, nonfer- 0.70 rous metal smelting and pressing, nonmetal coefficient mineral products, food processing, petro- 0.60 Gini leum refining and coking,chemical materials and products, and tobacco processing. 0.50 The coastal provinces have upgraded their industrial structures and begun to special- 0.40 ize in more advanced industries, including telecommunications and electronic equip- 0.30 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 ment, transportation equipment, general- share of exports in gross output and specific-purpose machinery, petroleum B. Foreign capital 1.00 refining and coking, electrical machinery and equipment,chemical industries,and fer- 0.90 rous metal smelting and pressing. Therefore, geographic agglomeration of labor-inten- 0.80 sive industries in the coastal region caused industrial diversification in the coastal prov- 0.70 inces in the 1990s, while the agglomeration of advanced industries has caused industrial coefficient 0.60 specialization along the coast recently. With Gini the upgrading of industrial structures in the 0.50 coastal provinces and the implementation of heavy industrialization strategies, the coastal 0.40 region probably will experience more indus- trial specialization. 0.30 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 To further probe the relationship between share of foreign capital in total capital industrial agglomeration and industrial specialization, I compare the temporal evo- lution of industrial specialization in select recently has industrial agglomeration in Chinese provinces. Figure 16.14 shows the Shanghai resulted in increasing industrial changes in industrial specialization in three specialization as capital- and technology- centrally administered municipalities. The intensive industries, including telecom- three municipalities diversified their indus- munications and electronic equipment, tries with the development of light industries transportation equipment, machinery, and in the 1980s. Since the early 1990s, Beijing ferrous metal smelting and pressing, have and Tianjin have become increasingly topped the industrial structure. specialized, as telecommunications and Figure 16.15 presents the temporal changes electronic equipment, transportation equip- in industrial specialization in select coastal ment, ferrous metal smelting and pressing, provinces, which have agglomerated many and chemical materials and products have industries during economic transition. gained prominence. Shanghai has been All coastal provinces underwent signifi- less specialized than Beijing and Tianjin, cant industrial diversification as industries and industrial agglomeration in Shanghai increasingly agglomerated in the coastal has caused industrial diversification. Only provinces in the 1980s and the early 1990s. 272 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 16.11 Relations between ratio of income tax and value added tax to sales revenue significantly more specialized. As marketi- (panel A) and ratio of total profits in sales revenue (panel B) and Gini coefficient of three-digit zation and globalization progress, advanced industries in 2004 industries will be increasingly concentrated A. Ratio of income tax and value added tax to sales revenue in the coastal provinces, which will also 1.00 become more industrially specialized. The inland provinces provide a rather 0.90 simple picture of the temporal changes in industrial specialization. Figure 16.16 0.80 illustrates the evolution of industrial spe- cialization in select western provinces. 0.70 Most western provinces have been more coefficient 0.60 specialized than the coastal region and also Gini have experienced significant industrial spe- 0.50 cialization. The specialization coefficient for Yunan, for instance, increased from 0.40 0.56 in 1980 to 0.72 in 2004. The coefficient for Qinghai increased from 0.64 to 0.79 in 0.30 the same period. As the economic transi- 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 tion has proceeded, labor and capital in ratio of income tax and value added tax in sales revenue the west have moved to the coast, generat- B. Ratio of total profits to sales revenue 1.00 ing strong incentives for local governments to develop industries based on natural 0.90 resources.Rapid growth of resource-based or resource-processing industries in the west- 0.80 ern provinces has made them increasingly specialized. This is also true for central prov- 0.70 inces, including Heilongjiang, Neimenggu, and Shanxi (see figure 16.17). The recent coefficient 0.60 increase in the specialization coefficients Gini of the central provinces of Hubei and Jilin 0.50 is largely owing to the agglomeration of transportation equipment. Anhui, Henan, 0.40 Hunan, and Jiangxi have not been the core manufacturing bases and have experienced 0.30 no significant increase in industrial special- 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 ization. Recently, as labor and resource costs ratio of total profits in sales revenue have skyrocketed in the coastal region, some By the middle of 1990s, Fujian and Guang- labor- and resource-intensive industries, dong provinces had the most diversified such as garments, shoe and hat making, and industrial structures, with specialization leather and fur products, started to relocate coefficients smaller than 0.50.Further indus- to the inland region. The coming spatial trial agglomeration in the coastal provinces restructuring of traditional industries will has caused more industrial specialization, possibly lower the level of industrial spe- particularly in Guangdong, Jiangsu, and cialization in the inland provinces, especially Liaoning. Shandong and Zhejiang, however, those near the coastal provinces. show no significant increase in industrial To summarize, as Chinese industries specialization. Guangdong and Jiangsu are became geographically dispersed in the two critical manufacturing provinces and 1980s, Chinese provinces began to experi- are among the top four provinces for many ence industrial diversification. Increasing industries. Due to the recent development industrial agglomeration since the late 1990s of heavy machinery and equipment indus- has caused more industrial specialization. tries, which has been initiated by local gov- As Chinese industries increasingly agglom- ernments, the two provinces have become erate in the coastal provinces, the inland Industrial agglomeration and economic performance in transitional China 273 provinces gain opportunities to develop Figure 16.12 Relation between ratio of state capital to total capital and Gini coefficient of resource-intensive industries due to market three-digit industries in 2004 forces, leading to more industrial special- 1.00 ization. Recently, the coastal provinces have also become increasingly specialized as more 0.90 capital- and technology-intensive industries shift to the coast. 0.80 Industrial agglomeration and labor productivity in China 0.70 Theoretically, as Porter (2000) argues, indus- coefficient trial agglomeration would improve the pro- 0.60 Gini ductivity of constituent firms, upgrade the capacity of participants in clusters for 0.50 innovation and productivity growth, and stimulate new business formation that sup- 0.40 ports innovation and expands the cluster. To justify the increasing industrial agglomera- 0.30 tion in China, it is necessary to inquire more 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 explicitly whether there are productivity ratio of state capital in total capital effects of industrial agglomeration. Figure 16.13 Relationship between industrial agglomeration (weighted Gini coefficients I conduct a statistical analysis based on across industries) and industrial specialization (weighted Gini coefficients across provinces) a simple production-function approach with two-digit manufacturing industries 0.59 as observations. The dependent variable is the log of gross industrial output per 0.57 worker, and the independent variables 0.55 include the log of capital per worker and the Gini coefficient of gross industrial 0.53 output. Considering the evolutionary nature of marketization and globaliza- specialization 0.51 tion in China, I expect to find an increas- ingly strong and significant relationship industrial0.49 between industrial agglomeration and labor productivity, controlling for capital 0.47 per worker. I run the regression analysis for years in which provincial-industrial 0.45 data are available. As shown in table 16.2, 0.45 0.47 0.49 0.51 0.53 0.55 0.57 0.59 0.61 0.63 0.65 there is an extremely significant coefficient industrial agglomeration attached to K/L, and capital per worker has become more important since the 1990s. economic justification for the formation of As expected, the variable of the Gini coef- industrial clusters in China. ficient has a positive coefficient, indicating The results in table 16.2 represent a highly that labor productivity effects will emerge aggregated level of investigation.It is also crit- as industries become increasingly agglom- ical to see whether the relationship between erated. The variable of the Gini coefficient industrial agglomeration and labor produc- turned from insignificant to significant tivity holds for all individual industries.To do in the middle 1990s, implying that the so, I evaluate models based on a production- increasing industrial agglomeration since function approach for two-digit manufactur- the middle 1990s has driven up labor pro- ing industries, with the observations defined ductivity in Chinese industries. The posi- in terms of provinces. I apply the location tive relationship between the Gini coeffi- quotient to measure the geographic agglom- cient and labor productivity also provides eration of industry j in province i, defined as 274 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 16.14 Temporal changes of industrial specialization in centrally administered economies, localization economies, and municipalities of China, 1980­2004 urbanization economies, controlling for 0.70 capital per worker, defined as follows: Kij 0.65 Qij Lij = f Lij , LnSIZEij , LnUPOPi , LQij , (4) 0.60 where Qij is the gross output of industrial j 0.55 in province i, Kij and Lij represent the total capital and employment of industry j in coefficient 0.50 province i, SIZEij is the average employment Gini per enterprise of industry j in province i, 0.45 LQij is the location quotient of gross out- put of industry j in province i, and UPOPi 0.40 is the total nonagricultural population in province i. 0.35 I perform the regression analysis for 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 each two-digit manufacturing industry in year each year from 1980 through 2004. (More Beijing Tianjin Shanghai detailed data are available from the author.) All models are fairly significant, with rela- tively high values of R2. For instance, the R2 Figure 16.15 Temporal changes of industrial specialization in select coastal provinces of China, 1980­2004 ranged from 0.24 for leather and fur prod- ucts to 0.93 for tobacco processing in 2004. 0.70 All industries but leather and fur products, cultural education and sporting goods, 0.65 chemical fiber, and garment-making indus- tries, have an R2 greater than 0.50 in the 0.60 models. I am particularly interested in the significance of the location quotient (LQ), 0.55 which captures the effect of localization coefficient economies. There are wide variations in the 0.50 Gini significance and magnitude of the regression coefficients on LQ. The relationship between 0.45 industrial agglomeration and labor produc- tivity differed significantly in the 1980s and 0.40 in the 1990s. Similarly, industrial agglom- 0.35 eration was not significantly associated with labor productivity in a number of industries 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 year in the 1980s. Significant positive relation- Guangdong Fujian Jiangsu ships only held for textiles, tobacco process- Zhejiang Shandong Liaoning ing, cultural education and sporting goods, petroleum refining and coking, chemical fiber, plastic products, electrical machinery LQij = OUTPUTij OUTPUTi and equipment, telecommunications and OUTPUT OUTPUT , (3) electronic equipment, and instruments and ij i meters, which were also the most agglomer- i i ated industries in the 1980s and mainly spe- where OUTPUTij represents the gross out- cialized in labor-intensive functions. They put of industry j in province i and OUT- were also the first group of industries to PUTi is the gross industrial output in allow nonstate-owned enterprises and to use province i. I then assume that labor pro- foreign investment. Marketization and glo- ductivity is a function of internal scale balization therefore stimulated the spatial Industrial agglomeration and economic performance in transitional China 275 agglomeration of these industries because Figure 16.16 Temporal changes of industrial specialization in select western provinces of geographic clustering was rewarding. The China, 1980­2004 petroleum refining and coking industry was 0.85 highly agglomerated and also productive in the 1980s because of internal scale econo- 0.80 mies.Ferrous and nonferrous metal smelting 0.75 and pressing were also highly concentrated in the 1980s, but geographic concentration 0.70 did not bring higher labor productivity 0.65 because this industry was tightly controlled by state-owned enterprises. 0.60 coefficient The 1990s saw an optimal picture of the Gini 0.55 relationship between industrial agglomera- tion and labor productivity, as most indus- 0.50 tries became increasingly agglomerated 0.45 and many more industries became liberal- ized and globalized. First, all industries but 0.40 food processing, beverage manufacturing, 0.35 tobacco processing, petroleum refining 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 and coking, medical and pharmaceutical year products, and ferrous metal smelting and Guizhou Yunnan Gansu pressing had a significant and positive rela- Qinghai Ningxia Xinjiang tionship between industrial agglomeration and labor productivity. The exceptions are Figure 16.17 Temporal changes of industrial specialization in central provinces of China, heavily protected, strongly favored by local 1980­2004 governments or controlled by the state- 0.80 owned enterprises, and fairly geographi- cally dispersed. Regression coefficients on 0.75 location quotients in petroleum refining and coking and tobacco processing have 0.70 turned from significant and positive to insignificant or negative since the 1990s. 0.65 The results suggest that government inter- 0.60 vention and local protectionism account for the improvement in labor productivity and coefficient 0.55 industrial competitiveness. Gini Second, the relations between indus- 0.50 trial agglomeration and labor productivity 0.45 remained positive and significant for labor- intensive industries such as textiles, cultural 0.40 education and sporting goods, chemical fiber, and rubber and plastic products in the 0.35 1990s.Many more traditional labor-intensive 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 industries have benefited from geographic year agglomeration since the early 1990s, with a Anhui Jiangxi Henan Hubei Hunan significant and positive relationship between Shanxi Neimenggu Jilin Heilongjiang industrial agglomeration and labor produc- tivity. These industries include food manu- why labor-intensive industries could ben- facturing, garments, shoe and hat making, efit from geographic clustering. The clusters leather and fur products, timber processing of labor-intensive industries in the coastal and furniture making, rubber and plastic region are characterized by a deeper division products, and nonmetal and metal mineral of labor across enterprises, which signifi- products. There are several critical reasons cantly cuts production costs and increases 276 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 16.2 Regression analysis of the relationship between productivity and industrial agglomeration for two-digit industries in China, 1980­93 Year Constant Gini Ln K/L R2 F B-P 1980 5.43*** 0.99 0.40*** 0.28 4.57 0.08 1984 7.74*** 2.00 0.09 0.15 1.99 0.43 1985 3.76*** 0.79 0.63*** 0.528 12.88 1.66 1988 1.02*** 0.12 0.71*** 0.60 16.79 5.57 1989 3.20*** 0.30 0.75*** 0.65 20.13 4.95 1990 3.20** 0.42 0.74*** 0.58 15.42 1.40 1991 2.47** 0.90* 0.79*** 0.75 33.26 4.83 1992 3.60** 1.52 0.63*** 0.27 4.06 9.18 1993 2.73 0.63 0.80*** 0.76 36.87 13.57 1994 2.26 1.04 0.83*** 0.78 40.49 11.72 1997 2.15 1.61** 0.79*** 0.76 36.66 10.11 1998 0.97 1.47 0.89*** 0.37 6.77 5.50 1999 1.03 1.18** 0.83*** 0.82 51.41 0.68 2000 0.49 1.38*** 0.88*** 0.80 45.81 0.85 2001 0.72 1.22*** 0.87*** 0.84 59.84 1.62 2002 0.50 1.10** 0.90*** 0.87 74.62 1.60 2003 0.46 0.97** 0.91*** 0.87 76.64 3.05 2004 0.51 0.67* 0.93*** 0.88 83.22 4.48 Source: Author's calculations. *** p < 0.01. ** p < 0.05. * p < 0.10. Note: Number of cases = 26. Results are corrected with heteroskedasticity. labor productivity of related enterprises. from insignificant to highly significant since The majority of workers in the coastal indus- the middle 1990s. The magnitude of regres- trial clusters are migrants, who are produc- sion coefficients on telecommunications and tive and hard working, which boosts labor electronic equipment, electrical machinery productivity. The labor pool in the coastal and equipment, and instruments and meters region also lowers labor-related costs and has increased significantly since the 1990s. improves labor productivity. The industrial During 1980­2004, all these industries clusters in the coastal region are particularly became more agglomerated. attractive to foreign investment,which brings The productivity effects of industrial capital, management, and advanced technol- agglomeration in these capital- and tech- ogy, resulting in higher labor productivity. In nology-intensive industries are derived addition, state capital has largely withdrawn from strong localized business linkages and from the labor-intensive industries, stimu- the use of foreign investments. Upstream lating fierce market competition and forcing firms are close to downstream firms, as related enterprises to be more competitive. this is their main source of demand, while Overall, market forces and globalization downstream firms want to be close to a large effects have driven labor-intensive industries number of upstream firms, because this is to concentrate in a few coastal provinces where intermediate inputs are cheaper. The due to the productivity effects of geographic geographic proximity of related firms raises clustering. industrial labor productivity by speeding Third, capital- and technology-intensive the process of matching downstream and industries such as general- and specific- upstream firms and by lowering transaction purpose machinery, transportation equip- costs. The capital- and technology-inten- ment, telecommunications and electronic sive industries have also used a significant equipment, electrical machinery and equip- amount of foreign investments, facilitating ment, and instruments and meters have industrial agglomeration. For instance, for- significantly benefited from geographic eign capital accounted for 62 and 48 percent agglomeration. The regression coefficients of total capital in telecommunications and on the location quotient of machinery and electronic equipment and in instruments transportation equipment have changed and meters, respectively; almost all major Industrial agglomeration and economic performance in transitional China 277 auto producers in China are foreign joint Overall, manufacturing industries are con- ventures. Many studies find foreign-invested centrated in the coastal region, including enterprises to be more productive than Fujian, Guangdong, Hebei, Jiangsu, Lia- domestic-owned enterprises. As economic oning, Shandong, Shanghai, and Zhejiang transition proceeds, market and global forces provinces. Inland provinces such as Henan, will play a larger role in driving industrial Hubei, Jilin, and Sichuan are also impor- agglomeration and allocating resources more tant locations for manufacturing industries. effectively.This simple statistical analysis sug- From the northeastern province of Hei- gests that industrial agglomeration has been longjiang to the southwestern province of pervasive because industrial agglomeration Yunnan, I could draw a line to separate the has productivity effects. However, industrial whole nation into two parts, with manufac- localization will not necessarily bring higher turing employment heavily concentrated in labor productivity if industries are heavily the right part. To the northwest of the line, protected or controlled by governments. most counties have large and medium-size enterprises with fewer than 5,000 manu- facturing workers, while those that have Industrial clusters in China: enterprises with more than 5,000 manufac- a county-level analysis turing workers are resource based. Several I now use employment data from the first clusters of manufacturing employment are economic census conducted at the end located along the coastal region, including of 2004 to examine some typical indus- the Yangtze River delta, the Pearl River delta, trial clusters in China. I first aggregate the the Shandong peninsula, and the Beijing- industrial employment by county and map Tianjin area. There are also some scattered the county distribution of aggregate manu- industrial clusters in the central provinces, facturing employment (see figure 16.18). such as in Henan, Hubei, and Sichuan. Figure 16.18 Spatial distribution of manufacturing employment in China, by county, 2004 Employment (thousands) < 5 5­10 10­50 50­100 100­200 >200 278 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 16.19 Spatial distribution of employment in telecommunications equipment, computers, and other electronic equipment in China, by county, 2004 Manufacturing Employment 0­1,000 1,001­5,000 5,001­10,000 10,001­50,000 50,001­100,000 For illustrative purposes,I map the county panies are mainly clustered in the Pearl River distribution of employment in one of the delta and recently moved to Kunshan and more successful clusters: telecommunications Suzhou in Jiangsu. Japanese electronic com- equipment, computers, and other electronic panies strongly favor the Yangtze River delta equipment. As figure 16.19 shows, employ- and Liaoning province. Korean electronic ment in telecommunications equipment, firms invest heavily in Beijing, Shandong, computers,and other electronic equipment is and Tianjin. The Xingwang Industrial Park clustered in the Pearl River delta and in areas in the Beijing economic and technology zone near Beijing,Shanghai,and Tianjin,but there houses a manufacturing cluster of mobile are also clusters in Chongqing, Dalian, Jinan, telecommunications equipment centered on Wuhan, and Xi'an. The electronic indus- Nokia,with more than 30 component suppli- try typically agglomerates in cities that host ers (Yeung and others 2006). The strong sup- ETDZs or high-tech industrial development plier-buyer relations and business networks zones set by the central and local govern- are essential to the success of electronic clus- ments. There are favorable financial incen- ters. The downstream and upstream firms in tives and policies to attract electronic-related the electronic industry locate closer to each companies to these development zones. The other and form the successful clusters. The industrial policies have played an essential existence of electronic clusters in some inland role in facilitating the formation of industrial cities can be partially related to the combined clusters in telecommunications and elec- effects of market forces and the past loca- tronic equipment. However, the expansion tional policies of the Chinese government in and development of electronic equipment promoting the growth of inland cities. clusters in coastal provinces are driven largely Industrial clustering has been the critical by foreign firms. Taiwanese electronic com- source of industrial competitiveness and has Industrial agglomeration and economic performance in transitional China 279 significantly stimulated industrial develop- industries with strong internal scale econo- ment in China. Market and global forces are mies, while the dispersed industries were responsible for the formation and success of resource-based or resource-processing indus- industrial clusters, particularly in traditional tries. In 2004 many agglomerated industries labor-intensive industries.Related firms clus- were highly globalized, while the dispersed ter together because clustering is rewarding. industries were localized, resource-based, Central and local governments in China have domesticmarket­orientedindustriesorprof- also played critical roles in facilitating indus- itable, strategic industries that were favored trial clustering by concentrating companies and protected by local governments. As in development zones, which are set up by expected, the globalized industries with the the central government or local governments least intervention, such as telecommunica- at different levels.Governments set up invest- tions and electronic equipment, instruments ment platforms to attract new companies, and meters, cultural education and sport- while market and global forces underpin the ing goods, garments, shoe and hat making, expansion of industrial clusters, especially in leather and fur products, chemical fiber, and advanced industries such as telecommunica- rubber and plastic products, have become tions and electronic equipment, transporta- increasingly and significantly more agglom- tion equipment, machinery, and electrical erated since the early 1990s, while domestic machinery and equipment. market­oriented and protected industries have shown no significant trend of central- Conclusions ization. The globalized and market-driven China'seconomicgeographyusedtobeheav- industries have shifted to the coastal prov- ily shaped by a socialist ideology that down- inces, including Fujian, Guangdong, Jiangsu, played agglomeration economies. Industrial Shandong, Shanghai, and Zhejiang. The location was planned by the government. empirical results indicate that sectors and China's economic transition has gradually spaces that have undergone economic liber- introduced global and market forces into alization are the most prone to the formation the economic system, while decentralization of agglomeration economies. Geographic has granted local governments the author- agglomeration of many labor-intensive ity and responsibilities for local economic industries such as garments and other fiber, development. Therefore, China now is a leather and fur products, and rubber and mixed economy in which socialist legacies plastic products, has gone beyond provincial and government intervention and plan- boundaries and promoted regional indus- ning exist side by side with global and mar- trial development. Some capital-intensive ket forces. Theoretically, global and market industries such as transportation equip- forces may foster the geographic clustering, ment, machinery, electrical machinery and which allows Chinese industries to exploit equipment, and chemical materials and locational and comparative advantages and products, which rely on regional business agglomeration economies, while decentral- linkages, have significant spillover effects ization may result in local protectionism beyond provincial boundaries. However, and a rational imitation strategy, which dis- industries dependent on localized business courages industrial agglomeration. linkages or protected by local governments Economic liberalization during the last are confined within a province. decades in China seems to have fostered both Industrial agglomeration appears to have the macroeconomic and local conditions caused the change in industrial structure in under which viable industrial agglomera- Chinese provinces. In the 1980s, as Chinese tions canemerge.Theempiricalinvestigations industries became dispersed, Chinese prov- show that Chinese manufacturing industries inces experienced industrial diversification. have become increasingly agglomerated since Increasing industrial agglomeration has the early 1990s. I find significant industrial resulted in gradual industrial specialization variations in the trend and level of industrial since the late 1990s. With labor and capital agglomerations. In 1980 the most agglom- flowing to the coast, the inland provinces erated industries were capital-intensive have been forced to develop industries based 280 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA on natural resources,leading to a higher level direction of efficiency in spite of provincial of industrial specialization. The agglom- protectionism. eration of labor-intensive industries in the coastal provinces before the middle 1990s Note allowed these provinces to diversify their Canfei He is associate professor in the Depart- industrial composition, while the recent ment of Urban and Regional Planning, College geographic agglomeration of advanced of Urban and Environmental Sciences, Peking industries has resulted in industrial special- University, Beijing. The author wishes to thank ization in the coast. Dr. Yukon Huang, Dr. Indermit Gill, Profes- The empirical analysis of Chinese indus- sors Masahisa Fujita, Keijiro Otsuka, Ming Lu, and others who participated in the World Bank tries also supports the argument that a workshop on the World Development Report positive relationship can be found between 2009 in Tokyo on November 29­30, 2007, for industrial agglomeration and labor pro- constructive comments and suggestions. ductivity in economies that were formerly dominated by central planning. The rela- References tionship has been increasingly significant Bai, Chong-En, Yingjuan Du, Zhigang Tao, as the economic transition proceeds, sug- and Sarah Tong. 2004."Local Protectionism gesting that economic reform has created and Regional Specialization: Evidence from the conditions for industrial agglomeration. China's Industries." Journal of International There are substantial industrial variations in Economics 63 (2): 397­417. the relationship. As most industries became Belderbos, Rene, and Martin Carree. 2002."The increasingly agglomerated in the 1990s, Location of Japanese Investments in China: stronger and more significant positive rela- Agglomeration Effects, Keiretsu, and Firm tionships between industrial agglomera- Heterogeneity." Journal of the Japanese and tion and productivity emerged, especially in International Economies 16 (2): 194­211. industries that have gained prominence since Brülhart, Marius. 1998."Economic Geography, economic transition. The heavily protected Industry Location, and Trade: The Evidence." or state-controlled industries have not signif- World Economy 21 (6): 775­801. icantly benefited from industrial agglomera- DRCSC (Development Research Center of the tion. Market and global forces not only have State Council). 2004."A Report on Local Pro- driven Chinese industries to agglomerate in tection in China [in Chinese]." References for the coastal provinces but also have improved Economic Research 18 (January): 31­38. labor productivity.Although decentralization Ellison, Glenn, and Edward L. Glaeser. has promoted local economic development, 1997. "Geographic Concentration in U.S. it has discouraged industrial agglomeration Manufacturing Industries: A Dartboard and sacrificed the labor productivity of Chi- Approach." Journal of Political Economy nese industries. However, as the economic 105 (5): 889­927. transition proceeds and domestic markets Fan, C. Cindy, and Allen Scott. 2003."Industrial become more integrated, interprovincial Agglomeration and Development: A Survey competition will become more fierce and of Spatial Economic Issues in East Asia and a force locally protected industries to be more Statistical Analysis of Chinese Regions." Eco- productive. As a matter of fact, provincial nomic Geography 79 (3): 295­319. governments have worked hard to pro- Fujita, Masahisa, and Dapeng Hu. 2001. mote the formation of industrial clusters to "Regional Disparity in China, 1985­1994: improve the competitiveness of local indus- The Effects of Globalization and Economic tries. Protected industries are fairly concen- Liberalization." Annals of Regional Science 35 trated in some counties or cities within a (1): 3­37. province. Provincial governments have also Fujita, Masahisa, and Jacques-Franēois Thisse. competed fiercely with each other to attract 1996."Economics of Agglomeration." Journal foreign investments into locally protected of the Japanese and International Economies 10 industries and make them more productive. (4): 339­78. Overall, market reforms and globalization He, Canfei. 2002."Information Costs, Agglom- have indeed pushed China's industries in the eration Economies, and the Location of Industrial agglomeration and economic performance in transitional China 281 Foreign Direct Investment in China." Regional in a Global Economy." Economic Development Studies 36 (9): 1029­36. Quarterly 14 (1): 15­34. ------. 2003."Location of Foreign Manufac- Qian, Yingyi, and Barry Weingast. 1997."Fed- turers in China: Agglomeration Economies eralism as a Commitment to Market Incen- and Country of Origin Effects." Papers in tives." Journal of Economic Perspectives 11 (4): Regional Sciences 82 (3): 351­72. 83­92. ------. 2006."Regional Decentralization and SSB (State Statistical Bureau). 2006. China Eco- Location of Foreign Direct Investment in nomic Census Yearbook 2004. Beijing: State China." Post-Communist Economies 18 (1): Statistical Press. 33­50. ------. Various years (1999­2004). Annual He, Canfei, Yehua Dennis Wei, and Xiuzhen Xie. Report of Chinese Industrial Statistics. Beijing: 2008."Globalization, Institutional Change, State Statistical Press. and Industrial Location: Economic Transi- ------. Various years (1981­2004). China tion and Industrial Concentration in China." Industry Economy Statistical Yearbook. Beijing: Regional Studies 42 (in press). State Statistical Press. He, Canfei, and Shengjun Zhu. 2007."Economic Thun, Eric. 2004."Keeping up with the Jones': Transition and Regional Industrial Restruc- Decentralization, Policy Imitation, and turing in China: Structural Convergence or Industrial Development in China." World Divergence?" Post Communist Economies 19 Development 32 (8): 1289­308. (3): 321­46. ------. 2006. Changing Lanes in China: Foreign Head, Keith, and John Ries. 1996."Inter-city Direct Investment, Local Governments, and Competition for Foreign Investment: Static Auto Sector Development. New York: Cam- and Dynamic Effects of China's Incentives bridge University Press. Areas." Journal of Urban Economics 40 (1): Venables, Anthony. 1996."Equilibrium Loca- 38­60. tions of Vertically Linked Industries." Interna- Kim, Sukkoo. 1995. "Expansion of Mar- tional Economic Review 37 (2): 341­59. kets and the Geographic Distribution of Wang, Jici. 2001. Innovative Space: Industrial Economic Activities: The Trends in U.S. Cluster and Regional Development. Beijing: Regional Manufacturing Structure, 1860­ Peking University Press. 1987." Quarterly Journal of Economics 110 Wei, Yehua D. 2000. Regional Development in (4): 881­908. China: States, Globalization, and Inequality. Krugman, Paul. 1980."Scale Economies, Prod- London: Routledge. uct Differentiation, and the Pattern of Trade." Wen, Mei. 2004."Relocation and Agglomeration American Economic Review 70 (5): 950­59. of Chinese Industry." Journal of Development ------. 1991."Increasing Returns and Eco- Economics 73 (1): 329­47. nomic Geography." Journal of Political Econ- Yeung, Henry Wai-chung, Weidong Liu, and omy 99 (3): 483­99. Peter Dicken. 2006."Transnational Cor- Krugman, Paul, and Raul Livas Elizondo. 1996. porations and Network Effects of a Local "Trade Policy and the Third World Metropo- Manufacturing Cluster in Mobile Telecom- lis." Journal of Development Economics 49 (1): munications Equipment in China." World 137­50. Development 34 (3, March): 520­40. Lee, Pak K. 1998."Local Economic Protection- Young, Alwyn. 2000."The Razor's Edge: Distor- ism in China's Economic Reform." Develop- tions and Incremental Reform in the People's ment Policy Review 16 (3): 281­303. Republic of China." Quarterly Journal of Eco- nomics 115 (4): 1091­135. Markusen, James, and Anthony Venables. 1999. "Foreign Direct Investment as a Catalyst for Zhao, Xiaobin, and Li Zhang. 1999."Decentral- Industrial Development." European Economic ization Reforms and Regionalism in China: A Review 43 (2): 335­56. Review." International Regional Science Review 22 (3): 251­81. Marshall, Alfred. 1898. Principles of Economics. Zhou, Huizhong. 2000."Fiscal Decentralization London: Macmillan. and the Development of the Tobacco Indus- Porter, Michael. 2000."Location, Competition, try in China." China Economic Review 11 (2): and Economic Development: Local Clusters 114­33. Capital allocation, regional specialization, and spillover effects in China Chong-En Bai and Xu Lin 17 With a record-high economic growth rate period; they also find evidence consistent and an important role in the world's econ- with both the market forces for specializa- omy, China has received ever-increasing tion and the forces for local protectionism research attention. An issue of great inter- against specialization. In a recent paper, Bai, est to researchers is the allocation of invest- Tao, and Tong (2008) document a U-shape ment across regions in China. Some studies, relationship between regional specializa- such as Boyreau-Debray and Wei (2005), tion and per capita gross domestic product argue that capital allocation across Chinese (GDP) in China, which is consistent with c h a p t e r provinces is becoming less efficient and that the finding in Imbs and Wacziarg (2003), the direction of capital flows is from regions which investigates cross-country data. with high returns to those with low returns. A closely related issue concerns the In contrast, in a recent study, Bai, Hsieh, and agglomeration effect among neighboring Qian (2006) systematically investigate the firms. There are tradeoffs regarding the spa- aggregate returns to capital in China and tial concentration of industrial activities. On find that they have remained high despite the one hand, agglomeration may induce one of the highest investment rates in the regional disparity in economic development. world. Furthermore, they study the pattern On the other hand, it may allow firms in the of investment allocation across regions and same industry to benefit from the proxim- find that the regional dispersion of returns ity of their peers. To better understand the to capital has decreased over time. tradeoffs regarding the spatial concentra- Another interesting issue is the degree tion of industrial activities, it is necessary of regional specialization in China. Has the to understand the agglomeration effect. degree of regional specialization increased There is a large literature on the agglom- or decreased as the economy has grown? eration effect. However, depending on the What factors determine the trend of spe- methodology used, data sets employed, cialization across regions? In an earlier and countries studied, the empirical results study, Young (2000) claims that regional vary greatly across empirical studies. Our economic structures in China are becom- knowledge about the agglomeration effect ing increasingly similar, which implies a rise in China is even more inadequate. Most of in local protectionism. In contrast, Naugh- the existing work in this regard focuses on ton (2003) finds evidence consistent with the effect of the presence of foreign direct increasing regional specialization in China investment (FDI) on the performance of using 1987 and 1992 input-output data.And domestic firms. Such a focus is useful if we Bai and others (2004) find that the degree of want to evaluate the effect of FDI, but it is industrial agglomeration in China changed not enough if we want to understand the from a decline between 1985 and 1988 to tradeoffs involved with the spatial concen- an increase afterward during the sample tration of industrial activities. Capital allocation, regional specialization, and spillover effects in China 283 In this chapter, we follow Bai, Hsieh, features in China's national account statistics and Qian (2006) and study the allocation and recent revisions to the statistics. of investment across regions in China. We also extend the work of Bai, Tao, and Tong Returns to capital. Following Bai, Hsieh, (2008) by using the most recent time series and Qian (2006), we calculate the real rate data from 1999 to 2003 to investigate recent of return to capital r(t) for each of China's trends in China's regional specialization.Our 28 provinces for the years from 1978 to 2005 results confirm that the efficiency of China's using the following equation:1 resource allocation has been improving and that market forces have played an increas- r(t) = i(t)-P^Y (t) ingly important role in China's economic development. = (t ) In addition, we study the spatial factors PK (t)K(t) PY (t)Y(t) behind firm performance to contribute + (P^K (t ) - P^Y (t )) - (t ). (17.1) toward our knowledge about the tradeoffs regarding the spatial concentration of Where i is the nominal rate of return, PY is industrial activities. We consider the effect the price of the output good, PK is the price of the proximity of peers on firm perfor- of capital, is the share of payments to mance and then explore how the effect capital in GDP, is the depreciation rate of depends on regional and industrial char- capital, P^Yand P^K are the percentage rates of acteristics and whether firms of different change of the prices of the output good and ownership, different sizes, and so forth capital, respectively, and K(t) denotes the enjoy the agglomeration effect to the same real value of the aggregate capital stock. degree. The rest of the chapter is organized as Aggregate output. To account for the follows. The following section addresses the possible bias in locally provided GDP, the allocation of investment across provinces National Bureau of Statistics adjusts the and regional returns to capital in China.This aggregate GDP based on nationwide eco- is followed by an analysis of regional spe- nomic censuses. Our estimation uses the cialization and an examination of the spatial revised national accounts data provided by factors behind productivity growth among the National Bureau of Statistics. Chinese firms. A final section concludes. Capital stock. Compared with the widely Returns to capital across used series for investment in fixed assets, the provinces series for gross fixed capital formation is a more accurate measure of the change in Chi- Bai, Hsieh, and Qian (2006) have studied na's reproducible capital stock. On the one returns to capital in China at length, espe- hand, the series investment in fixed assets cially aggregate returns to capital. They also includes the value of purchased land and provide some results regarding the alloca- expenditure on used machinery and preexist- tion of investment as well as the returns to ing structures, which should not be included capital across provinces. Most results pre- in investment data. On the other hand, the sented here closely follow their work. series may also understate aggregate invest- ment, because it is based on survey data for Methodology and data large investment projects only. In contrast, This section presents our methodology in calculating gross fixed capital formation, for estimating rates of return to capital; the value of land sales and expenditures on introduces data on aggregate output, capi- used machinery and buildings are excluded tal stock, and share of capital; and reports from investment in fixed assets, and expen- our findings about rates of return to capi- ditures on small-scale investment projects tal across Chinese provinces from 1978 to are added. Therefore, we use this series to 2005. We pay particular attention to special measure the capital stock and assume that 284 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 17.1 Returns to capital in China, by province, 1978­2005 price of structures and buildings and use the output price deflator of the domestic machinery and equipment industry for the 120 price of machinery and equipment. Before 100 1978, we simply use the price of the two types of investment goods. 80 Then we employ the standard perpetual year 60 a inventory approach to estimate the stock of the two types of capital. We initialize 40 the capital stock in 1952 as the ratio of percent 20 investment in 1953 (the first year for which investment data are available) to the sum 0 of the average growth rate of investment ­20 in 1953­58 and the depreciation rate. The depreciation rate for structures and for machinery is assumed to be 8 and 24 per- 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 year cent, respectively. Region: Eastern Central Western Share of capital. We calculate the share of capital in total income from the residual of Source: Bai, Hsieh, and Qian (2006). labor income. The National Bureau of Sta- Note: Each symbol represents the rate of return to capital of a province in the given year. Different symbols represent provinces from different regions (eastern, central, or western). tistics provides annual data on the share of labor for each province and each sector, Figure 17.2 Standard deviation of returns to capital across provinces in China, 1978­2005a which can be used directly to estimate the share of capital for each province. Returns to capital across regions 25 Figure 17.1 plots the returns to capital for each of China's 28 provinces from 1978 to 20 2005.Provinces are grouped into one of three year a regions--eastern, central, and western--as 15 shown in the figure.One striking feature pre- percent sented in the figure is the heterogeneity in the 10 regional returns to capital. As clearly shown, the returns to capital are generally highest in 5 the eastern region and lowest in the western region. However, the differences over time in the returns to capital across provinces are 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 shrinking. The convergence of the regional year returns to capital is also confirmed by figure Source: Bai, Hsieh, and Qian (2006). 17.2, which shows that the standard devia- tion of the returns to capital across provinces the share of investment in structures and is declining over the sample period. buildings and the share of investment in Therefore, contrary to the findings of machinery and equipment are the same as Boyreau-Debray and Wei (2005), the results those for investment in fixed assets. presented here demonstrate that the disper- For the investment price deflators, after sion in the returns to capital across regions 1990, the National Bureau of Statistics in China has been shrinking and there is reports separate price indexes for invest- no evidence that capital flows from regions ment in structures and buildings and for with higher returns to capital to those with investment in machinery and equipment. lower returns. In other word, China's invest- For 1978­89, we use the deflator of value ment allocation across regions has become added in the construction industry for the more efficient. Capital allocation, regional specialization, and spillover effects in China 285 Regional specialization ment spending is known for favoring local This section considers how the degree of firms and industries. Furthermore, local gov- regional specialization depends on various ernments with high ratios of expenditures to factors. We use panel data across 31 Chinese GDP are under financial pressure to practice regions to estimate the relationship between local protectionism and obtain fiscal revenue the degree of regional specialization and to maintain their large public sectors. Thus various factors. regions with higher ratios of local govern- ment expenditures to GDP are expected to Theory and hypotheses have more severe local protectionism. Local protectionism is a form of trade barrier.With There are a few theories about regional spe- higher trade barriers, the degree of regional cialization, and each of them implies empir- specialization is lower. ically testable hypotheses. The share of GDP from primary indus- tries is another proxy for the level of local Stage of economic development and size of protectionism. Like other planned econo- the economy. Using cross-country data, mies, China had national policies for devel- recent studies, such as Imbs and Wacziarg oping manufacturing industries at the (2003), find that the relationship between expense of primary industries--specifically, the degree of regional specialization and per artificially suppressed prices for the output capita income is U shaped. They offer two of primary industries but artificially inflated possible explanations. First, consumers tend prices for the outputs of manufacturing to demand a more diverse range of goods industries--before its economic reform in and services as their income rises, and this 1979 (the so-called price-scissors problem; implies a diversification of economic activi- see, for example, Sah and Stiglitz 1984). In ties if consumer demand cannot be met with addition, due to central planning, those imports from other countries due to high regions with high shares of GDP coming trading costs. Second, in the absence of per- from primary industries may not have been fect risk-sharing arrangements, it is risky for the ones that further processed the outputs countries to specialize in producing a small from primary industries and thus could set of goods and services, as predicted by the not take full advantage of their resource traditional theories of regional specialization endowments. Consequently, the price- (Kalemli-Ozcan, Sųrensen, and Yosha 2003). scissors problem led to severe misalign- To test whether this relationship holds for ment of economic interests among China's China's regional data, we include both per regions. Since China initiated its economic capita GDP and the square of per capita GDP reform in 1979, the price of products from in our regression. both primary industries and manufactur- Several studies, including Kalemli-Ozcan, ing industries has been increasingly deter- Sųrensen,andYosha (2003),argue that larger mined by market forces, but it takes much regions tend to have lower levels of special- longer to adjust the suboptimal geographic ization due to more diversified demand and location of manufacturing activities. In gen- the exhaustion of scale economy. To capture eral, manufacturing industries tend to have this effect, we include a region's total popu- higher value added than primary industries lation in the regression and expect it to have do. As a result, it is expected that, in regions a negative effect on the degree of regional with higher shares of GDP from primary specialization. industries, local governments place more restrictions on the sale of the output from Local protectionism. With fiscal decentral- their primary industries to other regions, ization,China'slocalgovernmentshavestrong and consequently the degree of regional spe- incentives to protect local firms and indus- cialization is lower (Bhagwati 1988). tries. However, the level and effectiveness of local protectionism depend on a number of Market competition. Market competition factors. One is the size of local government greatly limits the effectiveness of local pro- expenditures relative to local GDP. Govern- tectionist policies. To capture the domestic 286 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA competition from firms in other regions, we Then the localization curve for region j can construct a market potential measurement by be plotted by calculating the cumulative per- using the weighted (weighted by the inverse centage of output in region j (y axis) over the of distance between different provinces) aver- industries (x axis). The localization curve is age of GDP from other provinces. Compared the 45° line if every industry in a region with domestic firms from other regions, for- contributes the same share of output as the eign-invested firms and foreign imports pose whole country. And the localization curve is a greater threat to local firms. We use two more concave if a region's economic activi- measurements to capture the effects of com- ties are concentrated in only a few industries. petition from foreign firms. One is the share Then the area between the 45° line and the of annual FDI inflows in a region to its GDP. localization curve divided by the entire tri- The other is the distance of a region's capital angular area in which the localization curve to Hong Kong weighted by the percentage of is contained defines the Hoover coefficient China's exports going through Hong Kong. of localization, which is between 0 and 1. A It is expected that the degree of regional higher Hoover coefficient corresponds to a specialization will be higher in regions with higher degree of regional specialization. higher market potential or a higher ratio of FDI inflows to its GDP and in those regions Other variables. The following variables closer to Hong Kong. are used in this study: · HOOVERjt is the Hoover coefficient of Methodology and data specialization of region j in year t; This section defines the Hoover coefficient · rGOVT_GDPjtis the ratio of government of localization and other variables and pres- expenditure to GDP of region j in year t; ents summary statistics. · %PRIMARYjt is the share of GDP con- tributed by primary industries of region Hoover coefficient of localization. To j in year t; measure a region's degree of specialization in industrial production, we calculate the · MP stands for market potential and is Hoover coefficient of localization (Hoover the weighted (weighted by the inverse 1936) using output data for 32 two-digit of distance between different provinces) industries in 31 Chinese regions over the average of GDP from other provinces; period of 1999­2003. It is calculated based · rFDI_GDPjt is the ratio of annual flow of on the location quotient with respect to out- FDI to GDP of region j in year t; put, which is given by: · DIST_HKjt is the weighted distance to Hong Kong of region j in year t; Lij =OUTPUTij /OUTPUTj , (17.2) · pcGDPjt is GDP per capita of region j in OUTPUTi /OUTPUT year t; where OUTPUTij is industry i's output in · pcGDP2jt is the square of pcGDPjt; and region j, OUTPUTj is total output in region · POPjt is the population of region j in j, OUTPUTi is industry i's total output, year t. and OUTPUT is total industrial output of China. If Lij is larger than 1, then industry i Summary statistics. Figure 17.3 plots the has a higher percentage in region j than its average Hoover coefficients across all regions share in the total industrial output of China from 1999 to 2003. The simple average for and vice versa. all regions was 0.541 in 1999 and 0.572 in Analogous to the Gini coefficient for 2000. The simple average Hoover coefficients income distribution, to calculate the Hoover remained at that level until 2001 and then coefficient of localization, we first need to jumped to 0.582 in 2003. The weighted (by plot the localization curve for region j. regional industrial output) average across Given the location quotient of region j for regions demonstrates a similar time trend, all industries, i = 1,...,l, we rank industries with each weighted average Hoover coefficient by their location quotient in descending being about 0.04 larger than the correspond- order and obtain a sequence of industries. ing simple average in each year. By using data Capital allocation, regional specialization, and spillover effects in China 287 over the 13-year period of 1985­97, Bai, Tao, Figure 17.3 Average of Hoover coefficients across regions in China, 1999­2003 and Tong (2008) show that China's regions 0.64 have become more specialized in industrial production. Our result confirms that the 0.62 5 degree of regional specialization in China 2 has continued to grow in recent years. To see 3 0.6 4 the significance of regional variations in the degree of specialization,we plot the time aver- 0.58 5 age of the Hoover coefficients of specialization 1 2 for different regions. As shown in figure 17.4, 3 4 0.56 Beijing has the highest degree of specializa- tion, with a Hoover coefficient of 0.790, and 0.54 1 Qinghai has the lowest degree of specializa- tion, with a Hoover coefficient of 0.359. 0.52 The mean and rank of other variables across regions are presented in table 17.1.2 0.5 The ratio of government expenditure to 1999 2000 2001 2002 2003 GDP has a mean of 0.151, with that for Tibet Weighted average Simple average (0.521) being the highest and that for Hebei (0.078) being the lowest. And the share of GDP from primary industries ranges from Figure 17.4 Average (across time) Hoover coefficient in China, by region 0.017 (Shanghai) to 0.374 (Hainan), with a mean of 0.176. Market potential, the share of Beijing FDI in GDP, and capacity-weighted average distance to Hong Kong range from 2,764.0 Guangdong (Gansu) to 4,640.3 (Anhui),from 0 (Tibet) to Chongqing 234.734 (Jiangsu), and from 0.0004 (Beijing) to 0.0139 (Ningxia), respectively. The means Hebei are 3,656.51, 35.792, and 0.0058, respectively. With per capita GDP of 38,019.6, Shanghai Jilin ranks first among all regions. Guizhou has the lowest per capita GDP, at 2,957.6. The Jiangxi mean per capita GDP is 9,584.4.With regard Hunan to population size, Henan ranks first with a population size of 0.944, and Tibet ranks the Gansu lowest with a population size of 0.026. Table 17.2 summarizes the pair-wise correlation of Guizhou the key variables. Xinjiang Regional specialization Qinghai To understand the underlying property and 0 0.2 0.4 0.6 0.8 1 pattern of China's regional specialization, Series 1 we estimate the following model: Source: Authors' calculations. LogitHOOVERjt = 1 + 2GOVT_GDPjt Note: Each bar refers to the Hoover coefficient of one province. There are 31 provinces, and the names along the verti- cal axis are the 11 regions that encompass the provinces. + 3 %PRIMARYjt + 4rFDI_GDPjt where t is the time-specific effect, and jt is + 5 pcGDPjt the error term, and + 6 pcGDP2jt LogitHOOVERjt + 7MP + 8DIST_HK jt + 9POPjt + t + jt , - HOOVERjt HOOVERjt . (17.4) (17.3) = ln 1 288 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 17.1 Variable mean and rank Hoover rGovt_GDP %PRIMARY rFDI_GDP pcGDP POP MP DIST_HK Region Mean Rank Mean Rank Mean Rank Mean Rank Mean Rank Mean Rank Mean Rank Mean Rank Beijing 0.79 1 0.18 8 0.03 30 67.23 7 25,667.8 2 0.11 26 3,716.4 15 0.0004 31 Hebei 0.6 10 0.08 31 0.16 18 14.91 16 8,517.0 11 0.67 5 3,727.1 14 0.0021 28 Shanghai 0.68 3 0.14 14 0.02 31 78.58 6 38,019.6 1 0.13 25 4,630.9 2 0.0038 23 Jiangsu 0.58 14 0.22 3 0.27 3 234.73 1 13,312.0 6 0.60 8 4,115.8 5 0.0048 18 Anhui 0.57 17 0.10 24 0.22 9 9.94 21 5,413.4 25 0.62 7 4,640.3 1 0.0066 13 Henan 0.52 26 0.08 28 0.22 11 8.94 23 6,053.6 18 0.94 1 3,850.5 12 0.0045 21 Hainan 0.6 11 0.12 18 0.37 1 83.79 5 7,306.2 15 0.08 28 4,037.3 7 0.0067 12 Guizhou 0.52 25 0.20 7 0.26 5 3.23 29 2,957.6 31 0.37 16 3,426.3 21 0.0096 4 Tibet 0.56 18 0.52 1 0.27 2 0.00 31 5,418.4 24 0.03 31 3,191.3 25 0.0043 22 Gansu 0.53 22 0.17 9 0.19 15 4.95 28 4,236.8 30 0.26 22 2,764.0 31 0.0071 10 Ningxia 0.51 27 0.22 4 0.17 17 8.93 24 5,429.4 23 0.06 29 3,151.6 26 0.0139 1 Mean 0.57 n.a. 0.15 n.a. 0.18 n.a. 35.79 n.a. 9,584.4 n.a. 0.40 n.a. 3,656.5 n.a. 0.0058 n.a. Number of observations 155 n.a. 155 n.a. 155 n.a. 155 n.a. 155 n.a. 155 n.a. 155 n.a. 155 n.a. Source: Authors' calculations. n.a. Not applicable. Table 17.2 Pair-wise correlations between main variables Variable Hoover rGovt_GDP %PRIMARY rFDI_GDP pcGDP POP MP DIST_HK Hoover 1 rGovt_GDP -0.19 1 %PRIMARY -0.34 0.23 1 rFDI_GDP 0.43 -0.05 -0.08 1 pcGDP 0.55 -0.11 -0.68 0.50 1 POP 0.00 -0.41 0.08 0.06 -0.20 1 MP 0.37 -0.36 -0.08 0.36 0.43 0.09 1 DIST_HK -0.45 0.06 0.41 -0.22 -0.44 0.00 -0.36 1 Source: Author's calculations. The estimation results are presented in Table 17.3 Estimation results with dependent table 17.3. variable: LogitHoover The most interesting result is the Variable Beta t-value U-shaped relationship between regional GOVT_GDP -4.04E-06 -1.3813 specialization and per capita GDP, as %PRIMARY 0.25707 1.416 shown by the negatively significant (at the rFDI_GDP -0.000147 -0.73112 5 percent level) estimated coefficient of per pcGDP -1.54E-05 -2.2466** capita GDP and the positively significant pcGDP2 1.76E-10 2.0832** MP 1.94E-05 3.0433*** (at the 5 percent level) estimated coeffi- DIST_HK -1.49E-06 -1.2097 cient of per capita GDP square. Consistent POP -0.056435 -0.21264 with Bai, Tao, and Tong (2008), our result Year dummies Yes provides further evidence for the stage Number of of development theory. Another variable observations 155 that is also significant (at the 1 percent Source: Authors' calculations. level) is market potential, which is positive *** Significant at 1 percent. ** Significant at 5 percent. and consistent with our expectation that more severe domestic market competition of Bai, Tao, and Tong (2008). In particular, leads to a higher degree of regional spe- the estimated coefficient for the ratio of cialization. However, the other variables, local government expenditures to its GDP although statistically insignificant, do not (rGOVT_GDP) is negative, as expected, appear to be consistent with the findings and statistically insignificant. Thus a higher Capital allocation, regional specialization, and spillover effects in China 289 ratio of local government expenditures to of Chinese firms. We follow the empirical its GDP implies a higher degree of local framework in Combes (2000) and Cingano protection, although the relationship does and Schivardi (2004), making some modi- not appear to be significant. Also consistent fications to capture the special features of with our expectation, the coefficient for the the Chinese economy. weighted distance to Hong Kong is nega- tive, implying that regions that are subject Total factor productivity and to stiffer competition from firms with for- underlying spatial factors eign investment and foreign imports enjoy Due to data limitations, many empirical a higher degree of regional specialization. studies are based on employment growth; And the coefficient for a region's popula- that is, they assume that growth in produc- tion (POP) is negative, as expected. As dis- tivity is proportional to growth in employ- cussed in Bai, Tao, and Tong (2008), larger ment. However, as discussed in Cingano and regions have lower degrees of specialization Schivardi (2004), this assumption is rather due to more diverse demand. At the same strong, and studies relying on this assump- time, with the massive investments in infra- tion might suffer from identification prob- structure, the trading costs across regions lems. Therefore, we follow Cingano and are decreasing and so is the negative effect Schivardi (2004) and construct a measure of of the size of economy on regional special- local total factor productivity (TFP) as the ization over time. Contrary to our expecta- dependent variable. Specifically, we define tion and the results in Bai, Tao, and Tong TFP as Solow residual. To calculate TFP, we (2008), the coefficient for the share of GDP first estimate the following regression model from primary industries (%PRIMARY) is using pooled panel data: positive, and the coefficient for the ratio of annual FDI flows of a region to its GDP ln(profitit ) = + 1ln(Kit ) (rFDI_GDP) is negative. As pointed out by (17.5) Bai,Tao,and Tong (2008),the insignificance + 2ln(Lit ) + it . of the role of primary industries could be Then TFP can be obtained as due to the fact that primary industries have become more market oriented and, conse- sTFPit = ln(profitit) quently, the relationship between the size of the primary industries and local protec- - 1ln(Kit) - 2ln(Lit ). (17.6) tion has been weakened. And the role of firms with foreign investment in the whole Intraindustry spillovers. We use two vari- economy may be more complicated than ables to capture the intraindustry spillover we thought. effects on the localization economies. One is Nsr, which is the number of firms in the Spatial factors behind same sector s and region r; the other is Ns, which is defined as the number of firms in productivity growth the same sector s but not in the same region. Externalities among firms form from two As pointed out by Marshall (1920), exter- directions. One is localization economies, nalities can occur through three mecha- known as Marshall-Arrow-Romer (MAR) nisms: knowledge spillovers, labor pool- economies, where externalities come from ing, and learning. In this study, we do not other local firms in the same industry. The intend to separate these mechanisms from other is urbanization economies, known each other; instead, we use Nsrto capture the as Jacobs economies, where cross-fertiliza- effects of the spatial concentration of other tion from firms outside the same indus- firms from the same industry and use Ns to try generates externalities. In this section, demonstrate whether these spillover effects we use plant-level panel data to analyze operate locally or decay with distance. agglomeration effects in Chinese firms. Our purpose is to identify the spatial fac- Interindustry spillovers. To account for tors behind the local productivity growth spillover effects from firms outside the same 290 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA industry, we use two indicators to represent intraindustry and interindustry spillover product variety. One is Nr, the number of effects and firm assets, share of foreign capi- firms in the same region r but not in the tal, and share of state capital. same sector. The other is a Hirschman- Herfindahl index, defined as Summary statistics Our estimated TFP ranges from 14.063 to 14.137, with a mean of 14.104. The average Varietyr = 2 , (17.7) ,s L Lr,j number of firms from the same sector in js r - Lr ,s a region is 180.59, the minimum is 1, and where L is manufacturing employment. the maximum is 726. For number of firms from the same sector but not in the same region, the mean, minimum, and maxi- Scale of the local economy. As suggested in mum are 1,926.9, 3, and 4,193, respectively. Combes (2000), the scale of the local econ- And the number of firms in the same region omy affects the intensity of spillover effects. and from the same sector ranges from 38 to On the one hand, the level and quality of 5,791, with a mean of 2,992.2. The various spillover effects require a large enough num- indexes have a mean, minimum, and maxi- ber of firms. And large size of local markets mum of 0.092, 0.046, and 0.618, respec- helps to foster concentration of specialized tively. The mean level of local competition inputs and to develop market demand. On is 0.111. The labor force employed ranges the other hand, dense economic areas tend from 1 to 166,857 workers, with a mean to have higher rent and higher input prices, of 483.937. And the mean share of foreign as well as other negative effects such as con- capital and state capital is 0.174 and 0.452, gestion and pollution. We use the size of the respectively. The average population size is regional population to represent the scale of 55.145 million, the minimum is 2.478 mil- the local economy. lion, and the maximum is 95.847 million. And 64.24 percent of the firms are located Local competition. As discussed in Porter in coastal areas. Finally, the firm assets range (1990), local competition fosters innova- from Y 18,000 to Y 68.266 billion, with a tions and the adoption of new technology. mean of Y 126.28 million. However, if competition is too severe and The final sample consists of 45,093 firms, the return to research and development which covers 30 regions and 37 industries, (R&D) investment is too low, firms' moti- with a mean of 1,503.1 firms in a region. vations for R&D investment may be weak- ened. We define competition of sector s in Empirical Results region r as We estimate the following model: 2 ln(TFP)it = + 1Nsr + 2Nr + 3Ns Compr,s = L + 4N sr × assetit + 5N s ir,s Lr,s,i , (17.8) r,s × assetit + 6Nr × assetit where i is an index for the firm belonging to + 7N sr × foreignit + 8N s sector s and region i. L is labor employed. × foreignit + 9Nr × foreignit Other variables. Given the special features + 10N sr × statetit + 11N s of the Chinese economy, we also include the × stateit + 12Nr × stateit labor force employed at the firm level and + 13 populationr + 14Varietysr its square, share of foreign capital, and share of state capital in the specification. Further- + 15competitionsr + 16laborit more, to evaluate how the level of spillover + 17laaborit + 18 foreignit 2 effects differs across different types of firms, + 19stateit + dummies + it , we also include interaction terms between (17.9) Capital allocation, regional specialization, and spillover effects in China 291 where four time dummies and a region this type of spillover effects, larger firms dummy for coastal area are included to and firms with a higher share of foreign control for possible unobservable time and capital tend to benefit more, while firms region effects that might confound with spa- with a higher share of state capital tend to tial spillover effects. The estimation results benefit less. In addition, the TFP growth are presented in table 17.4. of a firm is positively correlated with the The results show significant spillover number of other firms in the same region, effects, both intraindustry and interindus- regardless of the sector, which indicates the try, among Chinese firms. Specifically, the presence of positive interindustry effects in number of firms from the same sector in the a region. In particular, smaller firms, firms same region has a positive effect on a firm's with a lower share of foreign capital, and TFP growth; and smaller firms that have firms with a lower share of state capital lower asset levels appear to benefit more appear to enjoy more positive externali- than larger firms. Likewise, firms with a ties from other firms in the same region higher share of state capital tend to benefit but not in the same industry. At the same more from the presence of other firms in time, variety, another indicator of interin- the same sector and the same region, while dustry spillover effects, shows a negative the foreign capital share does not affect the effect, although it is significant only at the magnitude of this type of spillover effects. 10 percent level. The size of local economy, As another indicator of intraindustry spill- which is captured by population size, has over effects, the number of firms from the a direct effect on TFP growth (at the 5 same sector in other regions also shows a percent level). And local competition also positive association with TFP growth. For shows a positively significant effect on TFP growth. As for the other variables, foreign capital share does not show any significant Table 17.4 Estimation results with dependent relationship with TFP growth, while state variable ln(TFP) capital share shows a positive effect at the 5 Variable Beta percent level. Interestingly, the labor force Nsr 1.43E-06*** employed by a firm exhibits a U-shaped Nr 2.41E-06*** relationship with TFP growth of the firm. Ns 1.88E-07*** Because the results are obtained after con- Nsr*asset -4.48E-12*** trolling for time and region unobservable Ns*asset 2.38E-13*** effects, we believe they provide sensible Nr*asset -2.84E-14*** estimates for spatial and other factors Nsr*foreign 1.75E-07 Ns*foreign 3.65E-07*** underlying TFP growth among Chinese Nr*foreign -3.06E-07*** firms. Nsr*state 6.09E-07*** Ns*state -9.68E-08*** Conclusions Nr*state -5.52E-08*** Our study, along with recent studies includ- Population 0.0001083** Variety -0.0003379* ing Bai, Hsieh, and Qian (2006) and Bai, Tao, Competition 0.0007005*** and Tong (2008), shows that China, a coun- Labor -2.89E-06*** try with a remarkable economic growth rate, Labor^2 2.17E-11*** has experienced improved resource alloca- Foreign -0.0001842 tion efficiency and exhibited an economic State 0.0001377** development trend consistent with that of Time dummies Yes other countries. Specifically, there is a con- Coast dummy Yes vergence among returns to capital across Number of 45,093 observations regions in China, implying that investment has not been flowing to regions with lower Source: Authors' calculations. *** Significant at 1 percent. returns to capital from those with higher ** Significant at 5 percent. * Significant at 10 percent. returns. Consequently, China's allocation of 292 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA investment across regions has not become 1. We include Hainan as part of Guangdong more inefficient. The U-shaped relationship and Chongqing as part of Sichuan. Tibet is not between regional specialization and per cap- included in our estimate of returns to capital due ita GDP demonstrates that, as an integrated to data limitations. 2. To save space, we only report the mean part of the world's economy, China follows and rank of the variables for some regions. The the same development trend for regional detailed information for all regions is available specialization as other countries. And fierce from the authors upon request. market competition has significantly limited the effectiveness of local governments' pro- References tectionist policies, which has helped to foster production specialization across provinces Bai, Chong-En, Yingjuan Du, Zhigang Tao, in China. These results suggest that market and Sarah Tong. 2004."Local Protectionism and Regional Specialization: Evidence from forces have played an increasingly dominant China's Industries." Journal of International role in China's economy development. Economics 63 (2): 397­417. The findings on agglomeration effects among Chinese firms show significant Bai, Chong-En, Chang-Tai Hsieh, and Yingyi intraindustry and interindustry externali- Qian. 2006."The Return to Capital in China." Brookings Papers on Economic Activity 2: ties. In particular, the number of firms from 61­101. the same sector, either from the same region or from other regions, has a positive effect Bai, Chong-En, Zhigang Tao, and Yueting Sarah on a firm's TFP growth.And the TFP growth Tong. 2008."Bureaucratic Integration and Regional Specialization in China." China of a firm is positively correlated with the Economic Review 19: 308­19. number of other firms in the same region, regardless of the sector. The strength of the Bhagwati, Jagdish N. 1988. Protectionism. spillover effect varies across different types Cambridge, MA: MIT Press. of firms. Specifically, smaller firms and firms Boyreau-Debray, Genevieve, and Shang-Jin Wei. with a higher share of state capital benefit 2005."Pitfalls of a State-Dominated Financial more from the presence of other firms in the System: The Case of China." Working Paper, same sector and in the same region. At the World Bank, Washington, DC. same time, larger firms, firms with a lower Cingano, Federico, and Fabiano Schivardi. 2004. share of state capital, and firms with a higher "Identifying the Sources of Local Productiv- share of foreign capital tend to benefit more ity Growth." Journal of the European Economic from the presence of firms from the same Association 2 (4): 720­42. sector but in other regions; however, larger Combes, Pierre-Philippe. 2000."Economic firms and firms with a higher share of for- Structure and Local Growth: France, 1984­ eign capital appear to enjoy fewer externali- 1993." Journal of Urban Economics 47 (3): ties from other firms in other sectors and in 329­55. the same region. Other factors, including Hoover, Edgar Malone Jr. 1936."The Measure- product variety, local competition, scale ment of Industrial Localization." Review of of local markets, share of state capital, and Economics and Statistics 18 (November): labor force employed, also have a significant 162­71. effect on TFP growth of Chinese firms. Imbs, Jean, and Romain Wacziarg. 2003."Stages of Diversification." American Economic Review 93 (1): 63­86. Notes Kalemli-Ozcan, Sebnem, Bent Sųrensen, Chong-En Bai is chair of the Economics Depart- and Oved Yosha. 2003."Risk Sharing and ment, and Xu Lin is assistant professor in the Industrial Specialization: Regional and Economics Department, both at Tsinghua Uni- International Evidence." American Economic versity in Beijing. We thank Yukon Huang and Review 93 (3): 903­18. other participants of the Tokyo workshop for their valuable comments and Jianhuan Xu for Marshall, Alfred. 1920. Principles of Economics. his excellent research assistance. London: Macmillan. Capital allocation, regional specialization, and spillover effects in China 293 Naughton, Barry. 2003."How Much Can Sah, Raaj Kumar, and Joseph Stiglitz. 1984."The Regional Integration Do to Unify China's Economics of Price Scissors." NBER Working Markets?" In How Far across the River? Paper W1156, National Bureau of Economic Chinese Policy Reform at the Millennium, Research, Cambridge, MA. eds. Nicholas Hope, Dennis Yang, and Mu Young, Alwyn. 2000."The Razor's Edge: Yang Li, pp. 204­32. Palo Alto, CA: Stanford Distortions and Incremental Reform in the University Press. People's Republic of China." Quarterly Journal Porter, Michael E. 1990. The Competitive Advan- of Economics 115 (4): 1091­136. tage of Nations. New York: Free Press. Coastal China's urban-rural spatial restructuring under globalization Yue-man Yeung and Jianfa Shen 18 In China's impressive dash to achieve eco- 2 million inhabitants also grew, from 7 to nomic development and modernization, its 21, and cities with 1 to 2 million inhabit- cities, especially those located in the coastal ants mushroomed from 8 to 30 during the regions, have acted as catalysts, launching same period. the nation on a trajectory of meteoric social Rapid urban restructuring both within transformation and economic uplift. cities and across regions has been the result Because of existing strengths such as of bold changes in policy. The first such pol- economic agglomeration and large concen- icy change was the decision of the central c h a p t e r trations of population, much of the early government to permit the decentralization impulse toward economic development in of authority to provinces, cities, counties, post-reform China was concentrated in the and even enterprise units. With the launch- coastal regions of the country, especially in ing of an open-door policy, the authority to the Pearl River delta, the Yangtze River delta, develop a piece of land was decentralized and the Bohai Bay area in northern China. to such an extent that every unit of author- Urban and regional development in these ity now vies for the opportunity to control three regions has centered on a few large development and generate revenue. Natu- cities such as Beijing, Guangzhou, Shang- rally, this has created excessive competition hai, Shenzhen, and Tianjin, because of their among different levels of government. The favorable initial conditions, coastal location, result has been wasteful redundancy and their attraction for foreign investors, and the inefficient land use practices. A footnote presence of strong state-owned enterprises to the decentralization of power to cities, and active municipal governments. which allowed them to make plans for their Development in these regions and cit- development, is the fact that certain cities, ies has been stimulated both by the forces like Shantou, Shenzhen, Xiamen, and Zhu- of globalization, notably through foreign hai, were designated special economic zones investment and international trade, and by (SEZs) in 1980 or thereabouts and given the internal forces, such as local development specific power to experiment with new poli- initiatives, decentralization, and marketiza- cies. For example, in 1987 Shenzhen held the tion. The strong force of urban agglomera- first land auction in post-1949 China. It was tion in coastal China has produced a model modeled after the Hong Kong system of gov- of development that is led by large cities. ernment ownership of land, with leaseholds At the macro level, the most important and granted by the state for a specified period obvious change is the exceptionally rapid of time in exchange for a fee. The land lease growth of China's cities during the reform model, having been implemented success- period. The number of cities in the coun- fully in Shenzhen, has since been widely try grew from 223 in 1980 to 649 in 2004. replicated across cities in China, even in far- The number of mega cities with more than away inland cities. The spread of this model Coastal China's urban-rural spatial restructuring under globalization 295 has been critical to the ability of Chinese mised more on market principles than on cities to pursue rapid physical growth and central planning. The result has been a pro- modernization. found change in the development landscape, Apart from the fact that the leasing of with the relative decline of the industrial land provided a major source of revenue north and northeast areas that had been for city administration, another factor that favored since the mid-1950s in previous changed the fiscal position of cities is the national development plans. Instead, the change, implemented during the reform decentralization of decision making has led period, in the central government's fiscal to the emergence of the coastal regions and relations with provincial governments as the south as new centers of growth, with well as with subordinate units such as cit- their cities powering ahead of the rest of the ies. Since 1978, the central government has country in development. An analysis of the progressively relaxed the highly centralized gross domestic product (GDP) of the vari- fiscal system, giving provinces and cities a ous provinces from 1978 to 1995 has borne much greater degree of freedom to pursue out this dramatic shift in the focus of devel- development. Various practices that have opment to the south (Lin 1999). This shift been implemented at different times and in has also led to a change in the relationship different provinces include the contracting between cities in China and in the propor- out of fiscal duties, the remittance of a fixed tion of cities of different sizes.Whereas in the proportion of a locality's fiscal surplus to the pre-reform period the urban hierarchy was central government, the sharing of revenues organized by vertical linkages and political between the central government and the functionality, cities are now shaped primar- provinces, the launching of tax-for-profit ily by horizontal connections and economic reforms, and the assignment of taxes. In the exchanges. The number of cities has vastly spirit of openness and decentralization, the increased. As noted, small and medium-size central government has taken a more liberal cities in particular have proliferated. The attitude toward fiscal administration and pace of the increase is astonishing. Conse- has fine-tuned or adjusted policies when it quently, since 1978, China's spatial trans- has perceived the need to do so. formation can be characterized as a reor- During the reform period, the coastal ganization of spatial relationships between provinces of Fujian and Guangdong northern and southern China, between large enjoyed preferential policies, while the three cities and small towns, and between cities provincial-level municipalities of Beijing, and the countryside (Lin 1999). Shanghai, and Tianjin were kept under The fourth factor follows from the rise tighter control. Since 1994, Shanghai has of small towns and cities: the mushrooming also enjoyed a "tax-sharing system" of the of township-and-village enterprises (TVEs) sort previously implemented only in Fujian across coastal China during the reform and Guangdong (Yeung and Sung 1996: period. In the reform period, rural urbaniza- 9). Shanghai's rapid development since the tion was led by the TVEs, with the active par- early 1990s can be traced in part to this fiscal ticipation and support of local governments. reform. The changes in China's fiscal system Their growth and expansion have been major since 1978 have led to a situation in which componentsof locallydrivenurbanization,or the tax revenues of a locality are strongly and urbanization from below. This has given rise positively associated with its level of devel- to the phenomenon of dual-track urbaniza- opment and rate of urbanization. Moreover, tion, a process in which the nonagricultural the fiscal reforms have created uneven fiscal population of cities has grown rapidly, com- relations between the center and different plemented by rural urbanization generated provinces, contributing to regional inequali- from below (Shen 2006).The rapid growth of ties (Shen 2005; Wei 1996). small towns and cities has, to a degree, been The third factor in the restructuring of fueled by the long-standing urban planning Chinese cities in the reform period is the guideline of strictly controlling the growth fact that urban development became pre- of large cities and promoting the growth of 296 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA small and medium-size cities. This planning struction and a wide variety of services, espe- guideline has not been adhered to strictly, cially in "3-D" occupations--that is, those as can be seen by the rapid growth of very that are dirty,dangerous,and difficult.Recent large cities, a result that is compatible with estimates have placed the temporary popula- economic logic and trends in urban develop- tion of China at around 120 million. These ment at the global level. migrants are not counted in official statistics The fifth and final factor accounting for on the hukou population, which means that the rapid change in the urban landscape in many urban population figures, especially contemporary China is the hukou (house- those for large coastal cities, clearly underes- hold registration) system,established in 1958 timate the real situation, because a quarter to to control population movements and essen- a third of the population of such cities can tially immobilize the urban and rural popula- be traced to this source. Cities like Shanghai tions. In 1984 the government began to relax and Shenzhen have introduced "blue stamp" what had previously been a rigid system. hukou to cope with some"floaters,"but many Since then, rural migrants have descended en cities have found these new migrants to be masse on coastal cities. Referred to as a tem- both a boon and a bane (Yeung 2002). porary or floating population,rural migrants The following sections examine the provide the crucial labor force for urban con- process of urban development and spatial restructuring in the economically most advanced regions of the country: the Pearl Figure 18.1 Three coastal regions in China River delta, the Yangtze River delta, and the Bohai Bay area (see figure 18.1). More than any other part of the country, these three coastal regions have been influenced by the forces of globalization, which have been accelerating since the early 1980s. The timing has been fortuitous for China, as this development has coincided with the early phase of the country's opening to the outside world. The conjuncture of these JJJ two processes has been highly beneficial for Liaodong China's rapid development. Tables 18.1 and BOHAI Peninsula 18.2 present a summary of the main demo- RING graphic and economic indicators for Beijing, TBND Guangzhou, Shanghai, and Shenzhen, four major cities in the three regions. In 2005 Shandong Peninsula each city had a population of more than 8 million. They are the most advanced cities in China, with per capita GDP well over the YRD national average. The rest of the chapter refers to these tables from time to time. Guangdong and the Pearl River delta In the early 1980s, the provinces of Fujian and Guangdong were chosen to carry out PRD experiments under China's open-door policy, largely because of their history of early contact with Western countries; thus they were considered to be more in tune with worldly developments than other areas Note: JJJ refers to the Beijing-Tianjin-Hebei (Jing-Jin-Ji) region; TBND refers of the country. These two provinces were to Tianjin Binhai New District. YRD refers to Yangtze River delta. PRD refers to Pearl River delta. allowed more flexibility in trying out new Coastal China's urban-rural spatial restructuring under globalization 297 Table 18.1 Demographic and GDP indicators of Guangzhou, Shenzhen, Shanghai, Beijing, and China, 1980­2005 Year Guangzhou Shenzhen Shanghai Beijing China Area (thousand square kilometers) 2005 7.4 2.0 6.3 16.4 9,600.0 Population (millions) 1980 5.02 0.33 11.57 9.04 987.05 1990 6.30 1.68 13.32 10.86 1,143.33 2000 9.95 7.01 16.21 13.64 1,267.43 2005 9.50 8.28 17.99 15.38 1,307.56 Population density (persons per square kilometer) 1980 427 170 1,870 551 103 1990 847 859 2,101 662 119 2000 1,338 3,591 2,556 831 132 2005 1,277 4,239 2,837 937 136 GDP (current price, yuan billion) 1980 5.8 0.3 31.2 13.9 454.6 1990 32.0 17.2 78.2 50.1 1,871.8 2000 249.3 218.7 477.1 316.1 9,800.1 2005 515.4 495.1 916.4 688.6 18,395.6 GDP per capita (current price, yuan per person) 1980 1,160 835 2,719 1,544 463 1990 5,418 8,724 5,891 4,635 1,644 2000 25,626 32,800 29,786 24,122 7,858 2005 53,809 60,801 51,461 45,444 14,040 Average annual growth rate of GDP (based on fixed price, percent) 1981­90 11.8 35.7 7.4 8.8 9.3 1991­2000 16.6 23.2 12.3 11.0 10.3 2001­05 13.8 16.3 11.9 12.0 9.9 1981­2005 14.1 26.6 10.2 10.3 9.8 Sources: Compiled or calculated using data from BMBS (2006); DCA and MPH (2000); GDPBS (1992, 2006); GMSB (2006); NBS (2006); SMSB (2001, 2006); SSB (2006); Guangzhou Economic Yearbook Editorial Committee (1983). Note: All population data refer to usual residents, except the 1980 figure for Guangzhou, which is based on hukou population. The 1990 figure for Guangzhou is from the 1990 census (GDPBS 1992: 132). policies, particularly with respect to luring enjoyed great fiscal autonomy by being foreign investment and promoting trade allowed to remit a lump sum to the central and development. The SEZs and develop- government for five years, much to the envy ment zones were allowed even greater free- of other provinces. In 1988 Guangdong was dom to pursue new development policies. allowed a fixed quota, with adjustments for Similarly, from 1984 onward, China's coastal growth, in which the central government's cities, including Guangzhou and Zhanjiang share was small. Shenzhen was exempt from in Guangdong, were allowed more leeway to remitting anything at all to the central gov- pursue development and experimentation ernment for 10 years, until 2003 (Shi 2003). (Yeung and Hu 1992).1 Within Guangdong, the most devel- Fiscal reform constitutes a critical oped area of the province consists of nine dimension of economic reform in China prefecture-level cities--Dongguan, Foshan, and has had a powerful impact on regional Guangzhou, Jiangmen, Shenzhen, Zhong- development. As early as 1979, Fujian and shan, Zhuhai, and part of Huizhou and Guangdong were candidates for fiscal exper- Zhaoqing--that form the Pearl River delta imentation. For the first time, Guangdong region (see figure 18.2). They make up 298 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 18.2 Share of secondary and tertiary industries in GDP, exports, and realized foreign capital in Guangzhou, Shenzhen, Shanghai, Beijing, and China, 1980­2005 Year Guangzhou Shenzhen Shanghai Beijing China Share of secondary industry (percent) 1980 54.5 26.0 75.7 68.9 48.2 1990 42.6 44.8 64.7 52.4 41.3 2000 41.0 49.7 46.3 32.7 45.9 2005 39.7 53.2 48.6 29.4 47.5 Share of tertiary industry (percent) 1980 34.6 45.1 21.1 26.7 21.9 1990 49.3 51.1 30.9 38.8 31.8 2000 55.2 49.6 52.1 64.9 39.3 2005 57.8 46.6 50.5 69.2 39.9 Exports (US$ billion) 1980 0.19 0.01 4.27 0.59 18.12 1990 2.36 8.15 5.32 1.12 62.09 2000 11.79 34.56 25.35 4.63 249.20 2005 26.67 101.52 90.74 17.10 761.95 Realized foreign capital (US$ billion) 1980 0.03 0.03 0.02 -- 1.98 1990 0.27 0.52 0.18 0.39 10.29 2000 3.12 2.97 3.16 1.68 59.36 2005 2.84 4.02 6.85 3.53 63.81 Source: Compiled or calculated using data from BMBS (2006); GMSB (2006); Guangzhou Economic Yearbook Editorial Committee (1983); NBS (1982, 1991, 2001, 2006) SMSB (2001, 2006); SSB (2006). Note: 1980 figure for Guangzhou exports is from Guangzhou Economic Yearbook Editorial Committee (1983: 557) and converted to U.S. dol- lars using the 1981 US$ to yuan exchange rate from NBS (1982). Realized foreign capital refers to 1981 for Shanghai in 1980 and to 1983 for China in 1980. Realized foreign capital for Shenzhen in 2005 excludes foreign loans. -- Not available. 23 percent of Guangdong's land area but the structural change of its economic base contribute 80 percent of its GDP, or one- from an agricultural to an industrial orien- tenth that of the nation. The Pearl River tation. The ratio of the contribution of the delta has achieved a level of urbanization primary, secondary, and tertiary industries of 72.7 percent (Ta Kung Pao, January 18, to Guangdong's GDP has changed sharply, 2005). These figures speak volumes about evolving from 33.2, 41.1, and 25.7 percent, the achievements that the Pearl River delta respectively,in 1980 to 6.0,51.3,and 42.7 per- has made after almost three decades of cent, respectively, in 2006 (GDPBS 2007). In openness. By now, the delta has developed 27 years,the secondary sector has become far into the country's largest production center more important, and the tertiary sector has for the electronics and information tech- grown rapidly, at the obvious expense of the nology industries, accounting for as much primary sector. In 1980­2005 Shenzhen led as 40 percent of the world's production of the province with a rising secondary sector some computer components (Chen and and a stable tertiary sector,while Guangzhou others 2003: 38). A large proportion of the experienced a significant increase in the share nation's production of home appliances of the tertiary sector (table 18.2). is manufactured in this province, such as electric fans (88.2 percent), air conditioners Density, agglomeration, and (38.2 percent), bicycles (35.7 percent), and economic growth refrigerators (25.1 percent; Chen and others Many factors are behind the rapid economic 2003: 25; also Enright and others 2003: 40). development that has occurred in the Pearl One of the most fundamental changes in River delta since the early 1980s. One is the Guangdong in the reform period has been agglomeration effects of towns and cities that Coastal China's urban-rural spatial restructuring under globalization 299 Figure 18.2 The Pearl River delta region and its constituent cities Pearl River Delta Watershed Prefecture-level City Boundaries Guangzhou Zhaoqing Foshan Huizhou Dongguan Shenzhen Zhongshan Hong Kong Jiangmen Zhuhai Macao Source: Yeung (2003: iv). have found a particular niche in economic (table 18.1). Guangzhou adopted a new production. Many towns in the region, espe- spatial policy of "expansion in the south, cially in the western wing of the delta, have optimization in the north, advancement in achieved success in concentrating on and the east, and linkage in the west." True to perfecting a single industry, a phenomenon the spirit of this policy, Guangzhou devel- that has been dubbed "one town, one indus- oped the mammoth new Baiyun airport in try" (Yeung, Shen, and Zhang 2005). Many the north in 2004 and gained direct access of these towns and cities have grown at an to the sea in the south by annexing Panyu astonishing rate, both in economic output in 2000. Toward the west, the Guangzhou- and population. New cities that had been Foshan twin-city region is being consoli- accorded special status, such as the SEZs dated. In an easterly direction, Guangzhou of Shenzhen and Zhuhai, have grown even has been expanding and consolidating its faster, with the former exploding from a bor- land transport network with Dongguan. der town of only 0.33 million inhabitants in Guangzhou has other infrastructure devel- 1978 to a metropolis of 8.28 million in 2005 opment plans that are envisaged to enhance (SSB 2006). The population of Shenzhen its pivotal geographic role in the province, grew rapidly in the 1990s, and its population through the purposeful construction of density reached 4,239 persons per square new expressways, railways, and light rail- kilometer by 2005, the highest among cities ways. Guangzhou and Shenzhen grew faster in mainland China (table 18.1). than Beijing and Shanghai throughout the Another factor spearheading urban- period of 1980­2005. By 2005, Guangzhou regional change is the role played by Guang- and Shenzhen had a higher per capita GDP zhou, the provincial capital. While it grew than Beijing and Shanghai (table 18.1). more slowly than Shenzhen in the 1980s, The third factor that has led to the Guangzhou caught up rapidly in the 1990s increasing density of human settlements 300 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA and to greater economic development is the in 2000­05. Exports from Shenzhen reached dual-track form of urbanization, which was US$101.52 billion in 2005, even greater than referred to earlier. The contribution of TVEs Shanghai. It is clear that a portion of foreign is a component of rural urbanization, with investment in Shanghai targets the domestic Hong Kong businessmen having invested in market of China, such as the industry of car no fewer than 66,000 such firms. manufacturing, while Shenzhen's industry is The fourth and last factor that has facili- mainly export oriented. tated the concentration of economic and An idea of how the Pearl River delta has urban activities in the Pearl River delta been developing into a densely populated, is massive and sustained investments in economically vibrant, and environmentally infrastructure. Expressways, railways, ports, sustainable region can be seen in figure 18.3, airports, power stations, and telecom- which shows an urban-regional plan for the munications facilities have developed at a delta until 2020. This plan was adopted in breakneck pace. 2005 by the Guangdong People's Congress. The region would not have developed so It envisages limiting the area of develop- rapidly had it not been for the fact that the ment in the delta to around 7,800 square open-doorpolicywasattractivetoHongKong kilometers, capping the population at about industrialists, who were looking for ways 65 million, and engaging in regional infra- to relocate their manufacturing operations structure planning to cope with a popula- away from the constrained environment of tion of 80 million. The guiding principle of the city-state, which was experiencing rising planning for coordinated development is to wages and land prices.The town of Shenzhen, "strengthen development centers and cre- located just across a narrow river from Hong ate spines and corridors." More specifically, Kong, offered a plentiful amount of land and "one spine, three corridors, and five belts" an almost inexhaustible supply of labor from is the broad spatial framework for coor- all over the country,along with favorable poli- dinating and maximizing the urban and cies for investors. Hong Kong entrepreneurs regional strengths of the delta (Yeung 2005). responded positively, moving their industrial The urban-regional plan was initiated by production operations en masse, initially to the provincial and central authorities and Shenzhen and then to the next town, Dong- designed to minimize the costly problem of guan, and elsewhere in the delta. A symbi- redundancy and overlap that has character- otic relationship was soon established, with ized some infrastructure and other devel- Hong Kong providing capital, modern man- opments in the delta's cities. As the plan agement skills, and information about global was undertaken and approved without the markets and the Pearl River delta offering participation of the authorities of the Hong cheap land, labor, and favorable economic Kong and Macao special administrative policies. Hong Kong has been serving as a regions, it is being updated by another more frontshopfromwhichbusinesspeoplefacethe holistic study involving the original delta world,handling production orders,designing cities plus Hong Kong and Macao. Called products, and making major investment and the Greater Pearl River Delta Urban Coor- management decisions, whereas the plants in dinated Plan, the study is scheduled to be the delta are engaged in the actual production completed in 2008. process. This mutually beneficial model is With Hong Kong and Macao having widelyknownasthe"frontshop,backfactory" become part of China in 1997 and 1999, arrangement. As many as 10 million work- respectively, the integration of these special ers are currently employed in these factories jurisdictions with the mainland is ongoing. in the Pearl River delta, which are financed In the first decade since Hong Kong's return and managed by Hong Kong interests. Such to Chinese sovereignty, the pace of integra- investment also contributes to rapid growth tion,whether measured by the flow of people, of exports.As shown in table 18.2,Guangzhou goods,capital,or information,has been rapid. and Shenzhen each attracted more foreign With ever more Hong Kong people traveling, capital than Shanghai in the 1980s and 1990s, working, and retiring in the Pearl River delta although they were overtaken by Shanghai and beyond, cities in the delta are increas- Coastal China's urban-rural spatial restructuring under globalization 301 Figure 18.3 Urban cluster-coordinated development plan of the Pearl River delta, 2020 Zhaoqing Foshan Guangzhou Zengcheng Huizhou Sandong Shunde Gaoming Guangzhou Heshan new town Changping Xiaolan Nansha Humen Songshanhu Jiangmen Zhangmutou Zhongshan Guangming Pinghu Yinzhouhu Qianhai Huiyang Zhuhai Zhuhai port zone Jinwan Macau Hong Kong Yantian Daya Bay Source: Yeung (2005: 9). ingly catering to their needs, in areas such as In ocean transport, container throughput in the provision of housing, medical care, work, thethree-portclusterofHongKong,Shenzhen, schooling, and so on (Yeung 2007). and Zhuhai is more than 30 million 20-foot Against this trend toward closer inte- equivalent units, surpassing any other port gration, some recent policies have fostered cluster in Asia (Chen and others 2003: 165). better economic relations between Hong Connectivity between Hong Kong and Kong and the mainland. Notable among the delta has continued to improve. The these are the Individual Visit Scheme and 5-kilometer western corridor link between the Closer Economic Partnership Arrange- Hong Kong and Shenzhen, consisting of a ment, which were announced in 2003 and bridge and a highway, was opened in July implemented beginning in 2004. 2007, followed the next month by the open- ing of a rail spur line to the border at Lok Improving regional transport Ma Chau. The regional express railway link- infrastructure ing Hong Kong with Guangzhou through Feverish construction continues in the Shenzhen is under construction and due region. In the four years from 1996 to 2000 for completion in 2010­14. In contrast to alone, Guangzhou invested Y 60.5 billion the situation before 1997, Hong Kong and in urban construction, more than twice its people clearly are increasingly taken into the amount of its investment in the 47 consideration in plans for the construction years prior to 1996 (Chen and others 2003: of large-scale infrastructure projects in the 108).2 Guangdong now has 3,140 kilome- delta. There is little doubt that infrastructure ters (2005) of expressways, most of them will continue to be an important avenue in the Pearl River delta. Guangzhou is the along which the density of development in hub of many of these expressways, which is the Pearl River delta will thicken and lead reinforcing the city's bid to become a lead- to more economic and social gains for the ing urban center. Within a radius of 100 people (Yeung and Kee 2007). kilometers, the delta has five large airports In the development of the Greater Pearl that together offer a daily total of 550 inter- River delta region--that is, the Pearl River national flights and 600 domestic flights. delta plus Hong Kong and Macao--the 302 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA concept of the"magic three hours" has been pool of talent, and superior infrastructure popularized, meaning that the aim is to facilities. This agglomeration will continue enable a businessman to travel from Hong until developed areas, usually led by central Kong to the Pearl River delta in three hours, cities, no longer enjoy economies of scale. allowing him to complete his day's work the At this point, faced with escalating land same day (Enright and others 2003). The prices and a rising cost of living, polariza- present direct rail link between Guangzhou tion begins to reverse. and Hong Kong, which runs 12 times a day In Guangdong, the transformation from in each direction, will be halved to less than centrally planned to market-oriented devel- an hour. The physical connectivity between opment was basically completed in all 21 cities in the delta will continue to improve prefecture-level cities by the 1990s. By the with the application of new transport and end of 1994,95 percent of all retail commod- telecommunications technology. ities, about 94 percent of agricultural prod- The area of Guangdong totals 177,901 ucts, and 93 percent of production materials square kilometers, representing 1.9 percent were regulated by market mechanisms.All of of China's total area. Within the province, this shows that the opening up of the market considerable intraregional differentiation occurred at a measured, but substantial, pace and disparities exist. The contrast between with regulatory controls (Zhou 1995). those areas of the province that are located The tendency has been for economic within the Pearl River delta and the periph- development in the Pearl River delta to be eral areas is striking, both in terms of level concentrated in the following six cities: of development and ecological endow- Dongguan, Foshan, Guangzhou, Shenzhen, ments. Within the delta itself, there is a Zhongshan, and Zhuhai. In 2005 these cities significant difference between the eastern accounted for 10.7 percent of the total area and western wings. The eastern wing, rep- of the province, but 37 percent of its hukou resented by Dongguan and Shenzhen, has population, 80.4 percent of its fiscal rev- experienced rapid development during the enues, and 70.5 percent of its GDP. In addi- reform period. This is primarily the result tion, the relative concentration of people, of the benefits derived from its geographic wealth, and goods in the province exceeded proximity to Hong Kong. The western wing, Guangdong's relative area by 26.3, 69.7, and which includes Zhongshan, Zhuhai, and 59.8 percent, respectively (GDPBS 2006). other cities, has lagged behind, although this Over the past five years,the composite devel- area was economically and culturally more opment of the province improved, although advanced than the eastern wing before the a weakening trend is probable in the future. onset of economic reforms. Their economic, However, the peripheral areas of the north social, and general conditions have reversed and the west, together with the mountain- since the 1980s, a result that is traceable to ous areas, were relatively weak over the past the presence or absence of the "Hong Kong five years but are expected to perform more factor." By the same token, much of the strongly in the future. In addition, consid- available land in the eastern wing has been ering the phenomenal growth of the prov- developed, leaving the western wing, espe- ince's GDP, with a 14.7 percent rise over the cially Jiangmen, a relative latecomer, with previous year, per capita GDP of US$4,915, much more room for development (Yeung, and a tertiary sector contributing 46.2 per- Shen, and Zhang 2005). cent to GDP in 2005, there is considerable Traditional studies on regional econom- scope for economic agglomeration before ics have postulated that, in the experience the stage of polarization reversal is reached. of Western countries, agglomeration effects Nevertheless, certain industries, especially take place in developed areas toward which labor-intensive ones in the electronics, labor, capital, and other factors of produc- telecommunications, and home appliance tion flow from less-developed areas. Central sectors, are beginning to relocate from the cities derive advantages on account of their developed core to the peripheral areas. economies of scale, the efficiency of their The end effect of the structural changes markets, circulation of information, large and geographic relocation of industries is a Coastal China's urban-rural spatial restructuring under globalization 303 noticeable shrinking of the development gap 2004. One out of three jobs in Guangdong between the Pearl River delta and periph- is held by someone whose"household regis- eral areas. This is a major move toward the tration"is elsewhere. The labor shortage has goal of building a xiaokang--or moderately prompted companies to begin moving their affluent--society, which was set out as a operations farther inland to provinces such national development objective in the Tenth as Hunan and Jiangxi,a process facilitated by Five-Year Plan, which concluded in 2005. the recent formation of the Pan-Pearl River In fact, by 1997, Guangdong had already Delta Framework. In the long run, China is reached a stage of development beyond that facing a labor shortage, as its population is of being able to meet basic needs. In 2003, aging rapidly (Yang 2005). 16 poor counties in Guangdong were able to The second factor that has elevated the shake off poverty; by 2005, 50 poor counties economic status of backward areas within in the mountainous region had reached xiao- Guangdong is the policy of paired assistance kang status (Chen and others 2003: 113). development, whereby an economically advanced city is paired with an economi- Unbalanced development and cally laggard city or county to encourage regional cooperation the former to provide assistance to the lat- For decades or more, physical and artifi- ter, in the form of fiscal allocations, policy cial barriers have separated the developed support, and technology transfers to alle- from the economically laggard areas within viate poverty and accelerate development. Guangdong. Until recently, it seemed that This policy has, in fact, been applied across the considerable regional disparities would the nation and has helped backward cit- be perpetuated. However, after almost three ies and regions to achieve rapid economic decades of reform, the barriers to factor progress. Especially prominent has been mobility have been coming down. the recent effort to pair up cities, and even The first barrier to fall was the mobility provinces, in the western region, where of people. Since 1984, people from the less- many minority groups live, with thriv- developed areas of the province and, in fact, ing cities and provinces along the coast. from other provinces, have been allowed to For example, the economically advanced move to more-developed areas, as the long- Jiangsu has been paired with the relatively standing hukou or household registration underdeveloped Guangxi under the paired system was relaxed. For the most part, rural assistance program. This is a crucial policy people have been allowed to leave rural areas aimed at spreading the positive effects of and move to cities to work, where they are rapid coastal development to other parts of considered a temporary or floating popu- the country, thereby minimizing the prob- lation. They have become a critical source lem of growing regional disparities within of the labor that has fueled the engine of the country (Yeung and Shen 2004). growth in the delta. For years, the recruit- In Guangdong, the policy of pairing a ment of rural labor has proceeded success- developed city with a developing city has fully. The process has helped to minimize been implemented since the mid-1980s. rural-urban and interregional disparities, During the past 3 years, 7 developed as rural laborers have remitted sizable sums cities in the Pearl River delta have pro- to their families back home. Lately, rising vided paired assistance to 20 counties wages and competition for labor from the in hilly areas, with outstanding results. Yangtze River delta region have posed new Over that period, Y 510 million was spent challenges for factories in the Pearl River under the program, and 221,000 work- delta. Many rural workers from inland ers from the hilly counties were gain- areas have been less inclined to travel long fully employed, earning a total income of distances to the coastal cities, preferring to Y 930 million. A total of 328 trade and eco- take advantage of improved opportunities nomic cooperation projects were proposed, back home. The migrant labor population 265 of which have been implemented, in Guangdong is huge, with an official esti- involving a total investment of Y 845 mil- mate placing it at 21.3 million in February lion. Under the current Eleventh Five-Year 304 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Plan (2006­10), the policy of pairing 7 had a superior agglomeration economy and developed cities with 20 backward counties embedded economic strength. With a popu- remains unchanged (Nanfang Ribao, June lation of 5.73 million in 1952, Shanghai's 17, 2006). per capita GDP (Y 590) was about five times Of the hilly and less-developed cities in the national average (Y 119) that year (NBS, peripheral areas of Guangdong, Qingyuan Department of Comprehensive Statistics is noteworthy. Richly endowed in natural 1999). The city was transformed from an resources, its rapid development is due in advanced service center to an industrial pow- part to its recent investments in infrastruc- erhouse during the decades of the 1950s to ture. The opening in 2004 of the expressway the 1970s in the Maoist period when China linking the city to Guangzhou at a distance was closed to the outside world. In the 1980s, of 60 kilometers has put it within a half- economic reform and foreign investment hour commute of the provincial capital,thus policies focused on Guangdong, and there making Qingyuan more accessible to other was no significant development in Shanghai, areas of the Pearl River delta. This greatly which had a GDP growth rate of 7.4 percent a enhances Qingyuan's position as a new focus year, below the national average (table 18.1). of development in northern Guangdong. The city's golden opportunity came in There has been a marked increase in foreign 1990 when the central government made the investment in the city, an enhancement of development of the Pudong New District in its strategic location as part of a north-south Shanghai a priority. Shanghai was desig- development corridor, and a renewed com- nated by the central authorities to become mitment to harness its plentiful resources in the leading economic center in China, and minerals, agriculture, and tourism. Conse- this resolution was backed up with the most quently, Qingyuan's GDP soared more than preferential policies for economic develop- 7.7 times between 1990 and 2005, reaching ment and foreign investment. In less than Y 32 billion in 2005. Foreign investment a decade, a new Shanghai emerged, with reached US$173.8 million in 2005, a nine- advanced infrastructure and facilities. fold increase from 1990. The investment in infrastructure has been fueled by favorable fiscal policies and Shanghai and the Yangtze land redevelopment in the urban area. River delta The reform of the land market has made it possible for the city government to collect The chapter now turns to Shanghai, the substantial revenues by converting indus- hub of the Yangtze River delta. This sec- trial land in the urban area to commercial tion attempts to show the contribution of and residential uses. The change from low- urban agglomeration to economic growth density to high-density land use has led to in Shanghai, Shanghai's spatial restructur- substantial increases in land value. Another ing, the role of development zones, and the major policy change on the part of the cen- economic diffusion from Shanghai to other tral government has been to allow Shanghai cities in the Yangtze River delta region. to retain more of its fiscal revenues to enable Shanghai is considered as an emerging the city to make major investments in eco- world city, and it is revealed that Shanghai's nomic restructuring and infrastructure. In major role in the delta is the diffusion of 1978 Shanghai's fiscal expenditures only human resources and provision of producer accounted for 13.6 percent of its total fiscal services. Foreign investors are still the major revenues, with more than 86 percent being source of capital and technology in delta. transferred to the central government. After the implementation of tax reforms and the Density, agglomeration, and launching of the Pudong development strat- economic growth egy,fiscal expenditures increased to 22.9 per- Shanghai has been the largest city in China cent of total fiscal revenues in 1986 and to and its leading economic center since the 38.1 percent in 1995. The figure remained at early twentieth century (Yeung and Sung around 36­40 percent from 1995­2006. The 1996). Throughout its history as a city, it has Shanghai government's fiscal expenditures Coastal China's urban-rural spatial restructuring under globalization 305 increased from Y 2.60 billion in 1978 to Y contributing to economic growth and to 181.38 billion in 2006 (SMSB 2007). Clearly, the rise in per capita income in Shanghai. Shanghai's dramatic development since the However, migrants have provided the bulk 1990s has been due to the decision to allow of the labor required to construct the city's the Shanghai municipal government to rein- many large infrastructure projects, which vest a greater portion of its fiscal revenues have been completed within a short period in the city rather than due to the transfer of of time. financial resources from the central govern- Many other Chinese cities have also ment to the city. grown rapidly in the reform period, due Shanghai has grown in size since the to a similar agglomeration effect. In 1978 1980s. The optimal size of Shanghai's pop- Shanghai's per capita GDP was Y 2,529, well ulation is an issue that scholars and the above that of other cities in China. The gap government have debated keenly, without between Shanghai and the rest of China coming to a definite conclusion (Zhou, narrowed significantly during the reform Yang, and Xiao 2005). The long-standing period. In 2005 Shanghai's per capita GDP official policy has been to control the growth was Y 51,461,3 well above the average for of large cities such as Shanghai. In the 1980s, mainland China of Y 14,040 and for Guang- a temporary population was allowed to dong as a whole of Y 24,438 (table 18.1). form in Shanghai. Since 1994, the Shanghai However, due to particularly favorable poli- government has adopted more relaxed and cies for parts of Guangdong in the 1980s and positive migration policies. Incentives have to the influence of Hong Kong, Shenzhen's been offered to encourage skilled and well- per capita GDP overtook Shanghai's in educated people from within and outside 1984. Guangzhou only overtook Shanghai in China to move to the city, while more and per capita GDP in 2005. Within the Yangtze more social services have been extended to River delta, Suzhou had a higher per capita include the temporary population. About GDP than Shanghai in 2005 (GMSB 2006; 30,000 people acquired the "blue-chop" JPBS 2006; NBS 2006; SMSB 2007; SSB hukou in Shanghai from 1994­2000, and 2006). These cases have three implications. about 20,000 of these acquired this status First, several economic centers may develop because of their investment in residen- in mainland China along with Shang- tial property (Kong 2001; Shen 2006). The hai. Second, the scale of these centers may "blue-chop" hukou was valid only in the expand further without significant negative city, while the normal"red-chop"hukou was externalities, as shown by Shanghai. Third, valid in the whole country. By the end of Shanghai's central controlling function is 2006, 0.12 million foreigners were living in still limited, and it has made limited contri- Shanghai (SMSB 2007). butions to its hinterland in the Yangtze River Shanghai's total population (its usual delta and the rest of China in terms of trans- residents) increased from 12 million in 1982 fers of capital and technology. Shanghai is to 18.2 million in 2006. Significant growth still building its capacity in these areas and came from the temporary population,which can only be considered an emerging world reached 4.7 million in 2006. With only 0.1 city. In the world urban system, Shanghai is percent of China's area and 1.4 percent of performing as a "satellite-type" base for FDI China's population, Shanghai accounted and as a regional headquarters for foreign for 5 percent of the nation's GDP, 12.2 per- companies (Huang, Leung, and Shen 2007; cent of its fiscal revenues, 27.5 percent of its Markusen 1996; Park 1996; Wei and Leung exports, and 10.2 percent of its foreign direct 2005). Foreign investors are still the major investment (FDI) in 2006 (SMSB 2007).At a source of capital and technology in these cit- fixed price, Shanghai's GDP grew 12.2 per- ies as well as in Shanghai. cent a year, and per capita GDP grew 10.0 FDI has contributed to Shanghai's dra- percent a year from 1991 to 2006. During matic economic growth since the early the same period, the city's exports grew 23.9 1990s (Wei and Leung 2005). Shanghai percent a year, while FDI grew 26.0 percent has become a hot spot for foreign invest- a year. It is likely that population growth is ment due to its advantageous geographic 306 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA location, solid economic foundation, busi- expected to enhance the city's competitive- ness culture, human resources, and favorable ness according to the theory of new eco- open polices. By 2006, total FDI in Shanghai nomic geography (Fujita, Krugman, and reached US$66.76 billion, distributed almost Venables 2001; Krugman 2007). equally between the secondary and tertiary The service sector picked up momen- sectors. Hong Kong, China, was the largest tum in the early 1990s when Shanghai was source of FDI, followed by Japan, the United designated to become the economic center States; Germany; Taiwan, China; Singapore; of China. In 1995 the central government and the United Kingdom. In 2006 foreign- proposed building Shanghai into an interna- funded enterprises accounted for 40.0 tional economic, financial, trade, and ship- percent of industrial output in the city. In ping center. The four-centers strategy is in addition, Hong Kong; Macao-, and Taiwan, line with the world city postulation, which China-funded enterprises also contributed emphasizes the growth of advanced producer 15.2 percent of Shanghai's industrial output services and the controlling function of the (SMSB 2007).As many foreign-funded enter- city in regional and global economies. A prises are engaged in outward processing and number of giant projects have laid the foun- assembly operations requiring large-scale dation for Shanghai to become a world city imports and exports, their share of exports (Shi and Hamnett 2002; Wu 2000;Yusuf and has increased greatly (table 18.2). To some Wu 2002).These projects include a new mass extent, the growth of export-oriented manu- transit railway system, the Lujiazui financial facturing in the Pearl River delta in the 1980s and trade zone, a new airport in Pudong, and 1990s was reproduced in Shanghai after and more recently the new Yangshan con- 2000. In 2006 foreign-funded enterprises tainer port. The completion of these plat- contributed66.9percentof totalexportsfrom forms facilitates the agglomeration of service Shanghai. In Shanghai, the share of exports industries, which further attracts capital, tal- due to outward processing also reached 56.2 ent, and businesses. Since 2000, the share of percent that year (SMSB 2007). the tertiary industry in Shanghai's GDP has As a major development strategy, the exceeded 50 percent (table 18.2). municipal government has made some The Shanghai government has adopted attempts to develop competitive pillar indus- various policies to attract transnational tries by nurturing indigenous enterprises, corporations to set up regional headquar- attracting FDI, and promoting Sino-foreign ters and research and development (R&D) joint ventures. In the Eighth Five-Year Plan centers in the city. In 2004 Shanghai had 86 period of 1991­95, automobile manufac- regional headquarters of transnational cor- turing, electronic and telecommunications porations, 105 investment companies, and equipment manufacturing, steel manufac- 142 R&D centers funded by foreign com- turing, petrochemical and fine chemical panies. There were 63 foreign banks and product manufacturing, power plant equip- financial companies and 24 foreign insur- ment and large-scale electric equipment ance companies in 2006 (Fan 2007). About manufacturing, and household electrical 80 percent of the top 50 banks in the world appliance manufacturing were identified as had set up branches in Shanghai. More than six pillar industries. These were revised in 400 of the top 500 transnational corpora- 2000 for the Tenth Five-Year Plan period tions in the world had invested in Shanghai. of 2001­05, when two industries remained Many domestic companies had moved their unchanged, three industries were renamed, headquarters to Shanghai, and more than one industry was deleted, and a new indus- 200 domestic companies had set up offices try was added (Lei 2005: 165). The value in Shanghai (Zhao 2005). The Bank of added of the six pillar industries increased China set up its Shanghai headquarters in from Y 46.81 billion in 1995 to Y 250.5 bil- 2005. Some 842 companies were listed on lion in 2006. Their share of the city's total the Shanghai Stock Exchange, with a total manufacturing value added increased to capitalization of US$918 billion, ranking the 51.9 percent in 2006, indicating increasing exchange the fourteenth largest in the world specialization and agglomeration, which are in 2006 in terms of stock market capitaliza- Coastal China's urban-rural spatial restructuring under globalization 307 tion (World Federation of Exchanges 2007). a university education, compared with only In 2006 the financial industry's share of 3.6 percent in the outer zone in 2000. The the city's GDP reached 7.8 percent. Clearly, population in the urban core declined 6.3 Shanghai is making progress in its bid to percent in the period of 2000­06, due to a become an advanced service center. Still, negative rate of natural increase and relo- there is much room to develop advanced cation of the population to other areas. In manufacturing operations in the inner and contrast, the population in the inner and outer zones of Shanghai. outer zones increased more than 22 percent in the same period, mainly due to the arrival Spatial restructuring and of a temporary population. development zones However, large-scale urban renewal and Shanghai has an area of 6,340.5 square land development have taken place in all kilometers encompassing 18 districts and three zones. Many factories were located in 1 county. Shanghai's population and its eco- the urban core of the city in the 1980s, but nomic activities have long been concentrated many of them have moved elsewhere since in the city's old urban districts. Based on dif- the early 1990s. Like many other large cities, ferent population densities in 2006, the city housing is in high demand in Shanghai, and can be divided into a core, an inner zone, the property market is booming. The price and an outer zone (see table 18.3). The core of new housing increased 65 percent from consists of 9 old urban districts with a popu- 2000­06. The price of residential land also lation density of 17,000­42,000 persons per increased 69.7 percent. This has not affected square kilometer. The inner zone consists of the price of land used for industrial and stor- 3 urban districts with a population density age purposes, which declined 9.2 percent in of 4,000­6,000 persons per square kilometer. the same period (SMSB 2007). The Chinese The outer zone consists of 6 urban districts government has taken many measures to and Chongming County, with a popula- cool down the hot housing market, but with tion density of fewer than 2,000 persons per very limited impact at the time of writing. square kilometer. This zone still has large With development of the commercial rural areas. The average population density and service sectors in the urban core, a great of Shanghai was 2,863 persons per square deal of industrial development has been tak- kilometer in 2006. ing place in the inner and outer zones of the The core only has an area of 289 square city. According to table 18.4, the urban core kilometers, but it accommodated 36 percent only accounted for 11.8 percent of indus- of the total population in 2006. The tempo- trial employees and 9.3 percent of industrial rary population is concentrated in the inner output in 2006. The value of industrial and outer zones, where most industrial jobs output increased 10.9 percent in the urban are located. They accounted for more than core in the period of 2001­06, but the 31 percent of the total population in the increase in the inner and outer zones in the inner and outer zones and 14.5 percent of same period was 193.6 and 236.5 percent, the total population of the urban core. The respectively. population in the urban core is the most There were 7 national development zones educated, with 16.9 percent having received and 24 municipal development zones in Table 18.3 Distribution of population in Shanghai, 2006 Indicator Core Inner zone Outer zone City Usual residents (millions) 6.50 6.00 5.65 18.15 Population density (persons per square kilometer) 22,446 5,113 1,158 2,863 Share of temporary population (percent) 14.5 32.9 31.0 25.7 Population growth, 2000­06 (percent) -6.3 23.9 22.0 10.6 Share of population with university education in 2000 16.9 9.4 3.6 10.9 (percent) Source: SMSB (2002, 2007). 308 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 18.4 Industrial distribution in Shanghai, 2006 Indicator Core Inner zone Outer zone City Share in the total of the city (percent) Employees 11.84 37.10 49.95 100.00 Industrial output 9.34 46.78 39.54 100.00 Total profit 14.60 56.29 27.99 100.00 Industrial indicators Industrial output per employee (yuan per person) 54,332 86,835 54,518 68,874 Industrial employees per 100 population 4.91 16.66 23.85 14.86 Change in industrial output, 2001­06 (percent) 10.85 193.58 236.49 181.22 Source: SMSB (2007). Note: Including all state-owned enterprises and nonstate-owned enterprises with sales revenues of more than Y 5 million. Shanghaiin2006.Therewere3,701industrial an area of 109,839 square kilometers with enterprises in these zones, accounting for a population of 93.23 million in 2005. It 25.7 percent of all industrial firms and accounts for about 1.1 percent of China's 34.9 percent of all industrial employees in territory and 7.1 percent of China's popula- Shanghai. These firms contributed 46.4 tion, while contributing 18.6 percent of the percent of industrial output and 68.5 per- nation's GDP. cent of the industrial exports of the city. Compared with Shanghai, other cit- Thus the industrial zone development pol- ies in the Yangtze River delta have a much icy has contributed significantly to indus- lower population density and GDP density trial growth in Shanghai. However, not all per square kilometer of land area (see table industrial zones have been successful so far. 18.5). In terms of per capita GDP, Shanghai Only a limited number of them, such as the is well above most cities in the region except 7 national development zones and 7 of the for Suzhou. The per capita GDP of a city in municipal development zones,have achieved the region generally declines as its distance significant industrial agglomeration, which from Shanghai increases. Shanghai's popu- enhances the efficiency and competitive- lation and total GDP are also far higher ness of industrial production. The other 17 than those of any other city in the region. municipal development zones are small in The second-largest city is Suzhou, which scale, contributing industrial output of Y 98 has less than half of Shanghai's population billion as a whole in 2006. Clearly, there is and GDP. In terms of industrial output and much room to improve the spatial distribu- exports, Shanghai's contribution is much tion of industrial zones, although this may greater than that of any single city in the be difficult to do under the existing decen- region. Thus Shanghai is strong in both tralized institutional framework of eco- services and advanced manufacturing, with nomic administration (Shen 2007; Zhang services making up 51.4 percent of the city's 2002). It has become a common practice for GDP in the first half of 2007. Other cities each city, district, or even town government in the Yangtze River delta are still indus- to set up industrial zones to attract foreign trial cities, with services making up less investment to their own territory. than 42 percent of their GDP in the first half of 2007. The exceptions are Hangzhou, Economic diffusion Nanjing, and Zhoushan. There is a close economic relationship Shanghai and other cities in the Yangtze between Shanghai and other cities in the River delta have close demographic, social, Yangtze River delta region. Many cities have cultural, and economic connections. In fact, grown along with Shanghai. The region many people in Shanghai are originally from consists of 16 cities, including the munici- Jiangsu and Zhejiang. Since the early twen- pality of Shanghai, 8 prefecture-level cities tieth century, Shanghai has been the domi- in Jiangsu province, and 7 prefecture-level nant service center in the region and indeed cities in Zhejiang. The region comprises the nation and has attracted capital and Coastal China's urban-rural spatial restructuring under globalization 309 Table 18.5 Population and GDP density in the Yangtze River delta region, 2005 Population GDP density density (yuan GDP growth (persons million rate first Population per square per square Per capita GDP (yuan half of 2007 City (million) kilometer) kilometer) GDP (yuan) billion) (percent) Shanghai 17.78 2,859 294 51,486 915 13.0 Jiangsu cities 43.57 898 60 33,587 1,463 15.6 Nanjing 6.86 1,042 73 35,147 241 15.6 Wuxi 5.57 1,163 117 50,353 280 15.4 Changzhou 4.11 939 60 31,712 130 15.5 Suzhou 7.53 887 95 53,473 403 16.0 Nantong 7.34 917 37 20,056 147 16.0 Yangzhou 4.51 680 28 20,444 92 15.5 Zhenjiang 2.96 769 45 29,448 87 15.3 Taizhoua 4.69 809 28 17,532 82 15.4 Zhejiang cities 31.88 578 37 31,922 1,018 14.8 Hangzhou 7.51 452 35 39,199 294 14.4 Ningbo 6.56 678 51 37,343 245 14.7 Jiaxing 4.00 1,021 59 29,021 116 14.0 Huzhou 2.72 467 22 23,703 64 14.5 Shaoxing 4.39 532 35 32,972 145 14.7 Zhoushan 1.03 712 39 27,333 28 17.1 Taizhoua 5.68 604 27 22,038 125 14.4 Yangtze River delta 93.23 849 62 36,431 3,396 14.5 China 1,307.56 136 4 14,002 18,308 11.5 Sources: Population, GDP, and land area data from JPBS (2006: 86, 510­13); NBS (2006); SMSB (2006: 3, 10­11); ZPBS (2006: 46, 603­05); growth rate of China from http://www.stats.gov.cn/was40/gjtjj_detail.jsp?channelid=75004&record=62; other growth rates from http://www. stats-sh.gov.cn/2005shtj/csj/sjxx/76.htm. Note: The growth rate of the cities in Jiangsu and Zhejiang is the unweighted average of their cities. The growth rate of the Yangtze River delta region is the unweighted average of Shanghai and the cities in Jiangsu and Zhejiang. The population data for Jiangsu and Zhejiang are based on a survey of 1 percent of the population on November 1, 2005. a. Two cities, Taizhou in Jiangsu and Zhejiang, have the same English spelling but different names in Chinese. talent from all over China. In the 1970s and In 2003 the Shanghai government 1980s, Shanghai played an important role announced the "173 project" to develop in the development of TVEs in Jiangsu and 173 square kilometers of low-cost indus- Zhejiang by providing technology, technical trial land in various industrial zones expertise, and subcontracting services. Since (Tan 2003). This project is considered to the early 1990s, advanced producer services have induced unnecessary competition such as financial services, trading, and port with other cities for FDI. We believe that and airport logistics have played an impor- Shanghai should focus on producer service tant role in facilitating the inflow of FDI and functions and high value added advanced industrial development in other Yangtze manufacturing. Industrialization based on River delta cities. Shanghai has been attract- cheap land and labor should be avoided ing the regional and national headquarters for long-term, sustainable development. of transnational corporations, R&D centers, Shanghai should continue to improve the and advanced manufacturing, while many quality of its human resources, infrastruc- other FDI and manufacturing operations ture, and institutional environment for have located in other Yangtze River delta investment, which will compensate for the cities. Shanghai is also an advanced center rising costs of land and labor in the city-- of higher education, and many young peo- only two out of many factors that figure in ple have found employment in the region's the investment decisions of transnational cities after receiving higher education and corporations. training in Shanghai. Thus the diffusion of Overall, other Yangtze River delta cities human resources and provision of producer have grown rapidly due to both the Shanghai services are two important functions of factor and other development conditions. Shanghai in the region. Their GDP growth rates ranged from 14 to 310 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA 17 percent in the first half of 2007, greater rally be expected to occur, its status as the than the 13 percent of Shanghai. In the long nation's capital has also meant that anything term, the per capita GDP of the various cit- that might affect political or social stability ies in the Yangtze River delta is expected to in the city has been a source of concern. converge (table 18.5). Any policy changes or experiments, such The current level of development of all as those that were allowed to be carried cities in the Yangtze River delta is signifi- out in Fujian and Guangdong in the early cantly higher than that of cities in many years of the launching of economic reforms, other areas in China. Yangtze River delta were frowned upon. Moreover, for years the cities and even Shanghai still have ample JJJ region has suffered from a national fis- land for further industrial and urban cal system in which Beijing, Tianjin, and development. With effective measures for the surrounding cities in Hebei typically controlling pollution and negative exter- were only allowed to retain a small portion nalities, these cities may expand further to of their fiscal revenues. After remitting the create millions of jobs for skilled workers required percentage to the central govern- and labor migrants. In the meantime, pub- ment, the fiscal resources that these cities lic investment should focus on enhancing retained were insufficient to pursue infra- urban infrastructure and social services to structure and other development projects. match the increasing demand of a growing Consequently, it is hardly surprising that per population for a high standard of living. capita government expenditures in Beijing, Otherwise, urban agglomeration will lead Hebei, and Tianjin were among the lowest to serious social and environmental prob- in the country, as all of these places suffered lems. Thus the key question is not whether from a heavy tax burden (Wei 1996; Wei urban agglomeration should be adopted and Jia 2003). In addition, the JJJ region has as a strategy for development, but whether paled in comparison to the Pearl River and suitable policies will be devised to prevent the Yangtze River delta regions with regard the emergence of serious social and envi- to foreign investment. Over the past three ronmental problems. decades, capital from Hong Kong, Taiwan, and elsewhere has poured into the other two The Bohai Bay region destinations because of their more favorable As the third major city-cluster region in policies, geographic propinquity, and cul- China, the Bohai Bay region has always tural affinity. By comparison, the JJJ region played a strategic role in the development of has remained a backwater, as revealed by northern China. Conventionally, the Bohai major economic indicators (see tables 18.6 Bay region refers to the area centered on and 18.7). Beijing is an outstanding case. Beijing and Tianjin, along with an agglom- Its GDP growth rate was below the national eration of eight cities that is sometimes average in the 1980s but has gained momen- called the Beijing-Tianjin-Hebei region, or, tum since then. It attracted much foreign for short, the JJJ (Jing-Jin-Ji) region, "Ji" capital and had a large share of exports in being the alternative name of Hebei. The 2005 (table 18.1 and 18.2). Beijing had the Greater Bohai region includes the JJJ region, highest share of tertiary industry among the together with the Liaodong peninsula and cities in mainland China in 2005 due to its the Shandong peninsula (figure 18.1). position as the national capital. Compared with the two more-developed The region, nevertheless, is about to regions that have been the focus of the dis- enter a new phase of rapid growth, as the cussion so far, the Bohai Bay region has national Eleventh Five-year Plan (2006­10) suffered from relatively slow growth since highlights the Tianjin Binhai New District China reopened to the world in the early (see box 18.1), as a new target for growth, 1980s. This is largely due to the presence of much as Pudong in Shanghai was designated Beijing, the national capital, which has been in 1990. Another powerful boost for the a double-edged sword. While Beijing is the region is the fact that the Olympic Games center of political power and, hence, a place were hosted by Beijing in August 2008. where economic agglomeration might natu- Beijing, Tianjin, and other host cities in the Coastal China's urban-rural spatial restructuring under globalization 311 Table 18.6 Comparison of the three coastal regions in China, 2005 Yangtze Pearl river Indicator river delta delta JJJa Subtotalb China GDP (yuan billion) 3,389.83 1,805.94 1,848.97 7,044.74 18,308.50 Economic structure Primary 4.1 2.8 7.0 4.5 12.6 Secondary 55.0 50.9 44.0 51.2 47.5 Tertiary 40.9 46.3 49.0 44.3 39.9 Per capita GDP (yuan) 40,612 41,990 24,772 35,146 14,040 Total investment in fixed assets 1,617.20 529.68 768.72 2,915.60 8,877.36 (yuan billion) Total consumption (yuan billion) 1,073.89 563.05 639.19 2,276.13 6,717.66 Exports (US$ billion) 275.96 227.12 67.68 570.76 761.90 Realized foreign capital (US$ 26.33 11.51 9.05 46.89 63.81 billion) Sources: JPBS (2007: 530); NBS (2007); NBS, International Statistical Information Center (2007: 4, 6, 13); see also table 7. a. Figures for the JJJ city-region are used here for comparison instead of those for the whole Bohai Bay region. b. All figures in the subtotal have been calculated by the authors. region have begun to reap the benefits of ment in the secondary and tertiary sectors the "Olympic economy," with rapid urban accounted for 24.6 and 68.6 percent of the (re)construction and upgraded transport labor force, respectively, in Beijing and for networks. The agglomeration and disper- 40.6 and 40.5 percent in Tianjin in 2005 sion effects arising from this new impetus (see table 18.8). The density of employment for the development of the JJJ region are in the secondary and tertiary industries in discussed below. both cities is far higher than in Hebei. For instance, in 2005, some 187,000 science and A high-density JJJ city-region technology personnel, or 62 percent of such Broadly defined, the Bohai Bay region con- workers in Beijing, were concentrated in sists of an extensive area encompassing Zhongguancun, the Silicon Valley of Beijing more than 30 cities surrounding the Bohai (BMBS 2006, 2007). According to the 2007 Bay. However, the focus of this report is on Chinese Cities Competitiveness Report, Bei- the JJJ city-region, with Beijing and Tianjin jing ranks fourth while Tianjin ranks twelfth as its core. The dominance and agglomera- nationally in competitiveness. Among all of tion effects of Beijing and Tianjin in politi- the Bohai Bay cities, they stand first and cal, economic, and transport development third, respectively, with second place going are revealed in the data presented in table to Qingdao (eleventh overall) in the Shan- 18.7. The two cities cover only 15 percent dong peninsula (Ni and others 2007: 3). of the area of the JJJ city-region, yet they Beijing has advantages in manpower, capital, account for approximately 35 percent of the science and technology, and infrastructure, population, 57 percent of the GDP, and 80 while Tianjin has potential strengths in capi- percent of the realized foreign investment. tal, science and technology, and infrastruc- Beijing is unrivaled as the business center of ture competitiveness. the region. It is the top location in the coun- The prominence of Beijing and Tianjin, try for the headquarters of domestic firms, however, has yet to bring better develop- with one-fifth of the top 500 Chinese enter- ment to the region, as the problems of over- prises locating their headquarters in that agglomeration have begun to afflict them.The city. In addition, 239 foreign R&D centers problem is more acute in Beijing, because, and 16 headquarters of foreign enterprises as the nation's capital from as far back as are located in Beijing. It is hardly surprising the Ming Dynasty (1368­1644), Beijing has that Beijing ranked first for three consecu- been a magnet for all kinds of activities. By tive years in the China headquarters econ- contrast, the pace of development in Tian- omy index.4 The economies of Beijing and jin has been slow in the past few decades, Tianjin are well developed and veer toward as it lies in the shadow of Beijing (Wei and high value added employment. Employ- Jia 2003). For instance, Beijing's airport, the 312 RESHAP ING ECONOMIC Table 18.7 Major indicators of the JJJ city-region, 2005 GEO Total Realized Economic structure (percent) investment in Total foreign GRAP Area (square Population GDP (yuan Per capita fixed assets consumption Exports (US$ capital (US$ Areaa kilometers) (millions) billion) Primary Secondary Tertiary GDP (yuan) (yuan billion) (yuan billion) billion) billion) HY Beijing 16,578 15.38 688.63 1.39 29.43 69.19 45,444 282.72 290.28 30.87 3.53 Tianjin 11,611 10.43 369.76 3.04 55.47 41.49 35,783 151.68 119.01 27.42 3.65 IN Shijiazhuang 15,722 9.61 178.68 13.87 48.45 37.68 19,370 92.90 60.60 3.72 0.44 EAST Qinhuangdao 7,467 2.88 49.12 10.44 38.75 50.80 18,087 16.49 14.70 1.84 0.24 Tangshan 13,206 7.26 202.76 11.65 57.30 31.06 28,466 63.57 46.86 1.20 0.50 Cangzhou 13,419 6.84 113.08 11.97 53.39 34.63 16,532b 35.45 25.01 0.63 0.14 ASIA Chengde 39,519 3.37 36.03 18.25 50.94 30.81 9,870 18.93 11.79 0.12 0.11 Langfang 6,330 3.96 62.12 16.22 54.10 29.69 16,200 34.16 17.82 0.59 0.26 Baoding 22,159 10.73 107.21 18.29 48.80 32.91 9,990 56.03 38.51 1.10 0.13 Zhangjiakou 36,829 4.18 41.58 16.20 44.72 39.08 9,876 16.78 14.61 0.20 0.04 JJJ 182,840 74.64 1,848.97 7.08 43.96 48.96 24,772b 768.72 639.19 67.68 9.05 Hebei 190,000 68.51 1,009.61 14.89 51.83 33.29 14,782 421.03 295.29 10.93 2.28 Sources: BMBS (2007: 46­49); Hebei Provincial Government Office (2007: 191­203, 471­79); NBS (2007); TMBS (2007: 70­76). a. Beijing and Tianjin are both municipalities, and the remaining eight are prefecture-level cities under the administration of Hebei province. b. The per capita GDP figures for Cangzhou and JJJ were calculated by the authors. c. Chengde's per capita GDP was US$1,205, and the exchange rate was US$1 = CNY 8.1917 Coastal China's urban-rural spatial restructuring under globalization 313 B O X 1 8 . 1 Tianjin Binhai New District: the third pole in China The Tianjin Binhai New District (TBND), strategically TBND has been allowed to experiment with reforms positioned as the only coastal window of Tianjin, in the financial sector, especially in dimensions not encompasses the Tianjin technological develop- presently offered in Shanghai and Shenzhen, with ment area, the Tianjin free trade zone,Tianjin port, a view to pursuing greater openness in China's and some other local administrative districts (figure financial sector. For example, a new commercial 18.1).The TBND is the third type of zone in China bank--China Bohai Bank--was established in 2005. for carrying out experiments in reform; as such, it It runs with a high degree of flexibility and freedom enjoys various preferential policies and rights (the that no other bank in China enjoys at this moment; other two types are the SEZs in Guangdong, Fujian, it provides various banking, investment, and insur- and Hainan and the Pudong New District in Shang- ance services. Also, the central government has hai). It is set to develop into the new economic initially granted the Tianjin branch of the Bank of center of North China and to become its most China the sole right to provide qualified individual populated area.The latest United Nations estimate direct investment services in the Hong Kong stock puts the combined nonagricultural population market, a new investment vehicle in China that is of Beijing and Tianjin at more than 18 million (UN expected to be implemented in the near future. Population Division 2006), but with the national For the construction of transport infrastructure, urbanization policy and the present Olympic fervor, besides the highway and railway linkages to other the Chinese government revealed that the figure cities of the JJJ city-region, the projects expanding has surpassed 20 million already.The TBND is now Tianjin's airport and port are the most significant. home to a hukou population of about 1.1 million Even though the work of expanding the airport is and will probably be one of the most popular urban ongoing, this has not affected its operations.The destinations for newcomers. airport continues to set records in the number of In the past few years, the authorities of the routes, frequency of flights, number of passengers, TBND have chosen to adopt some progressive strat- and volume of goods handled.The 2008 Summer egies in developing the zone's industrial and finan- Olympic Games in Beijing will surely provide the cial activities. For example, an A320 Airbus assembly airport with the opportunity to realize a new wave line has been set up next to the Tianjin Airport. of rapid growth and development. Immediate Also located in the zone are a large-scale (1-million- coordination with the Beijing airport appears to be metric-ton scale) ethane manufacturing enterprise, a matter of high priority. Similarly,Tianjin's port is a chemical industrial park, a high-tech textile indus- experiencing speedy growth in size and capacity, trial park, and more.With the attraction of these and its total throughput has soared since 2001. It is new manufacturing operations, the TBND is on the now the sixth-largest port in the world, the fourth- way to establishing itself as a strong industrial base largest in China, and the largest in North China.The in the region. Its stronger industrial presence is port has been allowed to set up the largest bonded reflected in the heightened contribution of second- port zone in China.The present imperative of Tianjin ary industries to GDP, from 66.9 percent in 2003 to port is to seek closer integration and coordination 71.5 percent in 2006 (TMBS 2007: 489). In relation with other ports in the Bohai Bay to consolidate its to Tianjin,TBND accounted for 43.9 percent of the hub status. Of all ports, Caofeidian port in Tangshan city's GDP, 67.4 percent of its exports, and 76.0 is the first choice in seeking greater integration. percent of its realized FDI in 2005, compared with Caofeidian has benefited from the integration of 40.6 percent, 62.2 percent, and 74.4 percent, Beijing Shougang (the country's best-known iron respectively, in 2003. However, a sound financial and steel manufacturing enterprise) and Tanggang, architecture and an efficient transport infrastruc- whose production line has been relocated to Cao- ture network are necessary to support healthy feidian. Caofeidian is likely to develop as a industrial development. In this respect, the modern heavy industrial port city in the future. busiest air hub in China, handled 48.74 mil- ing shortages of water, land, and energy; lion passengers in 2006, but Tianjin's airport population pressure; and so on (Li and Hu ranked only thirtieth, with a mere 2.76 mil- 2007: 175). With regard to the province of lion passengers. In rankings of competitive- Hebei--the immediate hinterland of Bei- ness, Tianjin's airport came fourth out of jing and Tianjin--poverty has become a five major airports in the region, clearly a challenge due to a "vacuum effect."A recent mismatch with its status as one of the four report has shown that some 2.73 million special municipalities in China (Chu and people in the peripheral areas of Beijing and Wu 2006). At present, the two cities, espe- Tianjin live in poverty, and such regional cially Beijing, are facing the consequences imbalance has been described as "modern of over-agglomeration, such as serious air mega cities and backward hinterland" by Lu pollution; worsening traffic jams;5 increas- Dadao (quoted in Li and Hu 2007). Thus 314 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Table 18.8 Employment in three cities of China, 2005 Greater Bohai Bay region. Among these, a Indicator Beijing Tianjin Hebei new Beijing-Tianjin expressway will shorten the travel time between the two cities to Total employed persons (1,000) Primary 622 805 15,623 about 1 hour from the present 2.5 hours, Secondary 2,264 1,733 10,484 and another five new expressways will con- Tertiary 6,318 1,831 8,566 nect the JJJ city-region. New 200 kilometer Total 9,204 4,269 34,673 per hour high-speed railways are now in ser- Employment density (persons per vice between Beijing and Bohai Bay cities,6 square kilometer) and a new intercity high-speed railway and Primary 37.52 69.33 82.23 a light railway will serve the route between Secondary 136.57 149.26 55.18 Tertiary 381.11 149.08 45.08 Beijing and Tianjin.It is estimated that oper- ation of the high-speed railway will contrib- Employment structure (percent) ute an extra Y 20 billion to China's GDP.7 Primary 3.8 18.9 45.1 Secondary 24.6 40.6 30.2 The lifestyles and travel patterns of Beijing Tertiary 68.6 40.5 24.7 and Tianjin commuters will change as a Source: NBS (2007: 127). result. A new airport will be built between Note: The data for Hebei province are used here instead of those for the eight cities, because of the availability of the Beijing and Tianjin, in keeping with the data. view that all airports in the region will have the competitiveness of other cities in the JJJ to be integrated. Tianjin Port, now under city-region is far lower than that of Beijing expansion, will serve as the seaport hub and and Tianjin. Even Tangshan, generally con- logistics center of North China. All of these sidered the third JJJ city after Beijing and projects will facilitate a smoother and denser Tianjin, ranked only forty-third among all flow of goods and people throughout the JJJ Chinese cities. city-region, consolidate the hub functions of Beijing and Tianjin, and increase their spill- Improving regional transport over effects. For other parts of the Greater infrastructure Bohai Bay region, the most spectacular In developing the JJJ city-region, shorten- project is the Bohai Strait channel project, ing travel distances and minimizing socio- which will provide a land link via a bridge economic disparities are high on the list of between the Liaodong peninsula and the priorities. Support from the central govern- Shandong peninsula. It will sharply reduce ment and effective local policies are equally the travel distance to about one-tenth of the important. Consequently, the top-down present distance of more than 1,000 kilome- national expressway and railway plans have ters and contribute to the integration of the called for building a comprehensive trans- region. This project will fill a missing link in port network in the region. This will be China's north-south coastal corridor and is complemented by local public works plans, similar to the three bridges over Hangzhou such as the new urban planning strategies Bay in the Yangtze River delta and the pro- of Beijing and Tianjin, the JJJ regional plan- posed Hong Kong-Zhuhai-Macao bridge in ning study, the JJJ transport infrastruc- the Pearl River delta. The integration of the ture development plan, and so on. A clear entire Chinese coastal region will be signifi- division of labor is envisaged for the cities. cantly advanced. For example, Beijing will play down its role In sum, the experience of the devel- as an economic center, with the intention of opment of the JJJ city-region reflects the decentralizing some major financial activi- view that core cities can, through their ties and heavy industries from Beijing to the agglomeration and dispersion effects, not Tianjin Binhai New District, Tangshan, and only decisively affect their own pattern of other cities (Li and Hu 2007). development, but also add to the strength Several projects to build expressways, of the whole region. Proper coordina- high-speed railways, ports, and airports tion, cooperation, and division of labor have been or are about to be launched in the among cities are critical to circumventing region. The main purpose is to strengthen the excessive agglomeration of core cities the coherence of the JJJ city-region and the and the hollowing out of peripheral areas. Coastal China's urban-rural spatial restructuring under globalization 315 Support from the central government and tics. First, Beijing, Guangzhou, Shanghai, local development strategies are equally and Shenzhen are mega cities, each with a critical to the success of regional develop- population of more than 8 million in 2005. ment in China. Above all, on the evidence From 1980­2005, their population grew deduced from the Bohai Bay region, trans- dramatically due to the inflow of migrants, port infrastructure is a forerunner to suc- mostly from rural areas. For example, popu- cessful urban-regional change and the tool lation in Shanghai grew from 11.57 million for enhancing the flow, communication, to 17.99 million in the period, and Shenzhen and integration of cities. grew from a small town to a city with a pop- ulation of 8.28 million. Second, these cities Conclusions are densely populated, with population den- This chapter has chosen the three most devel- sity over 1,277 persons per square kilome- oped city-regions of China to bring home ter, which may be underestimated, as some their experiences in pioneering rapid eco- counties and rural areas are included in the nomicdevelopmentandsocialmodernization city boundary. Third, these cities were more in the country over the past three decades. advanced and had higher per capita GDP at In the Chinese transition to rapid growth the beginning of the period than China as and development, Chinese cities, especially a whole. Their GDP grew faster than that those along the coast, have been catalysts of China as a whole, with the exception of for change. This is in large measure due to Beijing and Shanghai in the 1980s. With an the advantages of their geographic location, expanding population, the gap in per capita which provides easy access to and from for- GDP widened between these cities and the eign countries, and to their considerable rest of China, indicating the strong force of store of administrative, technological, eco- agglomeration economies. Finally, there is a nomic, and cultural experience. The growth tendency of convergence among these cit- of the three coastal regions as shown in this ies. The tertiary sector became important in chapter is led by the urbanization of large 2005, contributing more than 46 percent of cities such as Beijing, Guangzhou, Shanghai, GDP.All cities attracted large foreign invest- and Shenzhen.While the rapid growth of the ment in 2005, and their exports expanded coastal regions is planned and expected by greatly in the period of 1980­2005. To some Chinese economists, the rapid growth and extent, the growth of export-oriented man- expansion of large coastal cities are beyond ufacturing in the Pearl River delta in the the imagination of urban planners, as the 1980s and 1990s was repeated in Shanghai long-standing official policy is to control the after 2000. growth of large cities. In a word, the strong Nevertheless, the leading cities in the force of urban agglomeration has produced three regions also demonstrate major dif- a model of development in coastal China led ferences in terms of growth dynamics, by large cities. Large cities with powerful city economic structure, and degree of inter- governments and agglomeration advantages nationalization. First, there are differences have been very successful in attracting for- in growth dynamics. Beijing and Shanghai eign investment and promoting economic grew slowly in the 1980s, even slower than development under the triple processes of the national average. Their growth speeded globalization, decentralization, and mar- up after 1990. Guangzhou and Shenzhen ketization (Shen 2007). grew faster than Beijing and Shanghai from To conclude, we would like to draw the 1980­2005, although their growth slowed similarities and differences among the three down after 2000. By 2005, Guangzhou and coastal regions, focusing on Guangzhou, Shenzhen had a per capita GDP higher than Shenzhen, Shanghai, and Beijing, which that of Beijing and Shanghai. These cases are their leading economic centers. Guang- demonstrate that a few leading cities rather zhou and Shenzhen are in Pearl River delta, than one primary city may achieve high lev- Shanghai is in the Yangtze River delta, and els of development in a large country like Beijing is in the Bohai Bay region. These China. Their growth may not occur at the four cities share the following characteris- expense of other leading cities. 316 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Second, there are differences in economic Guangzhou is the provincial capital of structure. Shenzhen and Shanghai had simi- Guangdong province. It attracted more lar shares of secondary and tertiary indus- foreign capital than Shanghai and Beijing tries in GDP in 2005, around 50 percent in the 1980s and 1990s. But by 2005, it in each sector. Foreign investment plays an attracted the least foreign capital among the important role in the expansion of manu- four cities. Its exports also grew more slowly facturing industry in both cities. As the pro- than those of other cities in the period of vincial capital of Guangdong, Guangzhou 2000­05. Clearly Guangzhou is less export had a smaller share of secondary industry oriented than other cities. Beijing is the (39.7 percent) and a larger share of tertiary capital of China, and its economy, like that industry (57.8 percent) in GDP than Shen- of Guangzhou, is less export oriented. But zhen and Shanghai in 2005. As the capital of foreign investment and exports grew rapidly China, Beijing had the smallest share of sec- from 2000­05, indicating the growing influ- ondary industry (29.4 percent) and the larg- ence of globalization. est share of tertiary industry (69.2 percent) This chapter has shown that the Pearl in GDP. They experienced a different pace River delta, the Yangtze River delta, and of economic restructuring. Shanghai and the Bohai Bay area, in a sequenced pattern Beijing were industrial cities in 1980, and of reform and openness, have accounted their share of secondary industry in GDP for the bulk of the success of the country's declined sharply from more than 68 percent reform program, which has been carried to below 49 percent from 1980­2005. Their out over the past three decades. The Pearl share of tertiary industry in GDP increased River delta region was the first to adopt an sharply, from below 27 percent to more than open policy, and foreign investment from 50 percent in the same period. Guangzhou Hong Kong played an important role in the experienced similar changes on a smaller development of an outward-processing- scale. In Shenzhen the share of secondary based and export-oriented economy. Shen- industry in GDP grew rapidly at the expense zhen and Guangzhou outperformed other of agriculture, while the share of tertiary major cities in China in per capita GDP by sector in GDP was stable at 45­51 percent 2005. Shanghai in the Yangtze River delta in the period of 1980­2005. was China's major industrial base before Third, there are differences in the pace 1980. An open policy was implemented in and degree of internationalization. The Shanghai on a large scale only after 1990, Pearl River delta region was the first to and Shanghai underwent a dramatic trans- adopt favorable policies for foreign invest- formation, with significant expansion ment. Shenzhen led the country in attract- of advanced manufacturing and tertiary ing foreign investment and generating industries. Shanghai has a good foundation exports. Investment from Hong Kong played of human resources, R&D capacity, and an important role in the expansion of indus- business traditions. Transnational corpora- trial production and exports in the form of tions from Hong Kong and Western coun- outward processing. Shanghai came later tries made significant investment in Shang- in adopting an open policy. It has attracted hai. The Bohai Bay region is relatively less much foreign investment since the early advanced. But as the capital of China, Bei- 1990s, when the policy of developing the jing has considerable national resources, and Pudong New District was adopted, and its its service economy is well developed. Under export growth closely followed that of Shen- the general policy of economic reform and zhen in the period of 1990­2005. Exports opening, the Pearl River delta, the Yangtze from Shenzhen reached US$101.52 billion River delta, and the Bohai Bay region and in 2005, even greater than exports from their major cities have achieved significant Shanghai. It is clear that a portion of foreign development by making use of both domes- investment in Shanghai targets the domestic tic and foreign resources. They demonstrate market, such as the industry of car manufac- the important role that urbanization and turing, while Shenzhen's industry is mainly large cities have played in the development export oriented. of China. Coastal China's urban-rural spatial restructuring under globalization 317 Notes House of Guangdong), eds. 2000. Atlas of the Administrative Divisions of Guangdong Prov- Yue-man Yeung is emeritus professor, and Jianfa ince. Guangzhou: Map Publishing House of Shen is professor, both at the Chinese University Guangdong. of Hong Kong. Thanks are due to Gordon Kee for his research assistance in the preparation of Enright, Michael J., Ka-mun Chang, Edith E. this chapter. Scott, and Wen-hui Zhu. 2003. Hong Kong 1. These are 2 of 14 coastal cities in China and the Pearl River Delta: The Economic Inter- that were declared open and granted the author- action. Hong Kong: 2022 Foundation. ity to pursue favorable policies and given other Fan, Shaoqing. 2007."Overview of Financial advantages to attract foreign investment. Industry in 2006." (http://www.stats-sh.gov. 2. China's currency is the renminbi, and its cn/2005shtj/tjfx/ndxx/userobject1ai2618. currency unit is the yuan. html.) 3. US$1 = Y 8.19 in 2005. 4. See "Headquarter Economy Ranking, Fujita, Masahisa, Paul Krugman, and Anthony J. Beijing the Strongest in R&D [in Chinese]," Sep- Venables. 2001. The Spatial Economy: Cities, tember 18,2007 (http://big5.china.com.cn/news/ Regions, and International Trade. Cambridge, txt/2007-09/18/content_8909056.htm). MA: MIT Press. 5. For example, about 2.5 hours are needed GDPBS (Guangdong Provincial Bureau of Sta- to travel the 120 kilometers between Beijing and tistics). Various years (1992, 2006, 2007). Tianjin. On average, for about 6 hours a day Statistical Yearbook of Guangdong Province. the travel speed on the Beijing-Tianjin express- Beijing: China Statistics Press. way reportedly is less than 60 kilometers per GMSB (Guangzhou Municipal Statistics hour. See "Traffic Accidents Keep Increasing, Bureau). 2006. Guangzhou Statistical Yearbook Slowing Down the Traveling Speed in Jing- 2006. Beijing: China Statistics Press. Jin-Tang Expressway [in Chinese]." Novem- ber 17, 2004 (http://news.xinhuanet.com/news Guangzhou Economic Yearbook Editorial Com- center/2004-11/17/content_2227725.htm). Traf- mittee. 1983. Guangzhou Economic Yearbook fic accidents increased by double digits from 1983. Guangzhou: Guangzhou Economic 2001­03. Yearbook Editorial Committee. 6. The sixth acceleration of the speed of Chi- Hebei Provincial Government Office. 2007. na's railways was carried out nationwide on April Hebei Economic Yearbook 2006. Beijing: China 18, 2007. See "The Sixth Speed Acceleration of Statistics Press. Railway Brings Coordinated Economic Devel- opment in the Bohai Ring Region [in Chinese]." Huang, Yefang, Yee Leung, and Jianfa Shen. October 18, 2007 (http://www.gov.cn/jrzg/2007- 2007."Cities and Globalization: An Interna- 10/18/content_778773.htm). tional Cities Perspective." Urban Geography 28 7. See "GDP of TBND, December 2006 [in (3): 209­31. Chinese]." February 28, 2007 (http://www. JPBS (Jiangsu Provincial Bureau of Statistics). bh.gov.cn/jjfz/2007-02/28/content_9386550. Various years (2006, 2007). Jiangsu Statistical htm). Yearbook. Beijing: China Statistics Press. Kong, Yan. 2001. Reading Shanghai 1990­2000. References Shanghai: Shanghai People's Press. BMBS (Beijing Municipal Bureau of Statistics). Various years (2006, 2007). Beijing Statistical Krugman, Paul. 2007."The `New' Economic Yearbook. Beijing: China Statistics Press. Geography: Where Are We?" In Regional Integration in East Asia: From the Viewpoint Chen, Guanghan, Yunyun Zhou, Jiaan Ye, of Spatial Economics, ed. Masahisa Fujita, pp. and Fengxuan Xue, eds. 2003. Research on 23­34. New York: Palgrave Macmillan. Enhancing the International Competitiveness of the Greater Pearl River Delta [in Chinese]. Lei, Xinjun. 2005."A Study on the Key Indus- Guangzhou: Zhongshan University Press. tries and Development Path of the Advanced Manufacturing Industry in Shanghai in Chu, Yanchang, and Yuhua Wu. 2006."Compar- 11th FYP Period." In Building a Harmonious ative Study of the Competitiveness of Major World City: A Study of the Development and Airports in the Bohai Ring Region [in Planning Ideas of 11th FYP of Shanghai, eds. Chinese]." Zonghe Yunshu 8-9: 101­04. Ronghua Wang and Xuejin Zuo, pp. 161­87. DCA (Department of Civil Affairs of Guang- Shanghai: Press of Shanghai Academy of dong Province) and MPH (Map Publishing Social Sciences. 318 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Li, Shu, and Meijuan Hu. 2007."Beijing: Lead- SMSB (Shanghai Municipal Statistics Bureau). ing the Beijiing-Tianjin-Hebei Metropolitan Various years (2001, 2002, 2006, 2007). Region [in Chinese]." Chinese National Geog- Shanghai Statistical Yearbook. Beijing: China raphy 9: 172­81. Statistics Press. Lin, George C. S. 1999."State Policy and Spa- SSB (Shenzhen Statistics Bureau). 2006. Shen- tial Restructuring in Post-Reform China, zhen Statistical Yearbook 2006. Beijing: China 1978­95." International Journal of Urban and Statistics Press. Regional Research 23 (4): 670­96. Tan, Heng. 2003."Shanghai `173 Project.'" Markusen, Ann. 1996."Sticky Places in Slippery (http://news.sina.com.cn/c/2003-06- Space: A Typology of Industrial Districts." 12/10561162773.html.) Economic Geography 72 (3): 33­78. TMBS (Tianjin Municipal Bureau of Statistics). NBS (National Bureau of Statistics). Various 2007. Tianjin Statistical Yearbook 2006. years (1982, 1991, 2001, 2006, 2007). China Beijing: China Statistics Press. Statistical Yearbook. Beijing: China Statistics UN (United Nations) Population Division. Press. 2006. World Urbanization Prospect: The 2005 NBS (National Bureau of Statistics), Depart- Revisions. New York: United Nations. ment of Comprehensive Statistics. 1999. Wei, Yehua. 1996."Fiscal Systems and Uneven Comprehensive Statistical Data and Materials Regional Development in China, 1978­1991." on 50 Years of New China. Beijing: China Sta- Geoforum 27 (3): 329­44. tistics Press. Wei,Yehua, and Yanjie Jia. 2003."The Geographi- NBS (National Bureau of Statistics), Interna- cal Foundations of Local State Initiatives: Glo- tional Statistical Information Center. 2007. balizing Tianjin, China." Cities 20 (2): 101­14. Yangtze River Delta and Pearl River Delta and Hong Kong and Macao Special Administra- Wei, Yehua, and C. K. Leung. 2005."Develop- tive Region Statistical Yearbook 2006. Beijing: ment Zones, Foreign Investment, and Global China Statistics Press. City Formation in Shanghai." Growth and Change 36 (1): 16­40. Ni, Pengfei, and others, eds. 2007. Annual Report on Urban Competitiveness No. 5 [in Chinese]. World Federation of Exchanges. 2007. Beijing: Social Sciences Academic Press. "Domestic Market Capitalization." (http:// www.world-exchanges.org/WFE/home. Park, S. O. 1996."Networks and Embeddedness asp?menu=406.) in the Dynamic Types of New Industrial Dis- tricts." Progress in Human Geography 20 (4): Wu, Fulong. 2000."Global and Local Dimen- 476­93. sions of Place-making: Remaking Shanghai as a World City." Urban Studies 37 (8): 1359­77. Shen, Jianfa. 2005."Space, Scale, and the State: Reorganizing Urban Space in China." In Yang, Dali L. 2005."China's Looming Labor Restructuring the Chinese City: Changing Shortage." Far Eastern Economic Review 168 Society, Economy, and Space, eds. Laurence J. (10, January): 19­24. C. Ma and Fulong Wu, pp. 39­57. London: Yeung, Yue-man. 2002. Internal and Interna- Routledge. tional Migration in China under Openness ------. 2006."Understanding Dual-Track and a Marketizing Economy. Occasional Urbanisation in Post-Reform China: Con- Paper 129. Hong Kong Institute of ceptual Framework and Empirical Analysis." Asia-Pacific Studies, Chinese University Population, Space, and Place 12 (6): 497­516. of Hong Kong. ------. 2007."Scale, State, and the City: Urban ------. 2003."Integration of the Pearl River Transformation in Post-Reform China." Hab- Delta." International Development Planning itat International 31(3­4): 303­16. Review 25 (3): iii­viii. Shi,Wenjun. 2003."Epoch-Making Important ------. 2005. The Pearl River Delta Mega Fiscal System Reform [in Chinese]."Shuiwu Urban-Region: Internal Dynamics and Exter- Zhonghen (November): 12­14. (www.cnki.net.) nal Linkages. Occasional Paper 12. Shanghai- Shi, Yulong, and Chris Hamnett. 2002."The Hong Kong Development Institute, Chinese Potential and Prospect for Global Cities in University of Hong Kong. China: In the Context of the World System." Yeung, Yue-man, ed. 2007. The First Decade: The Geoforum 33 (1): 121­35. Hong Kong SAR in Retrospective and Coastal China's urban-rural spatial restructuring under globalization 319 Introspective Perspectives. Hong Kong: Yusuf, Shahid, and Weiping Wu. 2002."Path- Chinese University Press. ways to a World City: Shanghai Rising in an Era of Globalization." Urban Studies 39 (7): Yeung, Yue-man, and Xu-wai Hu, eds. 1992. 1213­40. China's Coastal Cities: Catalysts for Modern- ization. Honolulu: University of Hawaii Press. Zhang, Tingwei. 2002."Urban Development and a Socialist Pro-Growth Coalition in Shanghai." Yeung, Yue-Man, and Gordon Wai-man Kee. Urban Affairs Review 37 (4): 475­99. 2007. The Networked Region: Basic Infrastruc- ture Development in the Pan-Pearl River Delta Zhao, Hong. 2005. The Development Report of [in Chinese]. Research Monograph 70. Hong China's Headquarters Economy (2005­2006). Kong Institute of Asia-Pacific Studies, Chi- Beijing: Social Sciences Academic Press. nese University of Hong Kong. Zhou, Haiwang, Xin Yang, and Lichun Xiao. Yeung, Yue-Man, and Jianfa Shen, eds. 2004. 2005."A Study on the Population Growth Developing China's West: A Critical Path to in Shanghai in 11th FYP Period." In Build- Balanced National Development. Hong Kong: ing a Harmonious World City: A Study of the Chinese University Press. Development and Planning Ideas of 11th FYP of Shanghai, ed. Ronghua Wang and Xuejin Yeung, Yue-Man, Jianfa Shen, and Li Zhang. Zuo, pp. 215­32. Shanghai: Press of Shanghai 2005. The Western Pearl River Delta: Academy of Social Sciences. Growth and Opportunities for Cooperative Development with Hong Kong. Zhou, Yi. 1995."Guangdong's Openness and Research Monograph 62. Hong Kong Reform: Retrospect and Prospect [in Institute of Asia-Pacific Studies, Chinese Chinese]." (http://www.chinareviewnews. University of Hong Kong. com/crn-webapp/search/siteDetail.jsp?id=100 149497&sw=%E5%91%A8%E4%B9%89.) Yeung, Yue-Man, and Yun-wing Sung, eds. 1996. Shanghai: Transformation and Modernization ZPBS (Zhejiang Provincial Bureau of Statistics). under China's Open Policy. Hong Kong: 2006. Zhejiang Statistical Yearbook 2006. Chinese University Press. Beijing: China Statistics Press. A history of the Republic of Korea's industrial structural transformation and spatial development Sam Ock Park 19 The Korean peninsula, which is situated is divided into Yongnam (the southeast at the northeastern rim of the Asian con- region) and Honam (the southwest region). tinent, covers 221,000 square kilometers In general, the western part of Korea con- and includes both the Democratic People's sists largely of plains, while the eastern part Republic of Korea and the Republic of consists largely of mountains. Korea. Of that total, 99,000 square kilome- Historically, when agriculture was the ters constitute the Republic of Korea, the major economic activity, the western part of focus of this paper. Korea was relatively affluent, largely due to c h a p t e r The Korean peninsula historically served high agricultural productivity in the plains. as a land bridge over which Chinese culture However, since the early 1960s, Korea's eco- was diffused from China to Japan. Despite nomic geography has changed significantly, the significant influence of Chinese culture with rapid industrial development start- and the frequent invasions by foreign pow- ing in the southeastern and capital regions. ers, the Republic of Korea has managed to Construction of the Seoul-Busan express preserve a cultural and ethnic identity that is highway and the industrial development of different from that of either China or Japan, the southeastern region, centered on Busan, with a distinct language, alphabet, arts, and increased the spatial disparity between the customs (Lee and others 1988). axis of Seoul-Busan and the axis of the south- The province (do) is the country's larg- western and northeastern corners of Korea. est administrative unit. Presently, Korea has Korea has experienced remarkable eco- nine provinces, with names originating in nomic growth for the past five decades.Korea the early Chosun Dynasty.In addition,Korea was among the poorest countries in the world has one special city (Seoul) and six mega following the devastation of the Korean War cities (Busan, Daegu, Daejon, Gwangju, (1950­53). Korea's per capita gross national Inchon, and Ulsan) that have populations product (GNP) rose from less than US$100 over 1 million and are at the same adminis- (in 1996 US$) in 1960 to US$20,000 in 2007. trative level as provinces (see figure 19.1). This remarkable economic achievement, Traditionally, the Korean peninsula known as "compressed economic growth," was divided into three geographic regions: is closely related to the successful imple- north, central, and south. Due to the politi- mentation of export-oriented, sector-spe- cal division of the Korean peninsula in 1953, cific industrial development strategies and some of the northern part of the central the development of human resources since region and the whole of the northern region the launch of the First Five-Year Economic belong to the Democratic People's Republic Development Plan in 1962. of Korea. In addition, there are cultural and Korea's compressed economic growth geographic differences between the east and is dynamic, resulting, first, in the concen- west of each region. For example, the south tration of industry and population in the A history of Korea's industrial structural transformation and spatial development 321 capital region and, subsequently, in the con- Figure 19.1 Administrative divisions of Korea: provinces, mega cities, and provincial cities vergence of regional GDP per capita, as pro- vincial cities and rural areas began to evolve with the development of information and communications technology (ICT). Chunchon Considering the dynamics of the Korean economy and the country's rapid spatial development, this chapter investigates the Seoul KANGWON following issues: Inchon KYONGGI · The role of government policies in the NORTH transformation of industrial structure CHUNGCHONG and spatial development; Chongju SOUTH CHUNGCHONG · The spatial development of economic NORTH growth in terms of "distance" and "den- Daejon KYONGSANG sity"effects,agglomeration,factor mobil- ity, and technological development; and Daegu Chonju · The lessons and policy implications. NORTH CHOLLA SOUTH Ulsan In Korea, structural dimensions are KYONGSANG closely related to spatial dimensions, and Gwanju Changwon so this chapter examines industrial poli- Busan SOUTH Mokpo CHOLLA cies, spatial transformation, and industrial restructuring. The analyses at the national level are conducted mostly for the period from the 1960s to 2005. However, the analyses at the regional level are conducted mostly for the period from the 1980s to 2005 due to the availability of data Provincial Boundaries by regions and inconsistency of data for Megacities earlier years. CHEJU The basic unit of analysis is the province: 9 provinces and 7 major cities (Seoul and the mega cities), making 16 units. The 16 basic units of provinces and cities are clas- sified into 4 regions: capital, middle, south- 82 percent of the chemical industries, and west, and southeast (see figure 19.2). 90 percent of the electronic power­generat- ing facilities were located in the north when Industrial policies,structural Korea was liberated from Japanese occupa- tion in 1945 (Park 1981). To make matters changes of industry,and spatial worse, more than half of the manufacturing transformation facilities were destroyed during the Korean The history of Korean industrialization War, weakening the foundations for indus- is rather short. Manufacturing activities trial development. were mainly in handicrafts created by mid- Industrialization in Korea has progressed dle-class people in the Chosun Kingdom rapidly since the early 1960s, when the First (1392­1910). During the period of Japa- Economic Development Plan (1962­66) nese occupation (1910­45), Japan began was launched. The manufacturing indus- developing heavy and chemical industries try became the major driving force of the in the north (now the Democratic People's Korean economy for the next two decades. Republic of Korea) to make use of the abun- The share of manufacturing in total national dant hydraulic power and mineral resources production increased rapidly from 12.1 per- there. As a result, 90 percent of the minerals, cent in 1960 to 29.6 percent in 1980, while 322 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 19.2 Four regions, nine provinces, and major cities in Korea to the proliferation of industrial cities in the capital region, while the development of a large-scale industrial complex was the major source of growth for industrial cities in the southeast region. Korea's rural-urban transformation took place through the concentration of industry and population CAPITAL KANGWON REGION in major urban areas, on the one hand, and MIDDLE the decentralization of industry and popu- KYONGGI REGION lation from the large core cities to their NORTH hinterlands within the region, on the other. CHUNGCHONG Both processes are related to distance and density effects. SOUTH CHUNGCHONG Understanding the government's indus- NORTH KYONGSANG trial policy is a prerequisite to understanding the spatial transformation of Korea. Since SOUTHEAST the First Five-Year Economic Development REGION Plan was launched in 1962, the national NORTH CHOLLA government has sought to promote particu- SOUTH KYONGSANG lar sectors and locations. Export-oriented SOUTHWEST industrialization has been a major strategy REGION since the early 1960s, and the strategy was SOUTH CHOLLA fashioned to promote the most promis- ing industries at a certain stage. These were called "strategic industries." Labor-intensive industries were the key sectors for the Provincial Boundaries expansion of industrial exports before Regional Boundaries the early 1970s, while heavy and chemical > 500,000 population industries were the strategic industries for < 500,000 population the expansion of exports in the late 1970s CHEJU and early 1980s. The government's policy of support for heavy and chemical industries contributed to the evolution of the chaebol (business conglomerate) system by allow- the share of agriculture decreased from 39.9 ing chaebols to borrow foreign capital and to 14.6 percent (see table 19.1). granting them incentives to invest in heavy The "density" of a core city and the "dis- industries (Park and Markusen 1995). Since tance" to the core city were important in the the mid-1980s, high-technology industries industrialization phase, as rapid industrial- such as semiconductors have been increas- ization was accompanied by rapid urban- ingly favored. Since the 1990s, especially ization. The urbanization ratio (share of since the financial crisis in November 1997, cities with a population more than 20,000) the Korean government has sought to increased from 35.9 percent in 1960 to 78.6 promote the development of knowledge- percent in 1990. Most of the rapidly growing intensive industries with the intention of cities during the last four decades were in the opening up the country to trade and capi- suburban areas of Seoul and the southeast tal movements, restructuring the economy, region. In addition, most industrial cities including the financial sector, and making grew rapidly, suggesting that agglomeration the labor market more flexible. Regional or scale economies were important. Most innovation has been the key policy in the of the industrial cities in these regions were twenty-first century (see figure 19.3). rural agricultural areas in the 1960s. Along with sectoral policies, the national Industrial decentralization from Seoul government established several large indus- was one of the major factors contributing trial estates in the southeastern part of the A history of Korea's industrial structural transformation and spatial development 323 Table 19.1 Structure of production in Korea, by sector, 1960­2005 percentage of GDP in current prices Sector 1960 1970 1980 1990 2000 2005 Agriculture 39.9 31.1 14.6 9.0 4.6 3.4 Industry 18.6 28.4 41.4 44.7 42.8 40.4 Mining 2.3 1.3 1.4 0.5 0.3 0.4 Manufacturing 12.1 19.1 29.6 28.9 31.5 28.8 Construction 3.5 6.4 8.2 13.2 8.2 9.0 Utilities 0.7 1.6 2.1 2.1 2.8 2.2 Services 41.5 40.5 44.0 46.3 52.6 56.2 Total (GDP) 100.0 100.0 100.0 100.0 100.0 100.0 Source: Bank of Korea (various years). Figure 19.3 A brief history of Korean industrial policies · Industrial cluster 2000s · Innovative cluster · Regional innovation policy late · Restructuring of chaebols 1990s (conglomerates) financial crisis · New start-ups 1990s Industrial (venture firms) restructuring · Labor strategy early · Subcontracting 1990s · Foreign direct investment · Technological development 1980s · High-tech industrial development · Industrial dispersion · Spatial division of labor 1960s · Industrial complex development ~1970s · Import substitution; Export oriented Source: Author. country. The major new industrial cities without significant intraregional produc- or production agglomerations of Ansan, tion networks. Changwon, Kumi, Pohang, and Ulsan The industrial policies have had a sig- were created as a result of industrial poli- nificant impact on the structural and spatial cies implemented in the late 1960s and the makeup of industries. To show the struc- 1970s. Chaebols contributed heavily to the tural changes more clearly, industries can development and growth of industrial cit- be classified by level of technology or inten- ies by establishing large branch plants with sity of labor. In this chapter, manufactur- imported technology and borrowed foreign ing industries are classified, for the sake of capital. However, the idea of territorial pro- convenience, into five types: labor intensive, duction systems was not successfully imple- capital intensive, resource intensive, assem- mented in the early stage of development. bly, and other special types. The classifica- That is, at the initial stage, the industrial tion is based on the results of a factor analy- estates in the industrial cities had only lim- sis of manufacturing industries conducted ited local interfirm linkages and were just by Park (1993). Assembly-type industries agglomerations of production activities are mostly technology-intensive industries, 324 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA except for the manufacture of furniture try's industrialization, and, even though the (see table 19.2). Since the share of furniture share of assembly-type industries has risen manufacturing is very small, assembly-type slightly since the early 1990s,it remains lower industries are considered to be technology- than the share of labor-intensive industries. intensive industries. In Seoul, the structural changes within an The most distinctive structural change industry have been more intensive than the has been the shift from labor-intensive to structural changes among types of industry. technology-intensive industries. Labor- For example, within the apparel industry, intensive industries played a dominant role Seoul is highly specialized in fashion design during the phase of rapid industrialization and high-value-added products rather than in the 1960s and 1970s, but their share has standardized mass production. declined steadily since the late 1970s. On Spatial changes are closely related to the the contrary, assembly-type industries have industrial policies and structural changes of steadily increased their share since the late industries. In the early phase of industrial- 1970s. There have been slight changes in the ization, the export-oriented industrial policy use of resources and capital, but the most and heavy and chemical industrial policy significant structural shift at the national reinforced spatial disparities, with industries level has been from labor-intensive to tech- concentrated in the capital region and the nology-intensive industries. southeast region (Park and Wheeler 1983). In addition, there are considerable dif- The capital region increased its share of ferences in the structural changes occurring manufacturing employment in the 1960s and in different regions. The southeast region peaked in 1975, with 48.3 percent (see table became highly specialized in assembly indus- 19.3).The southeast region increased its share tries during the last three decades. The mid- sharply, peaking in 1980, with 40.4 percent. dle region began to transform from labor-in- Furthermore, the government's heavy tensive to technology-intensive industries in and chemical industrial development policy the 1990s. The changes to the middle region resulted in a spatial division of labor, with are closely related to the extension of indus- the headquarters of chaebols concentrated trial agglomeration from the capital region in Seoul and production functions decen- to the nearby Chungcheong region as well as tralized to other regions, especially the the development of Daeduck Research Park southeast. The high-technology industrial in Daejun in the 1990s. The shift to technol- policy resulted in a slight reconcentration ogy-intensive industries also occurred in the of production in the capital region in the southwest region, but the degree of the shift 1980s, due to the region's locational advan- is relatively small compared to the nation- tages. The concentration of high-technology wide trend. The southwest region is rather industries and advanced services, including specialized in resource-based industries. research and development (R&D) activities, Changes in the industrial structure of in the capital region intensified the spatial Seoul are significantly different from the division of labor in the 1980s (Park 1993). nationwide changes. Seoul specialized in The middle and southwest regions, labor-intensive industries early in the coun- which can be regarded as peripheral regions, Table 19.2 Share of manufacturing industry in Korea, by type of firm, 1981­2005 percent, unless otherwise noted Type 1981 1986 1991 1996 2001 2005 Resource based 16.5 15.4 14.9 16.8 15.9 15.1 Technology intensive (assembly) 27.6 33.7 41.8 47.8 51.3 57.6 Labor intensive 43.5 38.3 29.8 21.5 19.0 14.3 Capital intensive 4.5 4.3 5.0 8.8 8.6 8.1 Other 7.9 8.3 8.5 5.0 5.2 4.9 Korea, total number of firms 2,559,473 3,290,035 4,231,080 3,748,516 3,415,996 3,450,893 Source: KNSO, Census on Basic Characteristics of Establishments (various years). A history of Korea's industrial structural transformation and spatial development 325 Table 19.3 Share of manufacturing employment in Korea, by region, 1963­2005 percent Region 1963 1970 1975 1980 1985 1990 1995 2000 2005 Seoul 29.6 33.9 30.5 22.1 19.8 15.6 12.5 10.5 9.1 Incheon -- -- -- 7.7 7.2 7.8 8.3 7.8 6.9 Gyunggi 10.2 12.1 17.8 16.1 20.3 24.4 25.9 28.2 30.9 Capital region 39.8 46.0 48.3 45.9 47.3 47.8 46.7 46.6 47.0 Middle region 10.4 9.8 7.7 7.5 6.9 7.9 10.3 11.4 12.2 Southeast region 36.9 34.5 37.5 40.4 40.2 37.8 35.6 34.8 33.6 Southwest region 12.9 9.7 6.5 6.3 5.6 6.4 7.3 7.2 7.2 Korea, total number of 401,981 861,041 1,420,144 2,014,751 2,437,997 3,019,816 2,951,885 2,652,590 2,865,549 firms Source: KNSO, Census of Mining and Manufacturing Industry (various years). experienced a decrease in their share of man- in labor-intensive industries were either ufacturing employment until the mid-1980s, closed or restructured in the early 1990s. after which the middle region recovered According to research on industrial somewhat. The southwest region increased restructuring in the Asian newly industrial- its share of manufacturing employment izing economies of Korea; Taiwan, China; from the mid-1980s to the mid-1990s, but Hong Kong, China; and Singapore in the no significant change has occurred since the early 1990s,Korean firms pursued four strat- mid-1990s. Such regional changes are shown egies for industrial restructuring in the areas in table 19.3. of labor, organization, location, and tech- nology development (Park 1995). Flexible Firms' industrial restructuring, labor strategies, which focused on numeri- spatial changes, and globalization cal rather than functional flexibility, reduced In the 1990s Korea experienced two phases the costs of production by increasing the use of industrial restructuring: one was related of flexible labor, such as temporary work- to the labor movement in the late 1980s ers, part-time workers, and foreign workers. and one to the financial crisis in 1997. Subcontracting--the major organizational From 1987 to 1989, the country experi- strategy--reduced costs by outsourcing pro- enced severe labor disputes, and wages rose duction activities or separating some parts sharply. The sharp wage hikes, appreciation of production lines. Locational strategies of the won with regard to the U.S. dollar, focused on foreign direct investment (FDI), high interest rates, weak financial structure, which was mainly geared toward low-cost high turnover rates, and labor shortages in areas such as China and Southeast Asian production lines triggered the restructuring countries during the early 1990s. There was of firms, especially the labor-intensive small an increase in outward FDI from Korea dur- and medium enterprises (SMEs; see Park ing the 1990s until right before the financial 1993, 1994). crisis in 1997 (see figure 19.4). Along with Individual corporate strategies became efforts to reduce costs, technological devel- more important than ever for firms' sur- opments also were emphasized, and the vival and competitiveness as consumer number of firms involved in R&D activities markets became more diversified, technol- rose. About 19.9 percent of firms conducted ogy advanced rapidly, and product life cycles R&D activities in 1993, compared with became shorter. Before the early 1990s, the about 34 percent in 1996 (Park 2000). government played a critical role in the A second industrial restructuring industrial development and structural occurred after the financial crisis in 1997, changes taking place in Korea (Markusen when the central government took a leading and Park 1993; Park 1991). Since the liber- role in restructuring chaebol groups. During alization of labor in 1989, the strategies of this period, outward FDI stagnated, while private firms have been as important as the inward FDI increased dramatically. Since policies of the government, and many firms 2002, outward FDI has again increased. The 326 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 19.4 Inward and outward FDI in Korea, 1981­2006 cultural "distance" effects were important. US$ thousand Two-thirds of outward FDI to China went 18,000,000 to northeast China, focusing on Shandong province, which is near to and has close his- 16,000,000 torical linkages with Korea. 14,000,000 Changes in innovation systems, recent regional innovation, and 12,000,000 cluster policies $ 10,000,000 Innovation systems in Korea have changed 8,000,000 significantly during the last four decades. thousand In Korea the issues of innovation were 6,000,000 relatively neglected in the 1960s, because the ultimate goal was to provide a founda- 4,000,000 tion for industrialization. The government 2,000,000 took the initiative in the 1960s and 1970s by helping research institutes to take a leading 0 role in improving industrial technologies. 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Most firms were more interested in receiv- year ing technology transfers from the industrial Inward FDI Outward FDI countries than in conducting their own Source: Korea Trade Statistics (www.global.Kita.net). R&D activities. Universities, not firms, took the lead in the development of technology. Therefore, the national systems of innova- tion in the 1960s and 1970s were directed two periods of industrial restructuring sig- mainly by the government's science and nificantly contributed to the globalization of technology policy,which supported the trans- the Korean economy by strengthening the fer of technology to firms and the process relationship between Korea and other Asian of learning from imported technology. The countries in terms of trade and FDI. impact of inbound FDI on the development The two restructuring processes in the of innovation systems was not significant 1990s had a considerable impact on space during this period. and globalization in Korea. In the early Since the 1980s, however, the major role 1990s, many SMEs in labor-intensive indus- in R&D and innovation has shifted from the tries were closed or relocated to China government to private firms (Park 2001). or Southeast Asia, seeking low-cost areas. Many firms have established their own R&D Accordingly, regions that specialized in centers and significantly increased R&D labor-intensive industries, such as Seoul, expenditures. Private firms accounted for were under severe pressure to restructure. 56 percent of total R&D expenditure in 1981, Because the firms pursuing strong techno- surpassing the share of government expendi- logical development were mostly located in ture, and reached 81 percent in 1985 (MOST the capital region, the share of the capital 1990). In 1980 only 54 firms, most of which region did not decline in the early 1990s. belonged to chaebols, had their own R&D Because many new high-tech start-ups were centers, but the figure increased to 2,226 in established in the capital region, especially 1995 (KITA 1995, 1996). Chaebols aggres- in Seoul, there was a slight reconcentration sively established R&D centers in the early of industries for three to four years after the 1980s, but by the late 1980s, even the SMEs crisis. During the subsequent restructur- began to establish their own R&D centers. ing phase, scale economies and "density" According to a survey conducted right effects were significant for high-tech spin- after the financial crisis in December 1997, offs. Globalization progressed rapidly, with SMEs continued to be involved in R&D increasing outward FDI. Especially in the activities in the 1990s (Park 2000). Out of case of outward FDI to China, physical and 825 firms that responded to the question- A history of Korea's industrial structural transformation and spatial development 327 naire, 20 percent of firms conducted R&D of national industrial complexes, including activities in 1993 and 34 percent did so in the electronics and information technology 1996.All in all,the survey revealed that larger (IT) cluster in Kumi, the machinery clus- SMEs were more interested in R&D activi- ter in Changwon, the automobile cluster ties than smaller SMEs. However, among in Ulsan, the parts and components clus- the firms that conducted R&D activities, ter in Ahnsan, the parts and components smaller SMEs spent a higher percentage of of automobile and machinery cluster in total sales on R&D than larger SMEs, which Gunsan-Janghang, the photonics cluster suggests that a considerable proportion of in Kwangju, and the medical instruments these smaller SMEs are high-tech businesses. cluster in Wonju. Thirty-six mini-clusters Presently, more than two-thirds of all R&D emanating from the seven innovative clus- centers were established by SMEs. ters have been developed to promote col- The R&D activities of firms in the 1990s laboration and solve problems in industrial had some distinctive characteristics (Kim practices. 1997): (1) large firms of chaebols established Daeduck Research Park in the city of strategic alliances with worldwide high-tech Daejun has been supported by the spe- firms; (2) large firms, which mostly belong cial law to promote commercialization of to chaebols, were aggressive in establishing R&D and innovations and was renamed as foreign R&D centers and labs; and (3) due the Daeduck R&D Special District in 2003. to the difficulties in getting licenses for lead- Innovation clusters have been promoted ing-edge complex technology, large firms through support for strategic industries in sought to secure original technology by each region. High-tech IT clusters and local merging with or acquiring high-tech firms culture clusters have been supported through in the developed countries. collaboration among diverse economic Since the 1990s, regional innovation actors. And the development of clusters by networks have begun to evolve due to the private firms, such as Suwon's IT cluster by development of regional clusters of SMEs Samsung, Paju's semiconductor cluster by in technology-intensive sectors. The estab- LG-Phillips, and Pohang's material cluster lishment of science parks and high-tech by POSCO, has also been promoted. parks in areas outside the capital region, This strong promotion of balanced in addition to the Daeduck Research Park, national development seems to have had has contributed to the clustering of inno- an impact on the share of regional GDP vation networks since the 1990s. Starting compared to the share of population in the in 2002, the participatory government has capital region, a subject examined in the emphasized balanced national develop- next section. The regional innovation poli- ment and promoted regional innovation cies have certainly increased the density of and cluster policies (Park 2007). Regional regional innovation networks in the prov- innovation policies, which have been pro- inces. However, restrictions on the auton- moted as essential for balanced national omy of universities have exerted negative development, seek to integrate "talent," effects on the development of high-quality "technology," and "industry." Major poli- manpower, limiting the ability of Korea cies for regional innovation are seeking to innovate. to establish regional innovation systems, strengthen the innovation capacity of uni- Spatial development and versities in provinces, promote science and change technology in the provincial regions, and establish networks of industries, universi- During the last five decades in Korea, "dis- ties, and research centers. tance" and "density" effects have been sig- In addition to regional innovation poli- nificant in Korea. Population and economic cies, the policy of promoting innovative activities have become increasingly concen- clusters has been pursued strongly since trated in the capital region, which accounted 2002. Seven innovative clusters have been for only 28.3 percent of the country's total reformulated through the reorganization population in 1970 and for 48.2 percent in 328 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA 2005. Other regions, accordingly, experi- effect of distance from Seoul. Furthermore, enced a decline in their share of population the middle region, which is adjacent to (see figure 19.5). Gyunggi province, has declined in popula- However, there were meaningful excep- tion share, reflecting the effects of distance tions to and fluctuations in this trend. The from the capital region. southeast region's share of population The entropy index, which declines as remained steady during the 1970s, when disparity rises, suggests that the distribution heavy and chemical industries developed of population has become more regionally rapidly along the southeastern coast, but uneven over time (see figure 19.6). While its share of population has decreased the degree of spatial disparity of popula- slightly since then. The population share tion is lower than that of regional GDP and of the middle region declined steadily until of manufacturing, the spatial disparity of 1995, after which it increased slightly. The population is increasing, whereas that of southwest region, which was underdevel- regional GDP or manufacturing is not. The oped during the phase of rapid industri- entropy value in this chapter is measured alization, has seen a steady decrease in its as follows: share of the population, from 21.3 percent in 1970 to 11.7 percent in 2005, having n lost many residents to the capital region. H = - q (19.1) ilog2 qi , Seoul's share of population increased i steadily until the end of the 1980s, but has where H is entropy value; q is a set of declined since then, even though there has nonnegative numbers that sum to unity been no significant change in the absolute n number of population (see table 19.4). iqi = 1.0; n is the number of sub- Gyunggi province, which is essentially a groups (in this case, 16). If any q is equal to suburb of Seoul, has steadily increased its 1 and all other qs are 0, then H is equal to 0. share of population from 8.2 percent in For a given n, H is at its maximum when all 1970 to 22.0 percent in 2005, reflecting the qs are equal, so that Figure 19.5 Share of population in Korea, by region, 1970­2004 60 50 40 30 percent 20 10 0 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 year Capital region Middle region Southeast region Southwest region Source: KNSO, Resident Registration Population (various years). A history of Korea's industrial structural transformation and spatial development 329 Table 19.4 Share of population in Korea, by region, 1970­2005 Region 1970 1975 1980 1985 1990 1995 2000 2005 Seoul 17.6 19.9 22.3 23.8 24.4 22.9 21.4 20.8 Incheon 2.5 2.8 3.3 3.8 4.4 5.2 5.4 5.4 Gyunggi 8.2 8.9 9.9 11.5 13.9 17.2 19.5 22.0 Capital region 28.3 31.5 35.5 39.1 42.8 45.3 46.3 48.2 Middle region 19.7 18.2 16.5 15.1 13.9 13.2 13.4 13.2 Southeast region 30.4 30.5 30.5 29.8 28.9 28.6 27.9 26.8 Southwest region 21.6 19.8 17.4 15.9 14.4 12.9 12.5 11.7 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Korea, total 32,240,827 35,280,725 38,123,775 40,805,744 42,869,283 45,092,991 47,008,111 48,138,077 Source: KNSO, Resident Registration Population (various years). Figure 19.6 Entropy index 3.400 3.300 3.200 3.100 3.000 2.900 2.800 2.700 2.600 1955 1957 1959 1961 1963 1965 1967 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 Population GRDP Manufacture (value of products) Manufacture (wage) Manufacture (employment) Sources: KNSO (various years); author's calculations. H = -n n log21 than 5 percent, turning negative only twice: during the second oil crisis in 1979 and dur- n 1 = log2n . (19.2) ing the financial crisis in 1997. There have been some spatial changes in This trend of increasing population dis- regional GDP (see figure 19.8). The capital parity is closely related to the continuous region steadily increased its share of total concentration of population in the capital regional GDP until 1993, when it peaked at region, suggesting the significant effect of 49.1 percent. There were some fluctuations agglomeration economies and density. after that time: a decrease to 46.8 percent in 1998 and then an increase to 48.6 percent Spatial pattern of regional GDP in 2002. There was an opposite trend in the growth in 1985­2005 southeast region, where the share of regional Sincethelaunchof theFirstEconomicDevel- GDP declined slightly from 1985 until 1993 opment Plan in 1962, the annual growth rate and increased slightly from 1993 to 1998. of gross domestic product (GDP) has been The share of the middle region remained impressive, as shown in figure 19.7. For the fairly steady, whereas the share of the south- most part, annual growth rates were more west region declined slightly. 330 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA Figure 19.7 Annual growth rate of GDP in Korea, 1960­2005 leader. Beginning in 2002, the participatory government offered diverse incentives for 15 industries to locate in provinces other than 10 the capital region, on the one hand, and imposed strong restrictions on the expan- 5 sion of economic activities in the capital percent 0 region, on the other. 1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 Moreover, spatial disparity, in terms of ­5 regional GDP per capita, declined continu- ously from the mid-1980s until the financial ­10 year crisis in 1997, showing a clear trend of con- Source: Bank of Korea (various years). vergence (see figure 19.9, panel B). In 1997 the southwest region reached 95 percent of Figure 19.8 Regional GDP in Korea, by region, 1985­2003 the national average, rising from 77 percent in 1985. However, there was a trend of slight 60 divergence after the financial crisis until 2002. Since 2002, the trend once again has 50 been toward convergence. Seoul's regional GDP per capita has 40 been continuously higher than the national average, even after 2002, when the capital 30 region's GDP per capita was lower than the percent national average. Seoul has attracted new 20 industries or restructured its economy to keep its comparative advantages. Seoul is the 10 center of new industries and innovations, taking a leading role in the spatial division of labor in Korea. 0 Accordingly, the relative stagnation of 1985 1988 1991 1994 1997 2000 2003 year regional GDP per capita in the capital region is not the result of the stagnation of Seoul. Capital region Middle region Southeast region Southwest region It is due to the stagnation of Gyunggi prov- Source: www.kosis.kr. ince. Per capita regional GDP in Gyunggi province has stagnated for the following reasons. First, the population growth rate is Regional GDP per capita, by region, high in Gyunggi because of the continuous however, shows an interesting pattern, with in-migration from the rest of the country, a general trend of convergence. There were while the location of industry in the capital two turning points (see figure 19.9, panel A). region is controlled by law. Second, the sta- First, the middle region overtook the south- tistics underestimate the amount of service east region in 1989, when President Tae-Woo activities in regional GDP because of diffi- Rho liberalized labor. During the phase of culties in measuring the production of small rapid industrialization in the 1970s, indus- service-related firms. Because service activi- try clearly was concentrated in the capital ties are concentrated overwhelmingly in the and the southeast regions. However, with the capital region, the capital region's regional high-tech industrialization that began in the GDP is underestimated, especially in Seoul. 1980s,especially after the labor disputes from 1987 to 1989, the southeast region lost its Spatial division of labor second-place position to the middle region. Large industrial firms or chaebols evolved Second, the capital region lost its primacy during the phase of heavy and chemi- in terms of regional GDP per capita to the cal industrialization in the 1970s. Large middle and southeast regions in 2003. Until firms located their headquarters mostly in 2003, the capital region had always been the Seoul and located their plants in the major A history of Korea's industrial structural transformation and spatial development 331 Figure 19.9 Per capita regional GDP in Korea, by region, 1985­2005 A. Amount (won million) 18 16 14 12 10 8 6 4 2 0 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 year Capital region Middle region Southeast region Southwest region B. Korea average in each year = 100 120 115 110 105 100 95 90 85 80 75 70 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 year Capital region Middle region Southeast region Southwest region Korea total Source: www.kosis.kr. industrial estates outside Seoul, especially government provided support mainly to in the southeast region. With the trend of large firms because the scale of investments spatial separation between headquarters required in these industries was exception- and manufacturing plants, about 85 percent ally large. During the 1970s, loans to heavy of the headquarters of firms that had spa- and chemical industries accounted for tially separated plants were located in Seoul, more than 70 percent of all government while about 87 percent of the plants that funds made available to manufacturing had spatially separated headquarters were (Park 1981). These government loans were located outside Seoul (Park 1985). provided at very low interest rates, often During the 1970s, to develop heavy negative in real terms. In addition, govern- and chemical industries in the 1970s, the ment supported the rise of large firms by 332 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA establishing highly protective barriers for oriented more toward production research. infant industries and by permitting mono- In addition, high-tech sectors in the capital polistic production to overcome the prob- region had much stronger local linkages lems of a small domestic market. than other areas for purchasing input mate- During the 1980s, when government pol- rials. Such differences in R&D activities and icy emphasized high-tech industrial develop- the degree of local linkages of high-tech sec- ment,anotherspatialdivisionof laborevolved tors indicate that spatial divisions of labor beyond the division between the headquar- intensified significantly during the 1980s ters' control functions and the plants' pro- (Park 1993). duction functions. Because the capital region Since the early 1990s, headquarters and had advantages in terms of skilled labor,tech- high-tech industries have tended to locate nology, engineering services, and so forth, it outside the capital region, especially in the became more specialized in high-tech indus- middle region. However, the spatial divi- tries and R&D activities, while the rest of the sion of labor persists. In recent years, more country remained more specialized in the less than half of the headquarters of the top 100 high-tech industries. Even in the traditional firms were located in Seoul, even though sectors, such as textiles and apparel, the pro- some of them have decided to leave Seoul, duction of technology-intensive or high-val- and the share of manufacturing plants of ue-added products was concentrated in the the top 100 firms in Seoul has declined capital region. from 3.0 percent in 2002 to 2.3 percent in R&D activities in Seoul were oriented 2007 (see table 19.5). more toward basic and applied research, In addition to the concentration of which required more qualified manpower headquarters in the capital region, knowl- and information infrastructure, whereas edge-intensive activities are concentrated in R&D activities in the periphery were Seoul, suggesting that distance and density effects are important in the advanced ser- vices. However, the capital region's share Table 19.5 Share of headquarters and plants of top 100 firms in Korea, by region, 2002 and 2007 of the headquarters of the top 100 firms percent declined from 79 percent in 2002 to 72 HQ of top 100 firms Plants of top 100 firms percent in 2007, and its share of patents Region 2002 2007 2002 2007 declined from 80.6 in 1995 to 75.3 percent in 2005 (see table 19.6). Seoul has special- Seoul 66.0 58.0 3.0 2.3 Incheon 4.0 5.0 5.6 5.8 ized in advanced production services, such Gyunggi 9.0 9.0 16.1 14.5 as design, advertising, legal and financial, Capital region 79.0 72.0 24.6 22.5 management consulting, and ICT services. Middle region 6.0 6.0 22.3 21.5 For example, Internet domains are highly Southeast region 12.0 17.0 39.0 38.9 concentrated in Seoul, as is the transaction Southwest region 3.0 5.0 14.1 17.0 Korea, total number of firms and plants 100 100 305 311 of business-to-business (B2B) e-commerce (Park 2004). Source: Maeil Economic Daily (various years). Dynamics of Seoul and the Table 19.6 Share of industrial patents in Korea, by capital region region, 1995, 2000, and 2005 percent Population growth in Seoul was rapid in the 1970s and 1980s but has stagnated in Region 1995 2000 2005 recent years, whereas population growth Seoul 57.8 43.7 44.9 in Gyunggi province has been rapid since Incheon 4.6 5.1 4.2 Gyunggi 18.2 24.9 26.2 the 1990s, due to the influx of in-migration. Capital region 80.6 73.7 75.3 Regional GDP seems to be related to popu- Middle region 5.2 8.4 8.2 lation growth. During the recent years of Southeast region 12.5 14.3 12.8 growth, Gyunggi's regional GDP per capita Southwest region 1.7 3.5 3.7 has been lower than the national average. Korea, total number 52,449 106,363 162,844 of patents During its period of growth in the 1980s, Source: MOST (various years). Seoul's regional GDP per capita was lower A history of Korea's industrial structural transformation and spatial development 333 than the average of the capital region, Seoul, especially from the Gangnam area, though still higher than the national aver- where informal and formal networks are age; as Seoul's population growth has stag- intensive. The Gangnam area is the center of nated, its regional GDP per capita has risen breakfast meetings and evening gatherings, and is now higher than the average of the where new ideas, information, and knowl- capital region (see table 19.7). edge are exchanged, transferred, and gener- Seoul is the home of many new start-ups ated (Park 2005). There are many high-class in the technology-intensive sectors or newly hotels and restaurants, and Gangnam is the emerging industries, even though many center of new fashion, knowledge creation, firms have relocated to adjacent Gyunggi and innovation. province. In the late 1980s and early 1990s, Third, path dependence and path cre- new high-tech firms that were established ation are important spatial processes in the in Seoul relocated to or established branch capital region. The extension of high-tech plants in Gyunggi province, while maintain- industrial agglomeration to the middle ing their headquarters in Seoul, creating a region along the Seoul-Busan expressway is spatial division of labor in high-tech sectors a spatially path-dependent process, on the within the capital region. Gyunggi province one hand,and the development of Gangnam has specialized in high-tech industries and as a new core in Seoul and the development has hosted a number of high-tech agglom- of a new R&D center in Hwasung, Gyunngi erations since the 1990s, whereas Seoul has province, are examples of the creation of specialized in advanced services, such as new spatial paths, on the other hand. computer software, engineering, legal ser- vices, and high-tech start-ups. Development of ICT and Three factors have been significant in spatial changes this spatial restructuring: distance, den- The number of ICT-related firms has grown sity of network, and path dependence and considerably in recent years, and almost creation. First, the movement of industrial half of all ICT firms (44.8 percent) were agglomeration from the central part of concentrated in Seoul in 2003. The degree Seoul before the early 1960s to the inner of concentration in Seoul varies by type of city of Seoul in the 1970s; the movement firm. Firms in the ICT-related service sec- of high-tech industrial agglomeration from tor and in wholesale and retail trade are far Seoul in the 1980s to the nearby suburbs in more concentrated in Seoul than firms in Gyunggi province in the 1990s; and the cur- ICT-related manufacturing. During the last rent extension of high-tech agglomeration decade, IT-related service industries have to distant suburbs of Gyunggi province and tended to concentrate in Seoul, whereas IT- the nearby middle region, such as the city related manufacturing industries have dis- of Cheonan. The extension of agglomera- persed to other regions. tion has a wave-like pattern, with "distance" The impact of ICT on the Korean econ- being the key factor. omy can be examined through the creation Second, the density of networks and of Internet domains, e-commerce, and vir- collaboration is critical for innovation and tual networks in the rural areas. The Inter- the generation of new firms and industries. net infrastructure in Korea has been well New industries in Korea always evolve from developed since the late 1990s. By the end of Table 19.7 Regional GDP per capita in the capital region of Korea, 1985­2005 million won Region 1985 1990 1992 1993 1997 2000 2003 2005 Seoul 5.4 8.5 10.0 10.7 13.2 13.7 15.4 15.8 Gyunggi 6.4 9.0 10.0 10.2 11.1 12.2 13.2 15.0 Capital region 5.7 8.7 9.9 10.4 12.2 12.7 14.0 15.1 Korea, total 4.9 7.8 8.9 9.4 11.7 12.3 14.0 15.2 Source: www.kosis.kr. 334 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA June 2007, there were 34.43 million Inter- 79.5 percent share of the total firms operat- net users, and the rate of Internet usage had ing B2B e-commerce sites in Korea in 2003 risen to 75.5 percent (see www.nida.or.kr). (Park 2004). If we consider only the public In general, the younger generation shows a B2B e-marketplaces, in which many sell- much higher usage rate than older genera- ers and many buyers can conduct transac- tions, and the usage rate of males is some- tions, the degree of concentration in Seoul what higher than that of females. The data is overwhelming, at 84 percent (Choi 2003). on Internet usage suggest that there is no The predominance of Seoul in the loca- significant difference in access to the Inter- tion of public B2B e-marketplaces might net by regions (that is, between the capital be related to the clustering of ICT firms, and other regions), but rural and urban ICT-related spin-offs, and innovative entre- areas differ somewhat, largely as a result of preneurs and knowledge workers, especially the age distribution of the population. In in Gangnam area (Park 2004). ICT-related 2000 the proportion of persons belonging firms and advanced producer services are to the older age groups in rural areas was strongly concentrated in the Gangnam dis- 17.9 percent, compared with only 4.3 per- trict within Seoul (Park and Choi 2005). The cent in the cities (see isis.nic.or.kr). overwhelming concentration of the dot-kr Even though the regional disparity in domain, B2B e-commerce, and B2C e-com- usage rates is not significant, the dot-kr merce in Seoul suggests a strong tendency of domains are concentrated mostly in the ICT firms to cluster in Seoul, even though capital region. According to a survey by the the Internet infrastructure (transferability) Korea Network Information Center (see is present throughout the country. www.nida.or.kr), in August 2007, Seoul had 58 percent and the capital region had 78 per- Virtual innovation networks in cent of the total number of dot-kr domains peripheral areas in Korea. The share of Seoul and the capital Althoughmanypeopleregardtheimportance region has increased slightly in recent years of networks in spatial clustering and disper- (see table 19.8). The higher concentration of sal as applicable only to advanced economies, dot-kr domains in the capital region com- box 19.1 shows that new economic spaces pared with the share of population or Inter- can also occur in peripheral areas. net users suggests that the concentration The case of Sunchang suggests a new might be related to other factors that attract paradigm in the organization of production ICT-related firms in Seoul. The number of systems and economic spaces, with the cre- B2C (business-to-consumer) e-marketplaces ation of new ideas based on intensive local is also highly concentrated in the capital and nonlocal networks. Sunchang has insuf- region, with Seoul having about 73 percent ficient high-quality manpower, but it has of the total number (Choi 2003). networks of advanced services and top-level Firms operating B2B e-marketplaces are scientists and engineers. These networks do even more concentrated in Seoul and its not represent actual clusters of advanced ser- surrounding areas. The capital region had a vices; rather, they represent a virtual innova- tion network that has allowed the transfer of knowledge and innovation. High-quality Table 19.8 Share of dot-kr domains in Korea, by region, 2001­07 Internet infrastructure has enabled high- percent tech engineers and scientists to meet peri- Region 2001 2003 2005 2007 odically, while paved roads and easy access Seoul 58.50 55.60 56 57.60 to highways have made Sunchang accessible Incheon 1.20 1.30 3.40 3.40 to major regional cities and Seoul. Gyunggi 2.00 2.00 17.30 17.10 Capital region 61.70 58.90 76.70 78.10 Middle region 11.80 12.10 5.90 5.80 Policy implications Southeast region 21.70 24.40 13.00 12.00 Southwest region 4.70 4.50 4.40 4.10 Industrial and innovation policies need to Korea, total number of domains 457,450 611,548 642,770 928,177 keep pace with the transformation of soci- Source: www.nida.or.kr (National Internet Development Agency of Korea). ety and economy. In the knowledge-based A history of Korea's industrial structural transformation and spatial development 335 B O X 1 9 . 1 The case of Sunchang Sunchang-gun in Jeonbuk province is one of the traditional taste.Daesang Company has invested in most underdeveloped regions in Korea (Park 2005). advertising and R&D activities and has introduced Sunchang is located in a mountainous, peripheral automated mass production.Daesang Company area in southwestern Korea where the popula- illustrates the possible link between the production tion has been declining steadily since the 1970s. technology of a large company and traditional local Agriculture is the key economic sector, producing culture and resources.That is, the codified knowl- principally tobacco, red pepper, and diverse veg- edge of the Daesang Company is linked to locally etables and fruits.Gochoojang (a thick soy paste embedded knowledge and resources. mixed with red pepper) is a product of Sunchang, Second, the county of Sunchang designed famous for its distinct taste, which is acquired in a complex to gather the traditional gochoojang the fermentation process.Traditionally, most of the makers together in one place.The county allowed households in Korea made their own gochoojang. skilled persons, who have a license to make tradi- However, nowadays most households buy gochoo- tional gochoojang and more than 10 years of expe- jang in the market, and the brand name"Sunchang rience in making it, to move their household into Gochoojang"has become famous. the complex.As a result, 54 households live in the Since the late 1990s, the production system of complex and make their own specialized traditional Gochoojang has changed significantly as a result of gochoojang, selling their products in the local mar- two developments.First, Daesang Company, one of ket and through the Internet to consumers in large the large companies in the food industry of Korea, cities.In addition, the county recently established established a branch plant in Sunchang to produce a Research Institute of Soybean Paste for continu- standardized gochoojang under the brand name ous development of several types of soybean paste "Sunchang Gochoojang,"with quality control and a industries. information society of the twenty-first circulation,"which encourages talented peo- century, a new industrial policy is needed ple to live in the provincial areas. Qualified that focuses on the supply of qualified high schools and good living and service human resources. A policy of simply relo- environments are needed if regions outside cating industrial firms may not be enough the capital are to attract and retain talented to spur regional development in the pro- people. Specifically, university-industry col- vincial areas. Accordingly, Korea needs inte- laborative networks should be strengthened grated regional innovation policies, which in the provinces, and at least one good high can be improved to promote spontaneous school should be nurtured in each province. regional development. Future policy should Second, regional innovation and clus- seek not only to promote balanced national tering strategies should be promoted with development but also to provide an engine regional integration of the central city and for sustainable development with the inter- hinterlands. The regional innovation poli- play of scale economies, factor mobility, and cies in Korea have been promoted based development of ICT. on the administrative boundaries of 16 cit- First, attracting talented persons to the ies and provinces. Special cities and prov- provincial regions is critical, as regional inces have promoted regional innovation development depends largely on the creation policies separately, and there has been no of knowledge and capacity to innovate. To regional integration between a central city entice such talent to areas outside the capital, and its surrounding province(s). To pro- it is essential to promote the interaction of mote regional innovation successfully, a all economic actors in the region and to cre- central city (special city) and its hinterlands ate an environment in which creative people (provinces) should be integrated to generate and scientists can thrive. A policy of merely synergy effects and scale economies.Accord- relocating R&D centers and firms will not ingly, a wide-area regional integration plan, sufficiently improve the innovation poten- which considers a central city and surround- tial of peripheral regions. "Brain drains" of ing province(s) as one regional innovation the past, which represent out-migration of system, should be developed. Strategic talented people toward Seoul and foreign industries and cluster strategies should also countries, should be transformed into"brain be promoted at the level of an integrated 336 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA wide area, not at the level of a province or competitive by supporting the sustainability city within a wide area. of their population and the development of Third, new strategies to diffuse innova- virtual innovation networks. tion to the level of counties and small cities within an integrated region are needed. For Note example, a "local learning festival" could be Sam Ock Park is a professor at Seoul National considered, in which firms, schools, public University. institutes, all levels of students, and resi- dents from counties or cities get together to References display their creative and cooperative power. To counter the recent trend of high school Bank of Korea. Various years. National Indictors students avoiding engineering school and on National Account. Seoul: Bank of Korea. the natural sciences, efforts should be made Choi, Ji-Sun. 2003. Public B2B Electronic Mar- to inspire them to study science and engi- ketplaces: A Spatial Perspective. Ph. D. dis- neering and to help them to understand sertation, Department of Geography, Seoul the importance of scientific technology and National University. enterprise in their daily lives through fun Kim, Hyun Sook. 1997."Innovation Systems opportunities such as symposiums, work in and Science and Technology Policy in Korea." scientific laboratories, and programs that In Technology Capacity and Competitiveness of expose students to science and open them Korean Industry, ed. Keun Lee and others, pp. 123­66. Seoul: Kyungmoonsa. to the exchange of ideas. Regional innova- tion should be promoted, as should net- KITA (Korea Industrial Technology Associa- works joining firms, universities, research tion). 1995. Statistics of Industrial Technology. institutions, and government agencies. Seoul: KITA. Fourth, considering the demographic ------. 1996. Directory of Korea Technology trends of low birth rates and population Institutes, 95/96. Seoul: KITA. aging, a new system should be established KNSO (Korean National Statistical Office). to use talented people in the provinces to 2005. Results of the EC Statistical Survey. assist Korea's aging society. As most longev- Seoul: KNSO. (www.nso.go.kr.) ity is occurring in the provinces, efforts are ------. Various years. Census on Basic Charac- needed to attract "retired brains" to prov- teristics of Establishments. Seoul: KNSO. inces and involve them in contributing to ------. Various years. Census of Mining and local development. Attracting retired brains Manufacturing Industry. Seoul: KNSO. to the provincial regions could enhance the ------. Various years. Resident Registration innovation capacity of regions and help Population. Seoul: KNSO. to solve regional problems, including the shortage of qualified manpower. Lee, Chan, Ki-suk Lee, Yeon-ok Kim, Sam Ock Park, Man-ik Hwang, Woo-kung Huh, and Finally, beyond the regional industrial Dong-won Park. 1988. Korea: Geographical strategy and cluster polices, sustainability of Perspectives. Seoul: Korea Educational Devel- rural areas should be considered. Continu- opment Institute. ous out-migration of population from rural Maeil Economic Daily. Various years. Annual areas can result in the deterioration of rural Reports of Enterprises. Seoul. economies. Virtual innovation networks using local resources and culture should be Markusen, Ann, and Sam Ock Park. 1993. promoted in remote areas. Regional integra- "The States as Industrial Locator and Dis- trict Builder: The Case of Changwon, South tion of diverse areas within a region with an Korea." Economic Geography 69 (2): 157­81. emphasis on sustainability should be actively promoted, in addition to balanced national MOST (Ministry of Science and Technology). development. Spatial disparity among 1990. Science and Technology Annual 1990. Seoul: Sin Jin Business Affairs. regions has declined somewhat, but dispari- ties among localities within a region are still ------. Various years. Research of R&D Activity. problematic.Some rural areas can be globally Seoul: Sin Jin Business Affairs. A history of Korea's industrial structural transformation and spatial development 337 Park, Sam Ock. 1981. Locational Changes in ------. 2001."Regional Innovation Strategies Manufacturing: A Conceptual Model and Case in the Knowledge-Based Economy." GeoJour- Studies, Ph. D. dissertation, University of nal 53 (1): 29­38. Georgia, Athens, GA. ------. 2004."The Impact of Business-to-Busi- ------. 1985."Industrial Location Policies in ness Electronic Commerce on the Dynamics Major Metropolitan Areas of Korea." Journal of Metropolitan Spaces." Urban Geography 25 of Korean Planners Association 20 (2): 202­20. (4): 289­314. ------. 1991."Government Management of ------. 2005."Network, Embeddedness, and Industrial Change in the Republic of Korea." Cluster Processes of New Economic Spaces In The State and the Spatial Management of in Korea." In New Economic Spaces: New Eco- Industrial Change, ed. G. J. R. Linge and D. C. nomic Geographies, ed. Richard Le Heron and Rich, pp. 74­87. New York: Routledge. James W. Harrington, pp. 6­14. Aldershot: ------. 1993."Industrial Restructuring and the Ashgate. Spatial Division of Labor: The Case of the ------. 2007."Regional Innovation Policies for Seoul Metropolitan Region, the Republic of Maximizing Endogenous Regional Devel- Korea." Environment and Planning A 25 (1): opment Capabilities." In Balanced National 81­93. Development Policy in Korea: Theory and ------. 1994."Industrial Restructuring in the Practice. Seoul: Presidential Committee on Seoul Metropolitan Region: Major Trigger Balanced National Development. and Consequences." Environment and Plan- Park, Sam Ock, and J. S. Choi. 2005."IT Service ning A 26 (1): 527­41. Industries and the Transformation of Seoul." ------. 1995."Seoul, Korea: City and Suburbs." In Service Industries and Asia-Pacific Cities, In Asian NIEs and the Global Economy, ed. Peter W. Daniels, Kong C. Ho, and Tom A. Industrial Restructuring, and Corporate Hutton, pp. 301­20. London: Routledge. Strategy in the 1990s, ed. Gordon Clark and Park, Sam Ock, and Ann Markusen. 1995. Won Bae Kim, pp. 143­67. Baltimore, MD: "Generalizing New Industrial Districts: A Johns Hopkins University Press. Theoretical Agenda and an Application from ------. 2000."Innovation Systems, Networks, a Non-Western Economy." Environment and and the Knowledge-Based Economy in Planning A 27 (1): 81­104. Korea." In Regions, Globalization, and Park, Sam Ock, and J. O. Wheeler. 1983. Knowledge-Based Economy, ed. John H. "Industrial Location Policies and Dunning, pp. 328­48. Oxford: Oxford Manufacturing Employment Change: University Press. The Case of ROK." Regional Development Dialogue 4 (2): 45­64. SECTION IV Lessons from experience: reshaping economic Conclusion: lessons from experience geography in East Asia Yukon Huang and Alessandro Magnoli Bocchi 20 This chapter provides an overview of the both bilateral and multilateral. Transport development of East Asia over the last costs between countries fell, allowing for few decades, as seen through the lenses of greater specialization and, because trade in economic geography (WDR 2009). The intermediate goods is especially sensitive to main message is that the region's impres- transport costs, radically altered trade pat- sive economic performance can be under- terns. Trade with neighbors became more stood within the context of--and, in turn, important than trade with others outside the was shaped by--the dynamics of spatial region. Intraregional trade now accounts for c h a p t e r economics. As discussed in a recent study between 55 and 60 percent of all imports and (Gill and Kharas 2007), East Asia is the exports in East Asia, approaching that in the fastest-growing region in the world, with European Union (EU) and the North Amer- many countries achieving impressive gross ican Free Trade Agreement (NAFTA). The domestic product (GDP) growth (see table driving force has been growth in the trade 20.1). The major driving force for this per- of intermediate goods, which accounted for formance was regionalism, as East Asian about 60 percent of total intraregional trade economies took advantage of globaliza- in 2005. Driven by booming intraregional tion by rapidly expanding trade, especially trade of manufactured goods, industrial- with each other. This was facilitated by the ization spread across East Asia. Productive remarkable growth of China and the promi- activities became geographically concen- nent role that it played in regional produc- trated in each country, reinforcing the lead- tion-sharing networks. ing role of industrial agglomeration in the development process. Context: the rise of regionalism Today, the pattern of East Asian trade and the role of production- is complex and multidirectional, with the sharing networks transformation revealing the importance of scale economies. The reduction of trans- As Hamaguchi discusses in chapter 1, eco- portation and communication costs has nomic geography--through the opposing enabled firms to cut production processes forces of concentration and the dispersion into "pieces of tasks" and to allocate each of economic activity across countries--has "piece" to the most suitable location given played a key role in shaping these trends. factor price differences, a process dubbed To begin with, East Asian countries low- "fragmentation." This has lowered costs ered their international transaction costs and, with the shift to more efficient produc- through trade policies facilitating imports tion centers, allowed firms to benefit from of intermediate goods, favorable treatment economies of scale. Thus "fragmentation" of foreign direct investment (FDI), and the in East Asia explains much of the rise of development of infrastructure. This encour- production-sharing networks. aged a proliferation of free trade agreements, Lessons from experience: reshaping economic geography in East Asia 339 Table 20.1 Key indicators for growth, urbanization, and income distribution in East Asia, by country, various years Gross GDP growth (average Urban population national annual percent) (percent of total) Theil index income per Country capita, 2006a 1960­80 1981­2006 1960 1980 2006 1990 2002 Lao PDR 1,740 -- 5.9 7.9 12.4 21.0 19.8 23.1 Vietnam 2,310 -- 6.8 14.7 19.2 26.9 22.4 25.4 Indonesia 3,310 6.0 5.3 14.6 22.1 49.2 20.6 23.8 Philippines 3,430 5.4 2.9 30.3 37.5 63.4 30.1 36.8 Thailand 7,440 7.5 6.0 19.7 26.8 32.6 39.2 34.2 China 4,660 5.5 9.9 16.0 19.6 41.3 21.1 35.8 Korea, Rep. of 22,990 7.8 6.8 27.7 56.7 81.0 17.0 17.5 Source: World Bank (2008). For the Theil index, see Gill and Kharas (2007). -- Not available. a. Purchasing power parity, current international $ (a hypothetical unit of currency that has the same purchasing power that the U.S. dollar had in the United States at a given point in time). Accompanied by increasing the results are viewed from a regional or a disparities national perspective. On a regional basis, during the period of 1990­2005, each econ- Everywhere in the world, economic develop- omy made progress toward catching up with ment is uneven across space. Some countries Japan, although the progress was speedier grow rich, while others stay poor, and when in some countries than in others: Singa- countries grow, prosperity does not come pore among the upper-income countries, to every place at the same time. People and China among the lower middle-income firms concentrate in prosperous areas, cities countries, and Vietnam among the low- quickly pull ahead of the countryside, and income countries (Hamaguchi in chapter the quality of life--in terms of consumption 1). Over time, the region has become more levels--improves in some provinces and diverse and less reliant on Japan, given the lags in others. Higher premiums are paid rise of agglomeration economies in many to skilled labor and to knowledge-intensive countries. Within each country, however, products. despite regional convergence, income dis- East Asia is no exception. A compelling parities have become more serious. Between feature of the East Asian growth experience 1990 and 2005, inequality appears to have over the last two decades has been the uneven increased in six of the seven countries stud- spread of economic benefits, especially ied in this book, the only exception being within countries. The same forces that con- Thailand. Increases have been especially tributed to rapid growth have also shaped its pronounced in China, but to a lesser degree unevenness. In almost all countries, growth also in Indonesia, Lao PDR, the Philippines, has been accompanied by the widening, or and Vietnam (table 20.1).1 at least the persistence, of disparities and inequalities across space, sectors, groups, and, ultimately, individuals.As a result, there The new economic geography are growing concerns about the equity and Until recently, the traditional trade theory, social cohesion of the development process. based on differing comparative advantages Governments uniformly face increasing as reflected in varying resource endow- pressures to deal with distributional issues, ments, relative factor prices, and technologi- which often have a spatial character; exam- cal advantages, could explain these trends in ples are the focus on narrowing urban-rural production and distribution rather easily. differences and on implementing targeted The first attempt to model the shifting regional programs for lagging areas. location of production in East Asia was put In East Asia, the distributional or equity forward in the famous"flying geese"pattern implications differ depending on whether of catch-up industrialization, with Japan 340 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA as the lead economy successively shed- public policies, the location of production, ding industries over time to less-developed and, in turn, trade and growth patterns. economies, first to the newly industrializing Everything starts with the desire of firms to economies(NIEs)suchasHongKong,China; concentrate production in one location so as Singapore, Republic of Korea; and Taiwan, to enjoy plant-level economies of scale and China, then to lower-income Association of to be near customers and suppliers to reduce South East Asian Nation (ASEAN) countries transport costs. Once a market has reached such as Indonesia, Malaysia, the Philippines, a certain scale, other firms locate there to and Thailand, and then to China. But one take advantage of market size, thereby giv- drawback of this model is that, although ing rise to "agglomeration economies." The it focuses on interindustry relocation and existence of a large manufacturing sector trade, it does not explain intraindustry trade represents an incentive for others to come, or why some industries move to low-wage reinforcing the original advantages. Factors countries, while others do not. Indeed, the of production, however, and especially labor, model suggests a predetermined homo- are not mobile across countries in the same geneous trajectory and a minimal role for way they are mobile within countries; thus policy. In contrast, the "new economy geog- cost structures may drive firms from larger, raphy" allows for less determinism. There is higher-wage areas to smaller, lower-wage more potential for multiple equilibria, and areas both within and across countries. The small changes in initial conditions may have lower the transport costs firms face, the less large effects. History and luck also play a role likely they are to congregate in one major in determining which cities or countries are center. selected as the location of firms. And given In this volume, the three spatial dimen- the presence of unexhausted economies of sions of development proposed in the scale, the selected areas will have a persistent World Development Report 2009 (WDR advantage into the future. 2009; World Bank)--density, distance, and East Asia's success is evident in the con- divisions--provide a conceptual framework centration of industrial agglomeration in for the geographic transformation of East large urban areas in each country, exempli- Asia.2 fied by the core economic regions in Japan For policy makers, the challenge is and in major production centers through- getting density right by fostering the appro- out the region. The transformation is from priate concentration of economic activities a one-dimensional flying geese pattern to to realize the potential benefits of agglom- a more complex pattern that encompasses eration economies. If this is done well, multiple technological centers built on economic growth will be driven by geo- concentrations of industrial and service graphically concentrated clusters, and liv- activities that foster economies of scale. ing standards between lagging and leading In this model, high transport costs can areas and between urban and rural areas affect the location of activities and pre- will converge over time. But the distance vent agglomeration economies from being between concentrated activities is also an realized. important factor, which can be addressed In contrast to previous studies on East both by favoring the mobility of labor and Asia, this volume takes a country-specific by reducing transport costs with infrastruc- approach, focusing more on what is happen- ture investment. In the process, any artificial ing within each country rather than what is or politically driven divisions--due to juris- happening across countries regarding the dictional boundaries, ethnicity, language, or impact of spatial factors. religion--can divide people, hamper eco- As indicated in the preface to this book, nomic activities, impede growth, and exac- the body of thought encompassed in the erbate social development. new economic geography constitutes the The country studies collected in this vol- analyticalframeworkof thisvolume,explain- ume illustrate how the role of economic geog- ing how spatial factors affect the course of raphy in shaping outcomes depends largely a country's development by influencing its on the stage of a country's development as Lessons from experience: reshaping economic geography in East Asia 341 well as on the nature of government policies. consistent with that of a rapidly transform- For this purpose, our discussion categorizes ing economy, with growth increasingly evi- countries into four groups, which reflect dif- dent in more urbanized areas that benefit ferent developmental stages: from industrial expansion. This process was supported by government policies, which 1. Vietnam and Lao PDR,low-income econ- allocated more fiscal resources for the more omies on a path of rapid growth; rapidly growing provinces, thus providing 2. Indonesia, the Philippines, and Thailand, further incentives for industrial develop- middle-income countries with a variety ment. Today, the agglomeration process is of growth experiences; being facilitated by Vietnam's active par- 3. China,themostpopulousmiddle-income ticipation in regional production-sharing country, which is transforming very rap- networks. FDI is also becoming significant, idly into a world economic power; and with more than half going to areas around 4. Republic of Korea, unique in moving Ho Chi Minh City in the southeast and from low- to high-income status in less another quarter going to Hanoi (and the than 50 years. surrounding Red River delta). At this stage of income,Vietnam's quality of infrastructure is still relatively underde- Vietnam and Lao PDR: veloped. Reducing distances through invest- emerging spatial patterns ment in transport and communications is at low income levels thus receiving high priority. Areas that are Much of Vietnam's economic history since along the coast or are more accessible to World War II is a tale of internal conflicts, international markets are being favored, as wars, and the legacies of central planning. is the expansion of the internal road net- By the mid-1980s, a very poor and largely work connecting the north and the south. agrarian-based economy was impatient As highlighted in Son's chapter, the highest for change. The market-oriented reforms priority at this stage is to strengthen access initiated in the mid-1980s revitalized agri- to resources and services. Thus he recom- culture and gave a new emphasis to indus- mends improving connectivity by elimi- trialization. GDP growth took off, averaging nating barriers and providing the neces- 7 percent a year over 1980­2006. As Son sary infrastructure to link rural with urban describes in chapter 7, a significant portion areas, labor-supplying regions with labor- of this success was due to an increase in agri- demanding ones, and national with interna- cultural productivity, but the largest share tional markets. Internal divisions in terms came from growth in services and industry. of provincial boundaries or restrictions on These outcomes indicate the increasing labor mobility are not seen as significant role being played by agglomeration effects barriers to the agglomeration process. How- associated with rising density in economic ever,Vietnam's geographic contours, specifi- activities. In Vietnam, this is illustrated by cally the concentration of the poor in the the gradual rise of the urbanization rate, mountainous areas in the northeast, north- which started from a low of 19 percent in west, and central highlands (many of whom 1980, began to increase in the 1990s, and are ethnic minorities), pose significant is now at about 27 percent. This is also challenges to developing workable solutions confirmed by the share of the labor force to reduce disparities in living standards in agriculture, which accounted for about across space. two-thirds of the total labor force up to The concentration of higher-productiv- 1990 but since then has declined to about ity activities in urban areas, in combina- 50 percent.Increasing density has been facil- tion with the problems of remote lagging itated by large-scale labor migration from regions with poor resource endowments, is rural areas, with the bulk of the movement becoming evident in the measures of social going to the southeast region, drawn by the and income disparities. These disparities growing commercial and industrial activity are widening between the urban and the around Ho Chi Minh City. This pattern is rural population, between the plains and 342 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA the mountainous areas, and among ethnic capitals: Jakarta, Manila, and Bangkok. groups. However, by international standards, And all three show wide spatial variations inequality in Vietnam is still relatively low, in the level of economic activity, resource and budget allocations for social programs endowments, and population settlements, have a significant redistributive impact.Given including the presence of ethnic minori- the experience of other countries, disparities ties. Indonesia and the Philippines are the may increase before eventually moderating. two largest archipelagic economies, and this Lao PDR shares some of the same his- appears to be a major factor in how their tory as Vietnam: suffering from conflict and geography has been influencing economic dealing with a legacy of centrally planned trends. For Thailand, the mountainous bor- policies. In chapter 3, Ohno cites how a low- ders that it shares with other countries and income land-locked country such as Lao the diversity of its local communities may PDR can deal with distance and divisions, have played a similar role. illustrating the case of remote villages that For all three countries, a strong urbaniza- are dependent on traditional hand-weaving tion process has been under way, contribut- activities.Although clustered in proximity to ing to rising density, more so in the Philip- each other, weavers may not have products pines and Indonesia than in Thailand. The that are specialized enough to find a niche in extent to which this has fostered agglomera- global markets. This presents a formidable tion benefits is less clear, perhaps due to the barrier to facilitating spatial connectivity. To lack of disaggregated regional data on the achieve agglomeration benefits from density, location and nature of economic activity these villages must deal with their isolation. needed to assess such relationships. Ohno documents the very high transaction costs for these isolated producers, which Indonesia essentially condemn them to low returns. In Indonesia, economic agglomeration is At this stage of development, he argues hampered by the geographic setting of the that "culture brokers" who can both inspire country. With its 13,000 islands, Indonesia informal, trust-based mechanisms at the is the world's largest archipelagic state and community level and assess the demands in one of the most spatially diverse nations. external markets can provide the necessary In the last two decades, there has been no connectivity to narrow the vast distances significant change in the concentration of involved. Reducing distances in this case is economic activity across the major island not just an issue of developing transport groupings. Economic activity has continued links, which may not be financially feasible to cluster around some key regional econo- in such circumstances, but also an issue of mies. With only 6 percent of the land area, linking isolated clusters of rural producers Java has remained dominant, accounting for to the preferences of more sophisticated 60 percent of the population and 52 percent buyers overseas. of GDP. Over the course of several decades, the Indonesia,the Philippines, group of top-performing regions has been quite diverse, as to location, size, and socio- and Thailand: diverse settings economic characteristics (as described by and varied outcomes in three Hill, Resosudarmo, and Vidyattama in chap- middle-income ASEAN countries ter 8). There has been no significant shift As relatively large and geographically diver- in the concentration of economic activity sified countries, the ASEAN-3--Indonesia, across the major island groupings, although the Philippines, and Thailand--share a how the mining sector is treated affects number of characteristics as middle-income results. In general, the poorest regions have economies. Over the past quarter century, performed about as well as the national their average GDP growth rates have been average. The better-performing regions are in the range of 3­6 percent, with Philippines typically those that are the most connected at the lower end and Thailand at the upper. to the global economy. Jakarta stands out All three economies are dominated by their as a special case, growing richer than the Lessons from experience: reshaping economic geography in East Asia 343 rest of the country over time and account- industries, but less evident in textiles and ing for the clear shift in economic activity chemicals, because their externalities are toward Java. Although two of the strongest dynamic (firms are less willing to move performers are resource-rich regions, the to locations without a prior history of the performance of this group of provinces has industry and hence no accumulated stock varied considerably, and there is no clear of knowledge). If externalities are localized, natural resource story. Conflict is particu- smaller cities are more likely to specialize larly harmful to economic development, as in just one industry or in closely connected illustrated in the case of Maluku since 1997 industries. However, if the externalities hap- and to a lesser extent Aceh. pen to be urban in nature, an industry will Income disparities using national indica- have to find a location in a diverse, large tors have increased steadily, but not substan- urban environment. Kuncoro suggests that tially, over the past decade and a half, but government policies should not interfere the picture is complicated by a large increase with the decisions formed by private incen- up to 1999 and a gradual decline since tives but that the most important interven- then. However, at the subnational level, the tion is to improve the quality of roads and growth pattern is more mixed, with dispari- cut travel costs between factory sites and ties among provinces declining steadily over markets or ports. Thus a policy of (a) mix- time and disparities among districts within ing infrastructure development in lagging provinces being more varied (see McCulloch regions to reduce distance and (b) using pri- and Sjahrir 2008).Analysis of growth trends vate investment incentives to reconcentrate shows some evidence of convergence, with industries (density) in smaller cities in lag- poorer districts growing faster than better- ging regions could meet the twin objectives off ones, but this may have been the result of efficiency and more balanced growth. of the financial crisis affecting the richer At the national level,especially for smaller districts relatively more than a structural cities and localities adjacent to major cities, change in growth dynamics. However, Hill, connectivity to more dynamic commercial Resosudarmo, and Vidyattama conclude areas is the most likely path out of poverty. that disparities are either declining or stable, However, improved spatial connectivity that depending on the production series used, reduces distances can also affect local activi- and that convergence either was not taking ties in rural areas. In chapter 4, Yamauchi, place or was relatively weak, depending on Muto, Dewina, and Sumaryanto explore the indicators chosen. this issue, examining how investments in Looking at firm-level production trends local roads can affect allocative efficiency for Indonesian manufacturing, in chapter 10 in Indonesia. Going beyond other studies on Kuncoro describes how density of industrial the distance factor, they assess the interac- activity evolved as the economy developed. tions involving choices among investments Liberalization of the economy brought in "local" roads connecting villages with economic activities to concentrate in a few "trunk"roads that lead to economic centers. places, but also fostered unintended nega- Their study also examines how investment tive externalities associated with agglomera- in household education affects such deci- tion. By improving roads in rural areas, the sions. The results show that investments to government made it possible for firms to improve the quality of local roads have a pos- reconcentrate in smaller, less expensive cit- itive impact on income growth as well as the ies,including those in low-income or lagging transition to nonagricultural activities. The regions in Java. Based on empirical exer- extent depends on the distance to economic cises conducted on Indonesia's four most centers and the degree of post-primary important industries, Kuncoro finds that household education. Thus investments in the nature of externalities and agglomera- education and in local infrastructure are tions favored industrial spillovers--that is, complementary in promoting growth and localization was stronger than urbanization are not competing choices. effects. The deconcentration process is evi- Given the diversity of needs across the dent in nonmetallic minerals and machinery major island groupings, the Indonesian 344 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA government launched an ambitious decen- decade, but questions linger regarding its tralization initiative in 2001, as analyzed by sustainability. Economic activity has been Arze del Granado in chapter 9.This initiative highly uneven and concentrated particu- could potentially have a major impact on the larly in Metro Manila. Together with the two pattern of urbanization, industrial agglom- adjacent regions, Metro Manila produces eration, and welfare objectives. His findings about 55 percent of the country's GDP. suggest that more-urbanized areas tend to Thus building more density is not seen as attract migration from less-urbanized areas a desirable objective. Instead, the pattern of but that, where urbanization is the most uneven development, coupled with a disap- advanced, congestion costs tend to push pointing performance in poverty reduction, the inhabitants to neighboring areas. Thus has focused the policy debate on whether a the forces that promote concentration tend more balanced approach is needed regard- to go only so far before reversing. In addi- ing the allocation of resources to address tion, the degree of public expenditures on disparities (as discussed by Balisacan, Hill, social services and infrastructure influences and Piza in chapter 11). Inequality in the the pattern of migration. The higher these Philippines is seen as a more serious prob- expenditure levels, the lower the outflows lem than in other East Asian countries, with of population toward larger districts--that inequality due largely to disparities within is, residents will move to districts with bet- regions rather than among regions. Relative ter services or employment opportunities. income differentials also drive migration, Employment growth is inversely related to with the bulk of movement going to the the distance between a district and its near- region around and including Manila and, to est higher-order urban center, and thus a lesser extent, a few other prosperous areas investments that reduce distance matter. But such as Cebu City. the impact of decentralization on industrial In recent years some regions have done expansion is more ambiguous. There is no quite well in attaining high per capita evidence that expenditures on social services income growth and poverty reduction,while will encourage more industrial concentra- others have experienced a decline in their tion in that area. This suggests that natural average per capita income and an increase advantages and production externalities in poverty. On average, most of the poor affect firm-level decisions more than the regions grew more slowly than the national impact of government expenditures. Nor average from 1985 to 2003, and there is no are special tax incentives to attract indus- firm evidence that incomes are converging tries likely to provide sustainable solutions. across provinces.3 Empirical tests on the importance of local- The poor performance in economic ization and urbanization externalities are growth and poverty reduction is related to inconclusive. This may be due to a lack of the large disparities in access to infrastruc- more disaggregated data, but it could also ture and social services across regions and mean that the full impact of agglomeration island groups and between urban and rural effects has not yet been felt in Indonesia; that areas. A widely held view is that develop- is, at this stage of development, the structure ment efforts have favored Luzon, particu- of industries may not have evolved enough larly the national capital region (Metro to secure all of the benefits that can accrue Manila), and discriminated against the to specialization. Visayas and, especially, Mindanao. This development pattern is seen as having led The Philippines to substantial spatial differences in access The Philippines, with more than 7,000 to economic opportunities, in rates of islands, is the second-largest archipelagic poverty reduction, and in the incidence of state in the world. For most of the past armed conflict. Similar to Indonesia, it is 25 years, economic growth barely exceeded also worth pondering whether the divisions the population growth rate, which con- inherent in being a nation of islands create tinued to expand rapidly at 2.3 percent a the same kinds of inefficiencies that result year. Growth has quickened in the present from international borders.4 Lessons from experience: reshaping economic geography in East Asia 345 The government's allocation of infra- 1980s, during which the Thai government structure funds has had implications for promoted industrialization and shifted regional development patterns. Following the policy emphasis from import substi- the dismantling of the old import substitu- tution to export promotion. Supported tion growth regime, the new driver of spa- by major infrastructure investments, the tial development patterns has been decisions manufacturing sector grew rapidly from regarding the location of export zones. In 27 to 38 percent of GDP from 1980 to 2005. the last two decades, the Philippine govern- The share of agriculture fell from 23 to ment (and donors) has been more inclined 13 percent over this period. Services remain to invest in internationally oriented infra- the largest sector, contributing half of the structure (ports, harbors, and associated country's GDP in 2005. Overall, GDP grew facilities) than in domestic transport net- at a rate of 6 percent a year for more than works and corridors. The effect has been 25 years, despite the impact of the Asian to reinforce the internationally connected financial crisis. As a consequence, the pro- enclaves at the expense of a denser set of portion of people living below the poverty domestic connections, a factor exacerbated line declined from 38 percent in 1990 to less by the regulatory barriers erected between than 10 percent in 2007. firms inside and outside the export zones. As discussed by Wisaweisuan in chapter Balisacan, Hill, and Piza stress that spa- 12, despite this impressive growth and pov- tial disparities need not reduce growth if erty performance, there are strong feelings they arise from efficiencies associated with about the extent of spatial disparities and an agglomeration. If so, the strategy to prevent ongoing debate about the role that market unreasonable spatial disparities during the forces or policies play in shaping outcomes. development process would be to improve Special emphasis is placed on urban-rural market links between leading and lagging as well as regional differences, with a par- regions through greater factor mobility, ticular attempt to determine whether the particularly of labor. This would need to growth-promoting policies as well as pub- be supported by improved social services lic finances have alleviated or aggravated that, given the fiscal constraints, need to be spatial disparities in Thailand. There is no targeted more efficiently. Infrastructure, clear evidence that spatial disparities have however, is seen as the glue that unifies become better or worse at the national lev- the national economy and the single most el--the Gini and Theil indexes show little important instrument of regional policy. change and possibly some improvement-- But where the infrastructure should be but perceptions persist about increasing located can have ramifications in balanc- disparities across regions and between rural ing the desirability of fostering links with and urban areas as well as within sectors international markets or strengthening ties and regions. with the lagging hinterland. Overall, the Over the past decade and a half, growth Philippines is seen as being deficient in in regional GDP of Bangkok has been the quality of its infrastructure, especially slightly lower than the national average, in its road network, and this has held back and some areas like the eastern and central efficient regional economic integration.5 regions have been doing much better. These The consequence is that the country has regions have benefited from a more active made little progress in national market industrialization process brought on by integration, as evidenced by the widening globalization and designation of some spe- variations in regional prices over the past cial export-processing zones. This has made decade and half. these regions a magnet for labor migration from other lagging areas. Thailand has Thailand benefited from the globalization process In Thailand, growth in density or economic more than Indonesia and the Philippines, concentration and poverty reduction went as indicated by the increase in the ratio hand in hand. The structure of the Thai of trade to GDP, which rose from 90 per- economy began to change in the early cent in 1990 to nearly 150 percent in 2005. 346 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA However, the consequence may have been by Deng Xiaoping's signal not just to allow an inevitable deterioration in income dis- but to encourage industrial expansion in the tribution; the poorest regions have tended major commercial centers along the coast. to grow slower than the national aver- The most notable policy manifestation was age, leading to a sense that disparities are the establishment of special economic zones increasing. Those regions, especially the (SEZs) in four cities along the southern coast, south, are highly dependent on agriculture, which then spread to the other coastal areas which makes it difficult for them to catch andeventuallytoallthecapitals(ChenandLu up. This suggests that Thailand is still in the in chapter 15). These SEZs laid the basis for stage of development where agglomeration a massive inflow of FDI, which even today is economies are strong and structural shifts concentrated largely in the coastal provinces. in production may still lead to increasing These incentive policies were complemented regional disparities. by fiscal reforms that gave preferential treat- ment to the coastal provinces of Fujian and Guangdong and ensured that these areas China: agglomeration,rapid would receive an increasing share of public growth,and major spatial investment relative to the inner provinces. consequences With the requisite resources and supportive China, the most populous and arguably the incentive policies, these coastal areas were most geographically diverse country in the then well positioned to benefit from the glo- world, reshaped its economic geography to balization process that has characterized East reverse half a century of economic decline. Asia's recent successes. History tells us that recent successes repre- Rising density brought forth agglom- sent a marked shift from failed policies in the eration economies, which uplifted the pro- pre-1980 "centralized planning era," which ductivity of Chinese industrial enterprises supported "balanced" growth and even the concentrated in the major commercial cen- spread of industrial capacity across regions ters along the coast. Although the coastal (as discussed by Yao in chapter 14). Over areas grew much faster than the inland much of this period, China shifted back and provinces, the inland provinces also per- forth from centralized and decentralized fis- formed much better than before; together cal policies in response to major economic with sustained growth in agriculture, this and political events, with the consequence resulted in near double-digit GDP growth that, by 1980, China was a very poor but for more than a quarter of a century. egalitarian society, with one of the lowest The pattern of industrial growth in this Gini coefficients in the world. process is broadly consistent with the tenets During the post-1980 reform era, prefer- of the new economic geography. According ential policies were sequenced to work with, to He in chapter 16, as market integration rather than against, differences in natural deepened, provincial industrial structures endowments and comparative advantages; initially became more diversified, but by the in the process, they fundamentally reshaped late 1990s the agglomeration process began to China's economic geography. These poli- take hold, with the effect of nurturing more cies had two distinct consequences: (a) they specialization. The relationship between reshaped the spatial dimensions of develop- regional specialization and per capita GDP is ment by increasing the density of industrial U shaped. Driven by market forces, both the activity and ratcheting up the urbaniza- very poor and very rich regions are now more tion process, and (b) in reducing distances specialized,with the more service-oriented or and breaking down divisions, they broke higher-technology industries concentrated in the gridlock in the mobility of factors and the coastal urban areas and the more heavily goods both internally between provinces protected and resource-intensive industries and urban and rural areas and externally remaining relatively more dispersed within between China and the rest of the world. provincial boundaries. Industries with less How was density encouraged? According local protection and government interven- to Yao, leadership was critical as exemplified tion typically are more exposed to external Lessons from experience: reshaping economic geography in East Asia 347 competition. Eventually, as agglomeration major commercial centers along the coast effects have taken hold, these industries effectively reduced distances, and, on the have become more regionally specialized as other, China embarked on an infrastructure- well as more productive. Over time, returns led investment program that was unprec- to capital across provinces have converged, edented in its scale. While labor migration although returns remain greater in the coast arose spontaneously, China's success in than in the west. This lends support to the dealing with the distance factor came from a view that the agglomeration process has not concerted strategy to improve connectivity become wasteful over time; rather, market through investments in transport and com- forces have been encouraging the more effi- munications. Initially, priority was given to cient allocation of resources (Bai and Lin in the coastal provinces, but starting in the late chapter 17). 1990s, the central and western regions began China's urbanization process was both to receive increasing attention. Over the past a consequence of and a factor contribut- decade, considering both highway and rail- ing to the agglomeration process (Yeung way investments, China spent more than 5 and Shen in chapter 18). Rapid urbaniza- percent of GDP annually on transport invest- tion and specialization drew their impetus ments amounting to more than US$100 bil- from the coastal areas. China's urbaniza- lion in 2006, which is roughly twice as much tion rates, which rose from around 25 per- as in other comparable countries. cent two decades ago to nearly 50 percent Reducing distances would not have had today, accelerated with the introduction of such a positive impact,given China's size and township-and-village enterprises, which decentralized administrative system, with- drew workers out of farm production and out major efforts to eliminate internal divi- inspired large-scale migration. Over time, sions. During much of the reform period, this spawned a huge cohort of migrant China was better integrated externally than workers,6 which now totals an estimated 140 internally. This was due largely to the pri- million to 150 million, heavily concentrated ority given to trade and globalization. Prior in the major commercial centers along the to 1990, internal divisions in the form of coast. As reforms deepened, administrative provincial boundaries essentially made each constraints on labor mobility were progres- province a separate fiefdom, effectively lock- sively alleviated. Although access to housing ing in resources (labor and capital), creating and social services remained unequal for protective markets for producers, and lim- migrants without urban residency status iting the choices for consumers. Empirical (hukou), they became less prohibitive over studies for that period showed widely diver- time. The main lesson here is that the forces gent availability of goods and huge regional of agglomeration are very powerful: despite price differentials. Subsequent price liberal- fairly stringent barriers discouraging migra- ization reforms and reductions in provincial tion, many have, in fact, chosen to move. barriers governing the movement of goods Over time, national market integra- and factors of production have done much tion facilitated labor and capital mobility to bring down internal divisions, but more across provinces and greater concentration work is needed to create a unified national of resources. This process allowed many market (Huang and Luo in chapter 13). Chinese companies to exploit economies Despite the surge in per capita incomes of scale (Chen and Lu in chapter 15; He in over the past several decades, spatial indica- chapter 16) As the coastal cities became more tors measuring income inequality--between linked to the global economy, the benefits coastal and inland regions and between became obvious, as exemplified by rapid rural and urban areas--deteriorated. But employment creation, competitive pressures China shows that there are"good"and"bad" on enterprises to restructure, and a much manifestations of inequality. The dynamics improved domestic and external financial of spatial disparities across subnational areas position (Yeung and Shen in chapter 18). have taken the form of a "race to the top." How did China reduce distances? On the Disparities are not the result of stagnant one hand, massive migration of labor to the income growth among certain segments 348 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA of society or regions but rather the con- between urban and rural areas. However, sequence of unusually high and sustained the gap in social welfare achievements by growth of coastal and urban areas fostered region, especially for health compared with by agglomeration economies. But growth in education, while narrowing over time, the inland provinces was not low by global remains large, as indicated in government standards, averaging around 8 percent policy statements. annually. Yet this paled in comparison with The principal instrument with which growth in the coastal provinces, which aver- to address spatial disparities in living stan- aged an impressive 12 percent during this dards is to give more priority to the poorer decade (Huang and Luo in chapter 13). inland provinces in the provision of social Although regional disparity is widely and infrastructure services through the considered as the key determinant in China, public expenditure program. China's capac- the rural-urban divide is the more important ity to use public finances for redistribution factor in shaping overall inequality. Changes purposes, characterized by distinct inter- in the Gini, which rose from 30 to 45 over governmental layers of responsibility for the the past 25 years, are closely associated with collection of revenue and the provision of changes in the urban-to-rural income ratio services, has been hindered by several fac- and the coastal-to-inland per capita GDP tors. The most critical has been the steady ratio. The challenge in reducing income decline in the ratio of revenues to expendi- inequality, however, is that as impressive as tures, from more than 25 percent in 1980 to growth in rural incomes has been (4 percent 12 percent in 1995.7 With this near collapse in annually for several decades), it is still much revenues, the country's capacity to redistrib- lower than growth in urban incomes, which ute resources through the budget in favor of for some periods was twice as high. Thus the poorer regions was limited. Beginning in the ratio of urban to rural incomes has been the mid-1990s,an improved revenue position rising steadily since the mid-1980s. Because led to a shift in the share of resources going urban-rural differences are greater in the to the inland provinces. While the trend in poorer inland provinces than along the recent years has been in the right direction, coast, this drives regional differences, espe- most observers feel that more needs to be cially because urbanization rates are about done (Huang and Luo in chapter 13; Yao in 65 percent higher along the coast than in the chapter 14; Chen and Lu in chapter 15). western region. Aside from shifts in the public expen- Within provinces, inequality within rural diture program designed to ameliorate and urban areas has accounted for a larger regional differences in social welfare, share of total inequality over time. With the incentives were also geared to provide a structural transformation of the economy, more balanced approach in development individual circumstances have gradually objectives. These concerns underpin several played a more important role in determin- regionally targeted development programs, ing income, including the high premium the first of which was focused on develop- accorded to education. This suggests that ing the western regions in the late 1990s,later increasing income inequality is, to some the northeast, and more recently the central extent, a consequence of the stage of China's region (Yao in chapter 14). What character- development: the growth process unleashed izes these targeted initiatives is the pragmatic competitive pressures and created greater approach to crafting strategies that reflect rewards for skilled workers. geographically differentiated needs as well as But if not addressed, rising dispari- political pressures. For the western region, ties--whatever their source--can lead government policies focused on developing to social and political pressures. China's more sustainable agricultural production policy makers have been sensitive to these systems that would preserve natural forests concerns. At the national level, this aware- and invest in irrigation systems suitable for ness is reflected in the recognition given dry land farming. These were combined to achieving more uniform and equi- with stepped-up efforts to facilitate labor table social outcomes across regions and migration by improving access to social Lessons from experience: reshaping economic geography in East Asia 349 services. These policies were complemented ter 19 as "compressed economic growth." by a ratcheting up in funding for infrastruc- This approach has concentrated industry ture--roads and rails--that improved con- and population in the capital region and nectivity to the rest of the country. For the achieved a major spatial transformation. northeast, the major challenge was to deal This process was broadly characterized with an outdated industrial structure based by the building of density to achieve the on state-owned heavy industries and under- associated agglomeration effects. The share lying geographic conditions (low popula- of industry in production increased from tion density and inappropriate positioning 19 percent in 1960 to 41 percent by 1980 and to tap national and global markets). The has since stabilized around that level. Since strategy that has emerged encompasses two 1980, the changes lie more in the nature of general concerns. One is to strengthen social the industrial structure, with a steady decline protection systems to deal with unemploy- in the share of labor-intensive industries in ment, pension, and retraining needs, includ- favor of the production of more technology- ing facilitating labor migration as needed. intensive parts and components for assem- The other takes more of a spatial perspec- bly and capital-intensive goods. In the early tive by looking at the three affected prov- stages of industrialization, there was a bipo- inces in a broader regional context to see lar concentration of industries separating the how existing industrial activities could be capital region from the rest of the country. reshaped to complement national and glob- Over time, this disparity was considerably ally linked production-sharing networks. reduced by forces that linked provincial cit- This approach includes recognition that the ies and rural areas with the more dynamic northeast is well placed to connect China centers and by the rapid dissemination of with external markets in Japan and Korea. information and communications technol- The strategy also includes recognition that ogy (ICT) services. This shift is mirrored by a some traditional agricultural areas could steady decline in the role of agriculture: from prosper with more support to realize their 40 percent of GDP in 1960 to 3 percent today. resource-based advantages. Work on strate- However, it is also worth noting that indus- gies for the central region is just beginning. trial restructuring in response to globalization As future growth dynamics are likely to be pressures was not an easy process; it required driven more by domestic consumption major shifts in the nature and location of than by exports, the region has significant activities; for some product lines, it entailed locational advantages as a centralized hub relocation to lower-cost centers abroad, such serving the surrounding, heavily populated as China. This process is now being nurtured provinces. This should provide a rationale by a persistent pressure to build new innova- for investments designed to improve con- tion systems (Park in chapter 19). nectivity with the surrounding regions. These trends are reflected in the sharp increase in the urbanization rate, which rose from 27 percent in 1960 to 57 percent Korea: from developing to by 1980 and to 81 percent by 2006. In the developed status and eventual process, the economic geography of Korea equalization in living standards changed dramatically. At the beginning, Much has been written about the Korean when agriculture was the major activity, the success story. The Republic of Korea is western part of Korea was relatively affluent unique in having gone from an income due to the fertility in the plains. However, of less than US$100 per capita in 1960 to the rapid industrial expansion in the capital US$20,000 in 2007 in the aftermath of the and southeastern region, supported by the devastating Korean War. Such a remarkable Seoul-Busan highway, created a disparity achievement in such a short time (by the between the axis formed by Seoul and Busan standards of economic history) is the con- relative to the axis formed by the southwest- sequence of an impressive export-oriented ern and northeastern corners of Korea. and sector-specific industrial development Seoul continues to play a dominant strategy, which Park characterizes in chap- role in the economy and continues to be 350 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA a magnet for attracting skilled labor and been remarkable in terms of its equity impli- resources; however, its share of the popula- cations (table 20.1). tion peaked more than a decade ago. Today, the migration to the capital region flows Lessons learned mainly to Seoul's surrounding areas, such The WDR 2009 concludes that spatial trans- as Incheon and Kyunggi, nurtured by the formations--rising density, falling distance, spillover of economic activities, due to the and dissipating divisions--will remain attractiveness of lower-cost settings. essential prerequisites for economic success These industrial spillover effects are now in the foreseeable future.8 Do these same being felt in other regions,notably the south- three principles apply to East Asia? east, which has benefited from being highly As argued in the WDR 2009, economic specialized in assembly-related production growth is seldom balanced, and efforts to and its position as a major port connecting spread it out prematurely may jeopardize key markets in China and Japan. The middle sustainable progress. Centuries of economic region has also benefited in taking up a large development in other parts of the world share of high-technology-intensive indus- show that spatial disparities in income and tries from the capital region. The southwest production are inevitable.However,as coun- region is more specialized in resource-based tries develop, a series of policies are enacted industries. Over time, the structure of indus- to make basic living standards more even trial organizations has also evolved, with the across space. concentration of headquarters and research The East Asian experience conforms to activities of major firms in Seoul but the these stylized facts. Plants have become big decentralization of production functions to exploit economies of scale, but places do to non-capital regions. In the earlier stages not have to be big to generate those econo- of development, the distance factor largely mies.The function of cities matters,not their related to improving connectivity through size. The medium-size cities are often large improved transport networks but, as Park enough for "localization" economies. As a stresses in chapter 19,the current challenge is city grows, localization economies become to develop ICT to improve the operations of less important, giving way to"urbanization" networks, as firms fragment and locate func- economies,which tend to generate,especially tions across different regions and increasingly in large cities, knowledge spillovers. Human rely on the transferability of services to be capital moves to where it is abundant, not competitive in a knowledge-based economy. scant. Falling transport costs make special- The consequence of these shifts in indus- ization possible and increase trade not only trial specialization and location over the past with neighboring countries but also across several decades has been a marked decline internal provinces. The move to density is in disparities due to the natural spreading quick at the local level, manifested in a rapid out of economic activities supported by rural-urban migration that accompanies the major infrastructure investments, improved sectoral shift from agriculture to industry. ICT, and stronger social services, notably Over time, this leaves an uneven landscape, education. Particularly striking is that the with people and production concentrated gap between per capita GDP for the capital in some places and not in others. Migra- region, which is the highest nationally, and tion, trade, taxes, and transfers influence that for the lowest region--the southwest-- the pace of convergence. As incomes rise, dropped from about 40 percent in 1985 to living standards converge between places about 10 percent by the late 1990s (Park in where economic mass has concentrated and chapter 19). More disaggregated data for places where it has not. The challenge for subnational administrative units indicate governments is to allow--even encourage-- that per capita GDP in the poorer areas is "unbalanced" economic growth and yet to typically no more than 20­25 percent below ensure "balanced" development in terms the national average. Other indicators such of quality of life and consumption levels. as the Theil coefficients also illustrate that As suggested by the WDR 2009, East Asia the Korean development experience has can do this through economic integration: Lessons from experience: reshaping economic geography in East Asia 351 by building density, reducing distance, and and those areas within countries that have eliminating divisions. been participating more actively in this pro- cess also have benefited the most in terms of Density and scale economies growth and employment generation. The studies in this volume show that, as a Broadly speaking, rising density allows a country grows richer,location becomes more country to realize the externalities of bring- important for economic production. Put ing producers together and linking produc- another way,as countries develop,production tion to dense consumer markets. Economic becomes more concentrated spatially, and development thus brings with it the condi- location matters less for families but more tions of even greater prosperity, creating a for firms. In the early stages of development, virtuous circle. Neighborhoods also matter. such as in Vietnam, this means a gradual A prosperous city like Seoul or Shanghai sel- movement of labor out of agriculture and dom leaves its periphery mired in poverty. into industry and services and the beginnings A province's prosperity is sooner or later of migration to the major production cen- shared with those nearby. Thus the evidence ters such as Ho Chi Minh City and along the shows that, over time, spillover effects occur, coast. The ASEAN-3 exemplify an agglom- nurtured in part by the higher congestion eration process that is well advanced, as indi- costs of large cities. cated by the dominance of their three major How these clusters form and whether capitals, which account for as much as half they become innovative enough to spur of national GDP. The same process has now growth can determine why countries pros- taken place in China,but over a much shorter per or lag behind, as Otsuka and Sonobe time horizon in terms of the concentration of discuss in chapter 2. They point out that activity along the coastal provinces. the formation of industrial clusters is As illustrated for Indonesia (Kuncoro vital for information spillovers and low- in chapter 10), China (He in chapter 16), ers transaction costs among enterprises and Korea (Park in chapter 19), as density and traders. As innovation possibilities are rises and agglomeration economies become enlarged, sustainable growth of industries more dominant, industrial structures evolve. becomes possible. Their case studies show Regions, and areas within provinces, show that such clusters have spurred indus- a distinct pattern in the structure of their trial development in a number of East industries, moving from a more diversified Asian countries, such as China, Japan, the base at lower income levels to a more spe- Philippines, and Vietnam. cialized structure at more advanced stages, Overall, the impact of density is illus- given the interplay of localization and trated by rising urbanization, which cor- urbanization externalities. Cities eventually relates well but not strictly with a coun- become more specialized, with some focus- try's stage of development and pattern ing on capital-intensive production, oth- of growth. For low-income countries ers focusing on more technology-intensive like Lao PDR and Vietnam, urbanization production, and yet others shaped by the rates cluster around 20­25 percent. The availability of resources and the proximity rate rises to around 40­60 percent for the to consumer markets. In the process there middle-income countries in ASEAN and is a role for both large diversified cities and China and then to as high as 80 percent smaller specialized cities. for Korea. Thus a distinguishing aspect of As discussed earlier, agglomeration econ- the East Asian success story and the role of omies, exemplified by the concentration economic geography is the rapid pace of of production along coastal areas and in urbanization. the major capitals of East Asia, have been nurtured by the forces of globalization, as Distance and factor mobility indicated by the rapid growth in trade of Location is the most important correlate of intermediate goods and the development a person's welfare, and choice of location-- of regional production-sharing networks proximity to markets and other producers-- (Hamaguchi in chapter 1). Those countries will often determine whether a firm succeeds 352 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA or fails. Reducing distances has been funda- chapter 4) and essential for low-income, mental to East Asia's economic success over landlocked countries (as indicated by Ohno the past several decades. The impact has in chapter 3 on Lao PDR). been felt at three levels: regional, national, and local. At the regional level, declining Divisions transport costs and improved logistics gave How significant are the divisions that sepa- rise to the production-sharing networks that rate nations for reducing economic effi- underpin the region's explosive growth in ciency? The WDR 2009 elaborates on these the trade in parts and components (Hama- issues: differing incentive regimes in the guchi in chapter 1). This has made possible form of divergent tariffs, tax, and investment the attainment of economies of scale even systems often distort the spatial location of for small enterprises and facilitated indus- firms and their access to markets. These trial expansion across a broader range of distortions lie behind the global efforts to technology-rich products not only in China achieve more integration through bilat- but also in the middle-income economies of eral and multilateral trade agreements and Southeast Asia and now even in low-income more comprehensive arrangements such as economies like Vietnam. the European Community and ASEAN. In At the national level, improved trans- this volume, however, the focus is largely on port links have influenced the location of internal divisions such as provincial bound- firms and promoted more diversification aries and the barriers that differing ethnic, and specialization in industrial structures. religious, and geographic circumstances As Huang and Luo discuss in chapter 13, create. China's impressive performance in expand- For the large and more diversified ing trade and restructuring its industries countries such as China, Indonesia, the was greatly facilitated by a massive invest- Philippines, and Thailand, internal admin- ment program that upgraded its transport istrative boundaries (for example, provincial network. This partly explains why China is and district borders) can matter as much ranked so highly in terms of its competi- as the existence of mountains, deserts, and tiveness. In contrast, as Balisacan, Hill, and islands. Internal administrative boundaries Piza note in chapter 11, the disappointing can lead to the same kinds of inefficiencies economic performance of the Philippines that occur with international borders, if may be partly due to its very low rankings in they lead to differential access to resources terms of the quality of its infrastructure. or impede the flow of factors of produc- Over time, the reduction of distance tion. Geographic formations can have a will reshape the industrial composition of similar effect in terms of discouraging fac- production and its location in major cit- tor mobility and creating protective barriers ies, as industries that are less dependent for local industries. As described in previ- on "national" versus "local" agglomeration ous chapters, these situations appear to have economies move to lower-cost centers. Also, played a significant role in shaping develop- industries relocate to gain better access to ments in countries like China, Indonesia, external markets or natural resources. This the Philippines, and Thailand. The case of pattern is especially true for China (He in large archipelagic nations like Indonesia chapter 16; Yueng and Shen in chapter 18), and the Philippines may present special but it is also exemplified by the experiences challenges, encouraging more self-sufficient of Indonesia, the Philippines, Thailand, and industrial structures for each of the major Korea (Hill, Resosudarmo, and Vidyattama islands, but in the process making it more in chapter 8; Kuncoro in chapter 10; Balisa- difficult to achieve agglomeration economies can, Hill, and Piza in chapter 11; Wisawei- and the benefits of specialization. Rigidi- suan in chapter 12; Park in chapter 19). Yet ties in the movement of labor can also be a the impact of reducing distances is not just problem, as illustrated by the unique hukou a force that shapes industrial expansion; it system in China. The solution lies partly in can also be vital at the local level in link- eliminating artificial barriers to the move- ing rural markets (Yamauchi and others in ment of goods and factors of production and Lessons from experience: reshaping economic geography in East Asia 353 partly in improving connectivity through appropriate policies, more inclusive develop- infrastructure investments. ment will come, sooner rather than later. Cross-border divisions are even more However, as seen throughout all the important. Two special cases are addressed papers in this volume, governments are in this volume. Bhaskaran, in chapter 5, under enormous pressures to deal with analyzes the potential benefits that could be perceptions--real or imagined--about gained from greater cooperation and cross- increasing disparities, which in most cases border flows of resources between Malaysia have a spatial dimension. At the national and Singapore. Both countries have done level this is exemplified by indicators like well over the past several decades, but both the Gini and Theil coefficient (table 20.1) would benefit greatly from a policy of more and by wide spatial variations in welfare seamless collaboration. This would create indicators (see figure 20.1). Disparities a much larger and diversified market and increase most rapidly in the earlier stages greater economies of scale. Given their dif- of development, typically below per capita fering endowments of land, capital, and ser- incomes of US$3,500, when the forces lead- vice-related skills,both would be able to real- ing to concentration in production are the ize efficiencies and more rapid growth, while greatest (WDR 2009). By the time countries avoiding duplication of costly infrastructure reach the upper middle-income levels of and wasteful competitive investments. around US$10,000, they tend to moderate. However, while the economic benefits are Most disparities are in some sense obvious, the politics of collaboration may undesirable from a social perspective. If result in an insurmountable barrier. unchecked, they can lead to pressures that As Rigg and Wittayapak discuss in chap- ultimately could jeopardize otherwise suc- ter 6, the subregion with the most signifi- cessful development processes, but there are cant cross-border spillover of poverty is the so-called "good" and "bad" disparities. In a Greater Mekong subregion, which includes successful transformation from developing Cambodia, Lao PDR, Thailand, Vietnam, to developed status, inequality almost always and Yunnan province in China. Given the surfaces, as the population moves from major logistical challenges presented by lower-productivity agriculture to higher- the numerous hills and rivers and the pres- productivity urban activities. There is then ence of numerous ethnic communities, the a lengthy period of rising inequality overall, concerned governments have launched a as the share of higher-paid urban workers comprehensive program to promote col- increases relative to the less-well-off rural laboration and ensure that the development population. Ultimately, however, with labor of vital resources with cross-border conse- migration and industrial spillover effects to quences such as water and energy is in keep- ing with the interests of everyone. At the Figure 20.1 Provincial disparities: human development indexes in East Asia heart of this initiative is the effort to promote more connectivity in the form of a network 1 of highways and improved border logistics so that cross-border trade and related pro- 0.9 0.91 duction are encouraged. These benefits can 0.82 0.8 be substantial, but, as the authors point out, 0.78 0.76 0.74 0.75 there are also costs involving those whose 0.7 0.67 situation might be made worse by the 0.6 0.60 0.61 0.59 integration process. 0.58 0.55 0.5 0.49 0.50 What to do: the role of government policies 0.4 The key message of the WDR 2009 is that 0.3 Lao PDR Indonesia Mongolia Vietnam Philippines Thailand China policy makers should aim to build density, (2000) (2002) (2002) (2001) (2003) (2003) (2003) reduce distance, and eliminate divisions. The Source: Gill and Kharas (2007). consequence will be uneven growth,but,with Note: The connected line shows the highest and lowest national values of the human development index for each country. 354 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA peri-urban and rural areas, disparities begin Fiscal programs to decline.As documented in the WDR 2009, The evidence from both the developed this has been the experience of developed economies and the recent experiences in countries. Also, the more successful a coun- East Asia show that the most successful try is, the shorter is the compression of this countries institute policies that make basic transition, as illustrated by the case of Korea. living standards more uniform across space. If this is the nature of"good"disparities,then Because budgetary resources are invariably "bad" disparities emanate from situations constrained--and richer regions tend to be where growth stagnates for long periods better able to collect revenues--the degree to of time or policies tend to restrict access to which governments can redistribute in favor opportunities and resources to a select few. of poorer regions varies considerably across The solution is a more inclusive development countries and over time. Moreover, for the process, to reach a broader range of society. larger countries, decentralization and the For most countries, significant regional role that subnational layers of government or locational aspects characterize dispari- play can be instrumental in implementing ties. As discussed in nearly all of the chap- these objectives. ters, governments often feel the need to Throughout East Asia, fiscal decentral- give priority in the allocation of resources ization has gained momentum over the to rural areas or lagging regions. In some past decade. While different structural and cases, ethnicity is a factor. The special needs political imperatives propelled the process of relatively poor ethnic communities that in different countries (ranging from regime inhabit isolated and often mountainous changes in Indonesia and the Philippines, to areas are exemplified by the situation in the the transition to a market economy in China hills and mountains of Vietnam, Thailand, and Vietnam), the share of subnational gov- Lao PDR, and the western region of China ernment spending has risen to significant, and in communities spread among the many though varying, levels (see figure 20.2). islands in Indonesia and the Philippines. However, while fiscal decentralization What is the message, then, about dealing has progressed, subnational fiscal disparities with the pressures for more"balanced"social remain persistently large.There are significant outcomes given the benefits of an agglom- differences in revenue capacity across local eration process that contains the promise of governments, reflecting the underlying large more rapid growth? The solution is to pro- variations in their economic and resource mote economic integration while also nur- base.Vertical imbalances between subnational turing more inclusive social development in revenues and expenditures are sought to be the process. The market forces of agglom- filled through transfers from the central gov- eration, migration, and specialization can, ernment,but such transfers have not gone far if combined with progressive policies, yield in addressing horizontal inequality. Transfers both a concentration of economic activity from the central government reduce the dis- and a convergence in living standards. parities in per capita revenues, but often not Governments typically have a range of by much (see figure 20.3). As a consequence, instruments with which to achieve this there are large disparities in per capita local objective. These broadly fall into two cat- government spending across lower levels of egories: promoting connectivity by link- government in countries like Indonesia (Arze ing markets and improving factor mobility del Granado in chapter 9), the Philippines and ensuring that all families, regardless of (Balisacan,Hill,and Piza in chapter 11),Thai- their location, are provided with roughly land (Wisaweisuan in chapter 12), and China equal access to social and public services. (Huang and Luo in chapter 13). In these country studies, two instruments The potential to use fiscal transfers and stand out in terms of how governments social expenditures to reduce disparities have been dealing with spatial factors: (a) thus depends partly on a country's overall fiscal programs that support infrastructure fiscal position, the rules that determine how and provide social services and (b) spatially resources are cascaded down to lower levels, targeted programs. and often the politics involved. However, Lessons from experience: reshaping economic geography in East Asia 355 decentralization can also be misused as an Figure 20.2 Share of subnational government expenditure in total government expenditure in instrument, for example, if resources are East Asia during the 1990s directed to encourage industrial expansion in 80 areas that are not suitable (Arze del Granado 72.3 67.4 in chapter 9; Kuncoro in chapter 10). 70 In addressing disparities, the focus is 60 usually on influencing the pattern of social 50 48.0 expenditures, such as health, education, and social protection. As indicated in the WDR 40 2009, such policies should be "spatially 30 26.0 blind," in the sense that the objective is to 24.8 25.2 provide the same level of social services to 20 15.8 all, regardless of location or, taking it one 11.2 10 step further, to aim for similar social out- comes for all, regardless of location. This 0 1998 2001 1990 2002 1992 2002 1990 2004 may mean providing extra resources to serve hard-to-reach or high-cost areas. Moreover, Indonesia Philippines Vietnam China a spatially blind policy does not necessarily Source: Gill and Kharas (2007). mean that programs are similar in design for all regions. The reality may mean that, Figure 20.3 Coefficient of variation in provincial per in diverse spatial settings, differentiated capita revenues before and after transfers in select approaches may be necessary to realize East Asian countries similar outcomes. 1 0.98 0.96 Infrastructure investments, particularly 0.97 0.92 those relating to transport, are usually not 0.82 0.79 0.8 0.74 perceived as a vehicle for addressing dispari- ties, although they are seen as fundamental 0.6 for reducing distances, promoting density, 0.42 0.4 and bringing down divisional barriers. As such, infrastructure is the basis for promot- 0.2 ing national economic integration. Often overlooked, however, is that such expen- 0 China Indonesia Philippines Vietnam ditures also can be an important factor in Before After reducing poverty and moderating dispari- ties. As exemplified in the case of China, Source: Hofman and Guerra (2005). improved connectivity, which facilitated the movement of rural labor to more productive employment opportunities, mostly in urban duction to areas where market forces would areas, is an effective way to reduce pov- render such activities uncompetitive or erty. However, the impact on poverty and unsustainable. These efforts often include regional disparities can also be more subtle. incentives to promote new industrial activi- Transport investments that reduce the logis- ties in lagging regions or to concentrate tics costs of traded goods whose prices are more production in rural areas. But they established by national or global markets also can involve special efforts to deal with allow wages to be higher for workers who the problems of slums in mega cities. are farther away from consumer markets. In Indonesia, the success of such efforts to encourage industries to relocate to less- Spatially targeted interventions congested areas or lagging regions has been Practically all developing countries have mixed.Similar efforts have been tried in other some spatially targeted interventions. The countries, including China and the Philip- WDR 2009, however, suggests that many of pines. Firms that depend on localization these programs turn out to be ineffective, externalities might find it attractive to move particularly if they attempt to redirect pro- from high-cost centers in major urban areas 356 RESHAPING ECONOMIC GEOGRAPHY IN EAST ASIA to smaller,specialized cities,but in some cases Future trends in spatial disparities incentives are not enough to overcome their and links with development reluctance to relocate (Otsuka and Sonobe Sustained growth eventually will raise all in chapter 2; Arze del Granado in chapter 9; incomes regardless of location and, in time, Kuncoro in chapter 10; He in chapter 16; Bai will provide access to social services and and Lin in chapter 17). equitable living standards. But the process Spatially targeted programs in many can be lengthy and often disruptive. In the East Asian countries are often designed to case of China, simulations suggest that, with provide special support for rural areas and good policies and continued rapid growth, may be warranted if there are biases favoring income inequalities may begin to decline major urban centers (Balisacan,Hill,and Piza in a decade or so but, with less success- in chapter 11; Wisaweisuan in chapter 12). ful policies and outcomes, may take several For low-income countries such as Vietnam, decades (Huang and Luo in chapter 13). the potential for increasing agricultural The reasons lie in the complexities of the productivity is often substantial, and thus growth process: the interplay between the programs to tap this potential are attractive. structure and spatial location of produc- At the same time, as the country develops, tion, the implications for labor mobility, and labor will gradually move out of agriculture, differential earnings of workers by industry and, as Son suggests in chapter 7, efforts to and location. strengthen connectivity between rural and Thus spatial disparities in production urban areas and to link factor and goods appear to be inevitable and, within coun- markets are attractive options. In middle- tries, even desirable, but policy makers in income countries where crop yields are East Asia can do better in moderating the already relatively high and land resources rising disparities in living standards that are scarce, the strategy should be to encour- accompany the growth process. Doing so age more off-farm employment and, if the means that they need to overcome the many employment opportunities exist, migration challenges in making their fiscal systems to urban areas. and institutions more effective in provid- Spatially targeted interventions may also ing basic services for a broader spectrum be needed to equalize opportunities to access of society, especially in disadvantaged areas, education and health services. The benefits while also improving connectivity between of such programs are enhanced by the fact leading and lagging regions. that a healthier and more skilled labor force In most of developing East Asia, income produces benefits that are "portable"; that inequality will likely continue to increase is, they move with the person and thus do in the future, although the rate of dete- not carry the same risks as investments in rioration appears to be moderating in the fixed assets. middle-income countries. As economies More controversial are targeted develop- develop, such disparities will diminish, but ment programs for lagging regions, which only slowly and never completely. This con- often are designed to concentrate more pro- centration should be viewed with patience, duction mass in those areas. The experience because it brings spatial efficiency. But with with such programs is often not satisfac- more inclusive public expenditure programs tory, especially if they run counter to mar- and stronger institutions--which reinforce ket forces and the limitations brought on by the forces of agglomeration, migration, and geographic endowments. But if programs integration--countries can be both spatially are designed to reflect the realities of the sit- efficient and equitable. uation in lagging areas and, as in the case of China, custom tailored because of differing Notes regional needs, they can be a sensible com- 1. For many countries, the key period for plement to other policies, which continue to widening inequalities was from 1960 to 1990, as promote growth-enhancing agglomeration they moved from low- to middle-income status. effects (Huang and Luo in chapter 13; Yao However, comparable data across countries are in chapter 14). not readily available for that period. Lessons from experience: reshaping economic geography in East Asia 357 2. See the editors' preface to this book for a 8. This section draws extensively on the editors' more detailed discussion of these concepts. preface in this volume. 3. In their analysis of growth determinants and impact on poverty, Balisacan, Hill, and Piza find that the low growth rates largely explain the References poor performance in poverty reduction, but even Gill, Indermit, and Homi Kharas. 2007. An East the growth achieved did not have as much of an Asian Renaissance: Ideas for Economic Growth. impact on poverty reduction in the Philippines Washington, DC: World Bank. as it did in other countries, suggesting that how Hofman, Bert, and Susana Cordeiro the targeted programs are formulated may need Guerra. 2005."Fiscal Disparities in to be reconsidered. East Asia: How Large and Do They 4. As noted in the literature, the mere exis- Matter?" In East Asia Decentralizes: Making tence of borders separating countries creates Local Government Work, chap. 4. Washington, inefficiencies, as exemplified in the studies on the DC: World Bank. implications for Canada and the United States. 5. The Philippines ranks 89 out of 102 coun- McCulloch, Neil, and Bambang Suharnoko tries regarding the quality of its infrastructure Sjahrir. 2008."The Determinants of Subna- performance in the Global Competitiveness Report tional Growth in Decentralized Indonesia." (World Economic Forum various years). Unpublished mss. World Bank, Washington, 6. Officially, these workers must retain their DC. official residency in their home province and, in World Bank. 2008. World Development Indica- most cases, leave their family behind. Normally, tors Database 2008. Washington, DC: World they return to their home province for major Bank. holidays and as needed. ------. 2008. World Development Report 2009: 7. This was typical of the problems that transi- Reshaping Economic Geography. New York: tioneconomiesfacedinmovingtoamoremarket- Oxford University Press. based economy,with revenues depending on taxes rather than surpluses of state-owned enterprises. World Economic Forum. Various years. Global In the interim, the revenue base of most centrally Competitiveness Report. Geneva: World planned economies typically collapsed. Economic Forum. Index Aceh, 122 economic diffusion, 308­310 advance-order system, 38, 39, 40, 45 high-density city-region, 311, 313­314 agglomeration and agglomeration economies, xvii, xix, xxi, indicators, 312 xxii­xxiii, 166, 340, 342, 351, 354 transport, 314­315 benefits, 167n.1 border zones, 353, 357n.4 categorization, 341 China and Japan, 9­11 China, 200, 225, 282, 301, 347, 348 cooperation, 96n.10 development strategy, 14­15 Mekong subregion, 84­85 effects, 168n.9 roads and, 95­96nn.6,7 Indonesia, 143­146, 156­168 Singapore­IDR, 75­76 Philippines, 345 trade, 84­85, 95­96nn.6,7 regional, 1­17, 23 BOTABEK, 157, 159, 160, 167 Vietnam, 341 business climate see also industrial agglomeration China, 231­232 aging, Vietnam, 113 Vietnam, 112 agriculture, 3­4 China, 199, 210, 212, 215, 255n.8 expenditures, 254, 255n.8 Cambodia fiscal transfers, 253 cross-border trade, 85 Indonesia, 143­144 migrants, 92 investment, 112 capital, 14 Philippines, 179 China, 282­293 prices, 252 returns to, 283­284 productivity, 3­4, 105, 210, 212, 215 state vs total, 273 Thailand, 345 stock, 283­284 value added, 111 chemicals manufacturing, 159, 160, 161, 162, 164, 165, 166 Vietnam, 101, 105, 111, 112 China, xviii, xxii­xxiii, xxi, 2 airports, China, 301, 311, 313 border effects, 9­11 see also transport capital allocation, regional specialization, and spillover effects, arbitrage trader, 35 282­293 assets capitalism, 223­224 China, 202 Central Rising program, 234, 235 Philippines, 179 city development, 242­247 Association of South East Asian Nations (ASEAN), 9, 342 cluster development, 27­30, 31n.10 coastal regions, 224, 262, 298, 311 cities, 317n.1 Badawi, Abdullah Ahmad, 66 history, 251 Bali, 121, 123, 133n.5 industrial agglomeration, 248 demographics, 125 map, 296 Bangkok, xx, xxii open cities, China, 218, 228 banking, China, 234 urban­rural spatial restructuring, 294­319 Beijing, 310­311, 311­312, 313, 314 cross-border relations, 10, 18, 85­86, 94 Bingo Working Clothes Cluster, 30n.3 demography, 196, 315 Bohai Bay region, 310­315, 316 development model, 223­226 density, agglomeration, and economic growth, 304­307 rationales, 224­226 development zones, 307­308 economic geography, 196­217, 226­230 359 360 INDEX China (continued) connectedness and connectivity, 48­49, 60, 96n.11 efficiency vs equality, 247­250 data and analysis, 49­50, 50­58, 64n.3 exports, 263, 271 empirical framework, 58­59 fiscal transfer, 250­253 Indonesia, 49­50, 50­58, 64n.3, 123 future, 213­216 Singapore­IDR, 75­76 geography, 196 consumption growth, 198­199, 289­292 Indonesia, 117 history, 196­198, 223­224, 225, 238n.4, 346 Thailand, 190 income, 12, 15 Vietnam, 112 industrialization, 23, 258 connectivity, 354 agglomeration, 242­247, 258­281 Korea, 350 autonomous municipalities, 242­243 Lao PDR, 342 coast, 242, 243 contractual arrangements Northeast, 242 China, 225­226 by province, 243 Lao PDR, 38­40, 43, 44, 45 Inequality, 13, 14 convergence lessons learned, 346­349 absolute, 129 map, 197 China, 219­221 Northeast, development strategy, 218, 230, 232­234, growth rate and, 221 234­235, 349 Indonesia, 126­127, 129 policies, 199­213, 218­240 cooperation adjustment, 250­253 cross-border, 96n.10 implications, 253­254 Mekong subregion, 92 productivity, 289­292 private and public sector, 81 regional disparities, 218­240 Singapore­IDR, 76­77 addressing, 235­238 subregional, 96n.11 reduction, 230­235 corridors, 95n.3 regions, 220­221, 285­289 construction, Indonesia, 115 returns to capital across provinces, 283­284 corruption, 176 traders, 95­96n.7 embezzlement, 45 trends, 213 costs, xix, 8­9 urban­rural development, 241­257 communication, 338 wages, 31n.11 congestion, 150n.16 West, development strategy, 218, 230, 231­232, 234 isolation, 86 Yangtze River Region, 27­30 transport, 338, 350, 352 cities culture brokers, 36, 45, 46, 342 China, 242­247, 250­251 currency open cities, 250­251 China, 317nn.2,3 ranking, 11­12 Lao PDR, 44, 45 see also urbanization; urban­rural inequities Thailand, 194n.1 cluster-based industrial development. See industry and customs-free zones, China, 228 industrialization clustering, xvii, 123 collectives, fake, 31n.10 Daesang Company, 334, 335 communication decentralization, 145, 147­149, 354 China, 206 China, 198, 200, 227, 258, 260­261, 266, 294­295 costs, 338 firms, 162­164 see also information and communication technology fiscal, xxii, 260 competition Indonesia, 115, 133, 133n.2, 135­143, 344 China, 260, 266, 285­286 Korea, 322 local, 290 policy, 146­148 Singapore­IDR, 71­75 deconcentration, 158, 160­161, 166­167, 167­168n.6, 343 concentration, 135 deindustrialization, 124 dynamics, 144 demographics. See migrants and migration; population Indonesia, 156­168 Deng Xiaoping, 224, 250, 346 manufacturing firms, 160, 162 density, xx, xxi, 156, 350, 351 see also agglomeration China, 197, 214, 346 congestion costs, 150n.16 Indonesia, 67, 343 Index 361 Korea, 322, 327­328, 333, 349 electronics sector, 4­6, 7 lessons learned, 351 employment, 150n.15, 216n.6 Thailand, 345 China, 29, 277­278, 300, 311, 314 development models and strategies density, 144, 150n.20 Chin, 307­308, 223­226, 238, 250 distance and, 150n.19 Central Rising, 238, 252 enclave activity, 84, 116­117 Go West, 238, 243, 251­252 Indonesia, 116­117, 144, 344 Reviving the Northeast, 238, 252 Korea, 325 Lao PDR, 33­47 manufacturing, 161, 162, 164, 277­278 Philippines, 169, 171­176 production, 29 political rationale, 225 Vietnam, 102, 104 Thailand, 193 see also labor force Vietnam, 111 environment and natural resources see also policy China, 231, 233 diffusion, China, 308­310 Indonesia, 131 disparities roads and, 88, 89­90 causes, 109­112 equality, efficiency vs, China, 247­250 Indonesia, 132, 343 equilibria, 340 provincial, 353 Ethiopia, leather-shoe industry, 25 regional, 132 ethnic groups Vietnam, 109­112, 113 Indonesia, 115, 132 distance, xx, xxi, xxii, 149, 156, 350, 351­352 Singapore­IDR, 77 adjusted, 216n.3 Vietnam, 113 China, 197, 205­206, 214, 238n.2, 276­277, 347 exchange rates benefit, 150n.18 China, 238n.3 bridging, 67 personalized vs impersonal, 45­46 penalty, 145, 150n.18 expenditures employment and, 150n.19 interdependence, 151n.26 incremental, defined, 154, 155n.1 spillovers, 150­151n.24 Indonesia, 48­64, 64n.5, 67, 137­140 externalities, 168nn.7­10 Korea, 322, 326, 327, 333 China, 214 lessons learned, 351­352 dynamic, 166 Vietnam, 341 Indonesia, 162­164 weaving clusters and, 40­44 positive, 167n.1, 214 district location, defined, 154 productivity and, 164, 165 districts, Indonesia, 150n.17 divergence, China, 219­221 growth rate and, 220­221 increasing, 219 "falling central" theory, 231 diversity, 145­146 FDI. See foreign direct investment Singapore­IDR, 71, 74 firms divisions, 350, 352­353 decentralization, 162­164 artificial, xx Indonesia, 156­168 China, 197, 214 location, 158­162, 167n.4 Indonesia, 67 productivity, 162, 156­168 lessons learned, 352­353 stock of, 162 fiscal programs, 354­355 lessons learned, 354­355 economic and technological development zones (ETDZ), 228, fiscal sharing, interregional, China, 227­228 250­251, 252, 260, 278 fiscal transfers, 241, 248, 252­253, 254, 255n.7, 354 economies of concentration, Indonesia, 135­152 growth and, 253, 254 economies of scale. See scale economies urban­rural development, 250­253 education, 46, 216n.6 flying geese model, xvii­xviii, xxi, 1, 339­340 China, 205, 235­236, 252 following goose, determination, 1 income and, 60­61, 63 food production, 159, 160, 161 Indonesia, 343 Vietnam, 104­106 migrants, 235­236 see also agriculture Thailand, 192­194, 194n.5 foreign direct investment (FDI), 12 362 INDEX foreign direct investment (continued) compressed, 349 China, 199, 225, 226­227, 231, 248, 259­260, 271, 277, 282, congestion, 150n.16 304­307, 309, 346 consensus, 223­224 Korea, 325­326 convergence, 221 Thailand, 190, 191 determinants, 177­179 Vietnam, 100­101, 113 divergence, 220­221 fragmentation, 6­9, 338 elasticity, 179 boundaries, 133n.1 homogeneous, 145 framework, 9 income disparities and, 248­250 Indonesia, 133 Indonesia, 122, 130­131, 132 fringe areas, defined, 154 Korea, 349 Mekong subregion, 85 Philippines, 169, 177­179, 344 GDP provincial, 130­131 China, 198, 201­205, 212, 214, 222, 244­246, 285, 297, 305, rate, 133n.7, 357n.3 308­310, 316 regional, 122 cities, 150n.14 Thailand, 184­185, 185­186 government revenue and, 201 urban­rural inequality and, 247­248 government spending and, 202, 203, 205 Vietnam, 100, 106, 107 indicators, 297 Guangdong, 296­304, 315 Indonesia, 116­118, 122, 128, 131, 150n.14 density, agglomeration, and economic growth, 298­301 industry and, 244­246 development and cooperation, 303­304 Java, 158 transport infrastructure, 301­303 Korea, 329­333, 350 Guangzhou, 316 manufacturing, 158, 159 natural resources and, 131 population density and, 247 health indicators regional, 122, 128, 285 China, 203, 205, 212, 252 rural­urban divide, 222, 295 Indonesia, 130 Thailand, 184­188, 345 Thailand, 192, 193 Vietnam, 100­101, 111 Vietnam, 103, 105 geography, xxiii, 19, 352 heddle, 46n.3 China, 226­230 Honda, Soichiro, 22 Indonesia, 115­121 Honda Motor Company, 22­23 logic, 81 Hong Kong, 300­301, 316 Philippines, 172 transport and, 302, 317n.5 globalization, 294 Hoover coefficient of localization, 286­287 China, 199­200, 206, 214, 242­244, 258, 259­260, 266, household responsibility system, China, 198 270­271, 279, 294­319, 347 hukou, 196, 208, 214, 235, 236, 296, 303, 305, 313, 352, 357n.7 industrial agglomeration, 242­244 human capital, 178, 179, 254, 213, 233 Korea, 325­326 see also labor force Thailand, 189, 345­346 human development index (HDI), 130, 212, 353 urban­rural spatial restructuring and, 294­319 human transformation, Greater Mekong subregion, 78­99 government, xxi China, 202, 238, 227­228 discretionary power, 238 income, xviii expenditures, 355 China, 210­212, 216, 219, 221­222, 241, 285, 347­348 Indonesia, 123 distribution, xix, 12­14, 184, 185­190, 194, 339 interaction, 150n.23 education and, 56 investment, 227­228 growth vs, 54, 59, 172, 177­179 Korea, 331­332 health indicators and, 103 loans, 331­332 Indonesia, 52, 54­57, 59, 143­144 spending, 181n.6, 191­193, 202 inequality, 11­14, 15, 15n.2, 157, 191, 210­212, 216, 2 Thailand, 191­193 19, 241 transfers, 228 Java, 157 Greater Mekong subregion (GMS). See Mekong subregion Korea, 349 growth, 339 nonagricultural, 36, 54­57, 59, 62­63 centers, xxiii Philippines, 177­179 China, 198­199, 214­215, 220­221, 247­250 regional, 11­14, 285, 356 Index 363 Thailand, 184, 185­190, 191, 194 Korea, 13 transfer, 3 Mekong subregion, 84, 81­83 urban­rural inequality, 221­222, 247­248, 249­250, 254 provincial income, 129 Vietnam, 102 regional, 125­132 Indonesia, xxi, xxii roads and, 87 agglomeration, 156­168 rural, 216n.5 data and data analysis, 49­50, 50­58 Thailand, 13 decentralization, 135­152 trade openness and, 96n.12 economic geography, 115­121 trends, 214­216 economies of concentration, 135­152 Vietnam, 109 empirical framework and methodology, 58­59, information and communication technology (ICT), 162­164 180­181n.5 firm location, 158­162 China, 232 firm productivity, 156­168 Korea, 327, 333­334, 349 lessons learned, 342­344 regional production, 6 manufacturing, 156­158 infrastructure, 135, 167, 167n.2, 355 map, 50 access, 169 policies, 156­168 China, 248, 300, 301­303, 304, 314­315, 231, 347, 349 provinces, 116 Indonesia, 63, 137, 149-150n.8 regional economic growth, 121­125 investment in, 95n.3, 149­150n.8, 150n.21 regional inequality and convergence, 125­133 Java, 157 variables and estimation results, defined, 154­155 Philippines, 169, 175­176, 180, 180-181nn.4,5, 345, 357n.5 village economies, 48­65 policy, 63 see also Java returns to investment, 49 industrial agglomeration, 340 spatial integration and, 93 China, 242­247, 258­281 spending, 141 county-level analysis, 277­279 Thailand, 186­187, 345 economic performance and, 258­281 Vietnam, 109­110, 113, 341 globalization, 242­244 see also roads industrial specialization and, 261­277 infringement, urban­rural income disparity and, 249­250 industrialization, 242­244 innovation, xvii, 3­4, 19­20, 21 scale economy in cities, 244­247 Korea, 326­327, 333, 334­336 theory, 258­261 multifacited, 25 see also agglomeration and agglomeration economies policy, 334­335 industrial concentration index, 161, 162 insiders, 36, 45 industrial specialization integration administration and, 274 border zones, 85­86 China, 261­277 de jure, 15n.3 industrial agglomeration and, 261­277 markets, 87 provincial, 269­273, 280 Philippines, 175­176 temporal changes, 275 quality, 83 industry and industrialization, 2, 14, 33, 261, 338 regional, 1­17 artisan manufacturing, 33 investment China, 23, 199, 218, 220, 233, 261, 264, 268, 269, 270 agriculture, 112 cluster-based development, 19­32 China, 213, 227­228, 237­238, 238n.1 favored and protected, 269 policy, 237­238 government support, 233­234 security, 76 Indonesia, 124 Singapore­IDR, 76 Korea, 321­324, 325­326, 349 Vietnam, 103, 108­109, 110, 111­112 labor-intensive, 323­324 Iskandar Development Region (IDR), 66­78, 78n.1 maps, 265, 267, 268, 269, 270 land development, 72­73 output, 264 isolation ranking, 261 cost, 86 spillover, 33, 289­290, 292 roads and, 88­89 technology-intensive, 324 solutions, 94 Thailand, 189 inequality, 216n.2, 353­354, 356n.1 China, 13, 214­216, 223 Jakarta, 117, 133n.4 Indonesia, 125­132 Japan, xvii, 1, 2 364 INDEX Japan (continued) change since 1986, 37 border effects, 9­11 contractual arrangements, 38­39 cluster-based industry, 30nn.3,6 history, 36­37 income, 12 industry profile, 36­37 income inequality, 13, 15 marketing, 37­38 motorcycle enterprises, 22 weaving techniques, 37 Java, 121, 133n.4 traders, 95­96n.6,7 demographics, 125 weaving clusters, 33­47 manufacturing, 156­158 lead goose, 1, 2 map, 158 legal issues regional comparison, 157 border zones, 86 see also Indonesia Singapore­IDR, 76 JJJ region. See Bohai Bay region see also policy; regulations Johor, 67, 71 lessons learned, 338­357 joint ventures, China, 306 disparities, increasing, 339 indicators, 339 new economic geography, 339­341 Kalimantan, 118­122 production-sharing networks, 338 Korea, Republic of, xxi, xxii, xxiii, 327­334 regionalism, rise of, 338 demography, 327­328 spatial patters at low income levels, 341­342 firms, 332 licensing, 158, 167, 167n.4 history, 320­321 linkages, 81, 112­113 income inequality, 15 living standards, xviii industry, 320­337 Vietnam, 101, 102­103, 104 inequality, 13 local government, China, 260 lessons learned, 349­350 local institutions, Thailand, 193 maps, 321, 322 localization, 146, 166, 167n.1, 350 policy, 321­327, 334­336 externalities, 163 population, 328­329 location, xxii, 167nn.4,5, 351­352, 354 R&D, 327, 332 changing, 25­30 manufacturing, 158­162, 166, 167n.4, 168n.7 labor force, 254, 259, 276, 356 China, 29, 200, 208­209, 246, 252, 303 Macao, 300 Korea, 325, 330­332 Malaysia, 70­71 Malaysia, 72 Maluku, 122 Philippines, 174­175 Manila, xx Singapore­IDR, 72, 74 manufacturing, 149n.1 Thailand, 91 China, 241, 261­264, 277­278 Vietnam, 101­102, 104, 113, 113n.3 employment, 277­278 see also employment; migrants and migration exogenous vs indigenous, 33 labor productivity Indonesia, 156­168 industrial agglomeration and, 275­276 Indonesia, 64n.4 Vietnam, 104 Java, 156­158 landholding Korea, 321, 324 China, 252, 254, 304, 307 location, 168n.7 Indonesia, 60, 62 machinery, 159, 160, 161, 162, 164, 165, 166 Philippines, 179 Singapore, 72 Singapore­IDR, 72, 74 Thailand, 186­187, 190 land lease model, 294­295 marginalization, roads and, 89 Lao PDR, xviii, xxi margins, social and spatial, 95 cross-border integration, 94 markets and marketization cross-border trade, 85­86 access, 94, 224­225 environment, 88, 89­90 China, 197, 199­200, 224­225, 242, 246, 254, 258­259, 265­266, lessons learned, 341­342 267, 270­271, 279, 285­286, 346, 347 policy, 90 competition, 285­286 roads, 87, 88, 89­90 economies, 357n.7 textiles failures, 19 Index 365 integration, 87, 88­89, 200, 346, 347 obstacles isolation, 88­89 Singapore­IDR, 78 overseas, 42­43 trans-border, 84 segmentation, 34­35, 246, 254 ocean transport, China, 301 spot-market transactions, 39, 40 oil, Indonesia, 115 tourist, 44 open-door policy, China, 198­199, 218, 221, 300 urban vs rural, 35 openness Mekong subregion, Greater (GMS), xxi, 79­80, 82­83 Indonesia, 123 economic progress, 82­83 Mekong subregion, 81­83 map, 80 Thailand, 190 openness and integration, 82­83 opportunism, 45, 46n.5 program, 81, 83 outsider hypothesis, 36 spatial integration and human transformation, 78­99 strategic framework, 80­81, 92 paired assistance development, 303­304 migrants and migration, xxii, 133­134n.10, 135­136, 148, paper manufacturing, 159, 160, 161, 162, 164, 165, 166 149­150nn.4,6,8­12, 276 Papua China, 196, 208­209, 213, 215­216, 235­236, 255n.2, 259, 296, budget, 133n.3 303, 305 growth, 122 income and, 61 spatial inequality, 131 Indonesia, 131­132, 136­143, 139, 140, 142­143, 344 see also Indonesia Korea, 350 PATANAS survey, 50, 64n.1 maps, 91, 106, 138, 140, 209 Pearl River delta, 296­304, 315, 316 Mekong subregion, 91­92, 96n.9 density, agglomeration, and economic growth, 298­301 net migration, defined, 154 development and cooperation, 303­304 Philippines, 174­175 map, 299 regional fiscal structure and, 140­141 transport infrastructure, 301­303 roads and, 90­92 urban clusters, 301 Singapore­IDR, 75­76 perceptions, 353 Thailand, 91, 92, 187, 345 performance, Indonesia, 132­133 Vietnam, 101, 106 Philippines, xxi, xxii, 170­171, 357n.5 1 million plus cities, defined, 154 administrative boundaries, 180n.1 minerals, nonmetallic, 159, 160, 161, 162, 164, 165, 166 economic indicators, 172 minimum distance to urban center, defined, 154, 155n.1 growth determinants, 177­179 mining, Indonesia, 116­119, 120, 121, 128, 129, 131, 132, lessons learned, 342, 344­345 134n.12 map, 170 mobility, 67, 90­92, 93, 96n.11, 351­352, 354 policy, 176, 179 China, 303 poverty reduction, 177­179 costs, 180 regional development patterns, 169, 171­176 gender and, 88, 89 social indicators, 173­174 Indonesia, 137­140 PODES (village census data), 50, 64n.1 lessons learned, 351­352 policy, 340 Philippines, 180, 345 agriculture, 252 see also migration; roads; transportation assistance, 303­304 motorcycle enterprises, Japan, 22 China, 196­216, 218­240, 248, 250­254, 303­304 multi-core system, xvii decentralization, 146­148 Myanmar, migrants, 92 fiscal, 141, 200­204, 236­237 geography vs, 229­230 Indonesia, 63, 115­116, 132, 146­148, 156­168 National Farmers' Panel. See PATANAS industrial, 30, 250­254, 321­327, 334­336 natural resources and environment infrastructure, 63 China, 231, 233 investment, 237­238 Indonesia, 131 Korea, 321­327, 334­336 roads and, 88, 89­90 lessons learned, 353­354 new economic geography, xvii, xix, 2­3, 144 manufacturing, 166, 167 China, 196­197 migration, 235­236 New Economic Mechanism, Lao PDR, 34 Philippines, 176, 179 nonpoor, Indonesia, 120 poverty reduction and, 177, 178­179 366 INDEX policy (continued) Korea, 323 preferential, 228­229 networks, 4­11 regional development, 250­252 spatial factors and, 289­290 role, 353­354 trends, 356 Singapore­IDR, 76, 77­78 Vietnam, 112 spatial disparities and, 169­182, 184­195, 196­197, 199­213 provinces, returns to capital, 283­284 spatial transformation and, 93­95 see also regions Thailand, 187­188, 190, 191, 194 public spending trade, 9 China, 237, 348 urban concentration and, 141­143 migration and, 140­141 urban­rural development, 250­254 Thailand, 191­192 Vietnam, 100­101, 110­111 putting-out system, 38, 40, 43, 44, 45 see also development strategies politics border zones, 85 railways, China, 206­207, 208, 314, 317n.6 logic, 81 see also transport patronage, Philippines, 176 rank size rule, 15n.5 population, xxii, 149n.5, 150n.13 recentralization, China, 198 agglomerations, 12, 13 recommendations China, 246­247, 297, 300 manufacturing, 167 convergence, 142 Vietnam, 112­113 GDP, 247 recovery, 194n.2 indicators, 126 red-cap enterprises, 31n.10 Indonesia, 124­125, 137, 142, 143 regions Korea, 327­329, 336 China, 218­223, 230­238, 248, 250­252, 282­293, 284, movement and concentration, 159­162 285­289, 348 Philippines, 174­175 development, 115, 234­235, 250­252 Vietnam, 107 disparities, xxi, xxii, 103­109, 125­133, 218­223, 230­235, see also demographics 248, 348 ports Indonesia, 115, 117­118, 121­125, 132 China, 314 integration, 14, 83, 95 open, 250­251 returns to capital, 284 see also transport specialization, 282­293, 285­289 poverty and poverty reduction Thailand, 190­192 density and, 108 Vietnam, 103­109 estimates, 181n.9 regulations income growth and, 177­179 China, 196, 208, 213, 214, 235, 252, 254, 296, 303, 305, 352, Indonesia, 115, 120 357n.6 Lao PDR, 84 Indonesia, 116 Mekong subregion, 84 Philippines, 176 Philippines, 169, 177­179, 344 residence, 196, 208, 214, 235, 252, 254, 296, 303, 305, 352, roads and, 87 357n.6 rural vs urban, 84 religion, Indonesia, 131 spatial component, 86 remittances, 94, 180n.3 Thailand, 184, 185, 186, 191, 193 resettlement, Lao PDR, 90 Vietnam, 106­107, 108, 109 resources price variation, Philippines, 176, 181n.7 Indonesia, 123 private vs public capital, Vietnam, 113 Singapore­IDR, 71 privatization, China, 232­233, 248 retailers, as urban-based traders, 45 production and productivity returns to capital, 283­284, 292n.1 China, 215, 289­292 returns to scale, 123 disparities, 143 revenues, 133n.2 empirical results, 290­291 reverse causality, 150n.22 externalities and, 164, 165 Riau, 120 firms, 162 roads and roadbuilding, 48­49, 64n.3, 83, 95n.3, 86­90, 167n.3 Indonesia, 151n.27 China, 206­207, 213, 232, 314 industrial agglomeration and, 275­276 cross-border, 83, 95­96nn.6,7 Index 367 income and, 55 potential synergies, 71­75 Indonesia, 50­51, 52, 59­61, 62, 63­64, 64n.6, 343 relocating activities, 72­73 isolation and, 88­89 success factors, 75­78 Java, 157, 158, 159 small and medium enterprises (SME), 19, 24 Korea, 336 Korea, 326­327 Laos, 87­88 Singapore­IDR, 74 people and, 89, 90­92 social capital, Java, 157 Philippines, 176 social conflict, Indonesia, 130­132 Thailand, 88 social disparities, China, 212­213 rural development social expenditures, China, 216n.2 clustering, 33­47 social indicators expenditure, 102 China, 212 industrialization, 33 Indonesia, 130 nonfarm industries, 33 Philippines, 173­174 Vietnam, 100­114 social services rural­urban inequality. See urban­rural inequality China, 203, 254 Philippines, 169, 180 social welfare, China, 348, 349 sanction mechanisms, 46 spatial disparities, 48 savings development and, 356 Singapore­IDR, 76 growth and, 185­186 Vietnam, 103 Philippines, 169­182 scale economies, xix, 3, 67, 259, 350, 351 Thailand, 184­195 China, 244­247 Vietnam, 100­114 cities, 244­247 spatial economics, xvii, xix industrial agglomeration, 244­247 spatial integration, 81 lessons learned, 351 Greater Mekong subregion, 78­99 local, 290 migration and, 92 Singapore­IDR, 71, 74 roads and, 89­90 self-employment, 36 success factors, 92, 93 density, 57­58 spatial transformations, 86­92 Indonesia, 56­62 Mekong subregion, 83­86 nonfarm, 59, 61­62 policies and politics, 92­95 roads and, 57 scales and sites, 83­86 Seoul, 332­333 spatially targeted interventions, 355­356 services, xxii special economic zones (SEZ), 198­199, 218, 224­225, access to, 192 226­227, 228, 297, 250, 346 China, 306 history, 251 Indonesia, 149­150n.8 specialization Thailand, 189, 192 China, 200, 346 SEZ. See special economic zones rates, 29­30 Shanghai, 225, 226, 304­310, 315, 316 spillover effects, China, 282­293 density, agglomeration, and economic growth, 304­307 state ownership, China, 232­233 development zones, and spatial restructuring, 307­308 subsidies, China, 252 economic diffusion, 308­310 suburbanization, Indonesia, 139 industry, 308 Sulawesi, 118­119, 121, 123 population, 304, 305, 307, 308­309, 315 economic rankings, 119­120 shocks, household-specific, 59 growth rate, 123, 133n.7 shoes, leather, Ethiopia, 25 Sumatra, 117­118, 120, 121, 123, 133n.6 Singapore, 67, 69­70 economic rankings, 119­120 Singapore­Iskandar Development Region (IDR), xxi, 66­78 Sunan model, 25, 30n.7 comparison, 68­69 Sunchang, 335 economic ties, 67­68 factors driving the relationship, 67­71 joint economy, 74 Taiwan, China labor, 72 cluster development, 25­27, 30­31n.8 obstacles, 76 machine tool enterprises, 22­23 368 INDEX targeted interventions, 354, 355­356 urbanization, xxii, xxiii, 2, 135, 145­146, 148, 149n.4, 166, 167n.1, tarmac bias, 89 339, 342, 350 taxation, 151n.25, 201­202, 210­211, 225, 227­228, 231, 232, 237, China, 198, 200, 208­209, 236, 254, 316, 347 238n.7, 252, 266, 272, 310 defined, 154 technology, 4, 6, 327 dual-track, 295­296, 300 China, 233 Indonesia, 136­140 Korea, 324, 333­334 Korea, 322, 349 telecommunications. See information and communication migration and, 236 technology urban­rural inequality, xxii, 3, 214, 216n.7 textiles manufacturing China, 210­212, 214, 215, 221­223, 241­257, Java, 159­166 294­319, 348 Lao weaving, 33­47 connectivity, 49 Thailand, xxi, xxii GDP and, 222 economic indicators, 185 globalization, 294­319 growth and, 185­194 growth and, 247­248 income inequality, 15 income, 186, 191, 210­212, 215, 254 inequality, 13­14 Indonesia, 131, 136, 137­140 lessons learned, 342, 345­346 living costs, 103 migrants, 91, 92 Mekong subregion, 84 mobility, 90­91 Thailand, 186, 191 spatial disparities, 184­195 Vietnam, 101­103, 113 transport issues, 88 urban sector, 94 Tianjin Binhai New District (TBND), 310, 313 businesses, 46n.4 township and village enterprises, 295­296, 248, 309 cities, number of, 294 trade, 4, 6, 338 city size, 254 China, 200, 215, 248, 259 concentration, reduction of, 141­143 China­Japan, 11 defined, 149n.2 fairs, 46 density, 136­143 intraregional, 4­5, 11 expenditure, 102 openness, inequality and, 96n.12, 200, 259 gravity, 138­139 policy, 9 restructuring, 294­295 Thailand, 189 scale economy, 244­247 theory, 339 traders, 35­36, 45, 46 variables, defined, 154 arbitrage, 35 variations, coefficients of, 130 commission, 41­42 very poor, 120 itinerant, 41­42 Vientiane, 46n.2 retailers as, 45 Vietnam, xviii, xxi Vietnam, 95­96n.7 cross-border trade, 85­86 weaving clusters and distance, 40­44 disparities, causes of, 109­112 transaction costs, 34 growth, 100 transport, 15n.4, 88, 317n.5, 352 history, 100 China, 204­208, 301­303, 314­315, 317n.5 lessons learned, 341­342 corridors, 95n.3 policy, 100­101 costs, xix, 3, 34, 49, 86, 338, 350, 352 recommendations, 112­113 equipment industry, 29 regional disparities, 103­109 gender and, 88, 89 rural development and spatial disparities, 100­114 infrastructure, 113, 314­315 rural­urban disparities, 101­103 Singapore­IDR, 75 spatial disparities, 100­101 Vietnam, 113 traders, 95­96n.7 see also roads village economies, Indonesia, 48­65 trends, disparity, 214­216, 356 visas, Singapore­IDR, 76 trust, 34­35 community-based, 34, 35, 43, 45­46, 46n.1 decline of, 35, 44­45, 46n.1 wages, China, 31n.11 economic uncertainty, 46 see also employment; labor force process-based, 34­35, 45 water development, Thailand, 193 Index 369 wealth, Indonesia, 120 xaiokang status, 303 weavers, 42­43 weaving, Lao PDR, 33­47 West Nusa Tengarra growth, 120, 121, 134n.12 Yangtze River delta, 27­30, 304­310, 315, 316 mining, 134n.12 density, agglomeration, and economic growth, 304­307 women, roads and, 87­88, 89 development zones, 307­308 wood products, 159, 160, 161 economic diffusion, 308­310 workforce. See labor force yarn-on-credit system, 38­39, 44, 45, 46n.4 ECO-AUDIT Environmental Benefits Statement The World Bank is committed to preserving Saved: endangered forests and natural resources. · 15 trees The Office of the Publisher has chosen to · 11 million BTUs of total print Reshaping Economic Geography in energy East Asia, on recycled paper with 30 per- · 1339 pounds of CO2 cent postconsumer fiber in accordance with equivalent the recommended standards for paper usage of greenhouse gases set by the Green Press Initiative, a nonprofit · 5,556 gallons of waste program supporting publishers in using fiber water that is not sourced from endangered forests. · 713 pounds of solid waste For more information, visit www.green- pressinitiative.org. reshaping economic geography in east asia "Having spent much of my career pioneering the principles that underpin the `new economic geography,' or spatial economics, it is gratifying to see how these principles are now being used to deepen our understanding of the most dynamic region of the world and its development process.... These studies illustrate how well the concepts underpinning the new economic geography can explain what is happening in East Asia. Despite a few exceptions, urbanization and related agglomeration benefits are part of the region's success story, along with the related pressures on policy makers to deal with increasing spatial disparities." From the foreword by Masahisa Fujita Reshaping Economic Geography in East Asia, a companion volume to the World Development Report 2009, brings together noted scholars to address the spatial distribution of economic growth in Asia. It reveals how the new economic geography is reshaping development objectives: from initiatives to foster growth via enhanced agglomeration and connectivity to the world economy, to programs that channel resources to lagging regions. Key themes include how East Asian governments have dealt with agglomeration economies, urbanization, and regional disparities; improving connectivity with infrastructure investments; and eliminating barriers both inside and outside borders to favor the movement of labor, goods, and services. This volume will be of great interest to readers working in the areas of economic policy, poverty reduction and urban-rural development strategies, and transport-led infrastructure policy. SKU 17641 ISBN 978-0-8213-7641-6