2009 55539 A B C D E Annual World Bank Conference on Development Economics Global People, Politics, and Globalization Edited by Justin Yifu Lin and Boris Pleskovic People, Politics, and Globalization Annual World Bank Conference on Development Economics--Global 2009 People, Politics, and Globalization Edited by Justin Yifu Lin and Boris Pleskovic Washington, D.C. © 2010 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 :: 13 12 11 10 This volume is a product of the staff of the International Bank for Reconstruction and Development / 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 judgment 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. Edited by Justin Yifu Lin and Boris Pleskovic Professional affiliations identified in this volume, unless otherwise noted, are as of the time of the conference, June 9­11, 2008. ISBN: 978-0-8213-7722-2 eISBN: 978-0-8213-8059-8 DOI: 10.1596/978-0-8213-7722-2 ISSN: 1813-9477 Contents ABOUT THIS BOOK ix INTRODUCTION 1 Justin Yifu Lin and Boris Pleskovic OPENING ADDRESS 11 Trevor Manuel OPENING ADDRESS 13 Justin Yifu Lin OPENING ADDRESS 15 Thabo Mbeki KEYNOTE ADDRESS 23 Michael Spence KEYNOTE ADDRESS 33 Bassma Kodmani KEYNOTE ADDRESS 41 Sunil Kant Munjal Trade and Investment Trade and Economic Performance: Does Africa's Fragmentation Matter? 51 Paul Collier and Anthony J. Venables Protectionist Policies and Manufacturing Trade Flows in Africa 77 Lawrence Edwards COMMENT Beata Smarzynska Javorcik 109 v vi | CONTENTS Crisscrossing Globalization: The Phenomenon of Uphill Skill Flows 115 Aaditya Mattoo and Arvind Subramanian COMMENT Beata Smarzynska Javorcik 141 Migration, Remittances, and the Transition from Foreign Aid The Aid-Migration Trade-off 147 Jean-Paul Azam and Ruxanda Berlinschi COMMENT Melvin D. Ayogu 169 Are Remittances More Effective Than Aid for Improving Child Health? An Empirical Assessment Using Inter- and Intracountry Data 173 Lisa Chauvet, Flore Gubert, and Sandrine Mesplé-Somps COMMENT Melvin D. Ayogu 205 The Role of Emigration and Emigrant Networks in Labor Market Decisions of Nonmigrants 209 Jinu Koola and Çaglar Özden COMMENT Jean-Luc Demonsant 226 Higher Education and High-Technology Industry The Role of Higher Education in High-Technology Industrial Development: What Can International Experience Tell Us? 233 Sachi Hatakenaka COMMENT Erik Sander 266 African Higher Education and Industry: What Are the Linkages? 275 Akilagpa Sawyerr and Boubakar Barry COMMENT Shahid Yusuf 294 An Arrested Virtuous Circle? Higher Education and High-Technology Industries in India 297 Rakesh Basant and Partha Mukhopadhyay COMMENT Pankaj Chandra 346 CONTENTS | vii Human Development Health and Socioeconomic Status: The Importance of Causal Pathways 355 Duncan Thomas The Household Impacts of Treating HIV/AIDS in Developing Countries 385 Markus Goldstein, Joshua Graff Zivin, and Harsha Thirumurthy COMMENTS T. Paul Schultz 406 John Strauss 412 Political Economy When Is Public Expenditure Pro-Poor? 419 Francisco Rodríguez The Political Economy of Public Service Provision in South Asia 443 Lakshmi Iyer COMMENT Ashutosh Varshney 463 CLOSING REMARKS 467 Alan Gelb CLOSING REMARKS 471 Fundi Tshazibana About This Book The Annual World Bank Conference on Development Economics is a forum for discussion and debate of important policy issues facing developing countries. The conferences emphasize the contribution that empirical economic research can make to understanding development processes and to formulating sound development policies. Conference papers are written by researchers in and outside the World Bank. The conference series was started in 1989. Conference papers are reviewed by the editors and are also subject to internal and external peer review. Some papers were revised after the conference, to reflect the comments made by discussants or from the floor, while most discussants' comments were not revised. As a result, discussants' comments may refer to elements of the paper that no longer exist in their original form. Unless other- wise noted, participants' affiliations identified in this volume are as of the time of the conference, June 9­11, 2008. The planning and organization of the June 2008 conference was a joint effort by the Government of South Africa and the World Bank. We gratefully acknowledge timely and valuable contributions made by the all the members of the Steering Committee and several anonymous reviewers. We would also like to thank Alan Gelb and Aehyung Kim for their insightful advice and Leita Jones, conference organizer, whose excellent organizational skills helped to ensure a successful conference. Finally, we thank the editorial staff for pulling this volume together, especially Stuart Tucker, Mark Ingebretsen, and Nora Leah Ridolfi from the Office of the Publisher. ix Introduction JUSTIN YIFU LIN AND BORIS PLESKOVIC The Annual Bank Conference on Development Economics (ABCDE) brings fresh, innovative perspectives to key problems of development. By providing a forum in which policy makers, academics and leading researchers focus on a common theme, ABCDE plays an important role in advancing debate and shaping the international development agenda. The 2009 ABCDE, held June 9­11, 2008, in Cape Town, South Africa, was devoted to "People, Politics, and Globalization." The program included a strong dose of empirical research on the experience of developing countries in all regions of the globe. Speakers addressed such subjects as trade and investment, higher education and high-technology industry, migration and remittances, the interaction between health and economic development, and the political economy of public expenditures. This volume includes selected papers from the conference, as well as keynote addresses by Michael Spence, chairman of the Commission on Growth and Develop- ment and 2001 Nobel laureate for economics; Bassma Kodmani, executive director of the Arab Reform Initiative; and Sunil Kant Munjal, chairman of Hero Corporate Service Ltd., the services division of the Indian automotive concern. Keynote Addresses In his address "Rethinking Growth: Learning from Experience and Adapting to New Challenges," Michael Spence discusses the subject of sustained, high, inclusive growth and summarizes the findings of the Commission on Growth and Develop- ment. The commission, under Spence's chairmanship, has gathered new insights into the fundamental dynamics of sustained high growth and has identified the policy measures that appear to underpin such growth. Justin Yifu Lin is Chief Economist and Senior Vice President for Development Economics at the World Bank. Boris Pleskovic is Research Manager for Development Economics at the World Bank. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank 1 2 | JUSTIN YIFU LIN AND BORIS PLESKOVIC Spence observes that in all of human history, only 13 economies have grown at an average annual rate of 7 percent or more for 25 years or longer--and all of these instances of sustained high growth occurred after 1950. The commission has dis- sected these cases to display their essential features and common characteristics, some of which derive from basic economics and others from effective government. According to Spence, the single most important shared characteristic of the 13 high-growth economies is engagement with the global economy. Rather than gen- erating new knowledge, these countries mastered existing knowledge and applied it in certain industries. Once they had developed a competitive advantage, they took advantage of the huge elastic demand offered by the global open economy, which allowed them to grow as quickly as they could invest. The commission has documented that these countries also maintained macroeco- nomic stability and had high levels of saving, which spurred investment. Spence observes that a high national saving rate--20 percent and higher--has been common among the high-growth countries. In most cases, he notes, competitive market mech- anisms, labor mobility, and rapid urbanization have been crucial. The commission's most surprising finding, says Spence, concerns the importance of noneconomic factors, particularly political leadership and effective governance. The key factor of policy success seems to have been adherence to a consistent growth strategy. The successful economies did not all have the same type of government, but they all had political leadership that focused on improving the lives of members of society as a whole. Political leadership that focused on its own interests or on the interests of a particular group did not achieve sustained high growth. Among other ingredients of successful policy, Spence lists sustained investment in three areas: infra- structure, health and early childhood nutrition, and education. The commission's findings do not include specific policy prescriptions. The commission did look at special challenges that face certain developing countries, including many countries in Africa. Small states, for example, have little ability to diversify their economies, and they face high costs with regard to government and public services. Countries rich in natural resources have a mixed record. Spence notes that because of the peculiar geography inherited from the colonial period, many African countries are small, many are landlocked, and although some have a rich natural resource base, others are resource poor. Middle-income countries, for their part, may experience difficulty moving from labor-intensive industry to industry fueled by knowledge, innovation, and human capital. Bassma Kodmani presents an overview of the Middle East, juxtaposing certain grim political and economic realities against real achievements in recent years that seem to be moving the Arab region into a new era. She cites a number of long-term dangers--the Arab-Israeli conflict, the humanitarian disaster in Darfur, the war in Iraq, the depletion of water resources, and agricultural dependency--but also points to high rates of economic growth in most countries in the region and to important governance reforms. Kodmani observes, however, that reforms have slowed in the last couple of years because of reluctance to shift power decisively from government to society at large. She maintains that it is time to adopt a new democratic paradigm INTRODUCTION | 3 based on strong and accountable public institutions and on such fundamental prin- ciples as liberty, respect for human rights, the rule of law, and social justice. Kodmani perceives a new awakening of Arab societies. Independent media have emerged and are forging horizontal links among citizens. Both the marginalized seg- ments of society and elites are challenging governments and are launching protests. Kodmani argues that the existence of independent media has made governments much more reluctant to suppress such movements. Generally speaking, she asserts, these movements have been very positive in that they are centered on social and polit- ical issues, not issues of identity or religion, and they are calling for peaceful, grad- ual change through legal means. Still, she warns, major social and political unrest is possible. In Kodmani's view, Middle Eastern society fears abrupt change and instability, and that fear is reinforced by the situation in Iraq, the threat from extremist groups such as those connected to Al Qaeda, and sectarian tensions in the region. As a conse- quence, political protest has generally remained peaceful, even though the space for legal protest is tightly restricted. This is true of Islamist movements, as well as secu- lar social movements. For three decades Islamists have built social networks and have made inroads into political life, and they are now demanding a share of power. But the taking of power by force has been limited to conflict areas such as Lebanon and Palestine. Kodmani argues that reforms in the Arab world are not sustainable without an underlying shift to a democratic paradigm. Only by recognizing this and creating a space for citizens to develop a home-grown agenda will a transition to a better future be possible. Specifically, Kodmani sees a need for a systemic approach toward reform, greater social accountability, and mechanisms for linking reform measures "from above" to the demands of society. She gives examples of two areas in which local groups have been modestly effective but foreign donors and multilateral insti- tutions have not: security reform and gender issues. What is most important, Kod- mani argues, is for Arab societies to be trusted to chart their own way, even if they choose a path with which foreign partners are not entirely comfortable. In his keynote speech, Sunil Kant Munjal traces India's history of successful reform. Since 1991, India has opened with remarkable speed to the global economy, but it has followed its own unique path. Indeed, Munjal states, it could not be oth- erwise, as there is no precedent for a democratic country with a population of 1 bil- lion transforming itself into an outward-facing market economy. Munjal emphasizes the crucial role of innovation in India's miraculous progress. India, he says, has learned much along the way--mostly from its own mistakes, but also from the outside world. With respect to the latter, Munjal emphasizes the role of India's "brain bank" of overseas Indians. India, he adds, continues to need foreign capital, but most of all, it needs the knowledge and networks that accompany for- eign capital. In addition to a great capacity for low-cost innovation, Munjal observes, India has a long history as a trading nation. That history was interrupted for several decades after the country became independent in 1947, when it embraced central planning 4 | JUSTIN YIFU LIN AND BORIS PLESKOVIC and began building dams, power plants, and steel plants. This direction may have been right at the time, says Munjal, but India held on to central planning for far too long. Only in 1991 did it make a dramatic and fateful move toward the market. Since then, India has firmly reestablished itself as a trading nation and has actively reached out to establish trade relations with many countries. Munjal gives many examples of success stories--in productivity, infrastructure development, and so forth--and cites some remarkable statistics: an average Indian today can spend twice what he or she could spend in 1985, and in the next 20 years the average Indian will be able to spend four times as much as now. Still, Munjal remarks, many problems remain: labor regulation is still too rigid, physical infrastructure is often lacking, and red tape and corruption persist. Munjal also points to striking regional differences within India and to the failure of the cen- tral and state governments to adequately provide basic social services such as primary education and health care. He notes that the private sector has stepped in to help-- in education and training; rural infrastructure, electrification, irrigation, drinking water supply, and many other areas--but that the private sector cannot be a substi- tute for government. Trade and Investment Paul Collier and Anthony Venables examine the economic consequences--especially the effects on trade--of the political fragmentation of Sub-Saharan Africa into 50- some states, mostly small. By contrast, in South Asia 95 percent of the population lives in just three countries: India, Pakistan, and Bangladesh. Sub-Saharan Africa and South Asia are populated on the same scale (roughly, 1 billion people in each) and as recently as 1980 had broadly similar levels of income and human development. Since then, however, the economic performance of the two regions has diverged sharply. Africa's share of world gross domestic product (GDP) declined in the 1980s and 1990s, while South Asia's nearly doubled. In 1980 Africa's share of world exports was nearly four times that of South Asia. Collier and Venables show that political geography has been a major factor in the regions' striking divergence over the last three decades. The political fragmentation of Sub-Saharan Africa pushes small national economies toward peasant agriculture and other economic activities in which scale is not so important. In addition, the authors observe, small markets by their nature tend to be less competitive than larger ones: monopolies, oligopolies, and rent-seeking are more common and make invest- ment less attractive. International borders, largely the legacy of colonialism, have led to an unequal distribution of natural advantages such as mineral wealth, coastlines, and natural harbors. Countries that are small, resource poor, or landlocked are at a tremendous economic disadvantage. Borders pose a formidable barrier to trade and labor mobil- ity and restrict the region's ability to pool risk and absorb external shocks. Large Asian countries do not face barriers of this sort, and internal migration and urban- ization have played an important role in their rapid growth. INTRODUCTION | 5 Collier and Venables point out that whereas the rise of large industrial clusters in Asia has fueled high productivity and growth, Africa's failure to develop such clus- ters has had major negative effects on economic performance. The authors attribute this failure to the small size of African cities (relative to Asian cities), which derives in large part from Africa's political fragmentation. Collier and Venables conclude that Africa would benefit greatly from political and economic integration. Previous efforts at integration have not yielded fruit in Africa, but the application of lessons from other world regions could help. The authors sug- gest that political and economic integration should go hand in hand; that integration is more likely to succeed if it grows from a small core of states, none of which is dom- inant; and that the economic agenda should be much broader than just trade policy, touching on integrated infrastructure, integrated investment and taxation rules, and so forth. Because these other areas of cooperation are much more likely to generate mutual gains than are trade agreements, their economic consequences reinforce rather than undermine the political process. Lawrence Edwards adds to the debate on trade policy in Africa, where trade has declined dramatically over recent decades. Analyzing the effects of tariff liberaliza- tion on trade flows in Africa since 1990, Edwards paints a mixed picture. Drawing on sector-level data, he reaches three conclusions: (a) tariff liberalization has taken place in many countries, but African countries are still marginalized in terms of world trade; (b) the effect of tariff liberalization has had a very small, but positive, effect on manufacturing trade flows; and (c) tariff liberalization has had a negative effect on the manufacturing trade balance, but again the effect is very small. Edwards presents data showing a considerable reduction in tariffs since the early 1990s. He observes, however, that African economies began from a high level and that their tariffs are still high relative to countries elsewhere in the world. He notes that other studies of trade liberalization in Africa have provided evidence of a wors- ening trade balance, which can lead to an external constraint on growth. Export per- formance in Sub-Saharan Africa showed no growth in the 1990s, and although exports have grown sharply since 2000, much of the increase is attributable to the boom in commodity prices. Edwards does, however, find some improvement in competitiveness. Aaditya Mattoo and Arvind Subramanian discuss what may be a new phenome- non: "uphill skills flows"--the export of skills, embodied in highly sophisticated goods and services or foreign direct investment, from poor to wealthy countries. The authors provide evidence of a marked increase in uphill skills flows between 1991 and 2005, including such recent notable examples as the takeover of the U.K. auto- motive firm Jaguar by the Indian car maker Tata, Brazil's export of aircraft, and India's export of services to developed countries. The authors provide evidence that outflows of foreign direct investment (FDI) to countries richer than the country of origin increased from 45 percent in 2003 to more than 70 percent in 2007. They find a similar pattern in the export of sophisticated, skill-intensive goods: the share rose from about 1 percent in 1980 to 10 percent in 2006. Uphill flows from China, Malaysia, and Mexico were a big part of this increase. 6 | JUSTIN YIFU LIN AND BORIS PLESKOVIC The authors posit some tentative hypotheses that might explain uphill skills flows. Certain countries--such as India and South Africa--may for historical reasons be fol- lowing atypical development patterns that emphasize technical capabilities and give rise to pockets of skill-intensive industry. It is also possible that some firms, even in developing countries, are so productive that they succeed in entering markets in developed countries. And, trade barriers may favor uphill skills flows, as rich coun- tries typically raise barriers against low-skill products rather than skill-intensive products. Migration, Remittances, and the Transition from Foreign Aid Jean-Paul Azam and Ruxanda Berlinschi argue that there is a trade-off between for- eign aid and migration. It can be demonstrated empirically, they assert, that rich countries in fact use aid to reduce immigration from poor countries. On the basis of a review of the literature on the economic effects of immigration, Azam and Berlinschi conclude that the effect on wages in the receiving country, whether positive or negative, is small but that remittances from migrants are now one of the main sources of external finance for developing countries. The authors point to a sizable constituency in rich countries that pressures governments to raise barri- ers against immigration. They then raise the questions of whether foreign aid reduces immigration and whether that is an unstated aim of foreign aid. It has been demonstrated again and again in recent years, Azam and Berlinschi claim, that foreign aid rarely succeeds in reducing poverty or fostering growth in poor countries. The authors ask what benefits, then, bilateral donors actually get from foreign aid and suggest that reduced immigration may be one. Applying econo- metric methods to data from 1995­2003, they conclude that foreign aid is an effec- tive tool for reducing migration to rich countries and that donors are in fact using aid as a policy tool for that purpose. Their further analysis shows that uncoordinated bilateral aid is just as effective as coordinated multilateral aid. Lisa Chauvet, Flore Gubert, and Sandrine Mesplé-Somps analyze the impact of remittances and foreign aid on infant and child mortality. Despite the increasing mag- nitude of remittances (in 36 of 153 developing countries, remittances exceed official aid), there has been little examination of their relative effects on human development, as measured by infant and child mortality. The authors' analysis of panel data on 109 developing countries and cross-country quintile-level data on 47 developing countries shows that remittances have a sig- nificant positive effect on children's health but that the effect is stronger in the richest households. The researchers discern no effect on the poorest households. When migration by physicians is included in the overall analysis of the costs and benefits of migration, the positive effects are weakened because the migration of physicians harms the overall state of children's health. The researchers also show that foreign aid to the health sector has little effect on health outcomes in devel- oping countries. INTRODUCTION | 7 g Jinu Koola and Ça lar Özden examine the effect of migration on employment in the sending country. In their view, previous studies showing that migration decreases employment in sending countries--possibly because those who stay behind receive remittances and have less incentive to work--do not fully recognize the dynamics of migration. In fact, they argue, if workers have the opportunity to work in wealthier countries, their opportunity cost simply becomes higher. Consequently, if they do not make a high enough salary on the local market, they stop working and prepare to emigrate. Working with panel data from two linked household surveys conducted in the Indian state of Kerala in 1998 and 2003, the authors present evidence that the drop in employment in the country of origin is more likely ascribable to the opportuni- ties provided by migration networks that have been established in receiving coun- tries by the migrants' families and communities. In 2003, according to the data, nearly 10 percent of Kerala's workforce--1.84 million people--was living and working in Persian Gulf countries; another million had already emigrated and had returned. Analysis of the panel data shows that social and communal networks, which can facilitate migration and lower its cost, are among the main determinants of migration. Higher Education and High-Technology Industry Sachi Hatakenaka examines the role that universities have played in the development of high-technology industries in a broad range of economies--the United States; Japan; Finland; Taiwan, China; the Republic of Korea; Ireland; India; Israel; and China--and draws lessons for higher education systems in other countries. In all these cases, the production of a critical mass of scientists and engineers was a pre- requisite for the birth of high-technology industry, but it happened in different ways. Whereas some economies produced generic scientists and engineers, others provided specialized training with an emphasis on a narrower set of practical skills. Hatakenaka develops an analytical framework differentiating higher education institutions along three dimensions: responsiveness to changing practical and indus- trial needs; the degree of commitment to fundamental science; and the level of selec- tivity in recruiting students and staff (in other words, whether the institution is an elite one or is, instead, open to the broader society). Although institutions of all types have facilitated the development of high- technology industry, "responsive" institutions appear to play a much more proactive and direct role in helping such industry emerge and evolve, through a variety of modalities--education, research, spin-offs, science parks, licensing, and enrichment through international experience. Such institutions, Hatakenaka observes, do not seem to emerge naturally. In all cases, governments have played a critical role in founding them and in influencing their missions and orientation. Hatakenaka comments that in many cases it was crucial that science and technol- ogy had a champion or champions on the national stage. Yet, she adds, in the nine 8 | JUSTIN YIFU LIN AND BORIS PLESKOVIC economies under discussion, it has proved difficult to maintain a national commit- ment to science and technology over time. Akilagpa Sawyerr and Boubakar Barry explore the relationship between knowl- edge production and economic development in the context of a strikingly different set of countries: those of Sub-Saharan Africa (excluding South Africa). Because the Sub-Saharan context differs so markedly from that of countries such as those repre- sented in the Hatakenaka study, Sawyerr and Barry focus on small and medium-size industry. Indeed, the authors remark, the region has little industry that is truly high technology. Sawyerr and Barry look at the supply of knowledge generated by African universi- ties, as well as the demand for this knowledge by industry, and enumerate weaknesses in both supply and demand. The authors observe that now, as in colonial times, African industry is dominated by low-level processing of natural resources and the production of simple consumer goods. This sort of activity, unlike high technology industry, does not feel keen pressure for new knowledge and therefore does not demand industry-relevant research from African institutions. In spite of a few com- mendable attempts to build bridges between universities and industry, enterprises show little awareness of the importance of science and technology to competitiveness. Moreover, on the supply side, say the authors, African universities are oriented away from science and technology and have little understanding of what industry needs. African universities do not offer proper postgraduate training in technical fields and are certainly not graduating the numbers of PhDs required for high- technology industries to take off. Generally, Sawyerr and Barry assert, universities in Sub-Saharan Africa have suffered for many years from neglect and lack of funding. Clearly, the universities are not up to the task of lifting African industry to a globally competitive level. Sawyerr and Barry emphasize the need for a supportive public policy framework to strengthen both the supply and demand sides. They recommend that each country establish an "observatory," involving industry, government, universities, and tech- nology institutions, to assist in understanding the issues and in building consensus. A second recommendation is to revitalize and strengthen Africa's universities. This process should include a special program for the strengthening of staff quality, as well as research and graduate study in carefully selected priority areas. To give the entire process the necessary weight and visibility, it must be championed at the highest political levels. Human Development Duncan Thomas comments on the close relationship between socioeconomic status and health. Poverty has been correlated with poor health again and again in studies around the world, but what to make of this correlation is still a matter of dispute. Thomas asserts that causality probably runs in both directions: poverty causes poor health, and poor health causes poverty. But, he notes, there may also be unobserved underlying factors that affect both health and socioeconomic status. INTRODUCTION | 9 Thomas cites several recent studies showing that early childhood nutrition has a major impact on human capital formation (cognitive development, schooling, employment, and so on) and on economic productivity throughout life, beginning with school attendance. The evidence in these studies is drawn from various histori- cal periods and from different parts of the world: contemporary rural life in Zimbabwe and Guatemala; the Spanish flu pandemic in the United States in 1918­20; malnutrition in the Netherlands during World War II; the Chinese famine of 1959­1961; and the 1974 floods in Bangladesh. Although the impact of disease is not as well researched as that of nutrition, it is nevertheless clear that successful treat- ment of infectious diseases is likely to have very significant positive externalities--for example, its effect on school attendance--and is in many cases inexpensive. Markus Goldstein, Joshua Graff Zivin, and Harsha Thirumurthy analyze house- hold survey data from several African countries to develop a fuller picture of the eco- nomic effects of a particular health treatment--antiretroviral therapy for adults infected with HIV. A number of studies provide compelling evidence of the efficacy of antiretroviral therapy, which has been delivered on a much larger scale in Africa in recent years. Households with one or more adults suffering from HIV/AIDS are likely to with- draw their children from school, reduce food consumption, and increase children's labor, among other effects. Once the parent dies, the children are, of course, in an even worse position. The authors' study clearly documents short-run benefits to the households of patients receiving antiretroviral treatment, including increased work hours, improved nutrition, and better school attendance. The research team demonstrates that the benefits from increased labor productivity alone outweigh the cost of treatment and that households in which the health of an HIV-stricken adult is improved have a much better chance of coping with poverty. According to the authors, only about 30 percent of the people in Sub-Saharan Africa who need this sort of treatment get it, which means that about 4 million persons are falling between the cracks. The study provides an added rationale for scaling up treatment programs; when viewed as investments, they offer long-term economic returns to society. The authors observe that although the short-term effects of the therapy are large and positive, much less is known about the long-term effects. Still, it is safe to say that antiretroviral therapy can be a critical tool for avoiding large intergenerational economic effects from HIV/AIDS. Political Economy Using panel data from the World Bank (548 country-year estimates from 100 coun- tries) and the International Monetary Fund's Government Finance Statistics, Francisco Rodríguez creates a new indicator for evaluating how the relation between poverty reduction and growth is affected by government spending and the extent to which a given country's fiscal policy favors the poor. Rodríguez begins by asking what makes growth "inclusive." To answer this ques- tion, he grapples with two sets of issues in political economy: the conditions under 10 | JUSTIN YIFU LIN AND BORIS PLESKOVIC which the poor can contribute to economic growth, and the conditions under which they can benefit from growth. He emphasizes the importance of evaluating not only the effect of public spending levels on the poor but also how these effects are linked to the economy's growth rate. To tackle these issues, Rodríguez develops a new measure derived directly from the effect of various fiscal policies on the level of poverty. Rather than rely on a priori theoretical assumptions, he constructs an empirically grounded estimate of how state actions affect the poor under varying economic conditions. Rodríguez's calculations demonstrate a highly significant interaction between expenditures and changes in income, implying that government spending under con- ditions of growth will reduce poverty more quickly and make shared growth more likely. His empirical work shows that the optimal mix of policies will depend on a country's level of income and that certain policies which protect the poor under cer- tain conditions may harm them if those conditions change. Lakshmi Iyer examines the quality of public service provision in South Asian countries and provides a useful survey of the theoretical and empirical literature related to collective action and provision of public goods. Iyer observes that school enrollment is low in developing countries generally and is lower still in South Asia. She notes, however, that rates vary widely among the countries of South Asia and even within countries: for instance, nearly every village in the Indian state of Kerala has a school, but only 39 percent of villages in Bihar have one. Moreover, educational quality varies widely across countries and within countries. Iyer cites statistics showing that on any given day 15 percent of teachers are absent in India's Maharashtra State, as are 42 percent in Jharkhand State. Parents everywhere want their children to be educated, but not all groups are effective in demanding public services. Iyer provides empirical evidence that this is the case in South Asia: for example, access to education and other public services is bet- ter in areas with a higher population of Brahmins (the elite of India's caste system). Among the factors that appear to limit the ability of local groups in South Asia to effectively demand public services are social heterogeneity, unequal land distribution, and small group size. Iyer's empirical analysis demonstrates that social heterogeneity, whether related to ethnicity, religion, or caste, makes groups in South Asia less able to demand adequate public services. Moreover, she contends, South Asia exhibits many of the same colonial patterns as in other parts of the world: areas with highly unequal land distribution tend to suffer from poor provision of public services. Finally, Iyer calls attention to the remarkable rise in the number of private schools in rural areas of South Asia and asks whether private provision could be a solution to problems with quality of services. Although private schools perform better, Iyer nevertheless gives three reasons why the state must remain the provider of health and education services: (a) private schools are, so far, limited to larger and wealthier vil- lages and do not provide universal coverage; (b) they are more expensive than pub- lic schools, and many poorer households already say they cannot afford to send their children even to public schools; and (c) a study from Pakistan suggests that private schools cannot function without teachers educated by local public schools. Opening Address TREVOR MANUEL The theme of this conference, "People, Politics, and Globalization," is very appro- priate. You may have seen this morning's newspaper headline, "City Refugees in Suicide Bid." Let me sketch for you the context. Three weeks ago, South Africa witnessed the horrid spectacle of poor-on-poor violence, when groups of young people in some cities turned on their neighbors, who happened to be foreign nation- als. In Cape Town, fortunately, no deaths resulted, but around 10,000 people--this is a disputed number--were displaced from their homes. A group of Somalis that met with the United Nations on Saturday insisted that although they had been offered refugee status in South Africa, they would prefer to receive asylum somewhere in the Northern Hemisphere. And because the United Nations could not support them in this way, some of them chose to take the suicide route yesterday, by plunging into the sea. It is a terrible story, but it speaks to the heart of our theme. The government has been resolute in admitting foreign nationals into South Africa as part of our commitment to African development. Many people have been displaced; Somalis have been displaced because of the failed state in their country. And the challenges of globalization are everywhere with us. Just over six years ago, we gathered at Monterrey, Mexico, to agree on a partnership that we fervently believed would alter the course of economic development into the future. African leaders endorsed this theme and introduced a program for continental transforma- tion: the New Partnership for Africa's Development (NEPAD). Acting on these commitments, many countries moved toward more stable macroeconomic outcomes, improved their capacity to deliver services, and announced pro-growth, pro-poor economic and regulatory policies. These achievements have given many countries their first opportunity to reap the benefits of both tough economic policies and globalization. At the time of the conference, Trevor Manuel was South Africa's Minister of Finance. He is now head of the National Planning Commission. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank 11 12 | TREVOR MANUEL There have been significant research initiatives, such as the Commission on Global Public Goods and, just recently, the Commission for Growth and Development. And I am pleased to see with us the chair of the Commission for Growth and Development, Professor Michael Spence, and the deputy chair, Danny Leipziger. The commission has undertaken a thorough analysis of high-growth patterns between 1950 and 2005 in an attempt to understand the features of such growth, the measure of interaction between countries, and the impact on the living standards of people in high-growth countries. Among the commission's observations is the reality of growing income disparities across the world occasioned by technology change, shifting relative prices, and globalization itself. The commission report draws attention to a finding of the 2007 Pew Global Attitudes Project survey: that support for globalization is flagging, especially among citizens of developed and some developing countries. The questions before this conference have to do with how to sustain the momen- tum of growth and ensure that the benefits of growth are far more inclusive than they have been to date. Emphasis should also be placed on that tangible set of global com- mitments that measure progress against want, the United Nations Millennium Devel- opment Goals. Our discussions are framed by the reality of rapidly rising prices for food and fuel and persistent difficulties in the financial sector. These three Fs result in a convergence that threatens to roll back many of the recent gains with respect to each of the dimensions of our theme: people, politics, and globalization. Although this conference is not intended to conclude with a declaration, we know that the shared observations that will emerge are keenly awaited. This year will see the United Nations advance further in its discussions of food security under the Millennium Development Goals and of financing for development. None of us can afford to ignore the harsh realities that threaten to erode the gains that we have, until now, taken for granted. The moment calls for us to dig deep within ourselves and advance a new, rational set of ideas to be pursued by thinkers and policy makers everywhere. Allow me now to introduce the brand-new chief economist of the World Bank, Justin Lin, who took up his position on June 2, 2008, and who will share with us his observations on the challenges before us. Opening Address JUSTIN YIFU LIN Mr. President, Minister Manuel, Professor Spence, ladies and gentlemen: good morning. As Mr. Manuel just mentioned, I am the new chief economist of the World Bank. In effect, this is my first public lecture as chief economist. At the outset, I would like to express my sincere thanks to the government of South Africa for its hospitality and its great contribution to the conference today. I also want to take this occasion to thank the organizers of this conference. You bring a wide range of inter- esting topics to each of our discussions. I am pleased to welcome to the conference today the 950 participants from 70 countries. Some of them are from academia and are going to give us their insights about the process of development. But, equally important, we have many participants from the policy sectors, the private sector, and civil society. Your participation will enrich our discussions. This is the first time that the Annual Bank Conference on Development Economics has been held in Africa. We know that Africa is at a very important stage of devel- opment. In the past decade 11 countries, representing more than half the population of Africa, enjoyed an average annual growth rate of 5.5 percent. That is the best performance since the 1970s. And, a more important number, 7 of those 11 countries are not oil exporters. This shows that Africa, like other places, is a land of hope. Certainly, it is our obligation to find a way to sustain this growth. The title of this conference is "Politics, People, and Globalization." From my point of view, these are the three most important dimensions of economic development because people constitute the purpose of development. It is our dream to have a world free of poverty. It is also our goal to give people the freedom to choose, the opportu- nity for prosperity. But people are not only the end of development; they are also the means, because the world's wealth is created by people, especially people with new, good ideas. Justin Yifu Lin is Chief Economist and Senior Vice President for Development Economics of the World Bank. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank 13 14 | JUSTIN YIFU LIN People alone, however, cannot create wealth. They need to have the opportunity to work, the incentive to work, and the ability to work. Whether they will have the opportunities, incentives, and abilities very much depends on the government. Mistakes can be made. In a developing country with little capital, a lot of resources, and a large population, if the government adopts a mistaken policy of encouraging the development of very capital-intensive sectors, even though its intentions are good, not enough jobs will be created for the people. Moreover, firms in those capital- intensive sectors are not going to be viable, leading to government protection, and finally generating distortions that hurt peoples' incentive to work. In that kind of situation, the best of intentions may lead to failed policies and bad results. Mistakes can be avoided. If the government adopts a policy of building up market institutions to facilitate the country in exploring its comparative advantage it can set the stage for economic development and job creation and enable people to share the benefits of development. Even then, the government has to provide education and health services so that people will have the abilities to participate. What I like to stress is the importance of the role of the government in the process of development. As Arthur Lewis observed, looking at history, a successful country always has a very intelligent government. The government can fail by doing too much or too little, and that depends greatly on the political process in the country. So, we also need to look at politics. In our modern world, a country cannot isolate itself. A country has a much better opportunity to prosper if it integrates itself into the world economy through the globalization process. This is demonstrated by the successful East Asian economies and by China after its transition in 1979. Certainly, globalization opens opportuni- ties, but it also poses a new kind of challenge. These include the food crisis that we observe today and the financial turmoil that we see in the United States. In this kind of situation, a country also needs to manage the globalization process, and the global community needs to cooperate to find a new framework for integrating developing countries into the globalization process. I am sure that the discussion in the coming two and a half days will enhance our insight on this important topic, contribute to our understanding of the nature and process of development, and enable us to approach the goal of a world without poverty. Opening Address THABO MBEKI Dr. Lin, Trevor Manuel, Professor Spence, distinguished delegates, and ladies and gentlemen: I am indeed honored to welcome you to the Annual Bank Conference on Development Economics, and we are very proud to host you here in Cape Town, in South Africa, and, indeed, in Africa for the first time. I would like to think that you have among you many South African and African development economists who are engaged in a very important subject: the challenges of development on our continent. The overarching theme of the conference is "People, Politics, and Globalization," but it is fair that what will be discussed under that topic are the three challenges of globalization, investment, and growth; human development for equitable growth; and the political economy of shared growth. These are interesting subjects. I come among you today as one of those much-maligned human animals described as politicians. Some in this audience will, perhaps subconsciously, see in the talking head currently standing at this podium an example of what in the United States came to be known as carpetbaggers. Whatever might be the truth in this regard, I would like to say that my government and I are indeed intensely interested in the outcomes of this important conference. The simple reason is that the matters on the agenda of the conference--globalization, investment, economic growth, human development, equitable growth, and shared growth--are matters of deep interest to all Africans, and I dare say, even to whoever might fit the contemporary edition of the peculiarly African carpetbagger. This morning I was reading a recent article by Jan-Peter Olters, who is described as a World Bank representative in Montenegro. You would never think that the World Bank would have a representative in Montenegro, but it has. Among other things, Olters wrote, The recognition of both globalization's inherent potential and the accompanying risk has become the starting point in the ongoing policy dialogue between national governments At the time of the conference, Thabo Mbeki was president of South Africa. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank 15 16 | THABO MBEKI and international financial organizations. In Robert Zoellick's words: "It is the vision of the World Bank Group to contribute to an inclusive and sustainable globalization--to overcome poverty, enhance growth with care for the environment, and create individual opportunity and hope." The World Bank's emphasis on social inclusion--apart from reasons valid in themselves--stems from global experiences that social tension and large income inequalities lead to lower rates of potential growth, weaken political cohesion, contribute to environmental degradation, and add considerable costs to societies in terms of forgone opportunities. The principal challenge of economic policy making thus consists of increasing the overall productivity of invested capital and employed labor with the instruments that governments have at their disposal: public institutions, laws, regulations, and mechanisms ensuring their rules-based application.1 As I read this, I remembered my own education in economics at an English university more than 40 years ago, when the economics faculty sought to drill into our heads a sound understanding of development economics. I recall that we learned to be supremely skeptical of the teachings of such economists as Peter Bauer and Milton Friedman, who, as I remember, were presented to us as what some today would char- acterize as "market fundamentalists," opposed to the very notion of development economics. What was happening then and later was captured subsequently by an African public intellectual, Thandika Mkandawire, at the time director of the United Nations Research Institute for Social Development (UNRISD): For two decades, starting from the beginning of the mid-1970s, the status of develop- ment economics in both academia and policy circles was not enviable. . . . The "pioneers" of development economics were forced into a defensive posture as they fended off accusations of providing the intellectual scaffolding for dirigisme, which had failed, as well as of downplaying the role of the market. The "death" of development economics was not merely an academic "paradigm shift." It was given official sanction by the United States government. The U.S. representative to the Asian Development Bank was reported . . . to have announced that the "United States completely rejects the idea that there is such a thing as development economics." . . . Development economics became, as John Toye remarks, "an Orwellian un-thing" in the eyes of the most powerful nation. The Spartan certainty of the ascendant neoliberalism as to what was required left no room for specialized knowledge of the problems of devel- opment. Mrs. Thatcher's strident "there is no alternative" was echoed in international financial organizations through a standardized set of policies that was applicable to all economies.2 With regard to the Peter Bauer of my student years, Wikipedia has this to say: Bauer revolutionized thinking about the determinants of economic advance. Indeed, the World Bank, in its 1997 World Development Report, reflected the point of view Bauer had been advocating for years, stating that the notion that "good advisers and technical experts would formulate good policies, which good governments would then implement for the good of society" was outdated: "The institutional assumptions implicit in this world view were, as we all realize today, too simplistic. . . . Governments embarked on fanciful schemes. Private investors, lacking confidence in public policies or in the stead- fastness of leaders, held back. Powerful rulers acted arbitrarily. Corruption became endemic. Development failed, and poverty endured." OPENING ADDRESS | 17 For Bauer, the essence of development was the expansion of individual choices, and the role of the state to protect life, liberty, and property so that individuals can pursue their own goals and desires. Limited government, not central planning, was his mantra. Bauer placed himself firmly in the tradition of the great classical liberals. I must presume that this conference has convened here in Cape Town because you have made the determination that the proclamation about the death of development economics, including the apostolic pronouncements in the 1997 World Development Report, amount to nothing more than an opportunistic advertisement by a commer- cial funeral undertaker driven by the objective to maximize his or her profit, as would any self-respecting carpetbagger. At the same time, it may be that you consider the fact that I have raised the ques- tions I have about development economics as being somewhat arcane or archaic. Let me explain myself. As I have already said, I have to earn my keep as an African politician. Almost by definition, and especially because I represent desperately poor communities that absolutely cannot lift themselves out of poverty without the assistance of the rich countries of our universe, I have an obligation to implement the advice of those with- out whose support my people cannot achieve progress and the necessary advance- ment toward meeting the celebrated Millennium Development Goals. The advice I get, which I must accept, is conveyed by well-funded and immensely educated civil society organizations and a very vocal media, which together serve as the vox populi and therefore the vox dei. The message is very simple and straight- forward: Long live Peter Bauer! To celebrate Peter Bauer without this being stated explicitly, which in any case would make no sense except to the helpless and trapped cognoscenti, we are told that we must limit state intervention in the economy and expand individual choice as part of the process of the great flowering of open democratic systems and the attendant and resultant putative exponential growth and development of the economy. We are told that, more broadly, we must aim to build a minimalist state that should focus on providing such public goods as the protection of life, liberty, property, and the environment, leaving all else to the market, except to the extent that the state must intervene as a regulator to correct the imperfect functioning of the market. We are also told that we must create maximum space for domestic and interna- tional entrepreneurs to invest and make profits, understanding that this will release the immanent national energies that will create the wealth required to achieve the objectives stated by World Bank president Robert Zoellick: "to overcome poverty, enhance growth with care for the environment, and create individual opportunity and hope." We are also told that we must trust and follow the advice we get from good advis- ers and technical experts, which, as a good government, we would then implement for the good of society, provided that what the advisers and experts advise is consis- tent with the preceding prescripts. 18 | THABO MBEKI We are also told that we must always bear in mind that, given the fact of global- ization, we will fail to attract the foreign direct investment we desperately need-- especially given that we are too poor to generate the investment capital needed to achieve the required rates of growth--unless we abide by the rules set by the inter- national capital markets and recognize that we are competing with other possible investment destinations. We are told also that we must take into account the fact that the overwhelming bulk of investable capital in the world is privately owned. Foreign investment for growth and development will therefore not come from official development assis- tance funds but from private investors whose central goal is not understanding unique national public imperatives but identifying profitable business opportunities. But above all, we must take into account the fundamental demands of the global economy: privatize, deregulate, open up to free trade. I am confident that the development economists and other participants present here today understand very well that the prescripts I have mentioned do not fully address the overall theme and subthemes of this conference on "People, Politics, and Globalization." The point I am making, however, is that the dominant, immediate, and material voice that bears on the African politician, including the talking head that is standing at this podium, is the voice that proclaims, insistently, Long live Peter Bauer! There are some in organized global human society who strive to achieve preemi- nence as authentic voices of the people by claiming that, without fear or favor, they present truth to power. In many instances, many of them fail to understand that the power against which they pose as heroes and heroines is little more than a subsidiary formation in a global power system whose pinnacle funds others who pride them- selves on the claim of representing so-called civil society and enables them to devote their considerable energies to a fight that targets shadows and the powerless. I am convinced that gathered in this hall today, at the very southern end of the African continent, are thinking human beings who will help us better understand what we need to do to liberate ourselves from poverty and underdevelopment, refus- ing to be influenced by propaganda and supposedly universal truths, which in many instances are illusions born of smoke and mirrors and intimidation. Given the sphere of human activity in which I am inevitably and necessarily involved every day, I cannot avoid the conclusion that we are involved in titanic bat- tles on two broad fronts of an epoch-making war. One of these battles pursues the objective of winning material and, therefore, objective short- and medium-term vic- tories on the broad front of the struggle against poverty and underdevelopment in the countries of the South and globally. The goal of the other struggle is to win a critically important subjective and pop- ular ideological and political contest, victory in which would enable all humanity to deploy the enormous human and material resources demonstrably available within global human society and so ensure the achievement of the historic objectives of development economics, as broadly defined by Nobel laureate Amartya Kumar Sen. With regard to this latter battle, I firmly believe that contemporary human society disposes of sufficient intellectual, capital, scientific, technological, innovative, and OPENING ADDRESS | 19 vocational skills to close the major fracture in global human society. By that, I mean poverty in the midst of plenty and a process of globalization that emphasizes and entrenches wealth inequalities rather than universal progress toward the goal of a more equal and prosperous human society. As an African, and given the agenda you have set yourselves, I would like to suggest that you have convened here at the Cape Town International Conference Centre to consider what should be done to secure success on both the war fronts we have identified. It is for this reason that I have said that my government and I are intensely interested in the outcomes of the conference. To have any meaning, development economics must relate not to the logical integrity of theoretical paradigms, important as this might be, but to the central task of achieving human development. Since life does not stand still to allow for philoso- phers to contemplate reality, this conference, even within the context of its agenda, will, I hope, consider a variety of matters that are of major importance to the peoples of Africa and other developing communities elsewhere in the world. As you know, there are some immediate and critically important challenges we face as a country and continent, and there are questions you must help us answer. One is: what interventions should we undertake to respond to the high and rising prices of food and fuel? Given the unavoidable inflationary impact of this price increase, which will inevitably reduce standards of living, in particular, of the poor, what immediate, medium-term, and long-term measures, including agricultural policies, should we institute to guarantee long-term and affordable food security? What contingency measures should we take to adapt to the consequences of climate change? How much reliance should we place on the prospects for implementing the Comprehensive African Agricultural Development Programme (CAADP) and, thus, insulating Africa from the threat of food shortages and unaffordable food prices? Will the outcomes of the recent Global Summit of the Food and Agriculture Organization (FAO) in Rome help address these issues? How long will the commodity boom last? How should we, as Africans, take advantage of this boom to guarantee ourselves sustainable development even when commodity prices decline? What practical measures should be instituted globally to integrate Africa further into the world economy, other than as a producer and exporter of raw materials? How can we attract investments that will enable Africa to export greater volumes of manufactured, value-added products? What measures should we take if the global economy experiences a significant slowdown that negatively affects African exports and investment flows into Africa? All of us recognize the critical importance of building the necessary human resource base to drive the process of sustainable African development to which we are all committed. We must therefore pose the question: what should be done to achieve this objective? We also accept that, as can easily be demonstrated with regard to other regions of the world since the end of World War II and earlier, Africa needs the support of the developed world to achieve take-off. What should be done to strengthen this 20 | THABO MBEKI development partnership and overcome irrational and persistent Afro-pessimism? What should be done to implement the now universally accepted vision and pro- grams of the New Partnership for Africa's Development (NEPAD), which is funda- mental to the realization of the Millennium Development Goals as they relate to Africa? We have all observed the recent exciting growth trends in Africa. For some, this may appear to be the result of the passing effects of a commodity price boom. These observers may expect African economies to slow and run into difficulty at the end of this boom, much as many countries, including South Africa, did after the boom of the 1960s and early 1970s. But I am certain that those of us who have looked more closely at Africa's development have seen evidence that the current opportunity to benefit from the commodity boom will not be frittered away, as it was before, at least not by all countries. There are various pieces of evidence that the current growth acceleration in Africa will be sustained by a number of countries, and perhaps by enough to ensure that the continental momentum is maintained. I am certain of the capacity of the conference delegates to think independently and to carry out their own investigations relatively uninfluenced by the mass media, enabling them to understand the reality that the African continent is involved in a historical structural process focused on its sustained and progressive political, eco- nomic, and social transformation. Necessarily, development economics must be an integral part of this process. I trust that the fact that the ABCDE Conference is meet- ing in Africa for the first time will inspire all the participants to take it as their spe- cial obligation to intervene in the African development process so as to add impetus to our continental drive to end our condition as the wretched of the earth. In the article that I cited earlier, Olters wrote, "The World Bank's emphasis on social inclusion--apart from reasons valid in themselves--stems from global experi- ences that social tension and large income inequalities lead to lower rates of poten- tial growth, weaken political cohesion, contribute to environmental degradation, and add considerable costs to societies in terms of forgone opportunities." You will recognize that I quoted this earlier. The African, and perhaps global, development challenge is about eradicating poverty, lessening income inequalities and other inequalities, strengthening social inclusion and political cohesion, reducing environmental degradation, and improving the capacity of individuals and society to take advantage of all opportunities to achieve development. In his 1998 Nobel lecture, "The Possibility of Social Choice," Amartya Sen said: If there is a central question that can be seen as the motivating issue that inspires social choice theory, it is this: how can it be possible to arrive at cogent aggregative judgments about the society (for example, about "social welfare," or "the public interest," or "aggregate poverty"), given the diversity of preferences, concerns, and predicaments of the different individuals within the society? How can we find any rational basis for making such aggregative judgments as "the society prefers this to that," or, "the society should choose this over that," or "this is socially right"? Is reasonable choice at all possible, especially since, as Horace noted a long time ago, there may be "as many preferences as there are people"?3 This learned paragraph from a treatise by a Nobel laureate seeks to communicate the message that it is possible, in the celestial world of pure intellectual discourse, to posit OPENING ADDRESS | 21 a circumstance of coterminous expression of billions of different individual thoughts about the same thing, equal in number to all living human beings. At a certain level, especially within the context of an extreme solipsistic view, this is, of course, a logi- cal possibility, but practically, within the objective world of social existence, it con- stitutes an impossible proposition. Nevertheless, it offers to all of us gathered here the possibility of advancing an entirely theoretical paradigm: that we have no intellectual obligation to take any position on any of the important matters on the agenda of the conference because there are as many preferences as there are people, and therefore, there is no logical possibility of making any rational policy proposals as a conference. The immediate reality, however, is that all of us, whatever our social circum- stances, know that the poor are knocking at the gate. If this gate does not open because we who have the key are involved in the challenging effort to consider the meaning and implications of social choice theory, among other intellectual pursuits, the masses will break down the gate. They will do this to challenge us to join them practically to answer the question: what should be done to give effect to the human dignity that is due to those whom the modern social order, in all countries, does indeed define as the wretched of the earth? I believe that is the fundamental question this Annual Bank Conference on Devel- opment Economics must strive to answer. I wish you success in your deliberations, and I formally declare this important conference open. Thank you very much. Notes 1. Jan-Peter Olters, "On the Agenda: Inclusive Globalization." World Bank News and Broadcast. http://go.worldbank.org/6OX5CZ3RN0. Originally published in Monitor (19 [911]: 32­33) April 4, 2008 as "Tema dana: globalizacija na korist svih." 2. Thandika Mkandawire, "The Need to Rethink Development Economics." Draft paper prepared for discussion at the United Nations Research Institute for Social Development (UNRISD) meeting on "The Need to Rethink Development Economics," September 7­8, 2001, Cape Town, South Africa. http://www.unrisd.org/80256B3C005BCCF9/ (httpPublications)/CE9095BA4A739828C1256BC90047D402?OpenDocument. 3. Amartya Sen, "The Possibility of Social Choice." Nobel lecture, December 8, 1998. http:// nobelprize.org/nobel_prizes/economics/laureates/1998/sen-lecture.pdf. Keynote Address MICHAEL SPENCE Good morning, ladies and gentlemen. It is an honor for me to be here at this con- ference. I first have to apologize for my appearance. My bag and I parted company some time yesterday. There is a saying that clothes make the man. If that is true, then my name is Roberto Zagha--at least for half of my body, the better-dressed half. What I would like to do today is to take a relatively high-speed tour of the report of the Commission on Growth and Development entitled The Growth Report: Strate- gies for Sustained Growth and Inclusive Development (CGD 2008). We have with us today the vice chairman, Danny Leipziger; the secretary, Roberto Zagha; and Trevor Manuel, South Africa's minister of finance and an important member of the commis- sion. We have been working for just over two years. The commission members are, you will note, predominantly political or policy people, very senior, and very experi- enced in developing countries. There are a couple of exceptions. I am one, and Robert Solow is another. But the members' experience was intended to be an important part of the work of the commission. We focused on what we called inclusive sustained high growth. We asked two things. How does sustained high growth work; that is, what are the fundamental dynamics? What are the policies, investments, and political underpinnings that enable this kind of growth? And we defined "sustained" to mean "over several decades." We used a benchmark of 25 years or more, and we arbitrarily picked a 7 percent growth because output or income doubles every 10 years at that rate. I will talk more about the important concept of inclusiveness later. Our primary target audience is basically political and policy leaders in developing countries. Our hope when we started this process was that we would be able to pro- vide some useful insights and perhaps a framework that would help in setting policy priorities and developing growth strategies in the specific country context. Now, Michael Spence is a senior fellow at the Hoover Institution and is the Philip Knight Professor Emeritus at Stanford University, both in Stanford, CA. In 2001, he was awarded the Nobel prize for economics. He is currently chairman of the Commission on Growth and Development. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank 23 24 | MICHAEL SPENCE nobody really cares about growth as such; my family doesn't get up in the morning and think, growth is what really matters in the world. But growth does seem to be an important enabler of things people really do care about. Bassma Kodmani, in her keynote address, spoke about the importance of poverty reduction and the achieve- ment of the Millennium Development Goals, and there are other very basic things that people care about--their families' health, and their capacity to be productively employed or to be creative, innovative, and entrepreneurial. Things like that are fun- damental reasons for being interested in growth as an instrument. We went about this in a fairly straightforward way. We asked our colleagues in institutions like the World Bank and in academic institutions to try, in a whole variety of policy areas, to give us an assessment of what we do and do not know. We discov- ered that there are many areas in which we lack the complete knowledge that would enable us to give confident policy advice, and we decided to talk about that. We held 12 workshops, with many members of the commission participating, and learned from them. It was a fascinating experience. We prepared a set of 25 country case stud- ies, and we produced a relatively short commission report (CGD 2008) highlighting the essential features of growth dynamics and the key policy ingredients that seem to underpin it. We are also publishing working papers and workshop proceedings that are in-depth attempts to understand various policy areas that relate directly to growth. We started out intending to be practical and nonideological; none of us wanted to debate some historical divide. We went into the undertaking with considerable humil- ity, with the understanding that our knowledge of growth is not complete and that our knowledge of development--that parallel process whereby societies and economies acquire capabilities that were not there before--is even less complete. We realized that in the course of growing, countries and institutions are learning from experience and adapting; it is a process and not just the application of a static formula. The report is therefore not a set of policy prescriptions; it is a frame of reference that attempts to get at two elements: the growth dynamics, and the leadership and policies and politics that support them. We believe very firmly that actual growth strategies have to be set at the country level and have to be context specific. We therefore looked at countries that had actually experienced sustained high growth. Only 13 economies are in that category, and those growth episodes have all occurred post­World War II, since 1950. The countries are rather diverse in size, location, and form of government. We believe that India and Vietnam are probably pretty close to achieving sustained high growth because of growth acceleration; it is a matter of time. And there may be many more. I think that the hope in this room is that the accelerations in growth, on a much broader front, that we have witnessed in the last 5 to 10 years may actually turn into sustained high growth. Characteristics of Sustained-High-Growth Economies We attempted to summarize what we thought were the common characteristics of sustained high growth. The first set consists of the fundamental economics, and the second set has more to do with politics and government. KEYNOTE ADDRESS | 25 The single most important shared characteristic of the successful economies (and, of course, there are many, many differences among them) is that they are engaged with the global economy. They took advantage of its knowledge, and they benefited from the catch-up effect, the increase in both the size and the scope of potential out- put, that comes from not having to develop all that knowledge but being able to import it. Second, the successful countries took advantage of the huge elastic global demand that allows a country to grow about as fast as investments can be made and areas in which the country has competitive advantage can be found. The successful countries all maintained their macroeconomic stability. Almost every time I am interviewed, I am asked whether the report is, or is intended as, an attack on the Washington con- sensus. And the answer is, no. There are some differences in philosophy and spirit, but there is no question that a stable macroeconomic environment and a number of other items in the Washington consensus are critical for sustaining the kind of invest- ment that supports high growth. The third obvious characteristic is that these economies had very high levels of saving and investment--on the order of 25 percent of gross domestic product (GDP) or more for overall saving and investment and 5 to 7 percent for the public sector component. They are also--and I think it is important to focus on the microeconomic aspects here--market economies that make use of price signals, decentralization, and incentives and in which reasonable definitions of property exist and determine resource allocation. They are also economies in the process of structural transformation, which is a chaotic process--what Schumpeter called creative destruction. Therefore, there is a great deal of competition and a lot of incremental productive employment creation. Later on in the process, there is a fair amount of destruction as well, as the sectors that are the driving forces of growth become uncompetitive because of changes in incomes and relative prices. These high-growth economies are characterized by resource mobility, especially labor mobility, and by rapid urbanization. A fair amount of the report consists of just trying to describe these trends. There are policies and investments (education, for example) that support resource mobility and policies (such as limiting competition) that interfere with it. The report talks about both. I have learned a tremendous amount from my colleagues on the commission and from my fellow academics in the course of these two years. But what perhaps sur- prised me most was how important are leadership, effective political governance, and effective government. They make it possible to implement as well as develop policies. Political leadership is absolutely crucial. Difficult choices have to be made. One concerns the general approach to growth. A second has to do with a kind of consensus- building process. At saving and investment rates of 25 percent of GDP and more, very significant intertemporal choices are being made about when consumption is going to occur. That requires a credible game plan which people can sign up for, and which they will sign up for if they believe it will benefit their children and grandchildren. Achiev- ing some degree of consensus and support for growth-oriented strategies is a nontrivial part of the process, not only of starting but also of sustaining growth. 26 | MICHAEL SPENCE We do not have complete knowledge of the impact of our policies. We need to exper- iment. If something is not working well, we need to stop doing it, fairly quickly. I sometimes use the metaphor of navigating with an incomplete map. The full transi- tion from relatively poor to advanced-country income levels takes a long time (over half a century) even at high growth rates and much longer when things slow down. Underpinnings of Growth Without trying to summarize the report, let me give you some quick assessments of some of the things we thought were important underpinnings of sustained high growth. Even though everybody thinks knowledge and the catch-up effect are important, we know less about them than we should. This is an area in which productive research could be undertaken. We do know that channels exist which are important. One is for- eign direct investment. Foreign direct investment is not usually a large fraction of total investment. Its importance does not come mainly from relaxing a constraint on invest- ment but, primarily, from the importation of productive knowledge, technology, and know-how, including knowledge of the global supply chains, that accompany the investment. There are other channels--foreign education, training, experience, and participation in a network--which are also vitally important in advanced countries that are growing by virtue of collectively generating new knowledge. On the demand side, basically from a developing-country point of view, the global market is essentially limitless. On the supply side, in the early stages of growth you have surplus labor, meaning labor that is somewhat underemployed in traditional sec- tors. This combination produces a period in which you have what a theorist would call a linear growth model. You don't turn the prices against you, and you don't turn the supply, the price of labor, against you by drawing it away from the traditional sec- tors. Another way to think about this is that the opportunity cost in the early-stage high-growth mode of using resources in high-growth sectors is relatively low. Leadership is crucial. The model or strategy has to be basically right. There are examples of very well-intentioned leadership that simply picked the wrong approach--for example, a closed-economy model or an import-substitution model. But communication is also terribly important. We are often asked what form of government is most conducive to growth. Our answer is that it is not the form of government or governance that is crucial for growth. The high-growth countries exhibit a considerable variety of forms of gov- ernment. What does seem to matter, and what they do seem to have in common, is a political leadership that has as one of its main objectives making the lives of pretty much everybody in the country better off. Alternatively, if you look at countries that fail or exhibit poor economic performance, it is very frequently because the govern- ment is representing either its own interests or those of some subgroup. That finding led us to believe in the crucial importance of inclusiveness. Structural transformation and competition are, I think, pretty well understood, but a fair amount of policy in both the developing world and the developed world KEYNOTE ADDRESS | 27 interferes with this kind of dynamics. There have been intellectual rationales for such policies. For example, in relatively small countries the argument would be that the country has to restrict competition in order to take advantage of scale economies. That sounds plausible, but it is a static argument, and when you look empirically, the dynamic effect of competition and of entry and exit on productivity growth is so big that it simply overwhelms the static efficiency gains. So, structural transformation that allows competition to work is important. And then the question is, what about people whose jobs may disappear? The answer is that ways have to be found to pro- tect people in the course of transitions, but without doing so by protecting the com- panies they work for or the jobs that they are currently in. That is difficult but not impossible to do. Inclusiveness is essential. Inclusiveness for us has three dimensions. One is income equality and reasonable amounts of equity. People understand that markets do not produce equitable outcomes; that is not what drives a market system. And they will accept that up to a point. But they apparently will not accept excessive income inequality. People also will not accept inequality of opportunity. Unlike ex post income equality, where people will tolerate some degree of inequality, equality of opportunity is essential, and its absence is highly toxic. Systematic exclusion on the basis of class, income, religion, ethnicity, tribal affiliation, or gender will lead to polit- ical conflict and distraction or worse. The coherence of policy and the sustained commitment that are needed for growth will be lost. It is not strictly economics that disrupts or derails the growth process. Finally, as I mentioned before, you need to protect people through the transitions, or they end up being sideswiped by the micro- economic dynamics. At a high growth rate, that happens a lot, and it is not a minor concern. I don't believe there is a great deal of controversy about high saving and investment rates. What is interesting is how difficult it is to actually achieve those rates, especially on the public sector side. The pattern of government consumption's crowding out investment on the public sector side is widespread in the developing world, and so it deserves to be a focus. In China and India, saving and investment rates were high throughout the transformation. The rates in China were fairly high right at the start of the reforms, in 1980; it was not a case of rates rising once people had enough income. Health is something people care about, quite independent of any effects on growth or productivity. But one channel seemed to us to be strikingly important for growth and equity, and that was early childhood nutrition and stimulation. Experts told us that a failure in this area produces a nearly permanent reduction in children's ability to acquire cognitive and noncognitive skills in school. That struck us as a very long-term problem for growth, and a deeply unfair situation that, if widespread, would mate- rially diminish the accumulation of human capital and the growth potential of the society. There is not much controversy about the importance of education, but we did focus on some issues in this area. One is that education is normally measured by years of schooling or by enrollments. That is not crazy, but years of schooling is an input, and measuring just that is not enough. According to the experts, research 28 | MICHAEL SPENCE appears to suggest that there is extremely high variance in the output--the actual acquisition of cognitive skills and education. And so a very high priority, and not an easy task, is the improvement of the quality of education in many countries. Another issue: What do the primary, secondary, and tertiary portfolios look like at various stages of growth? Do you focus on primary first, then secondary, then tertiary? The answer appears to be, no, or only partly. That is to say you need all three, although the relative size of the investment in each component will shift as incomes rise. Generally, the share of investment going to secondary and tertiary will rise with incomes. But the key point is that the starting point is not zero. Having said that, the time path of the shares in the high-growth cases varies considerably and that seems not to impair growth. The portfolio needs to make sense and not go to the extremes, but it does not have to be optimized. Labor markets are very complicated in developing countries, with formal and informal sectors and barriers to mobility of a wide variety of kinds. Labor markets are extremely important because problems there affect the fundamental dynamics that involve movement of labor from less productive to more productive jobs. In a surplus labor environment, the incentives for incumbents in the formal sector do not include enthusiasm for reforms that create extra competition from either the tradi- tional or the other informal sectors. It seems to us that it is therefore often politically very difficult to reform, in a simple way, the labor market so as to expand access. And therefore, we suggest experimenting with a separate track, a different route for people who are, in many countries, trapped or cut off from the growing and more modern sectors. Sometimes that trapping occurs because of underinvestment in human capital and other factors, so I don't want to suggest that labor market rigid- ity is the only problem. But it is a problem that merits creative policy discussion because it is so pervasive. Everybody knows the world is now 50 percent urban. Urbanization is the physi- cal process that accompanies the structural transformation of the economy. It is chaotic, and it sometimes has ugly sides that may make people want to resist it. But resisting urbanization is not a good idea from the point of view of growth. And so the question becomes, how can urbanization be made as effective a process as possi- ble, as unchaotic as possible? The main challenge is finding sources of finance for building the urban infrastructure that is required--sewer systems, water systems, and so on--because the normal financing mechanisms that we are used to thinking about are not available in the early stages of growth. For example, municipal bond markets do not exist in early-stage developing countries. The notion of growing first and dealing with the local environment--air and water quality and the like--later struck the commission as a very bad idea. First, that approach sets off a process of capital accumulation that needs to be undone later, and it is very expensive if the wrong houses are built, the wrong industries develop, and so on. It biases the structural evolution of the economy. Second, "grow first" almost always has an adverse distributional effect. That is, the worst effects of degraded environmental conditions are felt by the poor. On energy, there is a very widespread pattern (that does not include South Africa) of subsidization. This is understandable in political-economy terms, but it is a KEYNOTE ADDRESS | 29 very bad strategy that is becoming worse as the cost of energy rises. The commission took a clear stance on energy subsidies, which is that they basically have to go-- notwithstanding the political difficulties and the merits of an argument for maintaining subsidies for the most vulnerable parts of the population. The resources that go into energy subsidies ought to be directed toward remedying underinvestment on the public sector side, in infrastructure and education, for example. And at these energy prices, with all the other attendant problems, energy subsidies are much too costly. Furthermore, it is almost surely true that any component of a successful global program to mitigate climate change and global warming is going to have an energy- efficiency component. Commitment to a sensible approach to global climate change will require that energy subsidies disappear anyway. We tackled things that are really controversial, complex, and unsettled. We didn't shy away from them; decision makers don't get to shy away from them. You either have a market-determined exchange rate in the global capital markets or you are managing it in some way, but it is not a choice you can put off. We tried to approach the controversies, and the divergence of opinions within the commission itself, by talking about the risks and benefits of various kinds of policy actions. I hope we have produced at least a useful guide to decision making. Country Challenges and Transnational Cooperation We focused on four categories of countries that we thought had important and rather specific challenges: Sub-Saharan Africa, small states, resource-rich countries, and middle-income countries. The African countries constituted one group, certainly not because they are similar, but because the continent has a very peculiar configuration and set of characteristics--many people living in landlocked countries, and many countries with large amounts of resource wealth. In Africa, effectively building out infrastructure is, for the most part, not something that can be done by a single country. We focused on small countries, defined as those with 2 million or fewer peo- ple. Small states have many problems, including lack of economic diversification and huge per capita costs of governance. They tend to solve those problems effectively when they pool resources. In resource-rich countries, resource wealth, which should be an asset, has too often turned out to be a liability in political terms. This, too, is something that could be turned around. The middle-to-high-income transition countries are really very different in policy terms, in what drives the economy. A middle-income country is starting to be an economy whose dynamism and growth are driven mainly by capital, human capital, and knowledge-intensive sectors in the economy, including services. A necessary pol- icy shift is to give up the growth engines inherited from the earlier stages of growth. They have to be allowed to decline and disappear, and rising incomes reduce and eliminate the comparative advantage. That turns out to be harder than it sounds. One of the most common mistakes is finding a formula that works and then following it for too long. 30 | MICHAEL SPENCE Trends and Policy Agendas The report ends with some topics that I believe we all think are terribly important-- global trends that affect the opportunities and strategies of developing countries. One is a clear pattern of declining enthusiasm for globalization, as measured by several surveys, including the Pew survey in October 2007. There is a lot of talk about protectionism. People attribute many of the problems they have with income distri- bution to globalization even when it is not the only explanation. But in general, our conclusion was that we have devoted far too little attention to things like income dis- tribution and equity and protecting people during the transition. I don't think this problem is going to go away; in fact, it seems likely that it will get worse. It is not a closely guarded secret that the rising price of energy is, in part, driven by rapidly increasing global demand. And global demand is rising rapidly because of developing-country demand. The energy price issue is one that requires domestic policy attention in a wide range of countries, including the advanced ones. China and India have grown quickly, and China, in particular, is big enough to have actually changed the relative price of manufactured goods. There is a wide- spread question in the developing world: is the old growth strategy, based on labor- intensive, optimal use of the labor asset, going to work? Some people think that it will not. The commission concluded that if the relative price of manufactured goods has gone down, perhaps the social return to the old strategy will not be as high, but that doesn't necessarily invalidate it. In the 1980s, Bill Cline and others raised what is sometimes called the adding-up question: if everybody follows the same strategy of focusing on comparative advan- tage in the labor-intensive sectors in the early stages of growth, can the global econ- omy absorb all that product? That question has been reviewed by Bill and others (see Cline 2008), and the general feeling is that it is less of a problem than people thought. But there are new versions of the question that I will come to in a minute. Clearly--and nobody really anticipated this very effectively--we are in a period in which the relative prices of commodities in general are changing very quickly. That has caused all kinds of problems. Underlying this is a real change in the world, and it has to do with growth, not just numbers of people. The population of advanced and rapidly growing economies was about a billion 30 years ago; it is 4 billion now, and the number is rising. So, it is going to be a different world, with very different levels of demand for energy and much greater pressure on the environment. On the food price front, the report observes that there is an emergency response problem for poor people in a wide range of countries. That has to take priority. Assuming that protecting those people from what amounts to a large effective loss of income can be accomplished fairly quickly, the balkanization that has occurred in the food markets through export restrictions and the like has to be undone. The reason is that we need the high prices and the openness of the global market to induce a very large supply response in the case of food--as in the case of energy. Nothing that I have heard or read or that the commission was aware of suggests that we are any- where near our productive potential with respect to food. So I think if this problem KEYNOTE ADDRESS | 31 is managed correctly, the supply elasticity is big enough in the long run that the prob- lem actually might go away. That is, we could go back to a situation where the rela- tive price of food is not so high. In the meantime, we have the food emergency and a very difficult inflation problem all over the world, and it is worse in developing countries because the poorer a country is, the larger is the fraction of GDP going to food (and, I might add, energy). On the energy front it is a little different. Energy prices, in my judgment, are unlikely to go back to anything like the levels we saw before. (I didn't want to imply that food prices will definitely return to the old levels, but I think they could easily go down.) The elasticity of energy supply is limited, so we are going to have to bet on a large demand elasticity, drawing on energy conservation and alternative energy sources. This may be the ultimate adding-up question in the global economy. If demand does not respond to high prices, global growth will eventually slow, and the opportunities for developing countries will be substantially reduced as a result. The commission believes it would be unwise to think of what we are seeing now in terms of high price volatility as a one-time event. We are going into a world in which we will be facing relative price volatility for some time. An important part of a new growth strategy is going to have to be appropriate risk-mitigation strategies. I will skip over demographics and aging except to note that the world (including many parts of the developing world) is getting older but the poorer parts of the devel- oping world are getting younger. And so there are two problems: Is global growth going to slow because of the aging effect? And what is happening in the younger and poorer parts of the world? Our answer is that global growth does not necessarily have to slow down because of the aging, although policies and behavior will have to adapt to sustain growth. But in the low-growth parts of the global economy, there is a huge mismatch between where labor is available and where it is in demand. That is sometimes called the youth unemployment problem, and the commission suggests that there is no logical solution except for labor to move from excess-supply envi- ronments to high-demand locations. The implication is that we probably ought to spend some time thinking carefully about migration patterns that actually work, meaning that people are properly protected, and so on. Many of you have thought a great deal about global warming. World carbon diox- ide emissions are about 4.8 tons per capita per year. The scientists tell us that a safe per capita output would be about 2.3 tons per year. In energy-inefficient countries, the United States and Canada are at 20.6 tons per person per year. Europe with greater energy efficiency is in the 6-to-10-ton range. France is low because it has a lot of nuclear energy. China and the rest of the developing world are essentially at or below the current estimated "safe" level. But their total emissions are large because of the sizes of their populations. If growth in the global economy just stopped, we would have to reduce emissions globally by a factor of 2. But with growth in the developing world, global per capita emissions will double over the next 50 years in the absence of mitigation. So, how do we mitigate global warming and at the same time accommodate the growth of the developing world? The report begins to address this problem by suggesting ways of getting started. But a lot more work is 32 | MICHAEL SPENCE needed to create a shared understanding of the paths we collectively need to follow to achieve safe levels of CO2 emissions globally several decades out. This shared understanding will have to include differentiated roles for advanced and developing countries if it is going to achieve broad acceptance in the developing world. The last section of the report deals with global governance. In brief, it says that we are increasingly interdependent in multiple dimensions, with large implications for potential financial market turmoil, infectious disease, product safety, energy prices, demand and growth, and global warming. In all of these areas there are global agendas to deal with. It is a major challenge for the next generation. It is going to be difficult; nobody has a blueprint; and in the meantime there are new and rising risks in the global economy that accompany this interdependence and that need to be addressed with a view to development strategy. And with that I will stop. Thank you for your attention. References CGD (Commission on Growth and Development). 2008. The Growth Report: Strategies for Sustained Growth and Inclusive Development. Washington, DC: Commission on Growth and Development. http://www.growthcommission.org/index.php?option=com_content& task=view&id=96&Itemid=169. Cline, William. 2008. "Exports of Manufactures and Economic Growth: The Fallacy of Composition Revisited." Working Paper 36, Commission on Growth and Development, Washington, DC. Keynote Address BASSMA KODMANI I have chosen to speak about the challenges involved in emerging from authoritari- anism in the Middle East, while maintaining security. My perspective will be one that starts from the current reality (a grim situation, constituting a regression from what seemed to be a more hopeful picture two years ago) and goes on to focus on real achievements that appear difficult to reverse and that are moving the Arab region into a new era, provided these progressive trends are protected. The last four years have seen a convergence of serious, long-term dangers: · The Arab-Israeli conflict exploded once more in the summer of 2006. A disaster bordering on genocide is taking place in Darfur. The war in Iraq is causing huge anxiety across the region, particularly among Iraq's neighbors. · Water resources are being rapidly depleted. This situation carries the seeds of con- flict not only between states but also within countries, as the distribution of water becomes a major new source of inequality between rich and poor. · Among the world's regions, the Middle East is the most dependent on imports of food staples. It buys one-quarter of all cereals traded globally. The fact that the United Nations Development Programme (UNDP) has taken human security as the central theme of its recent Arab Human Development Report is an indication of these mounting and converging threats. Yet, at the same time, most countries in the region are witnessing high economic growth rates. And during the past five years, the key word in the mouth of Arab leaders has been "reform." There is much ambiguity and misunderstanding around that word. Retrospec- tively, we can consider the ambiguity constructive because it has allowed govern- ments to integrate the concept into their discourse and to engage in important changes--administrative modernization, the introduction of new legislation, the Bassma Kodmani is executive director of the Arab Reform Initiative, a network of independent research and policy institutes. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank 33 34 | BASSMA KODMANI rationalization of some practices, and a host of measures that governments felt would yield benefits for economic growth, the country's image abroad, and foreign invest- ments; satisfy outside partners; and renew their own control over society. Nor did "reform," as construed, require commitment by leaders to yield any of their prerog- atives or to relate to their societies in a different way. The motto of reform and good governance allowed governments and "societies" to travel some way along the road toward introducing change. But in the past two years there has been a general feeling that we are stumbling over a set of serious obsta- cles that we are endeavoring to identify and analyze. Were these impediments inevitable? Were they the results of circumstances, or of specific flaws in the approach? Some of the obstacles are traceable to external factors resulting from old and new sources of insecurity over which we may have limited control. But the current stale- mate is largely attributable to a reluctance on the part of governments and of some key outside players to formulate the objectives more boldly and to opt clearly for shifting some amount of power to societies. We should not have been surprised by the resistance of old authoritarian political systems or by their resilience and capacity for adjustment, which seem to assure the ruling elites perpetuity. I believe that the situation is ripe for formulating objectives more frankly. We need to shift to a new paradigm that states explicitly the objective of democratic transfor- mation. This does not require agreement on the exact outcome in terms of the model of democracy desirable in each national context. The definition I want to use here is based on very basic principles. From there, we can outline a common framework and develop the practical tools that derive from it. The key principles are--in addition to building strong and accountable public institutions--respect for rights and freedoms, the rule of law, and equity and social justice. These are not new, but the emphasis here is on the interconnectedness between them. I will provide some practical considerations for making this framework operational. To begin with, I discuss a major change that has occurred: the awakening of Arab societies. The timid progress that has been made was followed, however, by setbacks. I explore the reasons for these setbacks--fears within societies, and the top-down, government-to-government institutional approach that has been espoused so far in order to foster change. Finally, I argue that there can be no sustainable reforms with- out democracy. Although many believe this to be true, the debate has focused more on the definition of democracy than on which strategies to adopt, which actors to empower, and which institutions to promote. The Awakening of Societies A remarkable change has been the opening up of public space and the emergence of independent media. The new media--satellite television channels and, more impor- tant, local media, the Internet, and blogs--play a key role in establishing horizontal links among citizens. Such links had already existed in the social sphere (traditional KEYNOTE ADDRESS | 35 civil society is strong, although not identified as such by international agencies), but not in the political sphere. Independent media are now creating those links, which are among the first conditions for political mobilization. Societies are learning to challenge governments: they go on strike even when the law forbids strikes, and they organize demonstrations and sit-ins to protest price increases, lack of water, or loss of land. The protests are initiated by two types of actor: the poor and marginalized and the elites who give priority to political demands--constitutional reforms, free and fair elections, the abrogation of emergency laws, freedom of expression, the independence of the judiciary, and so on. Faced with such protests, governments are reluctant to suppress the move- ments, which they know are now covered by local and international television cameras. There are two very positive aspects about these movements. First, the expressions of discontent and the mobilization crystallize around social and political issues and much less around issues of identity and religion. Second, these movements, whether grassroots or elite based, are calling for peaceful, gradual change through legal means and a process controlled by the state. Governments are trying to close down these new spaces for expression and mobi- lization, but the protest movements are not receding. In fact, they are increasing with the deteriorating social conditions. In the face of the unbridled rise in prices of vital products, governments have come up with short-term solutions, at best. The specter of major social and political unrest is looming. The region has to cope with economic deprivation that affects at least 80 percent of the overall Arab population. Of 320 million Arabs, only about 5 to 20 percent at most (or about 50 million) benefited from the second oil boom in the first decade of the twenty-first century. The rest are poor, and their poverty has either increased or has failed to diminish, as has begun to occur in many emerging countries of the global South. In addition to economic deprivation, social and political disenfranchisement is increasing. These increased hardships are fueling the protest movements, but they are not the initial trigger. The reasons for the awakening are structural and will probably be dif- ficult to reverse. Peoples' state of mind is changing. The sense of powerlessness is waning, and it is widely recognized that the change in peoples' minds is a key driver of democratic transition, as has been witnessed over the past 30 years, first in south- ern Europe, then in Latin America and in Eastern and Central Europe. What Societies Fear As unhappy as societies are with their governments, they seek to protect state institutions and the stability of the political system. Scenes of instability in their immediate environment act as a deterrent against fostering abrupt change and shape the attitudes of opposition forces. Radical groups advocating violence find little support within society. In the Arab Republic of Egypt, for example, some jihadist 36 | BASSMA KODMANI groups that had used violence in the past have repented and have conducted public self-criticism. It is remarkable that political protest has remained peaceful so far, even though openings for legal protest are tightly restricted. The more we advance on the path of liberal economic development without explicit social contracts, the more vital do alternative networks of solidarity become for ever-larger numbers of vulnerable groups. These community-based support sys- tems that help meet people's basic needs and avert social explosions are almost all faith based--Muslim (Sunni or Shia), but also Christian. I have deliberately refrained from specifying the Islamist movements as a source of fear because Islamists are not seen as a danger by the large majority of the popu- lation. Islamists are not viewed as alien; they are a product of the social fabric, and they embody the values of the people. They have pursued a three-decade-long trajec- tory of building social networks and developing a political culture based on religious messages. They have evolved into political actors that are now knocking on the door of the state and demanding their share of power. When they do not get a response in the form of a political opening, they may be tempted to force their way in to conquer their share of power. But this outcome has been the exception rather than the rule; it is directly related to a context of conflict in which the movement is militarized and is leading a resistance movement. (Examples are Hezbollah in Lebanon and Hamas in Palestine, which turned their arms inward.) In other contexts, Islamists are joining multiparty coalitions, taking an active role in parliaments, and negotiating portfolios in government. This is true in Bahrain, the Arab Republic of Egypt, Jordan, Kuwait, Morocco, the Syrian Arab Republic, Tunisia, and the Republic of Yemen. A much more serious source of fear for governments and societies alike is the threat to the stability and integrity of states as national entities that emanates from conflict situations in neighboring countries and from violent extremist groups (for example, transnational jihadists networks connected to Al Qaeda). The Iraqi situa- tion is the most recent and the most serious source of insecurity. Countries to the south of Iraq (Saudi Arabia and the smaller Gulf states) fear a Shia state in the south of the country, while countries to the north (Syria and Turkey) fear the emergence of a Kurdish state in northern Iraq. Despite such fears, we have not seen any serious schemes by Arab governments to address the issue of sectarian diversity within their own boundaries. Saudi Arabia has only recently announced a plan for the development of its northeastern province, where the Shia are concentrated. The budget is US$300 million--an amount that is unlikely to change the conditions of the inhabitants in any significant way or convince them that their government cares to make them equal citizens, like the rest of the pop- ulation. More tragic is the behavior of Arab governments toward the crisis in Darfur and their protective attitude vis-ŕ-vis the Sudanese government responsible for the atrocities. Arab governments have not acknowledged that sectarian coexistence is an integral part of maintaining human security, although we can clearly see that it is pre- cisely on such issues that human security and national security intersect. Political authorities are not regulating intercommunal relations so as to build harmonious relations between communities, and societies are not equipped or KEYNOTE ADDRESS | 37 empowered to address the problem. Other institutions within society--mainly reli- gious institutions, Christian and Muslim alike--often take on this role, for which they are intrinsically ill-equipped. Unsustainability of Governance Reforms without Democracy The only way out of this intricate predicament, I would argue, lies in the approach toward domestic change. We need to shift from the objective of reforms in gover- nance to a more straightforward paradigm of democratic transformation as a way to open new options and expand the arsenal of instruments that can be mobilized. Acknowledging this will open the way for a holistic approach to change in the Arab world. There is a danger of losing a sense of the big questions that the numerous technical and institutional programs and measures collectively try to answer. With careful attention to strategies and modes of operation, we need to embark on this conceptual shift toward democratic transformation as an achievement of societies, without diluting it. Only when sharp contours are defined can we derive effective strategies. But promoting democratic transformation is first and foremost about opening a space in which citizens can develop a home-grown agenda and about accepting the fact that societies will develop their own creative paths of transition. It may well be that international organizations and outside donors will have as much to learn from Arab experiences (once success is achieved) as they have to offer in terms of experi- ences with patterns of democratization. For outside players to contribute in a con- structive way to such processes, a number of suggestions can be made concerning the activities to develop, the actors to involve, and the strategies to promote: · Devise mechanisms to link measures for reforming public institutions from above with efforts to respond to demands by societies, particularly respect for rights and freedoms and for equity and social justice. · Focus on systems of governance, on processes, and on the quality of the relation- ship between the ruling elite and society, rather than basing assessments on bench- marks that governments have learned to meet without changing the reality of their relation to their citizens. · Support the establishment of monitoring tools at all stages of the process, from the drive toward and the desire for participation in decision making through the actual decision-making process, the implementation of policies, the design of guarantees for their implementation, and the assessment of their impact. · Engage in focused research to identify areas in which the performance of public institutions can be changed through policy measures undertaken by governments and outside partner governments or institutions, as opposed to changes that can only occur through the participation of societal forces acting from below. · Assist social groups to develop monitoring tools to track variations (negative or positive) in the status of democracy, with a view to intervening in the formulation of policies. A key area for initial efforts in this direction is to encourage groups 38 | BASSMA KODMANI within societies to push their governments for access to information as a right. Legislation (freedom of information acts) and sustained pressure from citizen groups will gradually erode the culture of secrecy within most government institutions. Practical Examples Two concrete examples of an approach from below are described next. Security Sector Reform Security sector reform in the Arab world has been on the agenda of many foreign donors and multilateral institutions for the last three to five years. Yet societies are often unaware of such programs and are developing strategies of their own that remain largely disconnected from the approach taken by outside donors. The challenges in developing an indigenous agenda for security sector reform by foreign partners are threefold: 1. The current approach is strongly influenced by post-conflict situations in other parts of the world or in the Middle East itself (Palestine, Iraq, Lebanon) and is ill suited to countries that are undergoing comprehensive reform processes but are not emerging from a conflict. 2. The approach is largely technical, even though the area is one in which political considerations dominate because security institutions form the backbone of the political regimes and determine complex domestic equilibriums and, ultimately, the survival of the political systems. 3. The approach is driven from above and relies on the goodwill of governments that have learned to take what is offered by foreign partners (for example, training pro- grams in human rights for their police forces) without linking these programs to an overall ethical code of conduct for their security institutions. Meanwhile, societies are developing their own ways of pressuring governments con- cerning the behavior of security forces. In Morocco, human rights groups and groups advocating for the rule of law have seized the opportunity opened by the monarchy through the creation of the Equity and Reconciliation Commission to push for a public debate on the practices of the security institutions. The results are remarkable, as these groups have succeeded in maintaining their momentum and in opening sensitive files that have remained totally secret for several decades. In Egypt, where the government did not initiate anything from above comparable to the Moroccan opening, a public debate on the practices of security agencies has been forced on the government thanks to independent media (newspapers and television channels), new technologies (documentation of abuses of detainees' human rights through mobile phones, blogs, and Web sites), and the work of a few KEYNOTE ADDRESS | 39 filmmakers who have produced films that have been widely seen and discussed across the country. The debate is now open and ongoing. It is exposing the security sector and creating resentment within its ranks at being put in the forefront of the govern- ment's struggle against citizens, and it is forcing the government to investigate some of the most embarrassing cases of human rights violations. Although this is a process of trial and error, societies are gradually developing strategies that amount to the beginning of what can be called a home-grown agenda for reforming the most opaque institutions of authoritarianism. Gender Programs Gender is arguably the area that has received the greatest attention and the most generous funding from foreign partners. Yet most programs have been focused on a small group of women in each country, often from the elite and connected with or co-opted by the first ladies. Although these programs have the blessing of the governments, they are not always successful in fostering significant change, since they rarely challenge the social order. Experience shows that female academics from the elite and scholars in general have rarely proposed alternative, more socially grounded visions of gender issues distinct from those offered by Western institutions. The case of the Khul'a law enacted in 2002 in Egypt is one of the most telling examples of a successful home-grown strategy. The legislation was the result of a quiet 15-year struggle by a group of widely respected female and male jurists, soci- ologists, and civil society representatives who worked closely with religious scholars from the conservative Islamic institution Al Azhar to reach agreement on the inter- pretation of an old principle enshrined in Islamic sacred texts. The law amounted to a silent revolution in Egyptian society, as it allowed tens of thousands of women to obtain divorces and custody of their children, after decades of unsuccessful proce- dures in courts. The law was far from ideal, and the process for reaching this out- come was lengthy, but it seemed that this was the price for building a consensus and legitimacy allowing the law to be enforced. It was done by working from within society, integrating social realities and beliefs, working from Islamic sacred texts, and bringing along key actors such as the religious establishment. A second area in need of revisiting within many donor-driven programs on gender is that these programs tend to be designed to eschew political issues, and they often promote certain rights for women specifically without integrating the larger social and political context. Although every woman in the Arab world will agree that her status as a woman needs to be improved and her rights protected, many of the most popular figures among Arab women are those active in the larger struggles for the civil and political rights of citizens. These women are members of political parties or social movements. They have constituencies and credibility to preserve, and they are often reluctant to be associated with donors' strategies, which they see as potential- ly weakening their own strategy--one that seeks to challenge the government in a bolder fashion. Again, we see the need to trust societies to produce their own leaders and strategies. 40 | BASSMA KODMANI Conclusion The heat around the democracy promotion agenda has receded. Now is probably the right time to say that it should not be abandoned; otherwise, foreign partner gov- ernments and organizations run the risk of being perceived as opportunists who were attracted to a fantasy of the former U.S. administration, like a noisy toy that captured their attention for a short while. The moment is probably auspicious for renewing efforts toward democratization. But there are conditions that have to be met if these efforts are to succeed. A key requirement is to trust the societies. Gradual change toward democracy is mainly about creating a different type of relationship between a government and its citizens. Foreign donors need to constantly question their strategies to ensure that they are not increasing governments' distrust of (and sometimes contempt for) their citizens. Accountability is first and foremost to be promoted vis-ŕ-vis the citizenry rather than foreign donors (as the recent debate on the sovereign wealth funds of the Gulf States has shown, for example). Societies might well produce social and political models that foreign partners are not entirely comfortable with. But this is probably inevitable, and accepting the inevitable is the only way to make the best of it so as to tame radical groups and soften anti-foreign attitudes and ideas through constant engagement. Keynote Address SUNIL KANT MUNJAL I am honored to be invited to address this annual conference. I consider myself fortu- nate to be here, for a variety of reasons. South Africa is known as the country where the founding father of the Indian nation, Mahatma Gandhi, took his first public stand. This is also the land of Nelson Mandela, who showed that peaceful protest is as powerful a tool as war for winning an argument or a case. I also count myself fortu- nate to be among so many economists, and all varieties of economists. That reminds me of a story I heard the other day about an architect, a surgeon, and an economist. The surgeon said, "Look, we are the most important profession. God is a surgeon, because the very first thing God did was to extract Eve from Adam's rib." The architect said, "No. God is an architect. He made the world in seven days from complete chaos." And the economist said, "Who do you think created the chaos?" So, you are certainly very important people. I must admit that I never studied to be an economist. I probably should have. I am told that this is the only profession where you can say "trickle-down theory" with a serious face. The theme for this conference seems extremely apt for the world today. Certainly, for us in India it is very relevant. I was, however, surprised to get the invitation to the conference because although India is important, my understanding was that this is a conference normally addressed by government functionaries or senior economists. So I shall try not to talk just about industry but to present an overall view. Whatever you hear about India is true--and, conversely, the exact opposite is also true at the same time. India is not a simple country by any standard. It is said that in the course of a lifetime one might just begin to understand what India is about. And that is as true for we who live in India, as well. India is a country that can launch rockets and satellites but at the same time has around 200 million people who live in Sunil Kant Munjal is the chairman of several enterprises, including Hero Corporate Services, and is the president of the Medical College and Hospital, India. He is a member of the Prime Minister's Council of Trade and Industry in India. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank. 41 42 | SUNIL KANT MUNJAL abject poverty. Close to half the population of Mumbai, which is considered India's big financial center, is living in shanties. We have 28 states, 22 languages officially recognized by the government of India, and thousands of dialects spoken across the nation. People in different parts of India wear different clothes, speak different lan- guages, eat different food, and even look different. Very often, for some of us, when we travel across the country, the only common languages are cricket and Indian movies and, of course, every now and then, English. The economic historian Angus Maddison has estimated that five centuries ago India was one of two countries in the world that accounted for nearly half of the world's gross domestic product (GDP). Attracted by India's prosperity, Columbus set out and lost his way and got to America. Who knows, if he had had a good compass, he would have probably found his way to India before the British did, and I might not be here giving this speech today. The British spent a little over 200 years in India, and they ruled India. When they arrived, India had about 21 percent of world trade and Great Britain, as it was then known, had 1 percent. In 1947, when the British left, India had 1 percent of world trade and Great Britain had 21 percent. So you can see what has happened to India. Many talk of this as the rise and fall and now the rise of India again. Our first prime minister, Jawaharlal Nehru, said, "A moment comes which comes but rarely in history when we step out from the old into the new. When an age ends and when the soul of a nation, long suppressed, finds utterance." It did not quite hap- pen after 1947, when India became independent. India chose as its broad philosophy a strange oxymoron, a socialist democracy. We borrowed from the Russian model of the five-year plans, and the government built power plants, steel plants, and dams. That was probably necessary at the time because no private enterprise or individual was allowed to accumulate enough wealth to invest in such projects. But the government probably did not step back quickly enough. The 20 or 30 years of this policy became a very difficult time for private enterprise. In 1991 India made a dramatic move toward market reforms. Our current prime minister, who was then finance minister, in many ways was the leading architect of this reform program. It is popularly believed that India did not go happily toward reform. It went kicking and screaming, but we were in a situation where foreign exchange reserves were only enough for a few days of imports. In fact, India had to send out a planeload of gold as a pledge to be able to raise international currency to continue to pay for its imports. To India's credit, it has never, ever defaulted on any global commitment. I think that is saying something for a country that has gone through many crises. What is not so well known outside is that many of the reform programs that were kicked off in 1991 had actually been discussed, debated, and written up in the 1980s--which is probably why implementation was easier than one would have imagined. Don't forget, it was a dramatic turnaround of policy from what we had had until then. So, what has happened to the simplicities of reform in India? India's trade-to-GDP ratio has doubled over the past seven years, to almost 50 percent. As a trading nation, India is now firmly established at the top table of the World Trade Organization KEYNOTE ADDRESS | 43 (WTO), and trade agreements are currently in the pipeline with at least 10 potential partners, ranging from the Association of Southeast Asian Nations (ASEAN) to the European Union, the United States, and Canada. We already have signed agreements with Singapore, Sri Lanka, Thailand, and a host of other nations. In fact, my last trip here was when India signed what is known as the IBSA agreement between India, Brazil, and South Africa--three nations spread across different parts of the globe that are at a somewhat similar stage of evolution of their economies. All three are large, all three are complex, and all three are trying to reach across to see how we can build bridges between us. India is a keen participant in the WTO discussions (in fact, some say, too keen. But the reality, often not recognized, is that India is a nation in which 72 percent of the people live in rural areas and 54 percent depend on agriculture as their primary source of income. Agriculture contributes less than 18 percent of GDP, so a lot of these people are just subsisting on the farming they do; it is not really trade or business for them. The government therefore finds that it has a responsibility to protect that sec- tion of the population, and it is a large number of people. (Percentages are deceptive sometimes; don't forget, this is a country of more than 1 billion people.) We are keen to build a global world trade order, but in the meantime India, like many other nations, is reaching out rapidly and building bilateral, trilateral, and other trade relationships. Savings are now almost 35 percent of GDP, a jump of 8 percent- age points in less than six years. The investment ratio stands at 36 percent, up from 22 percent in 2004, and many profess that it will rise to 50 percent in the not too distant future. The interesting thing is not just what is happening inside India; it concerns Indians outside India. Boris Pleskovic and I were talking about how he has helped many Slovenians residing outside Slovenia to get together and build a kind of platform. If you look at how Indians have performed across the globe, you will find that among many of Fortune 500 companies, the chief executive officer level or the top tier of management will always have a few Indians. Among the major successful banks, many of the movers and shakers are Indians. In North America the largest single ethnic group of doctors is Indian. Someone sent me an e-mail the other day that referred to NRIs, nonresident Indians. At one time NRIs was said to mean "not required Indians," but that perception is dra- matically changing as India's economic situation improves. I think in many ways India is maturing as a nation, both politically and socially. It is reaching out to its people out- side, trying to build a bridge. The e-mail that I mentioned said that 33 percent of the people working in Microsoft are Indians; the number at Intel is similar. The figure that surprised me was for the U.S. National Aeronautics and Space Agency (NASA), which has about 20 percent Indians, and this is really high-technology work. These individu- als have gone out and done exceedingly well. So the question arises, why was it that 20 years ago Indians were doing well but India was not? It really had to do with the facilitating environment that we had built--or, rather, the nonfacilitating environment that we had built--through regulations. That is what began to change in 1991. The impact is such that I understand the national dish of the United Kingdom is now chicken tikka masala. So it is not just the doctors that we are exporting. 44 | SUNIL KANT MUNJAL India is working overtime to reduce its infrastructure deficit. Some problems per- sist today, but we are better off than we were 10 or 15 years ago. The main problems are lack of physical infrastructure, continued red tape in decision making in govern- ment, and corruption in public life. And it is interesting, when you look around, you find that this is almost a standard feature of emerging economies. The good thing is that in spite of all its diversity, in spite of its fair share of problems and its lack of physical infrastructure, India has continued to build growth at a rate that has gotten better, decade on decade. At one time, India was known as a nation that had what was called, rather uncomplimentarily, the "Hindu growth rate" of 2.5 to 3.5 percent. Every decade since, growth has gone up by at least a couple of percentage points. As of now, we have had four years of nearly 9 percent growth, though recently issued numbers show that growth will be between 7.8 and 8.5 percent. If you are used to 9 percent and you are aiming for 9.5 percent or higher, a growth rate of 8.5 percent can be experienced as a slowdown, and it can actually begin to hurt. Let me tell you a few more success stories, and then I will talk about what is unique about India's journey. An average Indian today can potentially spend double what he or she could spend in 1985. In the next 20 years, he or she will be able to spend four times as much as now. By the end of this decade, India is expected to have an urban population of 173 million--more than the population of several European nations put together. This is significant because urbanization rises with GDP per capita in a hockey stick fashion. And it has been seen in many economies that the moment you start to hit US$700­US$800 per capita income, the income rise is sus- tained and that, fortunately or unfortunately, leads to higher urbanization. There- fore, we need to focus not just on the 600,000 villages to improve rural areas but also on rapidly growing the urbanization that is required. A large section of our workforce has become productive, much more productive, in these last 15 years. Our dependency ratio has decreased from 1.3 to 0.8. This has partially to do with the demographics of the country: 54 percent of all Indians are below age 30, and 30 percent are below age 15. Over the next 20 years or so, as the entire developed world falls short of people in the working-age group by roughly 41 million to 42 million people, India will add almost the same number to the working- age pool. So, there is a dramatic opportunity in the form of more work moving to India and more people moving out of India, but the current state of global politics prohibits that movement in many ways because of increasing concerns over security and nationalism, as we have heard in two of the sessions at this conference. Professor Spence, for example, observed that many nations, including many champions of glob- alization, are turning more nationalistic than they have been for a long time. Mahatma Gandhi once said, "I do not want my house to be walled in on all sides and my windows to be stuffed. I want the cultures of all the lands to be blown about my house freely. But I refuse to be blown off my feet by any." This, in many ways, sums up the uniqueness of India. India opens itself to global influences to look at what is happening in the West and in the East, but it chooses to follow its own unique path. I suppose this is partly because no rulebook has been written for a democracy with a billion people trying to turn into a market economy. The other part is that we are a young, new economy, but we are a 5,000-to-7,000-year-old civilization, with KEYNOTE ADDRESS | 45 strong likes, strong dislikes, and a very strong culture. And this has been seen by many of the global companies that came into India. Many came to India saying, "We have a standard set of products and services that is wonderfully honed to serve all the global markets," and in many cases they fell flat on their faces. They did not allow for the customers' preferences, which were engrained through many generations. McDonald's made a change in their menu for the first time, I am told, when they came to India. And they were so successful with this change that they carried out this practice in many other nations, adapting to local tastes, flavors, and recipes. There is a whole host of examples of like this. The big move that has happened in India is that the pattern of reform is moving from the central government to the state governments. The central government backed off from many of its regular activities by stepping back, by reforming, and by reducing licensing requirements for industry and other economic activities. Many of these changes were readily adapted by state governments, as well, but not by all. In fact, in an interesting recent article an expert observed that if you draw a line approx- imately from the north to the south of India, between two cities he named, the areas to the left and right of the line are like two different countries. The left side, which is mostly the south, the west, and parts of northern India, is where the major growth is taking place. But although the economic numbers are high, literacy numbers are still low, and that is one of the issues that continue to dog India as a nation. There was a wonderful session this morning on higher education, and I know some of the Indian experts who are in this audience addressed that session. At the top end, India has done exceedingly well in education; people who have gone into avion- ics or aeronautics or software engineering or medicine obviously must have received a good education. Yet the numbers are not nearly enough for the size of our popula- tion. Now, this was a role clearly marked out for the government. Governments are required to take care of primary education, primary health care, security, monetary policy, and the like. Nevertheless, the government has not had the wherewithal to provide sufficient education to all the people, so the private sector willy-nilly stepped in and started playing a role. But the private sector cannot substitute for the government. There is a need for the government of the day--whichever it is. In the last 15 years we have had a series of coalition governments, with sometimes as many as 12, 14, or 18 parties in govern- ment together. These governments have been made up of parties from the extreme left to the extreme right and every color in between. The good thing is that all of them went down the same path of reform and of viewing growth as the only answer to all India's ills. The not-so-good thing is that they focused on the easy part of the reform. (I am saying easy now; obviously it didn't look easy when they were doing it.) Some of the work is still to be done. Also, the government did a wonderful job of trying to manage both the revenue deficit and the fiscal deficit, going as far as to write an act of parliament binding itself to improve the fiscal situation. They have done pretty well, barring the current problem that the whole world is seeing with high inflation and high prices for oil, food, and other base materials. The government also backed off from spending in some areas in which they ought to be spending--some of the major infrastructure projects. There are many private players, both Indian and 46 | SUNIL KANT MUNJAL global, who are building airports, ports, highways, roads, power plants, and so on in India. But it is necessary to ensure that the government remains a key player, not only in areas such as infrastructure but also in ensuring that independent regulation is set up which will allow the private sector and the new model of public-private enter- prises to work well together. In a diverse nation like India, consensus is required not just through a spectrum of political parties but also involving all sections of society. We have a very vibrant democracy. We have one of the world's freest presses. We have an active executive, judiciary, and civil society, and awareness among people in India is very high--so much so that one of our recent Nobel laureates, Professor Amartya Sen, wrote a book about it, The Argumentative Indian. It is said that the difference between India and China is that while in China, you make a decision today and start implementa- tion tomorrow, in India you make a decision and the debate starts tomorrow--and that debate could take months or years. That is a reality we live with. The good thing is that we have learned that it is possible even with such diversity to build consensus and to take action through consensus. Sean Casey, the very successful baseball player, said "All the world's a stage. And most of us are desperately unrehearsed." India, clearly, is one of those cases. We are learning by making our own mistakes. I doubt that we can actually learn from any- body else's, or from experts like the World Bank. Very often one hears criticism from within India that the World Bank tries to give everybody a standard prescription. The reality is that in India, a standard prescription is not going to work. And that has been proved over and over again. It is therefore important that all international, multilateral organizations learn to work with India and with Indians, who have built this expertise of consensus. In fact, we could provide expertise to many nations on how to run an election. (We know what has happened in countries like the United States and others.) India has, of course, done a tremendous job in this area. We have elected 2.2 million representa- tives in what we call the panchayati raj--village-level local self-government. Of these representatives, more than 30 percent are women. I do not know of any nation in the world that has come anywhere close to this truly democratic exercise. So, there is a tremendous amount of strength in how India functions as a nation. Europeans, for example, can take a lesson from India in how so much diversity actually works well together and still continues to be a nation. Our big issue is how to tackle poverty. Many people agree that our only answer is growth. Trickle-down growth, however, is not fast enough. People will not wait two generations, especially when they can see others who have become exceedingly suc- cessful, very wealthy. And India is producing many billionaires and millionaires. The poor will have patience only to some extent, which is why some of us, who would normally have objected, actively participated with the government to design four to five direct intervention programs, including programs on skill building, education, rural infrastructure, electrification, drinking water, and irrigation. These are some of the real and serious major issues that we have in India as a nation. What we have done in India over a short period of time appears to be almost mirac- ulous. There are many success stories that one can talk about, in telecommunications, KEYNOTE ADDRESS | 47 for example. From having the highest-priced wireless service in the world, India has gone to having the lowest-priced service. In three years' time India will have 95 percent geographic wireless coverage, which will be a unique distinction across the world. Not many years ago India probably had 2.5 million telephone connections. Now it adds between 8 million and 10 million new connections every month, which is more than the population of many small nations. We have done similar things in innovation. On the plane trip to the conference, I heard someone mentioned Amul, which is the brand name of a cooperative of dairy farmers. Hundreds of thousands of dairy farmers provide milk every day to this organization, which does the processing, the value adding, the distribution, and the branding. India is the world's largest producer of milk and this is our largest organ- ization. It is made up of millions of people. And there are other examples of innova- tion that has taken place in a very Indian way. Jaipur Foot is an organization that makes prosthetics. It is unique in the sense that a foot which in the United States would cost US$8,000 costs US$40 when supplied by Jaipur Foot, and its functional- ity is at least 200­300 percent better than can be found in the best organizations and best institutions in the United States. This is just one example of how India is build- ing a low-cost economy that can function. Of course, we realize that cost and low wages cannot be permanent advantages. Therefore, value-added productivity, effi- ciency, and research are the new watchwords in India. And we are conscious that we need to pursue these goals while ensuring that we do not pollute as much as many nations did while they were undergoing their development. India's capital utility efficiency is fairly high right now. A recent McKinsey & Company report stated that for every US$1,000 spent, India is able to get 20­22 per- cent more than in China on the physical infrastructure that is being built. That is important because we are very short of funding. Of course, unlike the case in many countries, much of the capital for infrastructure is coming from within India. But we do need foreign capital, along with knowledge and networks. That, in fact, is our greatest need today. But India is actively investing overseas, as well. In fact, India is now the second-largest foreign investor in the United Kingdom. We are also invest- ing in countries such as France, Germany, Italy, the United States, and a host of other places. India is uniquely positioned to serve as a lesson for many others who are attempting to dramatically change their economies. We are one of the rare economies that evolved directly from being an agrarian economy to being a service economy. More than 50 percent of India's economy is services--which is why some of us are pushing very hard to grow the manufacturing sector faster so as to be able to provide many more jobs for the millions of young Indians who are entering the job market every year. India has been able to learn much from the rest of the world. But really, the best learning comes from making your own mistakes and at the same time being cautious when you need to be. We carried out a very calibrated opening up of the capital account, for example. Many experts across the world objected to the slow pace. But now some of them are saying, "Hey, hang on here. These guys did not suffer like some of the other economies when they went through an extreme crisis. So maybe there is something here that is being done right." 48 | SUNIL KANT MUNJAL India offers a unique opportunity, to both Indians and the global population, to work in India and with India. We have learned many lessons, and we still have much to learn. But one thing we have learned is that if you have the right attitude, the right frame of mind, you are going to be able to find the right answers. I spoke of Mumbai a little while ago. Mumbai has an area called Dharavi that is probably the world's largest shantytown--certainly Asia's largest shantytown. Just now, while driving to the conference, I saw a shantytown right outside Cape Town. What was unique was that every home had a color television, most of the people living there had cell phones and other electronic gadgets, and most of them also had jobs. The India of the future that we hope to see is not one of shantytowns but of color televisions, jobs, and continuous prosperity, and concern for those around us. Trade and Investment Trade and Economic Performance: Does Africa's Fragmentation Matter? PAUL COLLIER AND ANTHONY J. VENABLES Most of the population of South Asia lives in one megacountry with a population greater than that of the whole of Sub-Saharan Africa. By contrast, the population of Sub-Saharan Africa is spread across more than 50 countries. Does this political frag- mentation have economic consequences? We suggest that both private economic activity and the provision of public goods benefit from powerful scale economies that confer advantages on the South Asian model. Paradoxically, although Africa has a greater need than other regions for supranational power structures, it has made far less progress toward regional unity. Sub-Saharan Africa and South Asia have many features in common and many that are strikingly different. They each contain roughly 1 billion people, and in 1980, their per capita incomes and human development indicator levels were broadly similar. But South Asia is dominated by India, a unified state of 1,130 million people, and Pakistan and Bangladesh, each with more than 150 million. By contrast, Africa, with a total population 60 percent the size of South Asia's, is divided into 54 states. Of these, even the largest, Nigeria, is smaller than the smallest of the big three South Asian countries and has a mere 14 percent of the region's population, whereas India contains 74 percent of the population of South Asia. About 95 percent of South Asians live in the big three, but the three largest countries of Africa account for only 28 percent of Africa's population. Indeed, the average African state has a population of only 17 million people, one-sixty-sixth the size of India's. Does this matter? The economic performance of the two regions has diverged sharply since 1980, as illustrated in figure 1. The regions had similar gross domestic products (GDPs) in 1980, but whereas Africa's share of world GDP declined until the late 1990s, South Paul Collier is director of the Centre for the Study of African Economies (CSAE) and professor of economics, Oxford University. Anthony J. Venables is director of the Oxford Centre for the Analysis of Resource Rich Economies (OxCarre) and professor of economics at Oxford University. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank. 51 52 | PA U L C O L L I E R A N D A N T H O N Y J . V E N A B L E S FIGURE 1. Economic Performance, Sub-Saharan Africa (upper) and South Asia (lower), 1980­2004 3.0 2.5 2.0 percent 1.5 1.0 0.5 0 80 82 84 86 88 90 92 94 96 98 00 02 04 19 19 19 19 19 19 19 19 19 19 20 20 20 3.0 2.5 2.0 percent 1.5 1.0 0.5 0 80 82 84 86 88 90 92 94 96 98 00 02 04 19 19 19 19 19 19 19 19 19 19 20 20 20 % of world GDP % of world exports % of world manufacturers Source: World Development Indicators 2007. Note: GDP, gross domestic product. Asia's share nearly doubled during the same period. Africa's share of world exports was nearly four times that of South Asia in 1980, but--South Asia's share having risen more than twofold and Africa's having been halved in the interim--the pro- portions are now similar. During the same years, Africa's share of world manufac- turing exports stagnated, and South Asia's more than doubled. The Sub-Saharan African data are dominated by South Africa, a single country that accounts for TRADE AND ECONOMIC PERFORMANCE | 53 around 20 percent of regional GDP, and the performance of the rest of Africa is, on average, worse. There are numerous reasons for this divergence, but in this paper we focus on one set of issues: has Africa's fragmentation into numerous small states contributed to its relatively poor performance? The relationship between country size and economic performance has been ana- lyzed by a number of authors, with mixed findings. Recent work by Rose on a sample of 208 economies (some, such as the Isle of Man, not independent) led him to the conclusion that an economy's population "has no significant consistent impact on its well-being" (Rose 2006, 501). By contrast, the growth literature has come up with evi- dence for a positive relationship between country size and growth. A 2005 article by Alesina, Spolaore, and Wacziarg provides a succinct summary of findings. The authors hypothesize that economic growth should be positively associated with country size and openness and negatively associated with the interaction of these two factors. In a sample of 104 countries, they find that these relationships are present and statistically significant. Furthermore, they are quantitatively significant: "for a country at the median level of openness ([the Republic of] Korea) the effect of multiplying the coun- try's size by 10 would be to raise annual growth by 0.33 percentage points" (Alesina, Spolaore, and Wacziarg 2005, 1530; see also Alcalá and Ciccone 2003). A lack of consensus in these cross-country studies is perhaps unsurprising. On vari- ous measures of per capita income, small countries are both at the top (Luxembourg and Lichtenstein) and the bottom (Burundi) of the ranking. Our approach is not to under- take aggregate cross-country studies but, rather, to investigate the impact of fragmenta- tion into separate states in a more microfounded way. We will focus on Africa and will often compare it with South Asia and with India, in particular. Why is it that a frag- mented (sub-) continent might be at a disadvantage relative to a more unified one? Conceptually, three distinct mechanisms generate costs of fragmentation. The first is that natural advantages are likely to be unevenly distributed among countries. The second concerns the loss of scale economies, at the level of the firm, the city, and the country as a whole. The third has to do with the loss of public goods as the scale of political cooperation is reduced. We will examine each in turn. Natural advantages such as oil deposits and natural harbors are unequally dis- tributed across space. If a continent is politically fragmented, the likelihood that natural advantages will be unevenly distributed among countries increases. Evi- dently, this implies inequalities among countries. Of greater pertinence for this inquiry is that these inequalities are likely to imply inefficiencies: average income is reduced by fragmentation. This effect occurs if there are diminishing returns to having a natural advantage, in which case the aggregate benefits from nature would be greater the more equally shared are these advantages. Natural advan- tages may be fixed or transient, and we consider each category. As examples of fixed differences in natural advantage, we apply Collier's (2007) classification of countries into resource rich, resource scarce, coastal, and landlocked. Transient differences arise as countries experience idiosyncratic shocks, with some countries benefiting and others possibly losing. We look at the exposure to export price shocks generated by the structure of commodity exports and the fact that political fragmentation might reduce risk pooling. 54 | PA U L C O L L I E R A N D A N T H O N Y J . V E N A B L E S Whereas the first mechanism that generates fragmentation costs arises from dimin- ishing returns and underlying or "first-nature" unevenness, the second has to do with increasing returns and the losses stemming from the inability of small countries to gain sufficient scale to work efficiently. One context in which this matters is urban- ization and the failure of Africa to develop highly productive urban centers of eco- nomic activity. Another is thick market effects and the failure of many small African economies to achieve the scale to offer predictable economic environments. The economies of scale problem is discussed in the second section, below. The third issue concerns the provision of public goods through politically organ- ized collective action. The free-rider problem is frequently so acute that it can only be overcome by the coercive power of a government to tax its citizens, thereby gen- erating financing for public goods. Fragmentation of a continent into countries is, first and foremost, political fragmentation. As such, it increases the costs of provid- ing public goods, implying that provision will be both less adequate and more expen- sive. This mechanism is the subject of the third section. Uneven Distribution of Natural Endowments The discussion begins with an examination of fixed differences in natural advantages and then turns to transient events such as shocks. Fixed Differences in Natural Advantages Fragmentation of a continent into countries means that geographically concentrated natural endowments such as mineral resources, coasts, and rivers are likely to be unevenly distributed among countries, and so it turns out in Africa. Table 1 shows export earnings from natural resources (as a share of GDP and per capita), together with Collier's (2007) classification of countries. Average resource export revenues per African citizen range from several thousand dollars per capita and more than half of GDP to close to zero in resource-scarce countries, some of which, such as Burundi, Malawi, Rwanda, and Uganda, are land- locked as well. This unequal distribution matters for two reasons. The first is that it maps into an unequal distribution of resource rents per capita. Within a country, resource rents (or at least those that find their way into public accounts) are likely to be spent through- out the country. The spending may not be spatially uniform; producing regions may be favored. (In Nigeria in 2005, federal transfers from oil revenue amounted to US$210 per capita in oil-producing states and US$70 per capita in the northwest.) But the distribution is far wider within countries than across international bound- aries (i.e., zero). The second reason is that it is not only equity that is damaged by unequal distri- bution of resource revenues. Since the economic impact of resource revenues is likely to be subject to diminishing returns, their unequal distribution also leads to an effi- ciency loss. A simple economic model makes the point. Suppose that every country TABLE 1. "First-Nature" Geography in Sub-Saharan Africa: Natural Resource Exports and Country Classification Value of exports of fuels, ores, and metals As percent Per capita Collier (2007) Country (year) of GDP (U.S. dollars) classification Equatorial Guinea (2005) 93.92 14,591 Resource rich Angola (2005) 72.16 1,471 Resource rich Congo, Rep. (2005) 71.46 1,182 Resource rich Gabon (2006) 55.90 4,071 Resource rich Chad (2005) 44.47 258 Landlocked Nigeria (2004) 40.94 214 Resource rich Botswana (2005) 34.74 1,977 Resource rich Guinea (2006) 24.40 88 Resource rich Congo, Dem. Rep. (2006) 24.34 34 Resource rich Mauritania (2005) 19.79 123 Newly resource rich Mozambique (2005) 18.71 62 Newly resource rich Zambia (2005) 18.32 116 Resource rich Sudan (2005) 13.50 102 Newly resource rich Côte d'Ivoire (2005) 12.52 108 Coastal Mali (2004) 10.65 46 Landlocked Cameroon (2005) 9.40 89 Coastal; formerly resource rich Sierra Leone (2005) 9.40 20 Coastal South Africa (2005) 6.81 351 Coastal Zimbabwe (2005) 5.75 15 Landlocked Niger (2005) 4.55 12 Landlocked Kenya (2004) 4.52 21 Coastal Senegal (2005) 4.30 30 Coastal Togo (2004) 3.90 13 Coastal Namibia (2006) 3.11 97 Coastal Ghana (2005) 2.50 12 Coastal Madagascar (2006) 1.70 5 Coastal Central African Republic (2005) 1.62 5 Landlocked Tanzania (2005) 1.58 5 Coastal Cape Verde (2006) 0.87 19 Coastal Swaziland (2002) 0.81 9 Landlocked Uganda (2005) 0.66 2 Landlocked Ethiopia (2004) 0.54 0.7 Landlocked Burundi (2005) 0.34 0.3 Landlocked Rwanda (2005) 0.33 0.8 Landlocked Burkina Faso (2004) 0.31 1.2 Landlocked Mauritius (2005) 0.17 9 Coastal Benin (2005) 0.11 0.6 Coastal Malawi (2005) 0.05 0.1 Landlocked Liberia Coastal Guinea-Bissau (1995) 0.44 0.9 Coastal Gambia, The (2003) 0.05 0.1 Coastal Comoros (1997) 0.03 0.1 Coastal Djibouti (1990) 0.01 0.1 Coastal Source: World Development Indicators 2007. Note: Diamond exports are included in the calculations for Angola, Botswana, the Democratic Republic of Congo, Guinea, and Zimbabwe. Gold exports are included for the Democratic Republic of Congo, Guinea, and Ethiopia. For Mali, export value of fuels, ores, and metals consists solely of gold; for Sierra Leone, it consists solely of diamonds. 55 56 | PA U L C O L L I E R A N D A N T H O N Y J . V E N A B L E S FIGURE 2. Income Loss from Uneven Distribution of Resources GDP foreign exchange labor constrained constrained average income of GDP A and B if merged B average income of A and B if separate A R* resource exports Source: Authors' elaboration. consumes and produces a single good that is nontradable. Production of the good uses foreign exchange (imported oil or equipment) and domestic labor in fixed pro- portions. The only source of foreign exchange is resource revenues, and labor is in fixed supply. Real income in such an economy is illustrated in figure 2, in which resource revenue is measured on the horizontal axis. If resource exports are less than R*, production is foreign exchange constrained, and real income is given by the upward-sloping section of the line (with slope equal to the foreign exchange content per unit of GDP). If natural resource earnings are greater than R*, the economy is labor constrained, thus fixing income; further resource earnings beyond this point are simply accumulated as foreign assets. As a simplest case, suppose that one economy has no resource revenue (i.e., it is at point A) and another has resource revenue and is at point B. The average income of the two countries is at the midpoint between A and B. Merger of the two economies would exactly double the income, as illustrated. This is a very clear-cut example. What insights does it provide into reality? There are two key elements to the argument. The first is that shortage of foreign exchange constrains production in economies that lack resource earnings. Many resource- scarce and landlocked African economies have extremely low shares of exports in GDP--less than 15 percent in eight of these countries. Accessing world markets has been particularly difficult for this group, and they are heavily aid dependent. If they were located within a single country, such areas would earn the resources to finance "imports" through intracountry trade. The argument must therefore turn on the fact that barriers to trade created by international borders are an order of magnitude greater than within-country trade barriers, and there is plenty of evidence that this is the case. The classic studies of the barriers created by international borders, compared with within-country trade costs, are based on trade between Canada and the United TRADE AND ECONOMIC PERFORMANCE | 57 States. McCallum (1995) and Helliwell (1997) show that exports by Canadian provinces to other Canadian provinces are about 20 times larger than their exports to U.S. states at the same distance. According to one study, the U.S.­Canada border is 7,000 miles wide, in the sense that it chokes off trade as much as would 7,000 miles of borderless distance. African borders are generally much more difficult to navigate than is the border between the United States and Canada. Limăo and Venables (2001) find that poor infrastructure is particularly important in choking off trade between African countries. The implication is that resource-scarce, landlocked regions face more acute problems in financing imports as separate nations than they would as regions of a larger country. The second part of the argument contained in figure 2 is that at some point there are diminishing returns to resource earnings. In the example given, the economy reached full employment, and no more labor was available to produce further income. But the argument is more general. What are the sources of diminishing returns to the value of resource revenues in resource-rich African economies? Often, there are constraints on the supply of particular nontradable services, such as con- struction services or specific labor skills. Spending bids up the price of these inputs but does not buy additional real services. More generally, spending from resource revenues will be met by a combination of increased output and crowding out of other expenditures. The expenditures that are crowded out might be exports, giving rise to "Dutch disease."1 Alternatively, monetary and exchange rate policy might be used to mitigate Dutch disease, in which case crowding out will affect domestic activities, quite likely investment. If these activities are particularly valuable (as would be the case if they are initially operating at a suboptimal level), crowding them out may actually reduce income. The key point for the present argument is that the smaller are resource revenues relative to the economy as a whole, the more favorable will be the balance between increased income versus crowding out of other expenditures. If two countries are merged, the supply curve of the merged country is the horizontal sum of the supply curve of each separately, so a given increase in demand will lead to a larger quantity increase and a smaller price increase in the merged economy than in a separate one. This argument focuses on income and expenditure, but other aspects of the "resource curse" (e.g., political-economy issues) may also exhibit increasing marginal cost, possibly meaning that the benefits of resource revenue do not just flatten out with respect to resource revenue but turn negative at the margin. In this case, the citizens of both coun- tries would gain from the merger and a sharing of the economic impact of revenues. It is not just natural resources that are unequally divided between countries; access to the coast is, too. Coastal economies in developing countries are much better placed to engage in producing manufactures for world markets than are landlocked coun- tries, and such activities are important drivers of growth (Jones and Olken 2005; Commission on Growth and Development 2008). These differential opportunities mean, once again, that the fragmentation of a region into separate countries may cre- ate both spatial inequalities and efficiency losses. A simple economic framework is help- ful. Suppose that economies can produce a good in which there are diminishing returns to labor--say, in agriculture, where a fixed supply of land generates diminishing 58 | PA U L C O L L I E R A N D A N T H O N Y J . V E N A B L E S FIGURE 3. Returns to Labor in Coastal (C) and Landlocked (L) Economies W* W* W0 wage in L, WL wage in C, WC labor in C labor in L Source: Authors' elaboration. returns. Figure 3 illustrates two such economies. The total labor force is the length of the horizontal axis, and workers are equally divided between countries, with workers in coastal economy C measured from the left-hand origin and those in land- locked economy L measured from the right. The value of marginal product of work- ers in each country is given by the downward-sloping lines, with the slope reflecting the diminishing returns to labor. If this were all, wages in each country would be equal (and low, W0) at the intersection of these lines. Suppose, however, that C, the coastal economy, can also undertake an export activity for the world market that does not run into diminishing returns. Given pro- ductivity levels, this activity can pay wage W*. Some of the labor in the coastal econ- omy will then move to this sector, and wages will rise until they reach W*. Country L, however, is landlocked and is unable to access this source of employment, so it is left with wage W0. Fragmentation into two political units means a fixed division of the labor force between countries, and the lack of migration between C and L creates inequality (the gap between W0 and W*), as well as a loss of efficiency and real income. Within a single economy, there would be internal migration of labor from L to C, so that the division of the total workforce would move in line with the hori- zontal arrows in the figure. Migrants would gain and would also bid up wages for those remaining until all workers received W*. Although landowners (owners of the sector-specific factor) in L would lose, there is an overall real income gain from the migration equal to the shaded triangle. Furthermore, if the export activity had increasing returns to scale rather than constant returns, the expanded employment would increase productivity and the wage, raising income in both C and L. What insights does this offer? Some Asian economies have witnessed massive migration to regions (usually, although not necessarily, coastal) that have succeeded in building up sectors supplying world markets. The best example is west-to-east TRADE AND ECONOMIC PERFORMANCE | 59 migration in China. Internal migration is also important (although poorly docu- mented) in India, both as seasonal migration and as part of rapid urbanization. Africa has not yet developed such magnets of employment, but the analysis indicates that if it were to do so, fragmentation would prevent it from fully realizing the benefits. There have been substantial migration flows in Africa, but they have often been problematic. Even where governments permit international migration, the noncitizen status of immigrants creates opportunities for the politics of xenophobia, which increases as the stock of immigrants accumulates. Since attractive political niches at some point draw politicians willing to occupy them, the opportunity for xenophobia is unlikely to remain unexploited. In turn, this political response exposes immigrants to violence and expulsion. The clearest instance of this depressing sequence arose from the sharp difference in natural advantage between Côte d'Ivoire and Burkina Faso. With its radically better economic opportunities stemming from its coastline and its rainfall, Côte d'Ivoire attracted massive immigration from Burkina Faso. Indeed, at one stage around 40 percent of the labor force in Côte d'Ivoire was Burkinabe. During the 1990s, this development was exploited by populist politicians and was instrumental in triggering political meltdown into coups and civil war, during which much of the migration was reversed.2 Nigeria in the 1970s provides a second instance of the same sequence. The discovery of oil created a sharp distinction in natural advantage for Nigeria vis-ŕ-vis Ghana. The consequence was mass migration from Ghana--at one point, around a fifth of the Ghanaian population had emigrated. Xenophobia grew in Nigeria, and when economic conditions deteriorated, the Ghanaians were formally expelled. South Africa witnessed its own backlash against immigration in 2008. In addition to constraining migration, a further adverse effect of fragmentation can be illustrated with reference to the circumstances shown in figure 2. Owners of the specific factor in the diminishing-returns sector in the coastal country (C) suffer an income loss as production of the export activity takes off. The diminishing-returns sector is most naturally thought of as agriculture. As manufacturing expands, the agricultural sector has to compete for labor and faces higher wages. Agriculture in country C contracts, returns to land in C fall, and C comes to import agricultural goods from country L and the rest of the world. There is a likelihood that this trend will create a lobby for tariffs on agricultural imports. Although such a move would raise the real income of landowners, it would have a negative effect on total income in the coastal economy. The political economy of protectionism, however, is well understood to favor small but cohesive interests over the general interest. The import tariff would also choke off manufacturing production in the coastal economy, possi- bly with damaging long-run effects on development. And if the coastal economy were the primary market for the landlocked country's agricultural exports, the tariff would also damage country L. The point is, of course, that the opportunity for using an import tariff, at least against imports from country L, exists only if C and L are sep- arate countries. Merger therefore removes a policy instrument that is likely to be misused in response to political-economy pressures. An example of manipulation of trade barriers as a result of political fragmentation is the banning of food imports 60 | PA U L C O L L I E R A N D A N T H O N Y J . V E N A B L E S from Uganda by President Daniel arap Moi of Kenya in the early 1990s in response to lobbying by business interests that were holding large stocks of food. Time-Variant Differences in Natural Advantage: Export Volatility The economic advantages and disadvantages of a particular location at a particular date are shaped partly by fixed geography and partly by the impact of short-run shocks. The arguments we have developed above apply to these transient effects, as well as to the permanent ones. Shocks create unevenness between areas, and the abil- ity to spread their impact (i.e., to pool risk), with both distributional and efficiency implications, is impeded by fragmentation into national units. Collier and Goderis (2008a, 2008b, 2009) investigate the consequences of com- modity export shocks for GDP valued at constant prices. This method abstracts from the income effect accruing directly from changes in the terms of trade and focuses on the consequences for real output. The authors find that shocks have asymmetrical effects on growth: adverse shocks significantly reduce output, but positive shocks do not have significant effects. The effect of adverse shocks is substantial. For example, for a typical African country whose commodity exports are initially around 35 per- cent of GDP, the consequence of a 30 percent fall in export prices would be to reduce growth in the following year by 3.6 percentage points. More generally, given the fre- quency of shocks and their cost, it is possible to estimate the discounted present value of the output losses that they generate. Using a 3 percent discount rate, the cost for individual countries is sometimes large: for example, for Nigeria it is 13 percent of one year's GDP. Yet summed over the entire continent, country by country, the cost is modest, around 4 percentage points of one year's GDP.3 For this paper we investigated whether these costs would have been reduced had Africa been divided into fewer polities.4 We examined the consequences of a United Africa, which provides an upper bound to the analysis, and of political regrouping into four regional blocs. We found that a United Africa would have reduced the costs by a mere 0.6 percentage points of the region's GDP. The gains from regional group- ings are more variable, the largest being from a United West Africa, where they would amount to 1.7 percentage points of GDP. The main reason why the effect is so limited is that the scope for risk pooling in Africa is modest. Virtually all countries are commodity exporters, and not only are the prices of most commodities highly correlated, but a single commodity, oil, dom- inates Africa's exports. In a United Africa, around 65 percent of commodity exports would be oil. Hence, pooling has little effect on the size of the average shock. The key potential gain therefore comes not from reducing the average size of shocks but from changing their distribution. Most African countries do not export oil, so that a United Africa would have fewer enormous shocks and more moderate-size shocks than in the distribution of country-specific shocks. Whether such redistribution mat- ters depends on the precise relationship between the size of shocks and their costs. The original specification of the work by Collier and Goderis is linear, and the costs of a shock are thus required to be proportionate to the size of the shock. In that case, redistributions that do not affect the average make no difference. TRADE AND ECONOMIC PERFORMANCE | 61 For this paper we therefore investigated whether the true structure of costs is non- linear, rising more than proportionately with the size of the shock, as seemed inher- ently likely. We found little basis for a nonlinear relationship. For example, when the square of the size is added to the regression, although the coefficient is negative, it is not close to significance. We conclude that, given Africa's export structure, the scope for reducing the cost of shocks through greater political unity is modest. But, as we will argue next, Africa's export structure may itself be a consequence of political fragmentation. Manufacturing and service exports benefit from scale economies that are less important for primary commodities. If political fragmentation has frustrated these scale economies, it may have locked Africa into dependence on primary com- modity exports. Scale, Density, and Increasing Returns The preceding section pointed to diminishing returns as a source of loss for a frag- mented continent with uneven distribution of resources or of shocks. We now turn to increasing returns and the cost of opportunities forgone by small countries. We focus first on a microeconomic mechanism that we think is particularly important in many small African countries--the fact that markets are too thin to be competitive. This constitutes a major obstacle to investment, both because of high prices for cap- ital goods and, more fundamentally, because of vulnerability to opportunistic behav- ior. We then discuss how the microeconomic advantages of scale combine to produce a positive relationship between city size and productivity and suggest that fragmen- tation has created an African city structure which is ill suited to reap these produc- tivity effects. Competition and the Operation of Markets Small economies are likely to have high levels of monopoly power. That makes them bad places for new investments. Incumbent firms have little incentive to expand out- put or to innovate, and they have strong incentives to deter entry and innovation by newcomers. Monopoly raises the prices of many intermediate goods, as well as of final output, thereby raising costs. And, most important, monopoly creates the potential for opportunistic behavior in transactions and thereby creates a difficult business environment, even for the monopolists themselves. Evidently, a small market is likely to be less competitive than a large one because, given some firm-level economies of scale, fewer firms will operate. The effect will be particularly pronounced in sectors that are closed to trade. For example, the typical African economy has a very highly concentrated banking sector--often, four banks dominate lending, and this is a sufficiently small number to enable collusive oligop- oly. The limited nature of the market also leads to a concentration of risk: banks are exposed to a high covariance of the risk of default. Transport is another sector that is sheltered from international competition and that is often highly cartelized. A recent study (Teravaninthorn and Raballand 2008) 62 | PA U L C O L L I E R A N D A N T H O N Y J . V E N A B L E S TABLE 2. Relative Price of Investment (variables in logs) Price of investment Price of investment (all countries) (population < 20 million) Real GDP per worker 0.27 0.30 ( 12.7) ( 8.8) Number of workers 0.046 0.071 ( 3.5) ( 2.6) N 163 83 R2 0.50 0.50 Source: Caselli and Feyrer (2007), from Penn World Tables. Note: GDP, gross domestic product. Numbers in parentheses are t-statistics. finds that the real costs of transport services in Africa are not abnormally high but that trucking firms are able to charge exceptionally high prices. Average prices per ton-kilometer are US$0.02 in Pakistan, US$0.05 in China, US$0.08 for the run from Mombasa (Kenya) to Kampala (Uganda), and US$0.11 for the run from Doula (Cameroon) to N'Djamena (Chad). Many African economies have restrictive regula- tory regimes and transport cartels. (Deregulation of the trucking industry in Rwan- da has been estimated to have reduced transport prices by 75 percent.) In some cases, a major factor supporting cartelization is a treaty structure between countries designed to protect the national trucking industry from competition from neighbor- ing countries. Monopoly power raises relative prices, and a key relative price in an economy seeking to grow is the price of investment in relation to the price of GDP as a whole. Given the level of saving, the higher is the relative price of investment, the less phys- ical equipment the saving will purchase. This effect is quantitatively important, as the price of investment (in relation to GDP as a whole) can be three or four times higher in developing countries than in high-income countries. Recent work by Caselli and Feyrer (2007) shows that much of the variation in the marginal physical product of capital across countries is in fact the result of this price effect. Why is there this price difference? Part of it is attributable to the Balassa-Samuelson effect; investment has high import content, and prices of tradable goods are relatively high in low-income countries. It may also be partly the result of thin markets and monopoly power in supply of equipment and investment goods, which may be a func- tion of country size. To investigate this relationship, we explored the impact of GDP per capita (for the Balassa-Samuelson effect) and country size (for the market power effect) on this relative price. The evidence is given in table 2, which reports the results of regressing the price of investment relative to the price of GDP on real GDP per work- er and on the number of workers (used as a measure of economic size). The extreme- ly strong dependence on output per worker is clear, and so too is the scale effect. Increasing the labor force by a factor of 20 reduces the relative price of investment by 13 percent. Restriction of the sample to countries with a labor force of less than 20 mil- lion produces a quantitatively larger effect. Recalling that India's population is TRADE AND ECONOMIC PERFORMANCE | 63 66 times larger than that of the average African country, which has a workforce of less than 10 million, the implications for differences in the price of capital are substantial. Thin markets and the resulting monopoly power increase the price of capital goods, but they have further pernicious effects. One is that they create an incentive for incum- bent firms to actively pursue strategies that deter the entry of new firms. (Strategies for entry deterrence may include the use of predatory pricing or the purchase of polit- ical influence.) From the perspective of firms that are already operating in a sector, if one firm devotes resources to keeping new entrants out of the sector, that is a public good. From the perspective of society as a whole, however, such behavior is evidently undesirable. In an industry with many existing firms, the free-riding problem implies that the returns to any one firm from such antisocial behavior are limited. But in a small market with an incumbent monopolist, all the benefits to the existing industry are internalized, and so the incentive to keep out new competitors is maximized. Small and thin markets are unattractive places to invest because investors are vul- nerable to "holdup"--opportunistic behavior by other firms with which they have to do business. Holdup refers to the possibility that once an investment has been sunk, the investor will face a monopsonistic purchaser of the output of the investment. Even if the purchaser and investor entered an agreement before the investment was undertaken, the purchaser may subsequently act opportunistically, breaking the agreement and offering a lower price. The investor will anticipate the possibility of holdup and so may not make the investment in the first place. One way to overcome the holdup problem is to make the ex ante contract legally binding, but even in coun- tries with strong legal systems, it is often impossible to write a contract with a degree of completeness that will rule out such opportunistic behavior. The other way is to make sure that there are many alternative uses for the output of the project. This is partly a matter of the specificity of the investment (it might be a machine for making specialist parts demanded by a single manufacturer) and partly a matter of the size of the market in which the output is sold. Holdup is likelier the fewer are the people competing for the output. This suggests that in small economies the threat of holdup may be a major deter- rent to investment. In agriculture, returns to investment are reduced if there is a monopsonistic grain merchant. In manufacturing, having few potential purchasers of output deters investment. The consequence is coordination failure: there is no incen- tive to enter on one side of the market until the other side has more firms, and vice versa. And in a small economy, even the return to the worst option--liquidating the investment--may be reduced by thin markets for secondhand capital equipment. Distress sales are likely to be more coincident because smaller economies are less diversified, further depressing the expected price. The holdup phenomenon applies not only to goods markets but also to labor markets; the incentive to undergo train- ing is reduced if the skill acquired can only be sold to one employer. These arguments point to the fact that smallness does not just create static monop- oly or monopsony power but also creates a fundamentally more risky business environment. Entry of new producers will be deterred by predatory behavior by incumbents and by the scarcity of outside options and the consequent vulnerability to predatory and opportunistic behavior. 64 | PA U L C O L L I E R A N D A N T H O N Y J . V E N A B L E S Productivity and City Size Productivity tends to be higher in large (or dense) clusters of economic activity. This is why cities form. Firms and workers locate to gain the benefits of this productivity advantage, despite the congestion costs and other diseconomies associated with large cities. A number of mechanisms drive this productivity effect. Some are narrowly technical--for example, the fact that dense activity economizes on transport costs and improves communications (and possible learning externalities) among firms, and between firms and workers. Others have to do with the impact of size on market structure and the intensity of competition, as discussed above. Still others concern political economy: a city with a large business sector is likely to have a strong busi- ness lobby, producing a business-friendly investment climate. The quantitative evi- dence of the productivity effect of city size comes largely from studies of cities in developed countries. Rosenthal and Strange (2004) report a consensus view that a doubling of city size is associated with a productivity increase of about 3 to 8 per- cent. This is a large effect: a size increase from 100,000 workers in a city to 3 million is predicted to increase productivity by more than 30 percent. Au and Henderson (2006) find even larger results for Chinese cities, where they estimate that moving from a city of 100,000 workers to one of 1.3 million workers raises productivity by 80 percent, although beyond this scale weak diminishing returns cut in. In this section we investigate the effects of Africa's fragmentation on its city struc- ture and argue that fragmentation has a negative impact on city size, producing a city structure that is weak compared with the structure in an integrated country such as India. We have argued elsewhere that Africa's failure to develop large clusters of eco- nomic activity has had major implications for its economic performance. Africa's fail- ure to enter world markets for manufactured exports is best understood by looking at the location of productive clusters of activity (Collier and Venables 2007). Many Asian cities have already gained a head start in these sectors and have grown highly productive clusters, creating a barrier for new entrants. Asia's initial advantage over Africa, when Asia first penetrated the global market for manufactures in the 1980s, may have been modest. Quite probably, the reasons for Africa's initial disadvantage have evaporated; for example, in the 1980s much of coastal Africa was beset by con- flict (as in Mozambique) or poor economic policies (as in Ghana). But as clusters have developed in Asia, the resulting scale economies have given that region a new and more formidable advantage. Africa may have missed the boat on industrializa- tion, unless trade policies of the Organisation for Economic Co-operation and Devel- opment (OECD) artificially create an offsetting temporary advantage for African manufactures that pump-primes the formation of clusters. Does the political fragmentation of Africa have any bearing on this? The obvious fact is that small countries generally have smaller cities, so some productivity bene- fits are forgone. We first investigate this point by employing a cross-country regres- sion to explore the determinants of city size. We use a large dataset of world cities and take as the dependent variable the population of the jth-ranked city in country i. We work with the top five cities in each country, which yields (with some missing values for very small countries) 521 observations. The explanatory variables are TRADE AND ECONOMIC PERFORMANCE | 65 TABLE 3. Determinants of City Population (all variables in logs) Without regional fixed Without regional With regional effects; income Variable fixed effects fixed effects per capita US$10,000 Country population 0.639 0.70 0.731 (23.3) (18.9) (13.7) Country area 0.169 0.107 0.085 (7.1) (3.9) (2.1) GDP per capita 0.27 0.27 0.102 (7.6) (7.6) (1.8) City rank 1.08 1.1 1.21 ( 25.0) ( 28.0) ( 25.0) N 521 521 325 R2 0.80 0.83 0.83 Note: Numbers in parentheses are t-statistics. country population, country area, country per capita income, and the rank of the city in the country, thus, ln(populationij) a b1 ln(populationi) b2 ln(areai) b3 ln(income pci) b4 ln(rankj). Table 3 presents results for specifications with and without regional fixed effects, for the world as a whole and for countries with per capita income of less than US$10,000. As is well known, national GDP per capita has a positive effect on city population. The city's within-country rank has a negative effect, as it must by con- struction. The estimated parameter in our central specification is 1.08, close to Zipf's law--the rank-size rule--which states that within countries the size of each city is inversely proportional to its rank in the city size distribution (see Gabaix and Ioannides 2004). Of most interest for our purposes is the fact that both country population and country area are highly significant determinants of city size. The sum of the coefficients on these two variables is 0.80, indicating that a merger of two similar-size countries--producing a doubling of population and area--would lead to a 75 per- cent increase in the size of the largest city. To see the quantitative implications, sup- pose that initially there are 10 separate countries, in each of which the largest city has a population of 3 million people. Combining these countries and letting city sizes adjust in line with the regularities given in table 3 yields a largest city of 19 million and a size distribution of city population of (for the first 10 cities, in millions) 19, 9.5, 6.3, 4.7, 3.8, 3.1, 2.7, 2.4, 2.1, 1.9. It is interesting to compare these calculations with actual city size distributions in India and Africa (table 4). Both the calculations and the actual data for India suggest that a large, integrated country, compared with Africa, has much larger cities in the top rank; slightly fewer in the upper-middle ranks (compare the numbers of Indian and African cities with populations around 3 million, as given in table 4); and many 66 | PA U L C O L L I E R A N D A N T H O N Y J . V E N A B L E S TABLE 4. Cities with Population Greater than 1 Million, India and Sub-Saharan Africa India Sub-Saharan Africa City Population City and country Population Mumbai 21,600,000 Lagos, Nigeria 10,100,000 Delhi 21,500,000 Kinshasa, Congo, Dem. Rep. 8,200,000 Kolkata 15,700,000 Johannesburg, South Africa 7,800,000 Chennai 7,850,000 Khartoum, Sudan 5,450,000 Bangalore 7,350,000 Abidjan, Côte d'Ivoire 4,225,000 Hyderabad 7,150,000 Durban, South Africa 3,600,000 Ahmedabad 5,650,000 Kano, Nigeria 3,600,000 Pune 4,625,000 Cape Town, South Africa 3,400,000 Surat 3,875,000 Accra, Ghana 3,350,000 Kanpur 3,475,000 Ibadan, Nigeria 3,200,000 Jaipur 3,050,000 Nairobi, Kenya 3,175,000 Lucknow 2,800,000 Addis Ababa, Ethiopia 3,100,000 Nagpur 2,700,000 Dar es Salaam, Tanzania 3,000,000 Patna 2,350,000 Luanda, Angola 2,875,000 Indore 1,870,000 Dakar, Senegal 2,550,000 Vadodara 1,870,000 Pretoria, South Africa 2,450,000 Coimbatore 1,820,000 Harare, Zimbabwe 2,200,000 Bhopal 1,810,000 Douala, Cameroon 2,000,000 Ludhiana 1,730,000 Maputo, Mozambique 1,820,000 Agra 1,700,000 Antananarivo, Madagascar 1,760,000 Kochi 1,660,000 Bamako, Mali 1,730,000 Visakhapatnam 1,610,000 Lusaka, Zambia 1,720,000 Meerut 1,600,000 Yaoundé, Cameroon 1,610,000 Asansol 1,580,000 Conakry, Guinea 1,600,000 Bhubaneswar 1,560,000 Kaduna, Nigeria 1,590,000 Nashik 1,550,000 Kumasi, Ghana 1,520,000 Chandigarh 1,520,000 Kampala, Uganda 1,490,000 Varanasi 1,470,000 Lubumbashi, Congo, Dem. Rep. 1,450,000 Kolhapur 1,460,000 Mogadishu, Somalia 1,410,000 Jamshedpur 1,350,000 Brazzaville, Congo, Rep. 1,330,000 Madurai 1,350,000 Lomé, Togo 1,320,000 Rajkot 1,320,000 Ouagadougou, Burkina Faso 1,260,000 Jabalpur 1,300,000 Benin City, Nigeria 1,180,000 Dhanbad 1,290,000 Port Harcourt, Nigeria 1,170,000 Amritsar 1,270,000 Port Elizabeth, South Africa 1,150,000 Allahabad 1,230,000 Freetown, Sierra Leone 1,110,000 Vijayawada 1,220,000 Cotonou, Benin 1,090,000 Srinagar 1,180,000 Maiduguri, Nigeria 1,040,000 Shambajinagar 1,170,000 Solapur 1,100,000 Thiruvananthapuram 1,100,000 Ranchi 1,090,000 Jodhpur 1,040,000 Guwahati 1,030,000 Tiruchirappalli 1,010,000 Gwalior 1,000,000 Source: "The Principal Agglomerations of the World," City Population Web site, http://www.citypopulation.de/ World.html. TRADE AND ECONOMIC PERFORMANCE | 67 more medium-size to large cities (e.g., with populations between 1 million and 2 million). This analysis suggests that the smaller size of African cities stems in large part from the fragmentation of the region. The preceding discussion and the evidence from developed-country studies (as summarized by Rosenthal and Strange 2004) sug- gest that city size may have had an adverse effect on the productivity of African manufacturing. Although size is not the only determinant of urban productivity, Africa's fragmentation and consequent urban structure may have impeded the devel- opment of major international manufacturing centers of the type that contributes to the performance of high-growth economies. Political Economy of Public Goods Above, we considered those advantages of being a large country that arise from greater diversity and greater geographic concentration of private economic activities. We now consider those that arise from government activities. A core function of a government is to supply public goods. By definition, public goods are subject to scale economies. The distinctive aspect of public goods that ensures scale economies is that consumption is nonrival--one person's consumption does not reduce that of another person. Many public goods, however, also have more conventional scale economies: their production technology has high fixed costs that can be spread over more consumers as scale is expanded. A radio station displays both types of scale economy: listening is nonrival, and once the fixed costs of transmission have been incurred, the number of hours broadcast per day is subject to much lower variable costs. As the number of consumers is progressively increased to the point at which it includes all the world's inhabitants, relatively few goods still have unrealized economies of scale. At some point, consumption ceases to be nonrival, and technolog- ical scale economies are exhausted. Conceptually, we can order all the goods that might potentially be provided publicly according to the minimum size at which all scale economies are realized. Along this ranking, we find first those goods that are non- rival only within the locality, then the nation, then the region, and, finally, globally. The supply of public goods generates acute collective action problems that, except in a few cases, require coercive powers of taxation to overcome effectively. The high- est level at which such powers are found is the nation. Hence, in practice, those scale economies that occur beyond the level of the nation-state are generally not realized. The failure to realize such economies is widely lamented in discussions of the underprovision of global public goods (Barrett 2007) and regional public goods (Cook and Sachs 1999). It has a powerful corollary through its implications for dif- ferences in country size. Since there are some public goods that are global, a fortiori there must be many more for which the minimum efficient size lies somewhere with- in the huge population range implied by the difference between the smallest African country and India. Within this range, the smaller is the country, the fewer goods can efficiently be supplied at the level of the nation-state. This consideration is reinforced 68 | PA U L C O L L I E R A N D A N T H O N Y J . V E N A B L E S when population is replaced by GDP as the size metric. Public goods are economic activities, and the relevant metric for scale is more likely to be the size of the econo- my than the size of the population. Since Africa is now the poorest region, its typical national economy is even tinier than is suggested by the size of its population. The economy of Luxembourg is roughly the same size as that of the five countries of the East African Community combined. It might seem that nations could be too large as well as too small for the efficient supply of public goods. If different communities have distinctive preferences for pub- lic goods, provision may work better if political decision units are small and so can reflect these differences. There is, however, an asymmetry. Once the optimal scale has been reached, a large state can always replicate this scale by decentralizing supply to subnational authorities. In other words, once all the scale economies have been reaped, further expansion can be under conditions of constant returns to scale. States that are too small do not have an equivalent option: a national government can choose to pass authority down, but it does not have the power to pass authority up. A large nation thus has an advantage over a small nation, and this advantage may become very pronounced by the time we reach the tiny states that are common in Africa. We now analyze the provision of three public goods that are fundamental to pros- perity: security, economic policy, and infrastructure. Security Security is the clearest case of a public good that is subject to scale economies far beyond the size of the typical small African state. Like radio, security benefits from both types of scale economy. Over a wide range, defense is nonrival--the same army that defends one community from rebellion can defend a proximate community. Over a very wide range, it benefits from scale economies: big armies usually defeat little armies, a proposition formalized in contest success functions. The sheer power of scale economies in security has repeatedly been revealed in the expansion of empires. Once a power gets a military advantage over its neighbors, it can expand almost without limit if it chooses to do so. Rome, the Mongol Empire, Russia, and the nineteenth-century European empires demonstrate that big is safe. The incidence of warfare in Africa has been far higher than in India, and differ- ences in scale in part account for this. Evidently, political union would have reduced the incidence of international war, but almost all of Africa's wars have been internal, and so the key issue is how union would have affected this risk. Statistical analysis of the risk of civil war finds that whereas population significantly increases the risk, the effect is substantially less than proportionate: a territory under a single polity has a lower risk than the combined risk for two polities were the single state split in half (Collier, Hoeffler, and Rohner 2009). The case for scale, however, is complicated by the trade-off with ethnic diversity. In general, in order to make a polity larger, it is necessary to take in additional social groups, and so diversity increases. Unfortu- nately, diversity heightens the risk of civil war. Were Africa to have been split into fewer countries, the adverse effect stemming from diversity might have more than offset the benefit from greater scale. The issue has recently been analyzed by TRADE AND ECONOMIC PERFORMANCE | 69 Wigstrom (2008), who has carefully investigated how mergers between neighboring African countries would affect ethnic diversity. It transpires that in some cases Africa's borders are so arbitrary that ethnic diversity would actually be reduced by merger. In these instances the scale and the diversity effects of political union work in the same direction, lessening the risk of civil war. Even were diversity to increase, Wigstrom finds that over a wide range, had Africa been divided into fewer inde- pendent polities, it would have had a lower risk of civil war. Although his analysis is of course hypothetical and cannot take into account many aspects of politics, it omits the consequences for peace of the economic benefits of scale discussed above. Since both the level and growth of income significantly reduce the risk of civil war, these economic effects of political union would have reinforced the effects discussed here. The small scale of African polities not only increases the incidence of war; it increases military spending while at peace. Because large armies tend to defeat small armies, small states tend to compensate by spending a higher fraction of GDP on the military. Furthermore, although a country can increase its own security by spending more on its army, this reduces the security of neighboring countries that feel threat- ened. Evidently, in response to a perceived increase in threat, the neighboring coun- try will need to increase its own military spending, producing an arms race. The essence of an arms race is that, if we think of military spending as producing securi- ty, an increase in spending by one country reduces the average productivity but increases the marginal productivity of its neighbor's spending. Collier and Hoeffler (2007) establish that neighborhood arms races have been common around the devel- oping world. For the present paper we have used their results to simulate the reduc- tion in average African military spending had there been a United Africa. We set the incidence of civil war equal to that estimated by Wigstrom (2008), eliminate the effect of neighborhood arms races, and set all other variables at their average actual values for the continent. The predicted level of military spending falls by a quarter, from 3.2 to 2.4 percent of GDP (calculation by Anke Hoeffler). Good Economic Policies and Governance A second fundamental public good is the provision of sound economic policies and accountable government. The choice between good and bad policies is typically deter- mined by a complex mix of influences. Among them are the interests of the elite, the political power of ordinary citizens, the degree to which both elites and citizens understand basic economic issues and so grasp how their interests are best served by policies, and, finally, the capacity of the civil service to design and implement policies. Scale may enter here through various routes. For example, the quality of the civil service can be higher in a larger society simply because it can be more selective. There is around 50 times more competition to become permanent secretary in the ministry of finance in India than in Africa, and so the quality will, on average, be higher. Another scale effect is that, paradoxically, a larger society can be better informed about economic issues than a small one. The key reason is the scale economies in the commercial media--radio, television, newspapers, and magazines. A large market will permit more media outlets to exist than a small market will. Serious discussion 70 | PA U L C O L L I E R A N D A N T H O N Y J . V E N A B L E S of economic issues within a society is highly dependent on the existence of specialist media. India has such media, but in Africa, only South Africa comes anywhere close to providing a market in which specialist journals are viable. For example, the Indi- an newspaper the Economic Times has a circulation of 1.2 million, which is suffi- cient to finance a staff of economically qualified journalists. With the same density of circulation, an economics newspaper in Zambia would have a circulation of less than 10,000 and so would not be viable. Without a specialist media, discussion in the society is likely to be less sophisticated, and the pace at which social learning takes place will be slower. In effect, the society needs a critical mass of educated cit- izens before social learning can be rapid. This may help explain why India reformed its economic policies before Africa did. A distinct reason why larger states may be able to reform faster is a corollary of their greater need to decentralize authority: more public goods reach the level at which decentralization is the efficient form of organization. Such decentralized authority introduces variation in strategies, and this in turn represents a source of learning. Small societies can, of course, choose to replicate the same degree of decentralization, but the increased opportunities for learning come at the expense of forgone scale economies in provision. India, with its federal structure, has clearly had a very wide range of experimentation, with some states pioneering in the provision of social services and others in encouragement of foreign investment. Indeed, the equivalent of learning can occur even if successful experiments are not copied. People and firms will choose to relocate to attractive areas, and this gradually shifts the weighted average of policies across the nation toward the most successful. Potentially, Africa's equivalent is that its division into so many nations enables policy variation at the national level. It is not clear, however, that a small country is in any better position to learn from other nations than is a large country. (India has clearly learned a great deal from China.) A related scale effect is the switch from discretion to rules in decision making. At its best, an intimate organization can tailor each decision to the needs of the individ- ual and the circumstance; decisions can be personalized. As the organization becomes larger, this style of decision making breaks down because micromanagement becomes overburdened and is replaced by rule-based procedures. Rule-based decisions are seldom as good as first-best discretionary decisions, but they are far better than either patronage-driven or idiosyncratic decisions. They also enable the government to have a credible commitment technology that may even dominate the best discretionary policies by providing an escape from the time-consistency problem. Hence, we might expect that public decisions in large societies lie within a narrower range than those in small societies. This is closely analogous to the difference between autocracy and democracy. Autocrats have the discretion either to be very good or very bad, whereas democracies are rule bound. Besley and Kudamatsu (2007) compared the economic outcomes for these two types of government and indeed found that democracy trun- cates the distribution at both extremes. These effects of scale are plausible, but is there any evidence that they actually matter? Chauvet and Collier (2008) analyze the preconditions for policy turnaround in countries that initially had very poor economic policies and governance, using the World Bank measure of economic policy and governance, the country policy and TRADE AND ECONOMIC PERFORMANCE | 71 institutional assessment (CPIA). This is a subjective, ordinal rating and thus has obvi- ous drawbacks, but most of the controversies in economic policy concern the higher range of these ratings, and policies and governance that are very poor are often unmistakable. Chauvet and Collier define a turnaround as the passage from a rating below a very low threshold to one above a benchmark that, although substantially higher, is still modest in relation to the rating of most developing countries. The authors' universe is all low-income countries over the period 1977­2005. From this group they select those countries in which for at least four consecutive years eco- nomic policies and governance were below the threshold. They find that population size is one of the preconditions for a turnaround that is statistically significant: the larger is the population, the higher is the probability that a country that initially ranked below the threshold will achieve a sustained turnaround. Equivalently expressed, large countries appear to experience more rapid social learning. For this paper we have used the coefficients of the Chauvet-Collier model to esti- mate how long India, on the one hand, and the average African country, on the other, would have taken to reform from a common initial position of poor policies to a com- mon improved position. We create two artificial countries, "India" and "Zambia." Both are identical in all characteristics except population; all other characteristics are set at the sample mean. "India" has the population of India, and "Zambia" has the population of the average African country, both entered into the regression as logs. We set their initial CPIA scores at 2.5, indicating very poor policies, and ask how long it would take to reach a score of 3.5. Since the regression model is log-linear, the conse- quences of the huge differences in population are likely to be exaggerated; the regres- sion line is trying to explain the countries within the range and is likely to fail at the extremes. Nevertheless, the predicted difference in the pace of reform is striking: "India" is predicted to bounce out of bad policies into reasonable ones in only 6 years, whereas "Zambia" is predicted to take around 60 years (calculation by Lisa Chauvet). Infrastructure Transport and power infrastructure are public goods with such strong scale economies that the typical African polity is too small to exhaust them. Indeed, Africa still depends on the transport infrastructure created during the period when its pres- ent polities were united into a few empires. Quite possibly, from the perspective of transport infrastructure, the key feature of colonialism was not that the empires were ruled externally but that they temporarily united Africa into a few large polities. The most obvious problem generated by political division is that many countries are land- locked. As shown by Limăo and Venables (2001), the transport costs faced by land- locked countries are strongly affected by the infrastructure spending of their coastal neighbors. Evidently, these benefits to the landlocked are externalities from the per- spective of the coastal countries; they are not internalized into the decision calculus, and so spending is suboptimal. The failure to internalize costs and benefits, however, extends far more widely than the plight of the landlocked. The recent discovery of large iron ore deposits in Guinea by Rio Tinto Zinc is a telling example. The exploitation of the deposits 72 | PA U L C O L L I E R A N D A N T H O N Y J . V E N A B L E S evidently requires investment in a mine, but the pertinent issue is the necessary invest- ment in transport infrastructure. Fortuitously, as a legacy from the age of empires, a railway already links the deposit with a deepwater port, Buchanan. But Buchanan is in Liberia, and the government of Guinea does not want to find itself subject to a holdup problem vis-ŕ-vis the government of Liberia. It has therefore insisted that the transport link be entirely within Guinea, which requires the construction of a new, dedicated railway and a new deepwater port. This decision has more than doubled the total investment needed for the project, adding around US$4 billion. Evidently, these additional costs will be fully passed on; the government has agreed with Rio Tinto Zinc that it will absorb them through a reduced flow of royalty payments. Hence, the costs are ultimately borne by the people of Guinea. The decision is also costly to the people of Liberia; in particular, the port of Buchanan loses what may have been its key opportunity for scale economies. The generation of electric power is also more costly if the market for power is politically segmented. Not only is the generation of power subject to scale economies, but noncoincident peaks in demand cannot be pooled. The resulting volatility in demand leads to installation of capacity that is idle most of the time and to energy rationing. Energy is a fundamental input into both resource extraction and manu- facturing. The recent power cuts in South Africa are reducing investment in resource extraction, and high energy costs risk making manufacturing uncompetitive. Potentially, the highest costs of political division arise from the interactions between transport and power. The resource extraction sector is highly intensive in both, especially in Africa, where mineral deposits are often far from the coast. If, as in Guinea, the ore is exported unprocessed, it has a low value-to-weight ratio, and so transport costs are high. Processing would reduce weight and therefore transport costs, but very large energy inputs are required. In Africa the obvious source of non- exportable energy is hydropower; the key input is rainfall on high ground, something the continent has in abundance. This nontradable energy can potentially be trans- mitted to resource extraction sites and used to process ore that is then cheaply trans- ported to the coast. The exploitation of such synergies may yield huge payoffs, but they would also entail huge investments. Unfortunately, almost all such opportuni- ties in Africa involve crossing frontiers between sovereign states. This returns us to the holdup problem discussed above, but with an added dimension. The holdup problem within a state can, at least in part, be addressed by law: the difficulty is that of writing a contract sufficiently complete to cover all eventualities. The holdup problem between states is radically more severe because the whole domain of inter- national law is fragile. Essentially, the concept of national sovereignty constitutes a barrier to the enforcement of any contract entered into by states. Policy Implications: The Need for Integration In aggregate, Africa is less populous and poorer than India, yet it is subdivided into around 50 independent states. We have suggested that this radical political subdivi- sion of an already small economy has inflicted a wide range of costs on African TRADE AND ECONOMIC PERFORMANCE | 73 citizens. The benefits of smoothing, with respect to both temporary shocks and underlying natural differences, are forgone. In the private economy, manufacturing and services have the potential for large-scale economies that are frustrated by polit- ical fragmentation. This skews Africa's comparative advantage toward those sectors and modes of production in which scale is less important, notably peasant agricul- ture. In the public sector, the lack of scale raises the costs of a wide range of public goods and thus accentuates the problem of undersupply intrinsic to low-income economies. These losses to the private and public economies are mutually reinforc- ing: low private incomes reduce state revenues and compound the underprovision of public goods, while the lack of public goods further reduces private incomes. The evident implication is that Africa needs a process of political integration. Such processes have been common in other regions; over the last two centuries many states have chosen to create legal structures at the regional level that curtailed their own sovereignty. During the nineteenth century, power within the United States gradually shifted from the individual states to the federation. This shift in the locus of decision making is even detectable in language: before the American Civil War the term "the United States of America" was treated as a plural noun, but afterward it became sin- gular. In the 1940s, the territories of British India decided that on independence they would divide into only two large polities, India and Pakistan, instead of reverting to the many small states that had preceded colonization. During the last half-century, the European Union has gradually expanded to 27 states that have agreed to limit sovereignty across a wide range of economic decisions. Unlike Indian politicians, African politicians chose to dissolve the federations forged by the empires: the colonial map of Africa was far less fragmented than the current configuration, with Nigeria (Africa's largest country) the sole exception. Post- colonial political fragmentation enormously increased opportunities for the political class, multiplying by 50 the number of ministers required to govern the territory. African governments have launched many initiatives aimed at greater regional integration. These include, at the political level, the Organization for African Unity and its successor, the African Union, and, at the economic level, an array of sub- regional trade arrangements too numerous to list. Yet the practical achievement in terms of economic integration falls far short of other regions. This is brought out by a comparison of Burundi with California, Maharashtra (India), and Germany. Viewed as economic units, California, Maharashtra, and Germany are each over a hundred times as large as Burundi. Yet as measured by autonomy in fiscal policy, monetary policy, trade policy, exchange rate policy, and the scope for judicial appeal against the government, the government of Burundi has radically more power than the governments of the other three. Since Burundi is such a small economic unit, its citizens' need for a supranational authority is much greater than is the case for Californians, Maharashtrans, and Germans, yet they have much less resort to one. One reason why African efforts at greater political and economic integration have to date yielded so little may be that these efforts have themselves not been integrated. The approach to political integration has been by means of a pan-African entity, the African Union, which is so diverse as to be unwieldy. For example, it includes North 74 | PA U L C O L L I E R A N D A N T H O N Y J . V E N A B L E S Africa, which identifies more closely with the Middle East than with Africa, and it requires agreement among 53 sovereign governments for any action. By contrast, the approach to economic integration has been too limited, predom- inantly involving subregional trade agreements. These deals are so numerous and uncoordinated that they are mutually incompatible: governments have signed up to commitments that simply cannot all be met. This situation generates confusion and undermines the credibility of trade policies. More fundamentally, subregional trade deals between low-income countries give rise to economic forces that are likely to cre- ate severe political tensions. As Venables (2003) shows, they generate economic divergence, with the poorest members of the integration scheme losing in relation to the least-poor members. This is in stark contrast to regional integration arrangements between high-income countries, which generate forces for convergence. Indeed, even Venables's analysis probably underestimates the forces of divergence released by regional integration among low-income countries, since it rests only on the implica- tions of comparative advantage. The forces unleashed by the scale economies dis- cussed above imply further forces for divergence--with, for example, the economi- cally largest cities, such as Nairobi, Johannesburg, and Lagos, gaining at the expense of initially smaller cities in their respective regions. Hence, the politics of Africa's subregional trade schemes are almost inevitably going to be fraught. A more promising alternative would be to learn from three aspects of successful integration in other regions. First, political and economic integration should go hand in hand; supranational entities have to acquire real sovereignty over particular domains of economic activity. For example, currently, despite the many subregional trade arrangements, each African country has been negotiating individually both with the World Trade Organization (WTO) and, for economic partnership agreements (EPAs), with the European Union. By contrast, the trade policies of each member country of the European Union are genuinely locked in common, and so negotiation at the WTO and for the EPAs is conducted by the European Commission rather than by each member country individually. Second, integration is easier if it grows from a small core of states among which none is too dominant. In Europe integration started with just 6 countries and has progressively expanded to 27. The United States grew from a core of 13 founding states to its present total of 50. In Africa the most promising such nucleus is the East African Community (EAC), which has a core of three similarly sized states and has already added two new members. The EAC has the advantage that it is building political institutions which might acquire real sovereignty over some aspects of poli- cy, alongside steps toward economic integration. Third, the economic agenda should be considerably broader than trade policy. There is scope for common rules on a wide range of economic policies--for exam- ple, on investment and taxation, which would enhance credibility. There is room for the provision of common infrastructure; the East African Community used to run an integrated rail system, and power generation and distribution would be better han- dled at the subregional level. Unlike trade agreements, these other areas of policy cooperation are likely to generate mutual gains, so that the economic consequences will reinforce rather than undermine the political process. TRADE AND ECONOMIC PERFORMANCE | 75 Finally, we might note that the political stresses produced by poverty have tended to be perverse. What is needed is a stronger impetus toward unity, but poverty is a fertile breeding ground for xenophobia and division. During 2008, there were riots in Johannesburg against immigration from Zimbabwe, as well as de facto ethnic par- tition in Kenya. The vision of African unity pioneered by Kwame Nkrumah and Julius Nyerere is in need of serious revival. Notes The authors thank Mauro Caselli, Lisa Chauvet, Anke Hoeffler, Hyesung Kim, and Jean-Louis Warnholz for statistical analyses. This work was supported by the BP-funded Oxford Centre for the Analysis of Resource Rich Economies (OxCarre) and by the Centre for the Study of African Economies (CSAE). This research benefited from funding by the UK Department for International Development (DFID) as part of the iiG, a research programme to study how to improve institutions for pro-poor growth in Africa and South-Asia. The views expressed are not necessarily those of DFID. 1. "Dutch disease" takes its name from the supposed effects of North Sea natural gas dis- coveries on the Netherlands economy. 2. See Collier (2009) for a detailed account of this sequence. 3. These output loss costs are likely to be underestimates because they omit longer-term effects. Collier and Goderis (2008a) find that unless governance is good, dependence on commodity exports has adverse effects on long-term growth. The mismanagement of volatility may be one of the routes by which poor governance has these adverse long-term effects. 4. Benedikt Goderis undertook the substantial work involved in these recalculations. References Alcalá, Francisco, and Antonio Ciccone. 2003. "Trade, Extent of the Market, and Economic Growth 1960­66." Economics Working Paper 765, Department of Economics and Business, Universitat Pompeu Fabra, Barcelona, Spain. http://www.econ.upf.edu/en/ research/onepaper.php?id=765. Alesina, Alberto, Enrico Spolaore, and Romaine Wacziarg. 2005. "Trade, Growth and the Size of Countries." In Handbook of Economic Growth, ed. Philippe Aghion and Steven Durlauf. Amsterdam: North-Holland. Au, Chun-Chung, and J. Vernon Henderson. 2006. "Are Chinese Cities Too Small?" Review of Economic Studies 73 (3): 549­76. Barrett, Scott. 2007. Why Cooperate? The Incentive to Supply Global Public Goods. Oxford, U.K.: Oxford University Press. Besley, Timothy J., and Masayuki Kudamatsu. 2007. "Making Autocracy Work." CEPR Dis- cussion Paper 6371, Centre for Economic Policy Research, London. Caselli, Francesco, and James Feyrer. 2007. "The Marginal Product of Capital." Quarterly Journal of Economics 122 (2, May): 535­68. Chauvet, Lisa, and Paul Collier. 2008. "What Are the Preconditions for Turnarounds in Failing States?" Conflict Management and Peace Science 25 (4, September): 332­48. 76 | PA U L C O L L I E R A N D A N T H O N Y J . V E N A B L E S Collier, Paul. 2007. The Bottom Billion: Why the Poorest Countries Are Failing and What Can Be Done about It. Oxford, U.K.: Oxford University Press. ------. 2009. Wars, Guns, and Votes: Democracy in Dangerous Places. New York: Harper- Collins. Collier, Paul, and Benedikt Goderis. 2008a. "Commodity Prices, Growth, and the Natural Resource Curse: Reconciling a Conundrum." Centre for the Study of African Economies, Oxford University, Oxford, U.K. ------. 2008b. "Structural Policies for Shock-Prone Commodity Exporters." Centre for the Study of African Economies, Oxford University, Oxford, U.K. ------. 2009. "Does Aid Mitigate External Shocks?" Review of Development Economics 13 (3): 429­51. Collier, Paul, and Anke Hoeffler. 2007. "Unintended Consequences: Does Aid Promote Arms Races?" Oxford Bulletin of Economics and Statistics 69 (1): 1­28. Collier, Paul, Anke Hoeffler, and Dominic Rohner. 2009. "Beyond Greed and Grievance: Fea- sibility and Civil War." Oxford Economic Papers 61 (1, January): 1­27. Collier, Paul, and Anthony J. Venables. 2007. "Rethinking Trade Preferences: How Africa Can Diversify Its Exports." World Economy 30 (8, August): 1326­45. Commission on Growth and Development. 2008. The Growth Report: Strategies for Sus- tained Growth and Inclusive Development. Washington, DC: World Bank. Cook, Lisa D., and Jeffrey Sachs. 1999. "Regional Public Goods in International Assistance." In Global Public Goods: International Cooperation in the 21st Century, ed. Inge Kaul, Isabelle Grunberg, and Marc Stern. New York: United Nations Development Programme and Oxford University Press. Gabaix, X., and Y. Ioannides. 2004. "The Evolution of City Size Distributions." In Handbook of Urban and Regional Economics. Vol. 4, Cities and Geography, ed. J. Vernon Hender- son, and Jacques-François Thisse. Amsterdam: North-Holland. Helliwell, John F. 1997. "National Borders, Trade and Migration." Pacific Economic Review 2 (3): 165­85. Jones, Ben, and Ben Olken. 2005. "The Anatomy of Start-Stop Growth." NBER Working Paper 11528. National Bureau of Economic Research, Cambridge, MA. Limăo, Nuno, and Anthony J. Venables. 2001. "Infrastructure, Geographical Disadvantage, Transport Costs and Trade." World Bank Economic Review 15 (3): 451­79. McCallum, John. 1995. "National Borders Matter: Canada­U. S. Regional Trade Patterns." American Economic Review 85 (3): 615­23. Rose, Andrew K. 2006. "Size Really Doesn't Matter: In Search of a National Scale Effect." Journal of the Japanese and International Economies 20 (4, December): 482­507. Rosenthal, S. S., and W. C. Strange. 2004. "Evidence on the Nature and Sources of Agglom- eration Economies." In Handbook of Urban and Regional Economics. Vol. 4, Cities and Geography, ed. J. Vernon Henderson and Jacques-François Thisse. Amsterdam: North- Holland. Teravaninthorn, Supee, and Gaël Raballand. 2008. Transport Prices and Costs in Africa: A Review of the International Corridors. Washington, DC: World Bank. Venables, Anthony J. 2003. "Winners and Losers from Regional Integration Agreements." Economic Journal 113 (490, October): 747­61. Wigstrom, Christian. 2008. "Does Size Matter?" Master of philosophy thesis, Oxford Uni- versity, Oxford, U.K. World Bank. 2007. World Development Indicators 2007. Washington, DC: World Bank. Protectionist Policies and Manufacturing Trade Flows in Africa LAWRENCE EDWARDS It is well known that Africa's share of world trade is declining. Policy prescriptions vary widely, covering tariff liberalization, policies to facilitate output growth, and improvements in trade-related infrastructure and institutions. This study evaluates trade performance and tariff liberalization in Sub-Saharan African countries since the early 1990s. It also estimates the impact of tariff liberalization on manufacturing trade flows in Africa and other developing countries between 1990 and 2004, using sector- level data and direct measures of tariff protection. Three key results emerge. First, African countries continue to be marginalized in world trade, although there is evi- dence of improved competitiveness, particularly in manufacturing. Second, tariff lib- eralization, starting in the early 1990s, has had a positive effect on manufacturing trade flows, but its contribution to the overall growth of trade has been small. African countries are no different from other developing countries in this regard. Third, tariff liberalization in developing countries is often associated with a worsening manufac- turing trade balance, but this is not necessarily so in Africa. The marginalization of Africa in world trade is well recognized. Early evidence from a World Bank research program in the mid-1990s (Amjadi and Yeats 1995; Amjadi, Reincke, and Yeats 1996; Ng and Yeats 1996) found that Sub-Saharan Africa's share in world trade declined dramatically between the 1950s and the early 1990s, from more than 3 percent to less than 1 percent. Views on the source of this deterioration vary widely. Policy prescriptions alternatively emphasize tariff liberalization (Amjadi, Reincke, and Yeats 1996; Ng and Yeats 1996), policies to facilitate output growth (Rodrik 1997), and improvements in trade-related infrastructure and institutions (Limăo and Venables 2001; Wilson, Mann, and Otsuki 2005; Djankov, Freund, and Pham 2006). This paper focuses, in particular, on the effect of tariff liberalization on trade per- formance in Sub-Saharan Africa. It begins with a comparative analysis of trade flows and trade liberalization in Africa during the 1990s. This was a period of considerable Lawrence Edwards is an associate professor at the School of Economics, University of Cape Town, South Africa. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank. 77 78 | L A W R E N C E E D WA R D S tariff reform for many developing and middle-income economies. How Africa fared relative to its counterparts is explored in some detail. Next, the contribution of tariff liberalization to changes in manufactured exports, manufactured imports, and the trade balance in developing economies, including Africa, between 1990 and 2004 is estimated. The focus on manufacturing partly reflects data constraints but also the emphasis in the literature on the association between manufacturing and growth acceleration, externalities, learning by doing, and returns to scale. An important focus of the paper is the effect of tariff liberaliza- tion on export performance through reductions in the cost of inputs. Manufacturing goods, which are relatively import intensive and are often more dependent on processed inputs than primary products, are more suitable for such an analysis. Where the paper makes a contribution in this field is that the empirical analysis draws on sector-level data (28 manufacturing sectors) and direct measures of tariff protection, as opposed to aggregate data and dummy variables for liberalization periods. Three key results emerge. First, African countries continue to be marginalized in world trade--even though there is evidence of improved competitiveness, particu- larly in manufacturing, and many countries in the region have experienced consider- able tariff liberalization, despite only limited offers in the Uruguay Round of the General Agreement on Tariffs and Trade (GATT). Second, tariff liberalization has had a statistically significant and positive effect on manufacturing trade flows in developing countries, but its contribution to overall growth in trade has been small. Africa is no different from other developing coun- tries in this regard, although high import tariffs on intermediate and capital inputs continue to restrict exports from several countries of the continent. Third, tariff liberalization is often associated with a worsening manufacturing trade balance in developing countries (although the evidence for Africa is weaker). The effect, however, is small, so tariff liberalization is unlikely to lead to a substantial increase in the trade deficit or to the emergence of an external constraint on growth. The next section reviews the existing empirical literature on the effect of trade bar- riers on trade performance in developing economies. A comparative analysis of trade flows and trade reform in Africa and other developing economies from the early 1990s follows. Various estimates are then presented of the relationship between trade liberalization and manufactured exports, manufactured imports, and the trade bal- ance for 50 economies, including 30 developing economies (10 in Africa), over the period 1990 to 2004. The paper concludes with a summary of the key findings and some policy recommendations. Trade Barriers and Trade Performance in Developing Economies In traditional trade theory, tariff barriers restrict trade flows. The effect of tariffs on imports is intuitive: by increasing the relative price of imported goods, tariffs induce consumption shifts toward domestically produced substitutes. The negative effect of import tariffs on export performance is not as well appreciated, yet as Lerner (1936) P R O T E C T I O N I S T P O L I C I E S A N D M A N U FA C T U R I N G T R A D E F L O W S I N A F R I C A | 79 pointed out, an import tariff can have an impact on export volumes equivalent to that of an export tax. By conferring protection on import-competing sectors, import tariffs draw resources toward these sectors and away from other sectors of the economy, including those that export. Import barriers, particularly on intermediate inputs, also raise production costs and reduce the profitability of export production. This is expected to be particularly important for exports of manufactured products, which use a high proportion of intermediate and capital goods in the production process, compared with goods pro- duced in the primary sector. The fragmentation and global outsourcing of production processes also require that manufacturing firms be integrated into global value chains, which, in turn, necessitates cheap access to imported intermediate inputs. Tariff liberalization is therefore expected to stimulate both import and export volumes. To some extent, these relationships are born out by the empirical evidence, includ- ing that for Africa.1 Rodrik (1997), for example, estimates that a 10 percentage point reduction in trade taxes is associated with a 17 percentage point increase in the ratio of trade to gross domestic product (GDP) in a sample of 37 Sub-Saharan African countries over the period 1964­94. Although he emphasizes the relatively greater importance of per capita income and geography as determinants of the region's trade flows, these results suggest that, on average, taxes on international trade (exports plus imports) in the early 1990s were associated with a 12 percentage point reduc- tion in the ratio of total trade of goods and services to GDP in Sub-Saharan African countries--a large value compared with the average trade-to-GDP ratio of 68 percent for these countries over the same period.2 Another relevant study is that of Santos-Paulino and Thirlwall (2004), who eval- uate the effects of trade liberalization on aggregate import and export growth and the trade balance, using a pooled sample of 22 developing countries covering the mid- 1970s to late 1990s. Import and export taxes, measured using collection duties, reduce import and export growth, but the effect is marginal. By far the largest impact on growth of trade volumes and the trade balance in their estimates is attributed to the dummy variable used to signify periods of trade liberalization. Africa is found to be particularly sensitive to trade liberalization, with import growth rising by 8.4 per- centage points (as against an average of 6.19 percentage points for all countries) and export growth rising by a lower 3.58 percentage points (compared with 1.56 per- centage points for all countries). Liberalization is therefore associated with a worsening of the trade balance in developing countries, which can lead to an external constraint on growth. The esti- mates suggest that the trade balance in developing countries worsens by more than 2 percent of GDP during liberalization periods. Similar estimates (2.8 percent of GDP) are found by the United Nations Conference on Trade and Development (UNCTAD 1999), using a sample of 16 developing countries over the period 1970­95. By con- trast, Parikh and Stirbu (2004) find a more mixed set of results, with the trade balance improving during liberalization in 10 of the 29 developing countries for which signif- icant results are obtained. There is, therefore, substantial heterogeneity in the response of trade flows to liberalization across developing countries. 80 | L A W R E N C E E D WA R D S These studies have some serious limitations. The first is the use of a dummy vari- able to estimate the effect of liberalization on trade flows. This holds as well for other studies in this field (Papageorgiou, Michaely, and Choksi 1991; Greenaway and Sapsford 1994; Bleaney 1999). Trade liberalization periods are frequently accompanied by numerous other policy reforms, the effect of which will also be attributed to the dummy variable. The result may be to bias estimates of export growth from trade liberalization downward and those for import growth upward if improved confidence in the country leads to capital inflows and an appreciation of the currency, as was experienced in a number of developing countries (Bleaney 1999; UNCTAD 1999). Dummy variables also do not adequately account for trade liberalization involv- ing different protection instruments (tariffs, nontariff barriers, export taxes, and trade subsidies) or variations in the pace and depth of reform across countries. Def- inition of the commencement of the liberalization period therefore depends crucially on the definition of liberalization employed and on the aspects of the reform process that are emphasized (Santos-Paulino 2005). A second limitation of these studies is that the mechanisms through which liber- alization influences trade flows, particularly exports, are not adequately specified in the trade functions estimated. For example, the effects of lower tariffs on imports, particularly on intermediate inputs, are not included in many of the export relation- ships. Yet the available evidence suggests that the effect of input tariffs on export vol- umes can be large. Collier and Venables (2007) estimate that the waiver on the rule of origin for textile inputs for some Sub-Saharan African countries is a significant source of the quadrupling of apparel exports to the United States in response to the African Growth and Opportunity Act (AGOA). Edwards and Lawrence (2008) ana- lyze 44 manufacturing sectors in South Africa from 1990 to 2002 and find that reductions in input costs from liberalization contributed significantly to improved export growth, particularly of noncommodity manufactures. This leads to the third limitation. Trade liberalization has nonuniform effects across sectors of the economy. The export sector consists of heterogeneous firms and industries, and the effect of liberalization on export performance is expected to dif- fer among them. Yet the bulk of the studies, in particular the cross-country studies, analyze aggregate export performance. It could be extremely misleading to draw implications from the aggregate cross-country studies about the likely impact of lib- eralization on developing economies without taking account of structural and com- positional differences between countries. There is a need, therefore, to supplement the aggregate analysis with firm- or sector-level studies. Trade Policy Reform and Trade Performance in Africa from the Early 1990s This section evaluates Africa's export performance and trade policy reform in the 1990s. The discussion provides the background for the empirical analysis that follows. P R O T E C T I O N I S T P O L I C I E S A N D M A N U FA C T U R I N G T R A D E F L O W S I N A F R I C A | 81 Export Performance The source of Africa's marginalization in world trade prior to the 1990s was two- pronged: "[Africa] experienced declining market shares for its major exports which, in turn, were of declining relative importance in world trade" (Ng and Yeats 1996, 8). To evaluate post-1990 trends, a similar decomposition is conducted using country- level data obtained from the World Development Indicators (WDI) database and product-level data from the United Nations Commodity Trade Statistics Database (Comtrade), SITC rev. 2, at the three-digit level.3 Only countries for which data are available in both periods are included in the analysis. Figure 1 presents a comparative perspective of merchandise exports from Sub- Saharan Africa in current prices and in constant prices between 1990 and 2006. The data are obtained from the WDI and cover 99 countries, 37 of which are in Sub- Saharan Africa. The country sample accounts for between 70 and 75 percent of the total available value of exports in each period; for Sub-Saharan Africa the figure is 93­97 percent. Sub-Saharan Africa is split into the Southern African Customs Union (SACU) and the rest of Sub-Saharan Africa (RSSA). The latter group is further sub- divided into low-income and middle-income countries.4 The continued marginalization of Sub-Saharan Africa in world trade during the 1990s is clearly depicted in figure 1. Export performance in all the country groupings for Sub-Saharan Africa was poor in relation to world performance during the decade, whether exports are measured in nominal or in real terms. By contrast, from 2000 on, the value of Sub-Saharan African exports (in U.S. dollars) grew sharply, and by 2006 the figure had almost tripled. Much of this growth is related to the commodity price boom. For example, the very strong export growth experienced by middle-income Sub-Saharan Africa outside the SACU is almost entirely assignable to the oil-rich countries of Equatorial Guinea and Angola, which experienced an oil price boom. Measured in volume, Sub-Saharan Africa's export performance after 2000 is substan- tially weaker.5 As a consequence, whereas Sub-Saharan Africa's share of world exports in value terms was around 2.7 percent in both 1990 and 2006, in volume terms its share declined from 2.6 to 1.6 percent over the period. The marginalization of Sub-Saharan Africa in world exports is more drastic when the region is compared with other devel- oping countries. Sub-Saharan Africa's share of developing-country exports falls from 15.6 to 8.0 percent when current U.S. dollar values are used and from 13.8 to 5.2 percent when export volume data (in 2000 prices) are used. The substantial heterogeneity in export performance across countries within Sub- Saharan Africa can be seen in figure 2, which presents average annual growth in mer- chandise export volumes by country between 1990 and 2006. In most countries of the region, export growth was relatively poor, exceeding the world average (7 per- cent per year) in 11 cases and the developing-country average of 11 percent per year in only 4 cases--Mozambique, Lesotho, Sudan, and Uganda. Further insights into the source of the region's marginalization in world trade can be obtained through a more disaggregated analysis of trade flows. Table 1 presents a constant-market-shares analysis in which export growth is decomposed into changes 82 | L A W R E N C E E D WA R D S FIGURE 1. Merchandise Exports: World and Sub-Saharan Africa, 1990­2006 current prices 800 700 600 500 1990 = 100 400 300 200 100 0 1990 1992 1994 1996 1998 2000 2002 2004 2006 constant prices 600 500 400 1990 = 100 300 200 100 0 1990 1992 1994 1996 1998 2000 2002 2004 2006 RSSA, middle income RSSA, low income SACU World Developing countries excluding SSA Source: Own calculations using data from World Development Indicators database. Note: RSSA, rest of Sub-Saharan Africa; SACU, Southern African Customs Union; SSA, Sub-Saharan Africa. Export volumes are converted to values using 2000 prices in U.S. dollars. P R O T E C T I O N I S T P O L I C I E S A N D M A N U FA C T U R I N G T R A D E F L O W S I N A F R I C A | 83 FIGURE 2. Average Annual Growth in Merchandise Export Volume, by Sub-Saharan African Country, 1990­2006 25 20 15 Developing countries (11%) 10 percent World (7%) 5 0 ­5 ­10 so e ca U ud o Bu Re an n rk pu da a lic Se aso M A ell li ad n es Ta asc la nz ar Sw T nia zil o Ve nd G rde Se mb a ne ia Ke gal ng Ga ya n So ot run . ut sw di Af a Be ica Zi Ma nin d bw i o, M 'Ivo e D au ire . ius ig . M N eria rit er Ca Gu nia er a bi andn a, a e Co ôte ba aw B u p N Rep yc Ma Za han h an m ine Le iqu n g a o, bo am w o Th S th Ca a og ag go B Re au ig in b n G R o pe a em rit r a a C m l F b h am oz Co M fri ng lA ra nt Ce Source: Own calculations using data from World Development Indicators database. attributable to growth in world exports, favorable or unfavorable structural concen- tration in products, and changes in relative competitiveness (Richardson 1971). The approach followed is similar to that of Ng and Yeats (1996). The data are drawn from UN Comtrade, and the classification is based on the three-digit-level SITC rev. 2.6 The sample includes 89 countries, 25 of which are in Sub-Saharan Africa. Looking first at total trade, if South Africa and Sub-Saharan Africa excluding South Africa had retained their 1990­91 market shares, the total value of exports in 2005­6 would have been 216 percent higher. Instead, the value of exports for South Africa and for Sub-Saharan Africa excluding South Africa rose by 127 and 188 per- cent, respectively. A substantial part of this shortfall (30 to 50 percent of initial export values) can be attributed to the region's concentration in products that grew relatively slowly during the period. For example, world trade in primary products and resource-based manufactures, which account for between 80 and 90 percent of Sub-Saharan Africa's exports, maintained a constant share in world trade, but high-technology products, of which the region exports very little, rose from 16 to 20 percent of total world trade. Declining competitiveness, measured by declining world market share at the product level, also contributed to the shortfall for South Africa ( 59 percent), but not for Sub- Saharan Africa excluding South Africa, where the competitiveness effect was positive, at 24 percent. As shown in row 5 in the table, a substantial portion (17 percentage 84 | L A W R E N C E E D WA R D S TABLE 1. Constant-Market-Shares Analysis: Sources of Export Growth, World Regions and Sub-Saharan Africa, 1990­91 to 2005­6 (change as percent of 1990­91 values, in U.S. dollars, except as otherwise specified) Developing High-income South Sub-Saharan Africa, countries (OECD) Africa excluding South Africa All products 1. Total change 492 155 127 188 2. Product composition effect 7 2 30 52 3. Competitiveness effect 283 58 59 24 4. World growth effect 216 216 216 216 5. Competitiveness in growing 196 41 38 17 commodities Initial trade value (billions of 404 2,453 24 19 U.S. dollars) Manufacturing 6. Total change 680 157 300 290 7. Product composition effect 12 2 22 61 8. Competitiveness effect 464 63 99 128 9. World growth effect 222 222 222 222 10. Competitiveness in growing 325 45 64 95 commodities Initial trade value (billions of 241 2,138 8 6 U.S. dollars) Source: Own calculations using UN Comtrade data at the three-digit Standard International Trade Classification (SITC) level and initial period shares as weights. Note: Low-income, middle-income, and developing-country results include Sub-Saharan Africa. The decomposition is calculated as q s0 Q [ i s0 i Qi s0 Q ] i 1 Qi si, where q is total exports of the focus country, s is the export share of the focus country, Q is total exports, the sub- script i refers to commodity, and the superscripts 0 and 1 refer to initial and final periods. The first term on the right- hand side is the world growth effect, the second term is the commodity effect, and the final term is the competitive effect. See Richardson (1971) for some of the shortcomings of this approach. points) of the improved competitiveness effect for that subgroup has been driven by a restructuring of exports toward rapidly growing products. The results for manufacturing are more positive. The world market share of South Africa and for Sub-Saharan Africa excluding South Africa rose during the period, as reflected in the 290­300 percent increase in export value compared with the 222 per- cent increase in world trade. Although poor product composition reduced growth, it was more than offset by improvements in competitiveness and by structural shifts toward rapidly growing products (row 10). The trends in Sub-Saharan African trade identified by Ng and Yeats (1996) in the four decades prior to 1990 are thus only partly replicated in the subsequent period. Although exports continue to be concentrated in products with stagnant shares in P R O T E C T I O N I S T P O L I C I E S A N D M A N U FA C T U R I N G T R A D E F L O W S I N A F R I C A | 85 FIGURE 3. Merchandise Exports as a Share of GDP: Country Average, by Income Group and for Sub-Saharan Africa, 1990­2006 0.45 0.40 ratio of exports to GDP 0.35 0.30 0.25 0.20 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 20 20 low and middle income high income SSA world Source: Own calculations using data from World Development Indicators database. Note: GDP, gross domestic product; SSA, Sub-Saharan Africa. The sample consists of 99 countries, 37 of which are in Sub-Saharan Africa. Values reflect the simple country average for the region or income group. world markets, the region has managed to increase its world market share in many products, including those that are growing rapidly. Two important caveats to this conclusion need to be highlighted. First, the world experienced a commodity price boom during the latter part of the period analyzed, and the surge had a considerable impact on the value of Sub-Saharan Africa's exports, as can be seen in figure 1. Growth in trade volumes remains relatively weak in Africa, and a fall in commodity prices, as has been experienced recently, will adversely affect Sub-Saharan Africa's share of world trade. Second, alternative indicators, such as the ratio of exports (or trade) to domestic GDP, lead to different interpretations of the extent to which Sub-Saharan Africa is marginalized in world trade. Figure 3 presents this ratio using merchandise exports for Sub-Saharan Africa and for income-group categories. What is striking is that the level and growth of the ratio of merchandise exports to GDP for the average Sub-Saharan African country is very similar to the average for all low-income and middle-income economies between 1990 and 2006. The region's trade performance from 1990 on is no different from that of other developing countries, once bench- marked against GDP.7 An implication is that Sub-Saharan Africa's declining share in world trade may largely be a consequence of relatively poor economic growth, as is argued by Rodrik (1997). Nevertheless, even with this indicator, the region remains less open than the world average. 86 | L A W R E N C E E D WA R D S Tariff Reform Comparisons of protection across countries are fraught with difficulties. These include the lack of consistent protection data over time, problems in calculating ad valorem equivalents for non­ad valorem tariff rates and nontariff barriers, and various biases associated with the aggregation process (Anderson and Neary 1994). Nonetheless, an evaluation of protection using nominal tariff data is possible and allows for an instructive comparative analysis of trade reform across countries. To pursue such an analysis, detailed most favored nation (MFN) and applied tariff data are obtained for 115 countries from UNCTAD's Trade Analysis and Informa- tion System (TRAINS) database for the period 1990 to 2006. The countries are made up as follows: low income, 26; lower middle income, 33; upper middle income, 31; high income countries not members of the Organisation for Economic Co- operation and Development (OECD), 14; and OECD countries, 11. The sample includes 29 African countries, some of which are North African. Only those coun- tries in which average protection in both periods can be calculated are included.8 An important limitation is that the tariff rates do not include ad valorem equiva- lents for non­ad valorem tariff rates. This limitation mainly affects agriculture, where non­ad valorem tariffs are used extensively, but it remains an important caveat affecting the interpretation of the data. Figure 4 shows the maximum, minimum, and simple average MFN tariff for various country groups for two periods: 1990­95 and 2004­6. The simple country average rather than the import-weighted average of the corresponding measure for the individual countries is presented. This avoids domination by any individual country--for example, South Africa in Africa, India in the low-income group, and China in the lower-middle-income group. The country-level observation is the import-weighted average tariff for the periods 1990­95 and 2004­6. A number of observations can be drawn from the data. There has been a con- siderable reduction in average MFN tariffs for all income groups since the early 1990s. Low-income economies experienced relatively large decreases in the aver- age MFN tariff, from 22 to 12.3 percent. Middle-income economies saw slightly smaller declines in MFN protection, although the initial tariffs were far lower than in low-income economies. MFN tariffs also fell in African economies, from 19 to 13 percent, despite the meager offers made in the Uruguay Round to reduce bound rates (Wang and Winters 1998). The decline in protection is roughly equiv- alent to that in middle-income economies but is less than in the average low- income economy. The average change hides substantial variation in the extent of tariff liberalization across countries. As shown in figure 4, the spread of tariff rates across countries declined in almost all groups, including Africa. The narrowing of the range is great- est for low-income countries, as a result of substantial reductions in the average tariff in Bangladesh, which fell from 76 to 18.2 percent over the period. The extent of liberalization across African economies also varies widely. Figure 5 presents the estimated percentage change in MFN tariffs for the sample of African economies. The extent of liberalization is calculated as (t1 t0) (1 t0), where t1 P R O T E C T I O N I S T P O L I C I E S A N D M A N U FA C T U R I N G T R A D E F L O W S I N A F R I C A | 87 FIGURE 4. Maximum, Minimum, and Average Country Tariff by Region and Period, Most Favored Nation (MFN) Tariffs. 1990­95 and 2004­6 80 maximum 70 average 60 minimum 50 MFN tariff 40 30 20 10 0 1990­95 2004­6 1990­95 2004­6 1990­95 2004­6 1990­95 2004­6 1990­95 2004­6 low income lower middle upper middle high income Africa income income (OECD) Source: UNCTAD Trade Analysis and Information System (TRAINS) database and own calculations. Note: OECD, Organisation for Economic Co-operation and Development; UNCTAD, United Nations Conference on Trade and Development. The sample consists of 115 countries for which data are available for both periods. The Africa sample includes North African countries. and t0 refer to tariff rates in the final and initial periods, respectively. This is there- fore an estimate of the percentage change in the tariff-inclusive border price. Large decreases (greater than 10 percent) in the import-weighted average MFN tariff are found in Morocco, Libya, Mauritius, and Côte d'Ivoire. These economies had average protection rates in excess of 20 percent and up to 45 percent, in the case of Morocco in 1990­91. By 2004­6 average protection had fallen to 19 percent in Morocco, 12 percent in Nigeria, and less than 10 percent in the others. Protection also rose in a number of African economies, with particularly large increases (more than 5 percent) for the Seychelles. Smaller increases were experienced in Ghana, Madagascar, and the Central African Republic. Some of this increase reflects changes in tariff rates, but shifts in the composition of import weights and the tariffication of nontariff barriers also affect the average. Overall, more than half (16) of the 29 African economies that were studied expe- rienced tariff reductions greater than the simple country average of all 115 countries in the sample (3.9 percent). This indicates that many African economies have opened up considerably from the early 1990s. The decline in tariffs, however, comes off a high base, and current protection rates remain relatively high. The average level of 88 | L A W R E N C E E D WA R D S FIGURE 5. Change in Import-Weighted Most Favored Nation (MFN) Tariff, Selected African Countries, 1990­95 to 2004­6 10 5 0 all economies (­3.9%) percent ­5 ­10 ­15 ­20 L o Cô Ma ibya d' ius N ire Bu K eria oz F a bi o Et que an a Ug da So Za da Zi Af ia ba a nz e Tu nia Ch ia t, Al ad ab ria ria a . ui i G ea Co Ca M n o, ro i Re on l A M G . of ca ag na Se pu r he c s l G law ng me al to M ep a lle yc bli M kina ny Rw pi m ric Ta bw o c am as ut mb s Re c oc n fri ad ha Ar ge o an ab ni te it a n as o R p r e ig Iv ur hi or h M yp ua Eg Eq ra nt Ce Source: UNCTAD Trade Analysis and Information System (TRAINS) database. Note: UNCTAD, United Nations Conference on Trade and Development. MFN protection, at 13.1 percent for African economies in 2004­6, exceeds that of low-income economies (12.3 percent), middle-income economies (9.2­10.2 percent), and OECD countries (3.1 percent); see table 2. This is also the case at the disaggre- gated country level. Average protection exceeds the simple country average of all low-income and middle-income economies (10.5 percent) in 19 of the 29 African economies in the sample.9 Up to now, the focus has been on MFN rates. Table 2 presents data on levels of and changes in applied tariffs that include preferences granted under the various pref- erential trade agreements. The data show more extensive liberalization from the early 1990s in response to the implementation of various preferential trade agreements during this period. Current applied rates are lower than MFN rates. The difference, however, does not appear substantial and does not alter any of the conclusions drawn on the basis of the MFN rates. The aggregate data, again, hide considerable variation in protection at the sector level. Of particular interest for this paper is the extent to which liberalization reduces the cost of accessing key intermediate inputs. To evaluate this, table 2 presents the simple average tariff and change in tariffs on consumer, intermediate, and capital goods for Sub-Saharan Africa and by income group. The end-use categorization is based on the broad economic category (BEC) classification of the United Nations TABLE 2. Simple Average Protection by End Use, Country Income Groups and Sub-Saharan Africa, 1990­95 and 2004­6 (percent) Capital goods Consumption goods Intermediate goods Passenger vehicles Total 1990­95 2004­6 Change 1990­95 2004­6 Change 1990­95 2004­6 Change 1990­95 2004­6 Change 1990­95 2004­6 Change Mean import-weighted tariff, MFN Low income 19.4 8.6 7.7 28.6 18.4 7.0 20.4 11.4 6.6 48.9 28.2 11.1 22.1 12.3 7.3 Lower middle income 13.2 6.7 5.3 21.3 17.4 2.6 13.9 8.7 4.2 41.9 28.2 6.5 15.3 10.2 4.1 Upper middle income 11.5 6.0 4.7 17.6 15.5 1.9 10.3 6.3 3.6 32.4 23.9 4.9 12.6 9.2 2.9 OECD 4.6 1.6 2.9 9.9 5.8 3.7 4.2 2.2 1.9 9.4 5.7 3.2 5.8 3.1 2.5 Sub-Saharan Africa 16.2 8.0 6.6 25.3 21.6 2.5 17.2 11.1 5.0 43.3 25.3 11.1 19.0 13.1 4.8 Simple average applied tariff Low income 18.6 8.2 7.5 27.0 16.4 7.4 19.3 9.7 7.2 47.1 27.2 10.6 21.0 10.8 7.6 Lower middle income 12.7 6.0 5.5 20.4 14.7 4.0 13.3 7.1 5.1 40.7 25.5 7.4 14.7 8.6 5.0 Upper middle income 11.3 4.6 5.7 17.1 12.4 4.2 10.1 4.4 5.1 32.3 18.4 9.3 12.3 7.0 4.7 OECD 3.9 1.2 2.6 8.0 4.4 3.3 3.4 1.8 1.6 9.0 4.9 3.6 4.7 2.4 2.2 Sub-Saharan Africa 16.1 7.7 6.8 24.8 20.0 3.4 16.8 9.4 6.0 43.2 24.9 11.3 18.7 11.8 5.6 Source: Own calculations using data from UNCTAD Trade Analysis and Information System (TRAINS). Note: MFN, most favored nation; OECD, Organisation for Economic Co-operation and Development; UNCTAD, United Nations Conference on Trade and Development. 89 90 | L A W R E N C E E D WA R D S Statistics Division. Passenger vehicles (BEC 51) are included as a separate item because they are both an intermediate good and a final consumer good. The data show that the decline in tariffs from the early 1990s in African coun- tries has been concentrated in capital and intermediate inputs. There was little lib- eralization of tariffs on consumer goods; they declined from 25.3 to 21.6 percent, which would have raised effective protection on consumer goods in many of the African countries sampled. Despite reductions of MFN tariffs on capital and inter- mediate goods, tariffs on these products remain high (8.0 and 11.1 percent, respec- tively). The tariffs continue to tax production and impose a barrier to exports, par- ticularly of manufactured goods, where inputs constitute a relatively high proportion of overall production costs. The impact on Africa may be particularly large because domestic alternatives to imported inputs are not available. For example, World Bank Enterprise Survey data indicate that between 30 and 65 percent of inputs purchased by African manufacturing exporters are imported (Edwards, Rankin, and Schöer 2008). Although exporters have access to import duty rebate schemes in some African countries, utilization of these schemes by firms is low (Chandra et al. 2001; Clarke 2005). Summary In conclusion, Africa presents a mixed picture with regard to trade flows and liber- alization from 1990 on. The share of Sub-Saharan Africa in world exports continued to decline from the 1990 level when measured in volume terms but remained stable when measured in current U.S. dollars. Although an unfavorable product concentra- tion adversely affected export growth, there is evidence of improved competitiveness, and the region has increased its world market share in many products, including manufactured products and products that are growing rapidly in world trade. In addition, export performance in the average Sub-Saharan African economy is no different from that of other developing countries when evaluated using the ratio of merchandise exports to GDP. The results for the post-1990 period therefore differ from those found by Ng and Yeats (1996) for the period prior to the early 1990s. In terms of protection, Sub-Saharan African economies have made reasonable progress in reducing MFN rates and simplifying their tariff structures. The extent of tariff liberalization exceeded the world average for over half of the African countries analyzed. Liberalization, however, occurred from a high initial base, and average tar- iffs remain relatively high, even on key intermediate and capital inputs. These may continue to pose a barrier to export growth. Econometric Analysis of the Effect of Tariff Barriers on Manufacturing Trade in Developing Economies This section estimates the impact of tariff liberalization from 1990 to 2004 on man- ufactured exports, manufactured imports, and the trade balance in African and other developing countries. The focus on manufacturing trade is restrictive in the case of P R O T E C T I O N I S T P O L I C I E S A N D M A N U FA C T U R I N G T R A D E F L O W S I N A F R I C A | 91 Africa, where manufacturing accounts for only 20­25 percent of exports; a high pro- portion of these exports is also resource based. The concentration on manufacturing in part reflects data constraints regarding the availability of tariff data for primary sectors over time. However, there are additional reasons why it is important. Growth in manufacturing is closely associated with growth accelerations in devel- oping countries, including those in Sub-Saharan Africa (Pattillo, Gupta, and Carey 2005; Johnson, Ostry, and Subramanian 2007). The potential for output growth in manufacturing production is often greater than for agricultural and resource-based exports, which encounter diminishing returns to scale because of limited endow- ments (Collier and Venables 2007). Furthermore, diversification into manufacturing production helps insulate resource-based economies from volatile natural resource prices and politically and economically disruptive rent-seeking behavior (Sachs and Warner 1999). Finally, manufactured exports are likely to grow faster than primary exports as the global economy expands because their income elasticity of demand is higher (Elbadawi 2001). As a consequence, policies to stimulate growth of manufac- tured exports in Sub-Saharan Africa have frequently been recommended (World Bank 2000). This does not imply that improved economic growth cannot or should not be achieved through expansion of primary product exports, as has been done in, for example, Chile. The recent commodity price boom and growth in China provide an important opportunity for African economies to expand trade in primary commodi- ties. Moreover, there is substantial scope for technical progress in agriculture, often more so than in manufacturing (Martin and Mitra 2001). Nevertheless, continued growth in manufactures will need to form part of the overall strategy for improving economic growth. Mayer and Fajarnes (2008), for example, show that if income growth in Sub-Saharan Africa is to achieve the Millennium Development Goals (MDGs), a ninefold increase in the region's manufactured exports will be needed. Data The main database used is the trade, production, and protection database produced by Nicita and Olarreaga (2006). It provides information on trade flows, various measures of tariff protection, and production for 28 three-digit ISIC rev. 2 manufac- turing sectors from 1976 on.10 The analysis in this section draws on a sample of 30 developing countries (of which 6 are in Sub-Saharan Africa and 4 in North Africa) and 20 OECD economies over the period 1990­2004. Country selection is based on the availability of tariff data. Countries were included only if at least six years (four years for African coun- tries) of tariff data were initially available in the period 1990­2004. To ensure as much coverage as possible, the MFN tariff data were then updated using Harmo- nized System (HS) six-digit-level tariff schedules obtained from the country pages on the World Trade Organization (WTO) Web site. Tariff data for South Africa for the entire period were obtained from Edwards (2005). Tariffs for missing periods that were bounded by available tariff data in the preceding and subsequent years were 92 | L A W R E N C E E D WA R D S interpolated using simple averages. Overall, this yielded an average of 10 years of tariff data for each developing country. The database has some important limitations (Nicita and Olarreaga 2006). It is not balanced, and data are missing for many countries in the early 1990s. The tariff data do not include estimates of ad valorem equivalents for specific and other non­ad valorem rates. This is particularly problematic for agricultural products but is expected to be less of a problem for manufacturing, on which this paper concentrates. Non- tariff barriers are also omitted. The analysis in this paper thus focuses entirely on direct tariff measures, and the potentially important effects of nontariff barriers are ignored. Additional variables relevant for the study were included in the database. They include a consumer price index (CPI)­based real effective exchange rate (REER) index, obtained from the International Monetary Fund's International Financial Sta- tistics, and updated industrial production indexes obtained from the Instat database of the United Nations Industrial Development Organization (UNIDO).11 To capture foreign demand effects, a sector-level export-weighted foreign GDP (gdpf ) index is calculated as 10 k gdpfijt ij GDPtk (1) k k where GDP is valued in constant 2000 prices and ij is the average share of exports of product i in country j to each of the top 10 trading partners (k) over the period 2001 to 2002. Finally, a variable to capture the effect of tariff liberalization on the cost of pro- duction is constructed. This variable, termed input tariff cost (tcost), is constructed as follows: tcostijt a n tnt ij (2) n n where t is the tariff rate and aij is the quantity of intermediate input n used in the pro- duction of one unit of i in country j. The value of tcost is therefore a measure of the proportion of total costs accounted for by tariffs on intermediate inputs. An input-output table for the United States consisting of 17 aggregated manufac- turing sectors, based on the three-digit ISIC rev. 2 code, is used to calculate the input n coefficients (aij). The input-output table is obtained from Nicita and Olarreaga (2006). The tariff data used to construct tcost are aggregated to be consistent with this classification.12 There are various caveats regarding the variable tcost. Most important, tariff pro- tection on agricultural and mining inputs is not included because the input-output table used does not contain input shares for these sectors. This is expected to particu- larly affect tcost values for agricultural product­intensive manufacturing sectors. Measurement of protection in the agricultural sector is, however, difficult given the extensive use of nontariff barriers and specific tariffs in the sector. Protection on mining products is generally low throughout the period and is not expected to have a P R O T E C T I O N I S T P O L I C I E S A N D M A N U FA C T U R I N G T R A D E F L O W S I N A F R I C A | 93 substantial impact on tcost. A second limitation is that the variable excludes duty rebates on inputs that may be granted to exporters. Despite the limitations, this database has a number of advantages. First, it includes direct measures of tariff protection. Second, sectoral data are included, enabling a more disaggregated analysis of the effect of trade liberalization on trade flows. Third, the inclusion of tcost allows for an alternative specification of the export relationship that captures the effect of tariff liberalization on input costs. Finally, the database covers a period in which considerable liberalization and changes in trade flows have occurred in many economies. This variation in the data lends itself to econometric analysis. Some of the key data related to trade flows, tariffs, and tcost are presented in annex table A.1. The trends in trade flows and tariff liberalization are consistent with the more detailed analysis presented earlier. Average tariffs fell in all country groups, including Africa, although tariff protection remains relatively high in that region. Tariff liberalization also reduced average production costs (tcost) in many countries, with particularly large decreases in the low-income country group (from 14 to 7 per- cent) and the lower-middle-income group (11 to 3.3 percent). The decline is reason- ably widespread, with 43 of the 51 countries sampled experiencing lower production costs from liberalization. Of the 30 developing economies covered, production costs decreased in 23 between the early 1990s and 2004. As noted earlier, manufacturing trade volumes increased for many economies from 1990 on, with relatively slower growth experienced in Africa. Aggregate import growth was also high, but it lagged export growth for low-income and lower-middle- income regions. The effect is an improvement in the real manufacturing trade balance in 21 of the 30 developing countries in the sample. The improvements in the trade balance for many of the developing countries point to a potentially important differ- ence from the studies of Santos-Paulino and Thirlwall (2004) and UNCTAD (1999), where the trade balance worsened after liberalization. Manufactured Exports A simple reduced-form imperfect substitution model, as discussed by Goldstein and Khan (1985), is used to estimate the export relationship. The estimated equation takes the form: ln(x)ijt 0 1 ln(gdpf )ijt 2 ln(reer)jt 3 ln(prod)ijt 4 ln(tcost)ijt ijt , 0, (3) where x denotes export volumes, gdpf is foreign GDP, reer is the real effective exchange rate, prod is an index of real industrial production (a proxy for produc- tive capacity), and tcost is a measure of input costs associated with tariff protection. The subscripts i, j, and t represent sector, country, and time, respectively.13 Various methods are used to estimate the relationship. One option is to pool the data and estimate the relationship using ordinary least squares (OLS). A shortcom- ing of this method is that time-invariant characteristics of sectors and countries are 94 | L A W R E N C E E D WA R D S likely to be correlated with the explanatory variables, leading to biased estimates. In an attempt to deal with this problem, two approaches are followed. In model A, dummy variables for country (cntryj), sector (isici), and year ( t) are included in the OLS regressions. Omitted variable bias may nevertheless persist. For example, low tariffs on mining sectors in Africa reflect a comparative advantage in these sectors arising from a relative abundance of natural resources. The high exports of these products will erroneously be attributed to the tariff variable in the estimates. Furthermore, as noted earlier, countries frequently implement other reforms, such as capital account liberalization, concurrently with tariff liberalization. To avoid these biases, model B includes country-by-sector specific effects (cntryisicij) and country- by-year specific effects (ctryyearjt). The error component of equation (3) is therefore specified as ijt t cntryisicij ctryyearjt ijt. (4) The results for the export relationship are presented in table 3. Separate estimates are presented for the full sample of countries and for developing countries. To investigate the African trade relationship, two additional regressions are estimated. First, a dummy variable for Africa ( 1) is interacted with the tariff variable in the developing-country regressions. Second, the export relationship is estimated for the Africa sample alone. All relationships are estimated using both mirror export data and own-country reported data. Note that in model B the reer variable is dropped because it correlates perfectly with ctryyearjt. Looking at the results for all countries (columns 1 and 5), the estimated export relationship appears to be well defined, with a high R2 and a number of significant variables of the correct sign. Export volumes respond positively to rising foreign income, a depreciation of the real effective exchange rate, and increased domestic productive capacity. Except for the reer, the estimated elasticities for these variables are largely unaffected by the choice of mirror export or own-export data. Turning to the trade liberalization variable, mirror export volumes are found to be responsive to tcost, and the estimated elasticity declines from 0.385 for model A (column 1) to 0.137 for model B (column 5). The latter, the preferred estimate, sug- gests that a 10 percent reduction in the input-weighted average manufacturing tariff raises manufactured export volumes by 1.37 percent. The results for tcost using own exports in model A are generally similar, but the relationship is not robust to the inclusion of the country-by-sector and country-by-year fixed effects. The stability of the relationship across regions can be seen in the developing- country and Africa results. There is some variation in the estimated relationships across country groups and across different models. The relationship between export volumes and reer, foreign GDP, and industrial production in developing countries (columns 2 and 6) is generally similar to the full sample results. Rising protection, measured using tcost, is also found to affect export performance neg- atively, and the coefficients are not significantly different from those of the full sample. TABLE 3. Determinants of Manufactured Export Performance, World, Developing Countries, and Africa Model A Model B All countries Developing Developing interaction Africa All countries Developing Developing interaction Africa (1) (2) (3) (4) (5) (6) (7) (8) Mirror export data gdpf 0.419*** 0.222*** 0.223*** 0.271*** 1.152*** 0.499 0.506 1.914*** (0.018) (0.023) (0.023) (0.045) (0.244) (0.313) (0.314) (0.526) reer 0.351*** 0.606*** 0.593*** 1.689*** (0.094) (0.139) (0.14) (0.374) prod 0.583*** 0.678*** 0.674*** 0.615*** 0.426*** 0.402*** 0.402*** 0.027 (0.046) (0.064) (0.064) (0.137) (0.033) (0.049) (0.049) (0.125) tcost 0.385*** 0.337*** 0.306*** 0.708*** 0.137*** 0.222** 0.247** 0.067 (0.031) (0.064) (0.067) (0.178) (0.042) (0.088) (0.098) (0.224) Africa 1.434*** (0.274) tcost*Africa 0.156 0.131 (0.102) (0.221) N 15,513 7,661 7,661 1,850 15,513 7,661 7,661 1,850 R2 0.793 0.769 0.77 0.759 0.949 0.94 0.94 0.928 F 628 338 333 101 137 97.4 97.3 56.7 Own-export data gdpf 0.367*** 0.175*** 0.176*** 0.197*** 1.042*** 0.349 0.340 1.08* (0.02) (0.026) (0.026) (0.058) (0.248) (0.326) (0.327) (0.62) reer 0.072 0.041 0.043 1.71*** (0.103) (0.156) (0.157) (0.495) prod 0.523*** 0.635*** 0.632*** 0.603*** 0.341*** 0.33*** 0.330*** 0.165 (0.051) (0.072) (0.072) (0.175) (0.034) (0.052) (0.053) (0.166) tcost 0.342*** 0.216*** 0.206*** 1.019*** 0.003 0.113 0.083 0.27 (0.033) (0.071) (0.074) (0.244) (0.043) (0.094) (0.101) (0.333) Africa 1.610*** (0.347) tcost*Africa 0.068 0.224 (0.124) (0.275) N 14,963 7,204 7,204 1,510 14,963 7,204 7,204 1,510 R2 0.762 0.743 0.743 0.666 0.949 0.939 0.939 0.901 F 506 274 271 52.8 132 90.9 90.8 32.3 95 Note: The coefficients for the various sector-, time-, and country-specific effects are not shown. Standard errors are in parentheses. gdpf, foreign GDP; reer, real effective exchange rate; prod, real industrial production; tcost, input costs associated with tariff protection. *p .1 **p .05 ***p .01 96 | L A W R E N C E E D WA R D S Looking at the export relationship in Africa, none of the interaction terms in columns 3 and 7 are significant, suggesting that the effect of tariff protection on exports in Africa is no different from that for other developing economies. The results for model A using the Africa sample (column 4) are also similar to those for the full sample of developing countries, although African exports of manufactured goods appear to be more than twice as sensitive to the reer. Lower input costs from trade liberalization are relatively strongly associated with improved export perform- ance in the Africa sample. However, most of the results for the Africa sample become insignificantly different from zero once country-by-sector and country-by-year fixed effects are included in the model (column 8, model B), and only foreign GDP remains significant.14 Much of Africa's export performance therefore appears to be explained by factors other than tariff protection. This may reflect the relatively low share of manufactur- ing trade in total trade in African countries, as well as a dependence on resource- based products within manufacturing. Exports of resource-based products are less likely to be influenced by tariffs on inputs than are downstream products that use a relatively high proportion of manufactured inputs in the production process. Never- theless, the results for the pooled sample of developing countries suggest that as Africa deepens its manufacturing base, continued protection, particularly on inputs, will constrain export growth. A more detailed analysis of the contribution of tariff liberalization to export growth in developing countries from 1992­95 to 2003­4 is presented in table 4. The decomposition is based on two estimated tcost elasticities for developing countries: 0.222 and 0.337. Export volumes in the average developing economy grew by 110 percent over the period. Production costs associated with tariffs fell by 47 per- cent for the average economy, which translates roughly into an 11­16 percent increase in export volume, depending on the tcost elasticity selected. For the average African economy, the decline in tcost is associated with an increase in export volume ranging from 11 to 17 percent. This is equivalent to 14 to 20 percent of the change in total exports in the average African economy over this period. Thus, although liberalization raises export volumes, its contribution to overall export growth in the 1990s appears to be relatively small. TABLE 4. Contribution of Tariff Liberalization to Changes in Manufactured Export Volumes, for Africa and by Income Group, 1992­95 to 2003­4 Average ln change Estimated change in exports Country group tcost Exports Elasticity 1 ( 0.222) Elasticity 2 ( 0.337) Africa 0.49 0.82 0.11 0.17 Low income 0.73 1.12 0.17 0.25 Lower middle income 0.47 1.18 0.11 0.16 Upper middle income 0.28 0.99 0.06 0.09 Developing 0.47 1.10 0.11 0.16 Note: Export and tcost values reflect the simple country average. P R O T E C T I O N I S T P O L I C I E S A N D M A N U FA C T U R I N G T R A D E F L O W S I N A F R I C A | 97 Manufactured Imports The same estimation strategy is employed to estimate the determinants of manufac- tured import volumes. The estimated import equation is: ln(m)ijt bo d1 ln(gdp)jt d2 ln(reer)jt d3 ln(1 tar)ijt ijt, d 0 (5) where gdp is domestic GDP, reer is the real effective exchange rate, and tar is the average nominal MFN tariff on output. The tariff variable is included as (1 tar) to capture the effect of tariffs on the tariff-inclusive import price. The variables gdp and reer are only defined at the country level and therefore drop out of the estimation in model B. The estimated elasticities for the import relationship are presented in table 5. Only the results using mirror import data are presented because the results using own- country reported data are qualitatively similar. The coefficients in model A are generally of the expected signs: import volumes are negatively affected by a real depreciation of the currency, a decline in GDP, and higher tariffs. Looking more closely at the results for developing countries (columns 2 and 6), tariff protection negatively affects import volumes in both estimates, with an average elasticity of 0.74. A 10 percent increase in the tariff-inclusive price of MFN imports (1 tar) is therefore associated with a reduction in import volumes of 7.4 percent. This is a relatively inelastic response to liberalization and implies that MFN tariff liberaliza- tion from 1992­95 to 2003­4 in developing countries raised import volumes by 6 percent, on average. That figure is a small proportion of the 88 percent increase in import volumes experienced by the average developing country in this period. African manufactured import volumes are not well explained by import tariffs. The coefficient on the interaction term in columns 3 and 7 is significant and positive. Combining the interaction coefficient with the direct coefficient on (1 tar) implies a very small negative impact ( 0.3) of tariffs on import volumes for Africa in model A and a close to zero effect in model B. In the African sample estimates (columns 4 and 8), almost all the coefficients are insignificant. The exception is for the tariff variable in model A, but even this variable is insignificant once the various fixed effects are included. These results suggest that import volumes in Africa are driven by other country-specific effects not included in the regression. Why tariffs matter less for imports in Africa is unclear, but it may reflect the high dependence by African industries on imported inputs, as well as large capital inflows through foreign aid. A similarly poor import response to relative prices and import taxes in Africa is found by Santos-Paulino and Thirlwall (2004). Manufacturing Trade Balance The final relationship analyzed is the effect of trade liberalization on the trade bal- ance. This relationship depends on the relative responsiveness of exports and imports to tariff protection. Much of the current literature on developing economies suggests a muted response of exports to liberalization in relation to the response in import volumes, leading to worsening trade balances (UNCTAD 1999; Santos-Paulino and Thirlwall 2004). 98 TABLE 5. Determinants of Manufactured Import Volumes, by Country Group and for Africa Model A Model B All countries Developing Developing interaction Africa All countries Developing Developing interaction Africa (1) (2) (3) (4) (5) (6) (7) (8) reer 0.544*** 0.767*** 0.773*** 0.035 (0.06) (0.086) (0.086) (0.206) gdp 1.23*** 0.757*** 0.664*** 0.00 (0.102) (0.155) (0.157) (0.427) 1 tar 0.513*** 0.714*** 1.121*** 0.931*** 0.478*** 0.765*** 1.450*** 0.096 (0.087) (0.11) (0.145) (0.16) (0.146) (0.166) (0.222) (0.262) Africa 0.205 (0.538) Africa*(1 tar) 0.808*** 1.546*** (0.189) (0.334) N 16,855 8,666 8,666 2,293 16,855 8,666 8,666 2,293 R2 0.846 0.821 0.822 0.804 0.935 0.925 0.93 0.912 F 978 534 528 167 107 83 89.2 53 Note: The coefficients for the various sector-, time-, and country-specific effects are not shown. Standard errors are in parentheses. Values for reer and gdp are only defined at the country level and drop out of the estimations in model B. *p .1 ***p .01 P R O T E C T I O N I S T P O L I C I E S A N D M A N U FA C T U R I N G T R A D E F L O W S I N A F R I C A | 99 To test this effect, the following trade balance equation is estimated: tbijt bo 1 ln(gdp)jt 2 ln(reer)jt 3 ln(1 tar)ijt 4 ln(gdpf )ijt 5 ln(prod)ijt 6 ln(tcost)ijt 7 ln(tot)ijt ijt (6) where, in addition to the variables already explained, tb is the log ratio of the value of exports to the value of imports, and tot is the terms of trade calculated as the unit value of exports relative to imports. The trade balance is expected to improve in response to a depreciation of the real exchange rate, a rise in foreign GDP, a decline in domestic consumption (proxied by domestic GDP), an improvement in the terms of trade (tot), or a rise in productive capacity (prod). Liberalization has an ambigu- ous effect. Rising tariffs on inputs, measured by the variable tcost, are expected to diminish the competitiveness of exporters and domestic producers and therefore reduce import and export volumes. Tariffs on final goods reduce imports but also reduce the incentive to produce for the export market. The sign can therefore be positive or negative. Table 6 gives the results for the trade balance using mirror import data. The results using own data are quantitatively similar, but fewer variables are significant. The trade balance relationship appears to be well defined, and most of the vari- ables are significant or of the correct sign, or both. A real depreciation, for example, improves the trade balance in developing countries, including those in Africa. An exception is the terms of trade in model A, which suggests that an improvement in the terms of trade worsens the trade balance. The expected positive coefficient emerges, however, in model B. Focusing on the tariff variables for developing countries and the full sample of countries, there is some evidence that higher input tariff costs (tcost) are negatively associated with the trade balance. The estimates from model A suggest that a 10 per- cent reduction in tcost is associated with an improvement in tb of 4.63 percent. This relationship, however, is not robust to the inclusion of the various fixed effects and is insignificant in all estimates of model B. Similar results are found for Africa. By contrast, a positive relationship between output tariffs and the trade balance is found in both models for developing countries (and in the full sample). The preferred results from model B (column 6) suggest that a 10 percent increase in protection measured by (1 tar) is associated with an increase in the value of exports relative to the value of imports in developing countries of close to 7 percent. The average decline in (1 tar) during the period 1993­2004 for developing countries is 8 per- cent, implying that liberalization will have reduced the value of exports relative to imports for the average developing country by approximately 5 percent. This rela- tionship, however, does not hold for African economies, as is shown by the insignif- icant coefficient for output tariffs in the African sample (column 8), as well as by the offsetting negative coefficient on the interaction term in column 7. Overall, the results for developing countries are broadly consistent with those of Santos-Paulino and Thirlwall (2004), who find that tariff liberalization may have a negative impact on the trade balance. There are, nevertheless, some important differ- ences. First, there is evidence that liberalization enhances export performance through lower input costs. Second, the effect on the trade balance is less than is suggested by 100 TABLE 6. Determinants of the Trade Balance, World, Developing Countries, and Africa Model A Model B All countries Developing Developing interaction Africa All countries Developing Developing interaction Africa (1) (2) (3) (4) (5) (6) (7) (8) gdpf 0.551*** 0.328*** 0.327*** 0.477*** 0.324** 0.454** 0.443** 0.163 (0.016) (0.021) (0.021) (0.041) (0.153) (0.225) (0.225) (0.361) gdp 0.863*** 1.245*** 1.155*** 1.81** (0.152) (0.26) (0.261) (0.852) prod 0.480*** 0.504*** 0.499*** 0.328*** 0.354*** 0.387*** 0.390*** 0.140 (0.042) (0.058) (0.058) (0.127) (0.021) (0.035) (0.035) (0.086) reer 0.784*** 0.696*** 0.687*** 1.128*** (0.083) (0.129) (0.129) (0.335) (1 tar) 0.785*** 0.476** 1.272*** 0.455 0.256** 0.71*** 1.401*** 0.106 (0.127) (0.189) (0.279) (0.266) (0.117) (0.184) (0.267) (0.275) tcost 0.484*** 0.463*** 0.553*** 0.782*** 0.027 0.011 0.083 0.160 (0.029) (0.069) (0.08) (0.174) (0.027) (0.072) (0.087) (0.16) tot 0.042*** 0.162*** 0.161*** 0.105*** 0.021*** 0.021* 0.021* 0.004 (0.013) (0.018) (0.018) (0.038) (0.007) (0.012) (0.012) (0.024) (1 tar)*Africa 1.277*** 1.339*** (0.344) (0.369) tcost*Africa 0.036 0.073 (0.107) (0.172) N 15,409 7,571 7,571 1,810 15,409 7,571 7,571 1,810 R2 0.385 0.512 0.513 0.55 0.927 0.919 0.919 0.923 F 98.9 101 98.8 36.3 92.8 70.3 70.3 51.7 Note: The coefficients for the various sector-, time-, and country-specific effects are not shown. Standard errors are in parentheses. Values for gdp and reer are excluded from model B because their effect is captured by the cntryyear fixed effect. *p .1 **p .05 ***p .01 P R O T E C T I O N I S T P O L I C I E S A N D M A N U FA C T U R I N G T R A D E F L O W S I N A F R I C A | 101 the authors' estimates, particularly those based on dummy variables for the trade lib- eralization period. Third, there does not appear to be a negative relationship between protection and the trade balance in Africa, although these estimates are generally poor. Other Factors The results presented indicate that tariff liberalization enhanced manufacturing trade flows in developing countries, including Africa, between 1990 and 2004, but not by much. This finding suggests a limited supply response to the changing incentives aris- ing from tariff liberalization. There is much evidence to support this view. In a com- prehensive review of economic performance in Africa, Collier and Gunning (1999) argue that distorted product and credit markets, high risk, inadequate social capital, inadequate infrastructure, and poor public services are key factors inhibiting African firms' investment responses to opportunities. Microlevel institutions that affect the cost of exporting are also important (Johnson, Ostry, and Subramanian 2007). For example, it takes, on average, 40 days to export from a Sub-Saharan African country, but only 11 days for OECD countries. A com- bination of factors is responsible: excessive tariff bands, a lack of electronic docu- mentation, undue numbers of required documents, inefficient customs systems, poor roads, port congestion, and corruption at the border (World Bank 2005). The avail- able empirical evidence indicates that many of these factors are important determi- nants of Africa's export performance and firm productivity (Limăo and Venables 2001; Clarke 2005; Eifert, Gelb, and Ramachandran 2005; Wilson, Mann, and Otsuki 2005; Djankov, Freund, and Pham 2006). Higher transaction costs associated with weak institutions and infrastructure facilities are found to be particularly detrimen- tal to exports of manufactured products (Collier 2000; Elbadawi 2001; World Bank 2000).15 Conclusion Sub-Saharan Africa's share of world export volumes has continued to decline since 1990. Yet there are signs of improvement. The post-2000 commodity price boom has benefited Africa, and Sub-Saharan Africa's share in the value of world exports in 2006 was equivalent to its share in 1990 despite the relatively poor growth in export volumes. There is evidence of improved competitiveness in some manufacturing sec- tors, and when measured relative to GDP, export performance in Sub-Saharan Africa matches that of other developing countries. There are, therefore, some important differences in the post-1990 period from the period prior to 1990 analyzed by Ng and Yeats (1996). Some of the difference in export performance can be attributed to the progress many Sub-Saharan African countries have made in lowering tariffs since the early 1990s. The econometric estimates presented in this paper indicate that tariff liberal- ization stimulates manufactured export volumes in developing countries. A key 102 | L A W R E N C E E D WA R D S mechanism through which export performance is enhanced is through decreases in the cost of production inputs. A policy response to stimulate manufactured exports would therefore be to reduce tariffs on key production inputs. Implementation of such a policy, however, involves a number of difficulties. Whether products are inputs or final goods is not always clear cut. Lower tariffs on inputs increase effective protection in downstream sectors, which may raise the overall level of distortions in the economy. An alternative option is comprehensive liberalization that includes reductions in tariffs on final goods. For example, an across-the-board reduction in tariffs reduces input costs without leading to further increases in effective protection. Improvements in duty-drawback mecha- nisms available in Africa would further reduce the antiexport bias of protection. African economies will also need to consider the possible implications of further tariff liberalization for the manufacturing trade balance. The estimates presented show that lower tariffs are associated with increased import volumes and a worsen- ing of the manufacturing trade balance in developing countries. The growth in exports from liberalization is often exceeded by growth in imports, and if this leads to a shortage of foreign currency, the growth process may be constrained. The prob- lem may not be as acute in Africa as in other developing countries, as the relation- ship between protection and the trade balance was not significant in the preferred estimates. In general, the estimated import and trade balance relationships for Africa were poor. Nevertheless, the trade balance results suggest that liberalization policies need to be enhanced through complementary policies that stimulate export growth. For example, a depreciation of the currency is shown to have strong positive effects on the trade balance, particularly through the export channel. Although tariff liberal- ization is expected to depreciate the currency and thus offset some of the import effect, in many developing countries the currency appreciated in response to other reforms implemented (Bleaney 1999; UNCTAD 1999). The appropriate sequencing of reforms to minimize the possibility of unexpected currency appreciation needs to be considered. Other empirical literature indicates that improvements in trade- related infrastructure and institutions can have a substantial impact on export growth in Africa. ANNEX TABLE A.1. Manufacturing Trade and Tariffs, Selected Countries, Early 1990s to 2004 (percent) Average annual Average MFN tariff growth, 1990­2004 (import weighted) Average tariff-related input cost Share of Exports Income group trade, 2004 Imports (mirror) 1990­95 2001­4 Change 1990­95 2001­4 Change Low income 2.1 8.3 11.7 34.2 18.7 11.6 14.1 7.0 50.7 Bangladesh 0.1 7.6 14.7 82.4 18.9 34.8 43.0 12.7 70.4 Côte d'Ivoire 0.0 3.4 2.3 23.0 10.8 9.9 6.9 3.6 47.6 Ghana 0.0 8.4 0.4 11.1 13.4 2.1 4.3 7.1 64.9 India 1.1 12.1 12.6 60.7 28.5 20.0 23.8 10.8 54.6 Indonesia 0.7 4.6 11.7 13.9 6.1 6.9 5.4 2.2 59.5 Kenya 0.0 5.3 9.6 25.5 13.1 9.9 9.4 5.7 39.5 Uganda 0.0 6.3 7.6 16.7 6.5 8.7 5.8 2.5 57.6 Lower middle income 12.9 14.2 17.3 25.0 8.7 13.1 11.1 3.3 70.3 Algeria 0.2 5.5 0.7 15.9 13.5 2.0 2.7 6.1 121.3 Bolivia 0.0 7.2 7.7 9.7 9.0 0.6 3.5 3.1 11.7 China 9.9 17.9 18.8 30.9 8.4 17.2 12.4 3.3 73.4 Colombia 0.2 9.0 11.0 11.7 11.3 0.3 4.2 4.3 2.6 Ecuador 0.1 10.8 10.9 11.0 11.0 0.0 3.8 3.6 3.3 Egypt, Arab Rep. of 0.1 2.0 12.2 23.8 12.8 8.9 9.8 6.2 36.3 El Salvador 0.1 13.7 18.7 10.8 9.7 1.0 6.0 5.9 1.1 Morocco 0.2 7.5 8.7 59.6 25.1 21.6 25.1 12.0 52.4 Peru 0.1 9.6 8.8 15.8 9.4 5.5 6.2 4.0 34.8 Philippines 0.8 11.8 15.8 17.0 3.2 11.8 6.8 1.2 82.3 Sri Lanka 0.1 8.1 12.6 27.4 7.5 15.6 13.7 2.4 82.3 Tunisia 0.2 6.6 11.1 27.9 23.2 3.7 11.3 10.7 5.3 Turkey 1.1 11.9 15.3 7.9 4.8 2.9 3.6 2.3 36.4 103 (continued) 104 ANNEX TABLE A.1. (Continued ) Average annual Average MFN tariff growth, 1990­2004 (import weighted) Average tariff-related input cost Share of Exports Income group trade, 2004 Imports (mirror) 1990­95 2001­4 Change 1990­95 2001­4 Change Upper middle income 11.6 11.5 11.5 11.9 9.5 22.1 4.2 3.2 22.6 Argentina 0.4 17.0 7.6 12.6 13.8 1.0 3.8 4.1 6.9 Brazil 1.0 9.2 8.0 17.9 11.8 5.2 5.4 4.3 20.5 Chile 0.3 8.2 9.1 10.9 6.8 3.7 4.4 2.7 38.2 Korea, Rep. of 3.4 8.3 11.6 9.5 5.3 3.9 3.5 1.9 44.5 Malaysia 1.9 9.4 13.5 9.7 5.1 4.2 4.0 2.0 48.7 Mauritius 0.0 3.7 4.7 26.2 10.4 12.6 11.6 5.7 50.8 Mexico 2.8 15.7 13.4 12.1 14.9 2.5 4.6 5.6 21.1 Poland 1.1 36.3 15.2 10.5 9.8 0.7 3.6 3.6 0.2 South Africa 0.6 8.1 8.7 16.3 8.9 6.4 4.0 1.9 52.2 Uruguay 0.0 7.2 3.7 9.8 13.0 2.9 3.5 4.6 32.0 OECD 73.5 6.2 5.8 6.0 3.1 2.7 1.9 1.0 46.4 Africa 1.4 6.2 7.5 24.6 14.1 8.5 7.6 4.9 35.5 Source: Updated data based on trade, production, and protection database (Nicita and Olarreaga 2006). Note: MFN, most favored nation; OECD, Organisation for Economic Co-operation and Development. P R O T E C T I O N I S T P O L I C I E S A N D M A N U FA C T U R I N G T R A D E F L O W S I N A F R I C A | 105 Notes 1. See Santos-Paulino (2005) for a review of studies relating to Africa and other developing economies. 2. The calculations were based on World Development Indicators (WDI) data. The average ratio of taxes on international trade to total trade in goods and services in Sub-Saharan Africa was 10.3 percent between 1990 and 1995. 3. SITC is the Standard International Trade Classification. 4. Economies are divided according to 2006 gross national income per capita. The groups are as follows: low income, US$905 or less; lower middle income, US$906­US$3,595; upper middle income, US$3,596­US$11,115; and high income, US$11,116 or more. Developing countries include all low-income and middle-income economies. 5. These results contrast starkly with those of Morrissey and Mold (2006), who argue that African exports perform relatively well when analyzed in terms of volume. 6. Home-reported export data for most Sub-Saharan African countries are notoriously poor (Yeats 1990), and mirrored export data were used for the earlier period. Because of unre- liable data over time, the following categories were excluded from the analysis: 286 (ores and concentrates of uranium); 323 (coal briquettes; coke and semicoke; lignite or peat; retort carbon); 333 (crude petroleum and oils obtained from bituminous minerals); 334 (petroleum products, refined); and 688 (uranium depleted in U235). Exports from oil- exporting economies are thus underreported. 7. The gravity models of Foroutan and Pritchett (1993) and Coe and Hoffmaister (1999) yield consistent results. Foroutan and Pritchett, for example, find that intra-African trade in manufactures in the early 1980s is unusually high in relation to that of low-income and medium-income countries, given Africa's characteristics. A recent extension of this work by Subramanian and Tamirisa (2003) finds consistent results for anglophone Africa but not for francophone Africa, which is found to trade less than predicted. 8. To maximize the number of available countries, the average for 1996­2000 instead of for 1990­95 is used for 38 of the countries, and data for 2001­3 rather than 2004­6 are used for 11 of the countries. 9. The focus on the average tariff ignores other indicators of protection. According to 2006 tariff data obtained from the WTO's World Tariff Profiles 2006 (WTO 2006), Sub- Saharan African countries, on average, have lower binding coverage (50.6 percent of lines versus 82 percent for all countries), higher bound rates (60 versus 35 percent); a lower proportion of lines subject to duty-free rates (15.2 versus 23.4 percent), and a higher pro- portion of international tariff spikes, that is, tariff lines with duties less than 15 percent (37.6 versus 20.4 percent). In other dimensions, the structure of tariffs is far simpler in Africa than in other developing economies. The average number of tariff lines is lower; there are far fewer distinct duties (51 versus 412); and the proportion of tariff lines with non­ad valorem rates is a quarter of the average for all countries (0.5 versus 2 percent). 10. ISIC stands for International System of Industrial Classification. 11. Real effective exchange rate data were not available for Argentina, Bangladesh, Brazil, the Arab Republic of Egypt, El Salvador, India, Indonesia, Kenya, the Republic of Korea, Mauritius, Mexico, Peru, Senegal, or Turkey. CPI-based REER indexes for these countries were calculated using CPI indexes, exchange rate data (from IMF, International Financial Statistics, various years), and bilateral export weights of the top 10 export partners. 12. The average sectoral tariff is used to calculate the production costs associated with tariff protection. An alternative approach is to calculate average tariffs on intermediate and capital goods using the broad economic categories (BEC) classification. 106 | L A W R E N C E E D WA R D S 13. It is possible that industrial production is itself a function of exports. In an attempt to deal with possible endogeneity bias, lagged values of prod were included in alternative esti- mates. The results are largely unchanged. 14. To test the sensitivity of the results to choice of tariff variable, additional estimates were conducted using the simple average MFN tariff on output and tcost calculated using import-weighted applied rates that account for tariff preferences. The results using the average MFN tariff on output are poor, but those using tcost calculated with applied rates are similar. 15. See also Wood and Mayer (2001), who emphasize the role of endowments relative to infrastructure. References Amjadi, Azita, and Alexander Yeats. 1995. "Have Transport Costs Contributed to the Rela- tive Decline of Sub-Saharan African Exports? Some Preliminary Empirical Evidence." Policy Research Working Paper 1559, International Trade Division, World Bank, Washington, DC. Amjadi, Azita, Ulrich Reincke, and Alexander Yeats. 1996. "Did External Barriers Cause the Marginalization of Sub-Saharan Africa in World Trade?" Policy Research Working Paper 1586, International Trade Division, World Bank, Washington, DC. Anderson, James E., and Peter Neary. 1994. "Measuring the Restrictiveness of Trade Policy." World Bank Economic Review 8 (2, May): 151­69. Bleaney, Michael. 1999. "Trade Reform, Macroeconomic Performance and Export Growth in Ten Latin American Countries, 1979­95." Journal of International Trade and Economic Development 8 (1): 89­105. Chandra, Vandana, Lalita Moorty, Jean-Pascal Nganou, Bala Rajaratnam, and Kendall Schaefer. 2001. "Constraints to Growth and Employment in South Africa: Evidence from the Small, Medium and Micro Enterprise Firm Survey." World Bank Informal Discussion Papers on Aspects of the South African Economy 15, Southern Africa Department, World Bank, Washington, DC. Clarke, George. 2005. "Beyond Tariffs and Quotas: Why Don't African Manufacturers Export More?" Policy Research Working Paper 3617, World Bank, Washington, DC. Coe, David, and Alexander Hoffmaister. 1999. "North-South Trade: Is Africa Unusual?" Journal of African Economies 8 (2, July): 228­56. Collier, Paul. 2000. "Africa's Comparative Advantage." In Industrial Development and Policy in Africa, ed. H. Jalilian, M. Tribe, and J. Weiss. Cheltenham, U.K.: Edward Elgar. Collier, Paul, and Jan Willem Gunning. 1999. "Explaining African Economic Performance." Journal of Economic Literature 37 (March): 64­111. Collier, Paul, and Anthony J. Venables. 2007. "Rethinking Trade Preferences: How Africa Can Diversify Its Exports." World Economy 30 (8, August): 1326­45. Djankov, Simeon, Caroline Freund, and Cong Pham. 2006. "Trading on Time." Policy Research Working Paper 3909, World Bank, Washington, DC. Edwards, Lawrence. 2005. "Has South Africa Liberalised Its Trade?" South African Journal of Economics 73 (4): 754­75. Edwards, Lawrence, and Robert Z. Lawrence. 2008. "South African Trade Policy Matters: Trade Performance and Trade Policy." Economics of Transition 16 (4): 741­68. P R O T E C T I O N I S T P O L I C I E S A N D M A N U FA C T U R I N G T R A D E F L O W S I N A F R I C A | 107 Edwards, Lawrence, Neil Rankin, and Volker Schöer. 2008. "South African Exporting Firms: What Do We Know and What Should We Know?" Journal of Development Perspectives 4 (1): 93­118. Eifert, Benn, Alan Gelb, and Vijaya Ramachandran. 2005. "Business Environment and Com- parative Advantage in Africa: Evidence from the Investment Climate Data." Working Paper 56, Center for Global Development, Washington, DC. Elbadawi, Ibrahim A. 2001. "Can Africa Export Manufactures? The Role of Endowment, Exchange Rates, and Transaction Costs." In Policies to Promote Competitiveness in Man- ufacturing in Sub-Saharan Africa, ed. Augustin Nsouli, Kwasi Fosu, and Aristomene Varoudakis. Paris: Organisation for Economic Co-operation and Development. Foroutan, Faezeh, and Lant Pritchett. 1993. "Intra­Sub-Saharan African Trade: Is It Too Little?" Journal of African Economies 2 (1): 74­105. Goldstein, Morris, and Mohsin Khan. 1985. "Income and Price Effects in Foreign Trade." In Handbook of International Economics, vol. 2, ed. Ronald W. Jones, and Peter B. Kenen. Amsterdam: North-Holland. Greenaway, David, and David Sapsford. 1994. "What Does Liberalisation Do for Exports and Growth?" Review of World Economics 130: 157­74. IMF (International Monetary Fund). Various years. International Financial Statistics. Washington, DC: IMF. Johnson, Simon, Jonathan Ostry, and Arvind Subramanian. 2007. "The Prospects for Sus- tained Growth in Africa: Benchmarking the Constraints." IMF Working Paper WP/07/52, International Monetary Fund, Washington, DC. Lerner, Abba P. 1936. "The Symmetry between Import and Export Taxes." Economica 3: 306­13. Limăo, Nuno, and Anthony J. Venables. 2001. "Infrastructure, Geographical Disadvantage, Transport Costs, and Trade." World Bank Economic Review 15 (3): 451­79. Martin, Will, and Devashish Mitra. 2001. "Productivity Growth and Convergence in Agricul- ture versus Manufacturing." Economic Development and Cultural Change 49 (2): 403­22. Mayer, Jörg, and Pilar Fajarnes. 2008. "Tripling Africa's Primary Exports: What, How and Where?" Journal of Development Studies 44 (1): 80­102. Morrissey, Oliver, and Andrew Mold. 2006. "Explaining Africa's Export Performance-- Taking a New Look." Center for Global Trade Analysis, Department of Agricultural Economics, Purdue University, West Lafayette, IN. https://www.gtap.agecon.purdue.edu/ resources/download/2643.pdf. Ng, Francis, and Alexander Yeats. 1996. "Open Economies Work Better: Did Africa's Protec- tionist Policies Cause Its Marginalization in World Trade?" Policy Research Working Paper 1636, World Bank, Washington, DC. Nicita, Alessandro, and Marcelo Olarreaga. 2006. "Trade, Production and Protection 1976­2004." World Bank Economic Review 21 (1): 165­71. Papageorgiou, Demetris, Michael Michaely, and Armeane M. Choksi. 1991. Liberalizing Foreign Trade. Oxford: Basil Blackwell. Parikh, Ashok, and Corneliu Stirbu. 2004. "Relationship between Trade Liberalisation, Eco- nomic Growth and Trade Balance: An Econometric Investigation." HWWA Discussion Paper 282, Hamburgisches Welt-Wirtschafts-Archiv, Hamburg. Pattillo, Catherine, Sanjeev Gupta, and Kevin Carey. 2005. "Sustaining Growth Accelerations and Pro-Poor Growth in Africa." IMF Working Paper WP/05/195, International Monetary Fund, Washington, DC. 108 | L A W R E N C E E D WA R D S Richardson, J. David. 1971. "Constant-Market-Shares Analysis of Export Growth." Journal of International Economics 1 (2): 227­39. Rodrik, Dani. 1997. "Trade Policy and Economic Performance in Sub-Saharan Africa." NBER Working Paper 6562, National Bureau of Economic Research, Cambridge, MA. Sachs, Jeffrey D., and Andrew M. Warner. 1999. "The Big Rush, Natural Resource Booms and Growth." Journal of Development Economics 59 (1): 43­76. Santos-Paulino, Amelia. 2005. "Trade Liberalisation and Economic Performance: Theory and Evidence for Developing Countries." World Economy 28 (6): 783­821. Santos-Paulino, Amelia, and Anthony P. Thirlwall. 2004. "The Impact of Trade Liberalisation on Exports, Imports and the Balance of Payments of Developing Countries." Economic Journal 114 (February): F50­F72. Subramanian, Arvind, and Natalia T. Tamirisa. 2003. "Is Africa Integrated in the Global Econ- omy?" IMF Staff Papers 50 (3): 352­72. UNCTAD (United Nations Conference on Trade and Development). 1999. Trade and Devel- opment Report, 1999. Geneva: United Nations. Wang, Zhen K., and Alan L. Winters. 1998. "Africa's Role in Multilateral Trade Negotiations: Past and Future." Journal of African Economies 7 (Suppl. 1): 1­33. Wilson, John S., Catherine Mann, and Tsunehiro Otsuki. 2005. "Assessing the Benefits of Trade Facilitation: A Global Perspective." World Economy 28 (6): 841­71. Wood, Adrian, and Jörg Mayer. 2001. "Africa's Export Structure in a Comparative Perspec- tive." Cambridge Journal of Economics 25 (3): 369­94. World Bank. 2000. Can Africa Claim the 21st Century? Washington, DC: World Bank. ------. 2005. Doing Business 2005. Washington, DC: World Bank. WTO (World Trade Organization). 2006. World Tariff Profiles, 2006. Geneva: WTO. Yeats, Alexander. 1990. "On the Accuracy of Economic Observations: Do Sub-Saharan Trade Statistics Mean Anything?" World Bank Economic Review 4 (2): 135­56. Comment on "Protectionist Policies and Manufacturing Trade Flows in Africa," by Lawrence Edwards BEATA SMARZYNSKA JAVORCIK Lawrence Edwards has written a useful paper that provides answers to three impor- tant questions. First, it finds that African countries continue to be marginalized in world trade. Second, it shows that own-tariff liberalization has helped African coun- tries increase exports but that the effect has been relatively small. And third, it con- cludes that tariff liberalization has not worsened the trade balance in the countries considered. Although providing those answers is valuable, the natural next step in the analy- sis is to examine the causes of this state of the world. In this discussion, I would like to highlight some factors that in my view are worth exploring. Looking Back Disappointing export performance by African economies has often been attributed to three factors: (a) geography--remote location relative to major export markets and lack of access to the sea, and thus to cheap sea transport; (b) high internal transport costs as a result of underdeveloped road and port infrastructure and to inefficiency and poor governance in the customs service; and (c) protectionism in export markets-- although this last has diminished in recent years. Edwards does review in detail the arguments for why each of these factors should matter, but it would be interesting to look back and quantify to what extent each has affected Africa's export performance. Looking Ahead It might be even more interesting to look ahead and consider two factors that are likely to influence Africa's trade in the years to come: global production chains and ethnic networks. This discussion will focus on each factor in turn. Beata Smarzynska Javorcik is a reader in economics at the University of Oxford and a research affiliate with the Centre for Economic Policy Research, London. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank 109 110 | B E ATA S M A R Z Y N S K A J AV O R C I K Global Production Chains Technological advances in information and logistics systems, leading to a decline in transport and communication costs and accompanied by a lowering of tariffs and other barriers to trade, have changed the economic landscape facing nations, indus- tries, and individual firms. Multinational corporations have served as the key agents in this transformation by creating international production and distribution net- works that span the globe and actively interact with each other. In fact, estimates sug- gest that about two-thirds of all world trade in the latter half of the 1990s involved multinational corporations (UNCTAD 2002, 153). Although so far the participation of African countries in the network trade associ- ated with multinational corporations has been limited, participation in global produc- tion networks could create valuable opportunities for African producers, particularly in the apparel and food sectors. It is helpful for the purpose of this discussion to follow Gereffi (1999) and classify networks into buyer-driven and producer-driven global commodity chains. Producer- driven networks are often coordinated by large multinationals. They are vertical, mul- tilayered arrangements, usually with a direct ownership structure that includes parents, subsidiaries, and subcontractors. They tend to exist in capital- and technology-intensive sectors that are often dominated by global oligopolies, such as aircraft, automobiles, and heavy machinery. Integration of a country into producer-driven networks usually requires significant inflows of foreign direct investment (FDI). This is illustrated in figure 1, which FIGURE 1. Foreign Direct Investment (FDI) and Exports of Producer-Driven-Network Products, Selected Countries, 2003 (U.S. dollars) 8 Hungary Czech Republic logarithm of FDI stock in manufacturing 7 Slovenia Croatia Bulgaria Poland Estonia 6 Lithuania per capita, end-2003 Romania Serbia and Montenegro Slovak Republic 5 Latvia Armenia Kazakhstan Russia 4 Macedonia, FYR Albania Belarus Moldova Ukraine 3 Azerbaijan 2 Kyrgyz Republic 1 0 2 4 6 8 logarithm of network exports per capita, 2003 Source: Javorcik and Kaminski 2006. Note: Macedonia, FYR, Former Yugoslav Republic of Macedonia. C O M M E N T O N E D WA R D S | 111 depicts FDI stock and network exports (both expressed in per capita terms) in Eastern European and Central Asian countries, many of which have been very suc- cessful in joining global production chains. Given the meager success of Africa in attracting FDI inflows, producer-driven networks are likely to play a limited role in Africa's exports in the near future. In contrast to producer-driven commodity chains, buyer-driven networks may offer greater promise to African countries. Buyer-driven networks are organized by large retailers, branded marketers, and branded manufacturers and do not necessar- ily require large FDI inflows. The products are designed and marketed by the buyer. They are typically labor-intensive consumer goods such as apparel, footwear, furni- ture, and processed food products. Buyer-driven commodity chains are characterized by highly competitive locally owned and globally dispersed production systems. Prof- its do not come from scale, volume, and technological advantage but, rather, from market research, design, and marketing. Buyer-driven production networks in the food and apparel sectors may be particularly relevant to African countries. Global food markets have undergone a rapid transformation in recent years, driv- en by changes in consumer demand, increased concern about food safety, and the rise of modern retail systems. Two trends in the food industry during the past two decades are worth mentioning. The first is the consolidation of food retailing. In 2001 just 30 grocery retail chains jointly took in more than US$1 trillion in revenue, accounting for about 10 percent of global food sales. Within this group, the top 10 retailers constituted 57 percent of the combined total (World Bank 2005, 27). The second trend is the increasing reliance of major retail chains on their own agents for sourcing and thus the declining importance of wholesale markets. Whereas in the past wholesale and terminal markets were responsible for 20 or more percent of food sales, their share of sales in industrial countries has dropped to about 10 percent (World Bank 2005, 27). The consolidation of food retailing has given market leaders extraordinary mar- ket and purchasing power and has resulted in a strong tendency toward global sourc- ing, the introduction of preferred-supplier arrangements, supply chain integration and rationalization, and lower average prices, but also lower variability in prices for contract or program suppliers (World Bank 2005). In addition, it has led to the emer- gence of numerous private sector codes of practice or other technical protocols that nowadays tend to play a dominant role in the market. There are several advantages of being a supplier servicing a supermarket chain. These include higher margins than in wholesale transactions, more consistent and more predictable demand, access to detailed information on changing developments and requirements within the market, detailed guidelines for operations and good practice, and the ability to enhance one's reputation by being a supplier to a major retail chain (Jaffee 2003). In the South African context, Barrientos and Kritzinger (2004) found that producers selling fruit to U.K. and European supermarkets were able to obtain more stable outlets for their produce. For instance, most supermarkets negotiated purchases six months in advance. Moreover, producers servicing super- markets on average received better prices than those selling on the open market. 112 | B E ATA S M A R Z Y N S K A J AV O R C I K But compliance with the standards imposed by supermarkets is costly. It requires investment in machinery and facilities (for instance, cold storage and stainless steel tables); improvements in sanitation levels, worker hygiene, and skills; and investment in obtaining a formal certification--for example, for Hazard Analysis and Critical Control Points (HACCP) systems. Such investment may be beyond the reach of smaller producers, who are often credit constrained. Supplying supermarkets may also involve an increase in variable costs, such as expenditure on microbiological test- ing. Timeliness is another important aspect of serving supermarket chains. If a ship- ment is delayed along the way and misses its vessel in the port, taking the next ves- sel might not be an option because the delay may result in deterioration of product quality and the shipment may no longer meet the required standard. The apparel industry is another sector in which production is increasingly distrib- uted across low-income countries by buyers searching for cheaper labor. As a result of Africa's preferential access to foreign markets, a significant amount of such pro- duction was moved from newly industrialized countries in Asia to Africa. Foreign direct investment from Asia, induced by the quota system of the Multifibre Arrange- ment (MFA) and the U.S. African Growth and Opportunity Act (AGOA), enabled rapid growth of the African apparel sector. One beneficiary was Lesotho, which, thanks to its cheap labor costs, was an ideal host for Asian capital seeking to avoid the textile quotas constraining exports from their home countries. Other African pro- ducers benefited from AGOA, which allows African countries that practice good governance and demonstrate respect for human rights to enjoy duty-free access to the U.S. market. In 2004, Sub-Saharan African exports of apparel to the United States exceeded US$1.5 billion (ILO 2005). The expiration on January 1, 2005, of the MFA, which had set quotas on world- wide textile and apparel exports and had given preferences to many developing African countries, unleashed a new wave of Chinese sales on the world market. The ILO, in its analysis of the post-MFA environment, reported that textile and apparel exports under the AGOA fell to US$270 million in the first quarter of 2005, as against US$361 million a year earlier. The 25 percent reduction contrasts with a 19 percent increase in China's exports for the same period (ILO 2005). After the expiration of the MFA, some Asian companies that had invested in Africa to take advantage of the quota began moving back to China in search of cheaper labor. Between January and March 2005, Kenya exported US$60 million of textile and clothing products to the United States. That was 13 percent, or US$9 mil- lion, less than the exports during the same period in 2004. Between October 2004 and May 2005, a loss of 6,000 out of 39,000 jobs was reported. In Lesotho, where the garment sector had accounted for more than 90 percent of the country's exports and was by far the largest single employer, 6,650 (out of 56,000) workers were ter- minated at the end of 2004, and 10,000 more were moved to short-term contracts (ILO 2005). Are these trends in the apparel sector likely to continue, or have they already stopped? What factors can explain why some African countries have been more suc- cessful than others in becoming suppliers to large food retailers? Many African policy makers would find answers to these and related research questions useful. C O M M E N T O N E D WA R D S | 113 Ethnic Networks A growing literature has documented a positive link between the presence of ethnic networks and international trade. The main premise of this literature is that interna- tional transactions are plagued with informal trade barriers, in addition to formal trade barriers such as transport costs and tariffs. These informal barriers include the difficulties associated with accessing information on potential market opportunities, consumer tastes, and the trustworthiness of buyers and with the enforcement of con- tracts across international borders. As argued by Gould (1994), Head and Ries (1998), Rauch and Trindade (2002), and Combes, Lafourcade, and Mayer (2005), the presence of people with the same ethnic or national background on both sides of a border may alleviate these problems. Their language skills and familiarity with a foreign market can lower communication costs. They can provide valuable informa- tion about market structure, consumer preferences, business ethics, and commercial codes, and their social links may decrease the costs of negotiating and enforcing a contract. In sum, business and social networks that span international borders can help overcome many contractual and informational barriers and stimulate interna- tional transactions. Yet the actual and potential benefits of social networks have not been examined in the African context. Do African producers find it easier to export their goods to Chinese and Indian markets not only because of product suitability but also thanks to the presence of Chinese and Indian diasporas in their countries? Have African pro- ducers been able to benefit from the presence of their nationals in industrial-country markets? In sum, although buyer-driven production chains and ethnic networks may pro- vide new export opportunities that can be harnessed by African producers and are thus worthy of researchers' attention, they should not divert our attention from the basics. High internal transport costs resulting from underdeveloped infrastructure and poorly functioning customs services may still be the first-order obstacle prevent- ing African producers from exploiting new export opportunities. References Barrientos, Stephanie, and Andrienetta Kritzinger. 2004. "Squaring the Circle: Global Production and the Informalization of Work in the Food Sector." Journal of International Development 16 (1): 81­92. Combes, Pierre-Philippe, Miren Lafourcade, and Thierry Mayer. 2005. "The Trade-Creating Effects of Business and Social Networks: Evidence from France." Journal of International Economics 66 (1): 1­29. Gereffi, Gary. 1999. "International Trade and Industrial Upgrading in the Apparel Commod- ity Chain." Journal of International Economics 48 (1): 37­70. Gould, David M. 1994. "Immigrant Links to the Home Country: Empirical Implications for U.S. Bilateral Trade Flows." Review of Economics and Statistics 76: 302­16. Head, Keith, and John Ries. 1998. "Immigration and Trade Creation: Econometric Evidence from Canada." Canadian Journal of Economics 31 (1): 47­62. 114 | B E ATA S M A R Z Y N S K A J AV O R C I K ILO (International Labour Organization). 2005. "Promoting Fair Globalization in Textiles and Clothing in a Post-MFA Environment." Presented at the Tripartite Meeting on Pro- moting Fair Globalization in Textiles and Clothing in a Post-MFA Environment, Geneva. Jaffee, Steven. 2003. "From Challenge to Opportunity. Transforming Kenya's Fresh Vegetable Trade in the Context of Emerging Food Safety and Other Standards in Europe." Agricul- ture and Rural Development (ARD) Discussion Paper 1, World Bank, Washington, DC. Javorcik, Beata S., and Bart Kaminski. 2006. "Linkages between Foreign Direct Investment and Trade Flows." In From Disintegration to Reintegration: Eastern Europe and the For- mer Soviet Union in International Trade, ed. Harry Broadman. Washington, DC: World Bank. Rauch, James, and Vitor Trindade. 2002. "Ethnic Chinese Networks in International Trade." Review of Economics and Statistics 84 (1, February): 116­30. UNCTAD (United Nations Conference on Trade and Development). 2002. World Investment Report: Transnational Corporations and Export Competitiveness. New York and Geneva: United Nations. World Bank. 2005. "Food Safety and Agricultural Health Standards: Challenges and Oppor- tunities for Developing Country Exports." Report 31207, Poverty Reduction and Economic Management (PREM) Trade Unit, and Agricultural and Rural Development Department, World Bank, Washington, DC. Crisscrossing Globalization: The Phenomenon of Uphill Skill Flows AADITYA MATTOO AND ARVIND SUBRAMANIAN This paper documents an unusual and possibly significant phenomenon: the export of skills embodied in goods, services, or capital from poorer to richer countries. We first present a set of stylized facts. Using a measure that combines the sophistication of a country's exports with the average income level of destination countries, we show that the performance of a number of developing countries, notably China, Mexico, and South Africa, matches that of much more advanced countries such as Japan, Spain, and the United States. Creating a new combined dataset on foreign direct investment (FDI) that covers greenfield investment as well as mergers and acquisitions, we show that flows of FDI to member countries of the Organisation for Economic Co-operation and Development from developing countries such as Brazil, India, Malaysia, and South Africa, as a share of those countries' gross domestic prod- uct, are as large as flows from countries such as Japan, the Republic of Korea, and the United States. Then, taking the work of Hausmann, Hwang, and Rodrik (2007) as a point of departure, we suggest that it is not just the composition of exports but their destination that matters. In both cross-sectional and panel regressions, using a range of controls, we find that a measure of uphill flows of sophisticated goods is significantly associated with better growth performance. These results suggest the need for a deeper analysis of whether development benefits might derive not from deifying comparative advantage but from defying it. The phenomenon of uphill flows of capital has been subject to much scrutiny in recent years (see, for example, Bernanke 2006; Prasad, Rajan, and Subramanian 2007; Caballero, Farhi, and Gourinchas 2008). Much of this literature has focused on financial flows (or foreign savings). Indeed, Caballero and his coauthors (2008) Aaditya Mattoo is lead economist in the International Trade Group of the World Bank's Development Economics Research Group. Arvind Subramanian is a senior fellow at the Peterson Institute for International Economics and the Center for Global Development in Washington, DC, and is senior research professor at Johns Hopkins University. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank 115 116 | A A D I T YA M AT T O O A N D A R V I N D S U B R A M A N I A N attempt to explain why developing countries export savings while simultaneously importing foreign direct investment (FDI). The assumption has been that the only gravity-defying flow is finance. But a number of recent high-profile developments raise the possibility of uphill flows in other dimensions. These flows run counter to the predictions of standard trade models in which developing countries primarily export unskilled products and are recipients of FDI. Examples are the takeover of the U.K.'s Jaguar company by a prominent Indian enterprise (Tata); the acquisition of IBM by China's Lenovo; Brazil's success in exporting small commercial aircraft to industrial countries; and the growing exports of skilled services from India and Israel to markets in member coun- tries of the Organisation for Economic Co-operation and Development (OECD). These developments have two things in common. The first is that skills, embodied in goods, services, or capital (in the form of entrepreneurial and managerial skills asso- ciated with FDI), are being exported. The second is that these embodied skills are exported from poorer to richer countries. On its own, the first feature, although inter- esting, would not necessarily run counter to the predictions of standard trade models. For example, if China were exporting sophisticated goods to and investing in Africa, that would not be inconsistent with their relative endowments. It is the movement of sophisticated goods and FDI from China to countries that have relatively more skills and capital that is noteworthy from a trade perspective. This paper is a first stab at doc- umenting and understanding this unusual, and possibly significant, phenomenon. How significant is the phenomenon? Figure 1 plots a measure combining the sophistication of a country's exports and the average income level of the countries that are destinations for sophisticated exports against per capita income for the years 1991 and 2005. (See the section "Consequences of Uphillness," below, for a precise definition of uphill exports.) Two features are noteworthy. First, there is an upward shift of the curve between the two time periods, suggesting that exogenous factors-- perhaps technology--are increasing the propensity of countries, especially those at lower income levels, to export sophisticated goods to rich trading partners. Particu- lar striking is that in 2005 the performance of a number of developing countries, including China, Malaysia, Mexico, the Philippines, and South Africa, surpassed in this respect the 1991 performance of a number of industrial countries with much higher per capita incomes. Even more striking is that a few developing countries (China, Mexico, and South Africa) match the contemporary performance of Japan, Portugal, Spain, and the United States. Figure 2 presents a similar picture for outward flows of FDI, including both merg- ers and acquisitions (M&A) and greenfield investments.1 On the vertical axis are FDI outflows from selected countries to OECD countries as a share of the sending coun- try's gross domestic product (GDP), averaged over the period 2003­07. This meas- ure of uphill FDI flows is plotted against the sending country's per capita income. Flows of FDI to OECD countries from developing countries such as Brazil, India, Malaysia, and South Africa as a share of their GDP are as large as flows from countries such as Japan, the Republic of Korea, and the United States. Taken together, these charts provide evidence of the "precociousness" of some developing countries in exporting skills in a manner associated with countries at much C R I S S C R O S S I N G G L O B A L I Z AT I O N | 117 FIGURE 1. Exports of Sophisticated Goods by Selected Economies, 1991 and 2005 21 export sophistication and income MEX IRL GBR of destination countries 20 ZAF DEU FRA CHN HUNGBRISR USA IRLPRTDEU ESP FRA JPN PHL MYS JPN USA THAMEXKOR ISR KOR ESP BRA PRT IND IDN MYS CHL CHN THA BRA IND PHL CHL 19 IDN 18 6 7 8 9 10 11 log per capita GDP (constant PPP dollars) 1991 2005 CHI 1991 observations CHI 2005 observations Source: United Nations Commodity Trade Statistics database (Comtrade). Note: GDP, gross domestic product; PPP, purchasing power parity. The figure plots a measure that combines the sophistication of a country's exports and the average income level of the destination countries for these exports (see the text). The plot lines represent the fit of the relationship between this measure and per capita GDP in 1991 and 2005. The fit is based on a larger sample of countries than those identified in the figure. The country abbreviations are BRA, Brazil; CHL, Chile; CHN, China; ESP, Spain; GBR, United Kingdom; HUN, Hungary; IDN, Indonesia; IND, India; IRL, Ireland; ISR, Israel; KOR, Republic of Korea; MEX, Mexico; MYS, Malaysia; PHL, the Philippines; PRT, Portugal; THA, Thailand; USA, United States; and ZAF, South Africa. higher levels of development. This phenomenon, of course, has not gone unnoticed. A number of studies have recently emphasized the growing sophistication of the export and production base of developing countries. For example, Schott (2007) has shown that China's export profile is becoming increasingly similar to that of many OECD countries (see also Hummels and Klenow 2005; Schott 2005). Ramamurti and Singh (2009) have documented FDI flows from developing to industrial countries. A related literature has focused on the direction of these export flows, but in a more normative context. For example, Samuelson (2004) and Krugman (2008) have examined the consequences of growth of U.S. imports of manufactured goods pro- duced in developing countries that compete with domestic U.S. production. There has also been some discussion in the popular press of inward flows of FDI from developing countries (for example, in connection with the Dubai Port episode), but it is primarily related to security issues. These are perspectives--even paranoid ones--on uphill flows from the top of the hill. The vast literature on the effects of global integration, through trade in goods and FDI, has focused primarily on flows to developing countries. For example, Coe, Helpman, and Hoffmaister (1997) high- lighted the effect of technology diffusion through imports of capital goods on the growth of developing countries, and Lumenga-Neso, Olarreaga, and Schiff (2001) 118 | A A D I T YA M AT T O O A N D A R V I N D S U B R A M A N I A N FIGURE 2. Defying Comparative Advantage: Foreign Direct Investment, 2003­07 0.05 SGP FDI outflows to OECD countries IRL 0.04 ISR as a share of GDP (percent) ESP GBR 0.03 FRA 0.02 DEU PRT ITA USA 0.01 SAU ZAF BRA KOR BGR MEX PHL IND HUN JPN PAN ARG MYS IDN CHN THA TUR POL 0.00 ROM CHL 7 8 9 10 11 log per capita GDP (2003 PPP dollars) Sources: Thomson Financial SDC Platinum database and Financial Times FDI Intelligence database. Note: GDP, gross domestic product; PPP, purchasing power parity. The figure plots FDI outflows from a country to OECD countries as a share of its GDP (averaged over the period 2003­07) against its per capita income. The sample comprises selected industrial and emerging-market countries. The economies shown are ARG, Argentina; BRA, Brazil; CHL, Chile; CHN, China; DEU, Germany; ESP, Spain; FRA, France; GBR, United Kingdom; HUN, Hungary; IDN, Indonesia; IND, India; IRL, Ireland; ISR, Israel; ITA, Italy; JPN, Japan; KOR, Republic of Korea; MEX, Mexico; MUS, Mauritius; MYS, Malaysia; PAK, Pakistan; PHL, the Philippines; POL, Poland; PRT, Portugal; ROM, Romania; SAU, Saudi Arabia; SGP, Singapore; THA, Thailand; TUR, Turkey; USA, United States; and ZAF, South Africa. emphasized the impact of direct and indirect imports from industrial countries. There is also a large literature documenting the effects of inward FDI (Borensztein, De Gregorio, and Lee 1998; Haskel, Pereira, and Slaughter 2002). Recently, Hausmann, Hwang, and Rodrik (2007) looked at the effects of the sophistication of a country's export profile on its own growth (see also Burgess and Venables 2004). In a similar vein, Feenstra and Kee (2004) examine whether diversity of export production can have productivity-enhancing effects. The effects of outward flows of FDI and skilled exports, and of the destination of these flows, have received less attention. Why should the destination of trade and FDI flows matter? Javorcik (2004) has shown that selling to foreign-owned firms located in a country has positive upstream productivity effects because of the possibility of induced technological and manage- rial improvements. In principle, these benefits can also arise from sales to foreign firms located abroad. Recently, De Loecker (2007), working with microdata on Slovenian firms, has demonstrated that productivity gains are higher for firms exporting toward high-income regions. Moreover, exports of goods to high-income destinations are frequently associated with participation in global production chains that confer important benefits (Hoekman and Javorcik 2006). C R I S S C R O S S I N G G L O B A L I Z AT I O N | 119 In this paper we first present some new data on developing-country exports of services, goods, and FDI and assess the extent to which these are going to richer countries. We then undertake a preliminary exploration of the consequences of uphill flows of embodied skills for growth of the source country. Here, we follow closely the work of Hausmann, Hwang, and Rodrik (2007). Data Our focus in this paper is on the direction of flows of embodied skills. For the three areas--FDI, goods, and services--for which we present some broad data, we need to explain how we define or illustrate the flow of skills. Our FDI data come from two sources. The Thomson Financial SDC Platinum database provides data on FDI taking the form of mergers and acquisitions. The Financial Times FDI Intelligence database provides similar data on greenfield invest- ments. These databases are described in detail in the annex to this article. Our trade data come from the United Nations World Integrated Trade Solution (WITS) database. We collected data at the five-digit level, largely because finer data-- say, at the six-digit level--really become available only in the late 1980s, and we are interested in checking whether the phenomenon of uphill flows is a feature of the his- torical data. For computational reasons, we collected data for every five-year inter- val and restricted the sample to countries that together account for about 90 percent of world trade.2 We draw on Hausmann, Hwang, and Rodrik (2007) to characterize skill-intensive products. These authors calculate a measure called PRODY, which is a weighted sum of the per capita GDP of countries exporting a given product and, thus, represents the income level associated with each of these goods. In this paper, we define--admittedly, arbitrarily--skill-intensive products as those that are either above the median level or in the top 25th percentile of PRODY for all products defined at the five-digit level of aggregation for 1990. Our services data come from the Balance of Payments Statistics database of the International Monetary Fund (IMF) and from the U.S. Bureau of Economic Analysis. Stylized Facts on Uphillness We first present some basic facts about flows of embodied skills. Foreign Direct Investment The share in world FDI exports of non-OECD countries for which data are available for the period 2003­07 is shown in figure 3. This share goes up from about 20 to 25 percent over the period under consideration. These figures show how developing countries are becoming increasing exporters of FDI, but they do not indicate the direction of the flows. Figure 4 isolates the direction 120 | A A D I T YA M AT T O O A N D A R V I N D S U B R A M A N I A N FIGURE 3. Share of Non-OECD Countries in World FDI Exports, 2003­07 27 25 23 percent 21 19 17 15 2003 2004 2005 2006 2007 Sources: Thomson Financial SDC Platinum database and Financial Times FDI Intelligence database. FIGURE 4. Share of Non-OECD Countries in World FDI Exports to OECD Countries, 2003­07 15 14 13 12 percent 11 10 9 8 2003 2004 2005 2006 2007 Sources: Thomson Financial SDC Platinum database and Financial Times FDI Intelligence database. C R I S S C R O S S I N G G L O B A L I Z AT I O N | 121 of flow of these skills by depicting the share of non-OECD countries in FDI exports to OECD countries and so provides a measure of uphill flows at the global level. This share increased from about 9 percent in 2003 to close to 15 percent in 2007. Exports of Goods We repeat this exercise for exports of sophisticated goods and find a similar pattern. The average income level of world exports of sophisticated products declined by a similar percent--about 10 percent--but over a slightly longer period (figure 5). Unlike the case of FDI, China is a big contributor to the decrease in the income of the source country for world exports of sophisticated products. Exclusion of China reduces the decline by nearly 5 percent (figure 5, panel b). In figure 6, we calculate the uphill flows of sophisticated products from non- OECD countries. For each country, the amount of exports of sophisticated goods to richer countries as a share of the country's total sophisticated exports is the magni- tude of uphill flows. These exports are added for all non-OECD countries. This share was about 1 percent in 1980; it was about 0.5 percent for highly sophisticated products (HSPs).3 The share increased to 10 percent in 2006 (over 3 percent for HSPs).4 As shown in figure 7, uphill flows were pronounced for China, Malaysia, and Mexico but much less so for Brazil and India. Services In services, we focus on exports of services other than transport and travel, that is, the category "other commercial" (in the United States, "other private") services, which includes most skill-intensive business services. Again, we find a decline, albeit slow, in the average income level of services exporters (figure 8). This trend suggests that developing countries are becoming increasingly important exporters of skilled services. Unfortunately, bilateral data on services trade are available only for OECD coun- tries, so it is not possible to construct measures of uphill flows analogous to those for goods and FDI.5 However, bilateral data available for the United States show that for some developing countries (for example, India and Malaysia), services exports as a share of GDP are flowing uphill (figure 9) Country Heterogeneity Although the phenomenon of uphill flows appears to characterize several developing countries, there is heterogeneity across them. It is not the case, for example, that countries that see uphill flows of sophisticated exports also see uphill flows of FDI. For example, in figure 10, uphill FDI flows are plotted against uphill sophisticated exports for 22 important emerging market countries for which data are available. There seems to be little correlation between the two. Indeed, there appear to be four distinct categories: (a) countries such as Israel and Malaysia do well on both counts; 122 | A A D I T YA M AT T O O A N D A R V I N D S U B R A M A N I A N FIGURE 5. Average Income Level of World Exports of Sophisticated Products, 1990­2006 a. Including China 26,000 income level (per capita GDP in 1990 PPP dollars) 25,000 24,000 23,000 22,000 21,000 1990 1995 2000 2002 2003 2005 2006 sophisticated products highly sophisticated products b. Excluding China 26,000 income level (per capita GDP in 1990 PPP dollars) 25,500 25,000 24,500 24,000 23,500 23,000 1990 1995 2000 2002 2003 2005 2006 sophisticated products highly sophisticated products Source: United Nations Commodity Trade Statistics database (Comtrade). Note: For definitions of sophisticated products, see note 3. FIGURE 6. Uphill Flows of Sophisticated Exports from Non-OECD Countries, 1990­2006 4.0 14.0 3.5 12.0 3.0 10.0 2.5 percent percent 8.0 2.0 6.0 1.5 4.0 1.0 2.0 0.5 0.0 0.0 1980 1985 1990 1995 2000 2002 2004 2005 2006 sophisticated products (left-hand scale) highly sophisticated products (right-hand scale) Source: United Nations Commodity Trade Statistics database (Comtrade). Note: For each country, the measure of uphill flows is exports of sophisticated goods to countries richer than the country itself as a share of its total sophisticated exports. These are added for all non-OECD countries. FIGURE 7. Uphill Flows of Sophisticated Exports as a Share of Source Country GDP, 1980­2005 0.15 percentage of GDP 0.10 0.05 0.00 1980 1985 1990 1995 2000 2005 India China Brazil Malaysia Korea, Rep. of Mexico Source: United Nations Commodity Trade Statistics database (Comtrade). Note: The measure of uphill flows is the value of exports of sophisticated products as a share of a country's GDP (all measured in current dollars). 124 | A A D I T YA M AT T O O A N D A R V I N D S U B R A M A N I A N FIGURE 8. Average Income Level of Exporters of Other Private Services, 1995­2006 0.61 0.60 index 0.59 0.58 0.57 1995 2000 2005 Source: IMF Balance of Payments Statistics data. Note: The "other private services" category comprises services other than transport and travel and covers most skill- intensive business services. We compute the weighted average of per capita GDP of the exporting countries, with the weights being the share of each country in total exports of other private services. FIGURE 9. Exports to the United States of Other Private Services as a Share of Source Country GDP, 1992­2006 0.006 0.004 percent of GDP 0.002 0.000 1990 1995 2000 2005 India China Brazil Malaysia Korea, Rep. of Mexico Source: U.S. Bureau of Economic Analysis. Note: See note to figure 8. C R I S S C R O S S I N G G L O B A L I Z AT I O N | 125 FIGURE 10. Uphill Flows of FDI and Exports of Sophisticated Goods, Averages for 2003­07 a. FDI 0.04 ISR 0.03 percentage of GDP ZAF MYS IND BGR 0.02 PHL BRA 0.01 ARG IDN KOR PAN MEX CHN TUR TWN CHL POL ROM THA HUN 0.00 0.00 0.01 0.10 0.11 uphill exports of sophisticated goods b. FDI in Manufacturing 0.020 ISR 0.015 percentage of GDP IND 0.010 MYS ARG MEX KOR ZAF 0.005 BRA PHL TUR CHN TWN IDN POL BGR ROM THA HUN 0.000 CHL 0.00 0.05 0.10 0.11 uphill exports of sophisticated goods Sources: United Nations Commodity Trade Statistics database (Comtrade), Thomson Financial SDC Platinum database, and Financial Times FDI Intelligence database. Note: In the first panel, uphill outflows of FDI, measured as FDI outflows of a country to countries with a higher per capita GDP purchasing power parity (PPP) than itself, and exports of sophisticated goods are expressed as a share of a coun- try's GDP. Uphill outflows of FDI in manufacturing and exports of sophisticated goods are expressed as a share of a coun- try's GDP. The economies shown are ARG, Argentina, BGR, Bulgaria; BRA, Brazil; CHL, Chile; CHN, China; HUN, Hun- gary; IDN, Indonesia; IND, India; ISR, Israel; KOR, Republic of Korea; MEX, Mexico; MYS, Malaysia; PAN, Panama; PHL, the Philippines; POL, Poland; ROM, Romania; THA, Thailand; TUR, Turkey; TWN, Taiwan, China; and ZAF, South Africa. 126 | A A D I T YA M AT T O O A N D A R V I N D S U B R A M A N I A N (b) Brazil and India have significant uphill flows of FDI but relatively small uphill exports of sophisticated goods; (c) China and some East Asian economies (Thailand and Taiwan, China) and Eastern European countries (Hungary) are exactly the oppo- site of Brazil and India, with large uphill export flows but limited FDI flows; and (d) some countries such as Chile, Poland, and Romania, score low on both counts. Notwithstanding the above, success in exporting sophisticated goods could be associated with a greater likelihood of investing in manufacturing. But this does not turn out to be the case. The best examples are Brazil and India, which are not big uphill exporters of sophisticated goods but score well on FDI in manufacturing (see panel b of figure 10). Preston Curves How recent is this phenomenon of uphill flows? We cannot carry out meaningful his- torical comparisons for FDI because the data do not allow us to go so far back, but we can attempt to answer this question for exports of sophisticated goods. To do this, we relate uphill flows to the level of per capita GDP of a country for three points in time (1986, 1996, and 2005) that are sufficiently far apart to allow changes to express themselves (figure 11). The noteworthy point that emerges is that the relationship shifts markedly upward in the most recent period for which we have data.6 The shift implies that over time, uphill flows are becoming more common across the income spectrum. We also find that the fit of the relationship between uphill flows and income tightens over time, suggesting that higher-income countries are likely to see more uphill flows. Consequences of Uphillness An obvious question is, do uphill flows matter for economic growth? Hausmann, Hwang, and Rodrik (2007) have argued that the structure of exports does matter for growth. In particular, countries that produce more sophisticated goods (defined as those produced by richer countries) are shown to be more likely to grow faster. But the focus of this paper is not so much the sophistication of exports as whether a country's export pattern defies comparative advantage. As argued earlier, it would not be sur- prising or at odds with the predictions of the standard trade models if a poor country exported relatively sophisticated goods to countries poorer than itself. We are there- fore interested not only in the sophistication of exports, but also in their destination. To pursue this question of whether comparative advantage­defying (alternatively, "uphill") exports have growth consequences, we adopt the basic cross-national regression methodology deployed by Hausmann, Hwang, and Rodrik (2007). Our results for the pure cross-section are presented in tables 1 and 2, and the panel regres- sions are contained in tables 3 and 4. We calculate two measures of uphill exports. In the first, we combine the Hausmann, Hwang, and Rodrik indicator of sophistica- tion (EXPY) with a measure of the average income level of the destination countries receiving such sophisticated exports; specifically, we add the log of the EXPY meas- ure and the log of the average income level of destination countries, and call this UPHILL1. This is the measure used in tables 1 and 3.7 FIGURE 11. Uphill Flows of Sophisticated Exports and Per Capita GDP, Selected Economies, 1986, 1996, and 2005 a. Without controls for area, population, and remoteness 10.5 uphill flows of sophisticated exports 10.0 9.5 9.0 6 7 8 9 10 11 log of per capita GDP (constant PPP dollars) 1986 1996 2005 b. With controls for area, population, and remoteness 10.4 uphill flows of sophisticated exports 10.2 10.0 9.8 9.6 9.4 7 8 9 10 11 12 log of per capita GDP (constant PPP dollars) 1986 1996 2005 Source: United Nations Commodity Trade Statistics database (Comtrade). Note: PPP = purchasing power parity. Uphill flows are measured as the average income level of all the destination coun- tries that receive a country's sophisticated exports (defined here as above-median PRODY exports), where the weights are each destination country's share in total exports of the sending country. The sample is kept constant for all three periods. Panel b includes, for each year for which the relationship is plotted, controls for area, population, and remote- ness, all in log terms. Remoteness, drawn from Berthelon and Freund (2008), is measured as j k where D is distance and there are k foreign countries. 1 remotej N GDP k a D k ij TABLE 1. Growth and Uphill Flows of Sophisticated Exports: Cross-Sectional Regressions Using Scale-Free Measure of Uphill Flows 128 (dependent variable: annual average growth rate, 1994­2003) Variable (1) (2) (3) (4) (5) (6) (7) Initial per capita GDP (log) 0.013* 0.007 0.007 0.003 0.008 0.004 0.009 (0.007) (0.005) (0.005) (0.005) (0.005) (0.007) (0.006) Uphill flows for 1990 (median sophistication) 0.017*** 0.011*** 0.014* (0.005) (0.003) (0.008) Years of primary schooling 0.004 0.015 0.015 0.003 0.015 0.011 0.015* (0.013) (0.009) (0.009) (0.011) (0.009) (0.010) (0.009) Capital stock 0.000 0.000 0.000 0.001 0.000 0.000 0.000 (0.006) (0.005) (0.005) (0.005) (0.005) (0.005) (0.005) Institutional quality (rule of law) 0.009*** 0.008*** 0.008*** 0.008*** 0.009*** 0.008*** 0.008*** (0.003) (0.002) (0.003) (0.003) (0.003) (0.003) (0.002) Uphill flows for 1990 (75th percentile sophistication) 0.011*** (0.003) Uphill flows for 1990 of sophisticated relative to 0.007* unsophisticated products (0.004) Sophistication of exports 0.013** (0.006) Average income level of destination of 0.011*** sophisticated (median) exports (0.004) Uphill flows for 1995 (median sophistication) 0.007** (0.003) Number of observations 58 56 56 60 56 56 55 Adjusted R2 0.260 0.321 0.303 0.224 0.308 0.205 0.328 F-test 5.60 7.19 8.33 7.01 6.01 5.80 5.00 Source: Authors' calculations. Note: Numbers in parentheses are robust standard errors. Columns (2)­(7) exclude China and Ireland. Column (7) is an instrumental variable (IV) estimation with population and area (logs) serving as instruments for uphill flows. *p .1 **p .05 ***p .01 C R I S S C R O S S I N G G L O B A L I Z AT I O N | 129 TABLE 2. Growth and Uphill Flows of Sophisticated Exports: Cross-Sectional Regressions, Sophisticated Exports to Richer Countries, Scaled by GDP (dependent variable: annual average growth rate, 1994­2003) Variable (1) (2) (3) Initial per capita GDP (log) 0.001 0.000 0.000 (0.005) (0.005) (0.005) Downhill export flows of sophisticated (median) 0.099** 0.087** 0.112** products (as share of GDP) (0.047) (0.035) (0.050) Uphill export flows of sophisticated (median) 0.263*** 0.122** 0.096* products (as share of GDP) (0.080) (0.048) (0.048) Years of primary schooling 0.001 0.009 0.009 (0.017) (0.013) (0.013) Capital stock 0.003 0.001 0.000 (0.006) (0.004) (0.004) Institutional quality (rule of law) 0.010*** 0.010*** 0.010*** (0.003) (0.003) (0.003) Exports of nonsophisticated (median) 0.024 products (share of GDP) (0.022) Number of observations 61 59 59 Adjusted R2 0.314 0.236 0.235 F-test 6.17 7.90 7.89 Source: Authors' calculations. Note: Numbers in parentheses are robust standard errors. Columns (2) and (3) exclude China and Ireland. *p .1 **p .05 ***p .01 A particular issue with the Hausmann, Hwang, and Rodrik approach and our adaptation of it is that the measures of sophistication and uphillness are not scaled. For example, these authors' EXPY measure captures the sophistication of an economy's export basket without taking account of the importance, relative to the size of an economy, of the exports of these products. The use of a scale-free measure entails both a benefit and a limitation. The benefit is econometric; there is less endogeneity bias. The downside is that the economic intuition is less clear. Our uphill measure, too, is scale free, capturing the importance of uphill flows in the export basket but not their economywide importance. We accordingly calculate a second measure, which is the share of exports of sophisticated products flowing uphill as a share of GDP. We calculate uphillness by simply adding up the exports that a country sends to trading partners richer than itself. This is called UPHILL2 and is used in tables 2 and 4. In column 1 of table 1, we present the basic results with controls for human cap- ital, physical capital, and institutions. Our measure of uphill flows is positively signed and statistically significant at the 1 percent confidence level. We find that China and Ireland are clear outliers. So in column 2 we drop them and find that the results remain unchanged. The coefficient suggests that a 1 percent increase in uphill flows 130 TABLE 3. Growth and Uphill Flows of Sophisticated Exports: Panel Regressions Using Scale-Free Measure of Uphill Flows) (1) (2) (3) (4) (5) (6) (7) Variable OLS OLS OLS OLS Fixed effects IV IV Per capita GDP (log) 0.006** 0.006** 0.005** 0.004* 0.033*** 0.015* 0.015* (0.003) (0.003) (0.002) (0.002) (0.007) (0.009) (0.009) Uphill flows (median sophistication) 0.011*** 0.011*** 0.001 0.024*** (0.003) (0.003) (0.005) (0.009) Years of primary schooling 0.009** 0.009** 0.009** 0.010** 0.040** 0.013 0.012 (0.004) (0.004) (0.005) (0.005) (0.017) (0.020) (0.020) Uphill flows (75th percentile sophistication) 0.010*** 0.025*** (0.003) (0.009) Uphill flows of sophisticated relative to unsophisticated products 0.009*** (0.003) Number of observations 267 267 266 267 267 256 255 Adjusted R2 0.076 0.106 0.090 0.080 0.117 0.005 0.045 F-test 7.07 4.44 4.74 3.79 7.96 3.25 2.79 Number of countries 65 Source: Authors' calculations. Note: IV, instrumental variable; OLS, ordinary least squares. Numbers in parentheses are robust standard errors. The instruments for uphill flows in columns (6) and (7) are population and remoteness (in logs). All columns except column (1) include time effects. Fixed effects are included only in column (5). *p .1 **p .05 ***p .01 C R I S S C R O S S I N G G L O B A L I Z AT I O N | 131 TABLE 4. Growth and Uphill Flows of Sophisticated Exports: Panel Regressions, Sophisticated Exports to Richer Countries, Scaled by GDP (1) (2) (3) (4) (5) Random Fixed Fixed Fixed Fixed Variable effects effects effects effects effects Per capita GDP (log) 0.007*** 0.063*** 0.063*** 0.060*** 0.042*** (0.003) (0.012) (0.012) (0.012) (0.008) Uphill export flows of 0.229*** 0.231*** 0.226** sophisticated (median) products (0.055) (0.085) (0.090) (as share of GDP) Downhill export flows of 0.024 0.056 0.063 sophisticated (median) (0.031) (0.049) (0.052) products (as share of GDP) Years of primary schooling 0.019*** 0.021 0.021 0.022 0.052*** (0.007) (0.019) (0.019) (0.020) (0.018) Exports of nonsophisticated (median) 0.012 products (as share of GDP) (0.035) Uphill export flows of 0.264 0.489** sophisticated (75th percentile) (0.197) (0.205) products (as share of GDP) Downhill export flows of 0.059 0.081 sophisticated (75th percentile) (0.127) (0.151) products (as share of GDP) Exports of 0.037 0.034 nonsophisticated (75th percentile) (0.034) (0.036) products (as share of GDP) Number of observations 258 258 258 258 258 Number of rcode 64 64 64 64 64 R2 0.13 Adjusted R2 0.288 0.285 0.271 0.171 F-test 5.92 5.19 4.82 5.59 Source: Authors' calculations. Note: Numbers in parentheses are robust standard errors. All columns except column (1) include time effects. *p .1 **p .05 ***p .01 could increase growth by about 1.1 percent a year. (The partial scatterplot of this regression is shown in figure 12.) In column 3 we use the alternative measure of sophistication based on a 25th-percentile cutoff of products. In column 4 we use our uphill flow measure for 1995 instead of 1990. In column 5 we disaggregate our uphill measure into the sophistication component and the destination component and find that each is significant with the same magnitude. (The equality of the two coefficients cannot be rejected.)8 In column 6 we subtract the destination income of countries receiving unsophisticated products from the destination income of coun- tries receiving sophisticated products. This is a kind of validation check. In all cases the coefficient on UPHILL1 remains significant, suggesting some strong association. 132 | A A D I T YA M AT T O O A N D A R V I N D S U B R A M A N I A N FIGURE 12. Cross-Section Regression: Scatter Plot of Growth on Uphill Flows, Selected Economies 0.03 KOR IND 0.02 GRC LKA MUS PAN ROM growth residuals CRI 0.01 MLT MYS IDN CHL PER TUR ESP FIN THA AUS SEN PRT CAN PHL ISL NLD GBR GTM USA PAK 0.00 SLV MWI SWE MEX TGO NZL ITA BRA COL DNK FRA SGP ISR AUT HND MAR JAM HKG NIC 0.01 DEU ARG JPN PRY KEN JOR 0.02 CHE MDG URY 1.5 1.0 0.5 0.0 0.5 1.0 residuals of uphill flows coef = 0.01146931, (robust) se = 0.00344606, t = 3.33 Source: Based on the regression in table 1, column 2. Note: Controls include initial income, institutional quality, primary school enrollment, and capital stock. Coefficient, 0.01146931; (robust) standard error, 0.00344606; t, 3.33. The economies shown are ARG, Argentina; AUS, Australia; AUT, Austria; BRA, Brazil; CAN, Canada; CHE, Switzerland; CHL, Chile; COL, Colombia; CRI, Costa Rica; DEU, Ger- many; DNK, Denmark; ESP, Spain; FIN, Finland; FRA, France; GBR, United Kingdom; GRC, Greece; GTM, Guatemala; HKG, Hong Kong, China; HND, Honduras; IDN, Indonesia; IND, India; ISL, Iceland; ISR, Israel; ITA, Italy; JAM, Jamaica; JOR, Jordan; JPN, Japan; KEN, Kenya; KOR, Republic of Korea; LKA, Sri Lanka; MAR, Morocco; MDG, Madagascar; MEX, Mexico; MLT, Malta; MUS, Mauritius; MWI, Malawi; MYS, Malaysia; NIC, Nicaragua; NLD, Netherlands; NZL, New Zealand; PAK, Pakistan; PAN, Panama; PER, Peru; PHL, the Philippines; PRT, Portugal; PRY, Paraguay; ROM, Romania; SEN, Senegal; SGP, Singapore; SLV, El Salvador; SWE, Sweden; TGO, Togo; THA, Thailand; TUR, Turkey; URY, Uruguay; and USA, United States of America. In column 7, to address the potential endogeneity of our uphill measure, we instru- ment for it with the log of population and log of area, as in Hausmann, Hwang, and Rodrik (2007). The first stage suggests that the instruments are reasonably but not exceptionally strong. In the second stage, the uphill measure has about the same mag- nitude and remains significant, albeit at the 10 percent confidence level. Of course, there are a number of issues with our estimation method: some of our right-hand-side variables are prone to endogeneity bias despite our use of initial rather than contemporaneous values; we may be omitting other variables; and our variables could be mismeasured. Our results should therefore be interpreted at this stage as conditional associations rather than as full identifications. In table 2, we use the UPHILL2 measure (recognizing that this may well add another layer of endogeneity bias). We introduce these measures in the cross-country regres- sions instead of their scale-free counterparts that we used earlier. We can either add the total share of sophisticated exports to GDP and the uphill share of that as two C R I S S C R O S S I N G G L O B A L I Z AT I O N | 133 variables or simply use the uphill and downhill shares of sophisticated exports. We do the latter. We find that the coefficient on the share of uphill products to GDP is significant (column 1, table 2) and remains so after excluding Ireland and China (column 2). In column 3 we also control for the share of total unsophisticated exports in GDP and find that this variable is not significant and does not affect our uphill flow measure.9 Given the limitations of the above analysis, we turn to panel estimations in tables 3 and 4.10 In table 3, we use the scale-free measures, and in table 4, we use the meas- ures scaled by GDP. Instead of going through all the columns as was done above, we highlight the key findings. When we use the scale-free measures (that is, UPHILL1), we find that uphill flows are significant except when we add country fixed effects (column 5). But instrumental variable (IV) estimations (in this case, with population and remoteness of a country from the world's center of gravity as instruments) yield very strong first-stage results, with correspondingly strong and statistically significant coefficients for uphill flows in the second stage (columns 6 and 7). When we use the UPHILL2 measure (which is scaled by GDP), we find that uphill flows are statisti- cally significant (columns 2, 3, and 5), even after adding country and time effects. An issue we attempted to explore in greater detail was the PRODY measure. One could also try to get a measure of "sophistication" of products by, for example, using the level of education, or research and development (R&D), in the exporting coun- try rather than per capita income. For each product, we constructed a weighted aver- age of the exporting countries' secondary school enrollment ratio or spending on R&D as a share of GDP. The results were very similar to those from PRODY.11 For example, in table 1, when we replaced the uphill measures based on PRODY with those based on education and R&D, the coefficient on the uphill measure was cor- rectly signed and significant. The reason, of course, is that the income-based measure and the measures based on education and R&D are highly correlated, and the dif- ferences are not large enough to conclude that it is education, not per capita GDP, that is the more accurate measure of sophistication. Discussion and Limitations of the Analysis This paper is a first attempt to document a possibly new phenomenon, which we call uphill flows of skills. We presented a set of stylized facts relating to uphill flows of goods, services, and FDI and preliminary estimates of the consequences of these flows. We have not examined the determinants of these flows or elaborated on the possible channels through which these flows might have growth consequences. We offer next some suggestions with regard to these two issues. Explaining Uphill Flows Uphill flows raise some interesting theoretical questions. First, and most obvious, they seem to defy the prediction of the pure Heckscher-Ohlin model, where trade is 134 | A A D I T YA M AT T O O A N D A R V I N D S U B R A M A N I A N determined by relative factor endowments. Second, although such flows could be seen as a manifestation of intraindustry trade, driven by economies of scale and imperfect competition, this type of trade has typically been predicted between coun- tries at similar levels of development (Helpman and Krugman 1985). Two possible explanations for uphill flows--one domestic and one international-- suggest themselves. Within developing countries, for example, there could be atypi- cal patterns of development as a result of historical factors and policy actions. Two good examples are India and South Africa, both of which have exhibited skill-intensive patterns of development (see Amin and Mattoo 2006; Kochhar et al. 2006). In the Indian case, this was because higher education was favored at the expense of basic education, while in South Africa, apartheid and labor market policies played a role. Recent research shows that some of these larger developing countries are investing proportionately more in technical education than are poorer or richer countries (Sequeira 2003). If such policies are then overlaid on regional disparities, it is possi- ble for pockets to emerge within developing countries that are sufficiently endowed with skills or are sufficiently developed to explain the observed patterns of "criss- crossing globalization." In other words, the inconsistency of uphill flows with theory may be more apparent than real if we were to think of countries like China and India not as single units but as heterogeneous economic units (or regions) with widely differing relative factor endowments (Subramanian 2007). It is also possible for the relevant heterogeneity to emerge at the firm level. For example, Melitz (2003) allows for firm-level heterogeneity in productivity and fixed costs of exporting and shows that only the most productive firms export. Helpman, Melitz, and Yeaple (2004) demonstrate, in turn, that among the firms that serve for- eign markets, only the most productive engage in FDI. It is conceivable that some firms, even in developing countries, are so productive that they can incur the fixed costs of exporting and investing abroad. Furthermore, if the fixed costs of penetrat- ing foreign markets vary across destinations--say, by per capita income of the desti- nation country--it is possible for productivity differences across developing country firms to result in the phenomenon of "uphill" flows that we document. External policies could be another cause of uphill flows. One factor may have been international patterns of protection, in particular, rich-country barriers against imports of less-skill-intensive products, and developing-country barriers against imports of more-skill-intensive products. Thus, the larger developing countries may have been inhibited from exploiting their natural comparative advantage by exporting less-skill- intensive products to richer countries and more-skill-intensive products to poorer coun- tries. Put differently, if there is learning by doing, it may be that increases in uphill sophisticated exports have been possible because protection allowed domestic produc- ers to catch up with foreign producers in terms of competitiveness. Uphill Flows and Growth Standard theories of trade--Heckscher-Ohlin, intraindustry, and even the new het- erogeneous firm­based models--primarily see the gains from trade in static welfare C R I S S C R O S S I N G G L O B A L I Z AT I O N | 135 rather than dynamic growth (Bernard et al. 2007). Our results are more in the spirit of endogenous growth theories, which see trade as affecting the incentives and oppor- tunities for dynamic benefits such as technology acquisition and learning by doing. Although a large part of the benefits of trade has traditionally been seen as access to imports and inward FDI, there is a growing recognition that exports and outward FDI may also confer important benefits. We have not examined in any detail the channels through which uphill exports of sophisticated goods and services affect overall economic performance. One possibility is that our measure of destination may actually capture a finer degree of product differentiation, in horizontal or vertical terms. For example, Schott (2005) established that even when developing-country exports fall within the same product categories as rich-country exports, they tend to have lower unit values and may be located lower on quality ladders. In other words, what we identify as uphill flows may just be an alternative or complementary measure for product quality and sophistication. Our findings could then be seen as adding to the evidence that such quality matters for economic performance (Hausmann, Hwang, and Rodrik 2007). Another possibility is that final exports of sophisticated goods by a country may reflect merely its comparative advantage in the final assembly stage rather than a deeper sophistication in its production processes. For example, a significant proportion of China's uphill exports of sophisticated goods contains imports of sophisticated com- ponents from rich countries. On the one hand, this could indicate that we are mis- measuring sophistication. On the other hand, our measure could capture the extent of a country's participation in modern global production chains, which confer benefits in terms of knowledge of markets, just-in-time capability, improved production technol- ogy, and so on. Thus, what we capture--imperfect though it undoubtedly is--may provide clues about an additional channel through which the impact of global inte- gration is felt. As noted at the outset of this paper, there is now increasing evidence supporting the existence of these channels (Javorcik 2004). In principle, these benefits can also arise from sales to foreign firms located abroad. Furthermore, uphill flows could affect growth through induced changes in economy- wide skill acquisition and, hence, in long-run endowments, creating a self-reinforcing and virtuous cycle. India again provides a relevant example. Educational attainment in India, especially at the primary and secondary levels, was disappointing until the early 1990s. In the past 15 years, however, educational indicators have improved markedly. Although greater government attention has been important, a key change has been increased demand for education because of higher returns to human capital, which in turn is a consequence of increased skill-intensive and uphill specialization. (The derived demand for skills and, hence, education is arguably a function not just of what is sold, but also to whom it is sold.) This demand has elicited a supply response, largely from the private sector, that has led to a more rapid spread of edu- cation and skills (Kremer et al. 2006). In summary, if there are benefits from uphill flows, in some circumstances signif- icant development benefits might derive not from deifying comparative advantage, but from defying it. 136 | A A D I T YA M AT T O O A N D A R V I N D S U B R A M A N I A N Annex. Data on Foreign Direct Investment To what extent do we see uphill flows of foreign direct investment (FDI) in the avail- able data, and how have these flows changed in recent years? To pursue this ques- tion, we examined merger and acquisition (M&A) FDI data from the Thomson Financial's SDC Platinum database from January 1995 to December 2007 and data on greenfield investment from the Financial Times FDI Intelligence, which is a private organization that compiles proprietary data on such investments. Data Sources The World Investment Report (WIR) database of the United Nations Conference on Trade and Development (UNCTAD) includes coverage of both total FDI and M&A inflows and outflows for each country, but the published dataset does not break down these flows on a bilateral basis--data on countries of origin are not available for inflows, and data on destination countries are not provided for outflows. Although some UNCTAD-based datasets used by other researchers have endeavored to create this bilateral breakdown, these datasets generally examine FDI stocks rather than flows and have reliable data across a broad range of countries for only a few years, generally between 2003 and 2005. By contrast, reasonably comprehensive and highly granular coverage is available for M&A and greenfield FDI in the form of commercial financial databases. Such databases report information at the individual transaction level, enabling analysis on three principal axes: source countries of flows, destination countries of flows, and industry sectors of flows. For this analysis, the SDC Platinum database was chosen for its comprehensive dataset, including hundreds of thousands of cross-border M&A transactions from 1985 up to the present. The FDI Intelligence database produced by the Financial Times has tracked green- field foreign direct investment throughout the world since 2003. Greenfield direct investment is defined as the expansion or creation of physical facilities in any location other than the headquarters of a company. For each greenfield investment project, the database has the actual or estimated investment in dollars terms and the actual or estimated jobs created from the project. Every project is assigned to a source market and a destination market and is also disaggregated to the level of an industry sector, industry cluster, or business activity, in increasing order of disaggregation. The database is continuously updated and currently holds data on more than 78,800 projects. For the purposes of the paper, we focused on the period 2003­07. Taken at the industry sector level, this period gives 35,045 source-destination-industry observations totaling US$4.3 trillion in value. Collapsing across industry sectors to arrive at aggre- gate numbers for source markets yields 9,263 bilateral greenfield investment projects over this period, for a total of 132 source markets and 184 destination markets. Combining the greenfield FDI data with the M&A data from the Thomson Finan- cial SDC Platinum database yields 10,457 bilateral recorded investment projects in either or both categories, with a total value over the whole period of US$7.5 trillion. C R I S S C R O S S I N G G L O B A L I Z AT I O N | 137 Timeframe In seeking to examine uphill flows of FDI, we note that the years of greatest interest are evidently the most recent ones. Whereas the major East Asian countries have had a significant presence as exporters of FDI for some time, only since the turn of the millennium have the four BRIC countries (Brazil, Russia, India, and China) joined them in this regard, and only since 2002 have net FDI outflows for these four coun- tries combined amounted to more than 2 percent of total world FDI flows. Major oil- exporting countries such as Mexico, Saudi Arabia, and the United Arab Emirates have joined these ranks even more recently. The overall period chosen for analysis for this study was thus that covering the years from 2003 to 2007, inclusive. Data Coverage For the purpose of M&A analysis, only completed transactions where transaction value was disclosed and recorded, and where the stake acquired in the target com- pany met or exceeded 10 percent, were included. Accurate recording of transaction values is clearly essential to any calculation of flows, while stakes below 10 percent are considered too small to be classified as FDI under most definitions. Including only disclosed-value transactions eliminates a little over half the transactions recorded in the database, since many transactions are for unlisted companies or for other reasons do not face strict disclosure requirements. The dataset resulting from those selection criteria includes some 37,963 deals totaling US$8.4 trillion in value. Comparison of the dataset resulting from this selection with M&A data and total FDI data provided in aggregate form in UNCTAD's WIR demonstrates that the over- all transaction coverage provided by the SDC Platinum database over this time period is strong. Only between 2000 and 2002 is the total value of M&A transactions reported in the SDC database below that reported in the WIR; in those years, cover- age remains above 80 percent, while in all the remaining years, the SDC dataset cap- tures a bigger total transaction volume than that reported by the WIR. Although the overall volume of transactions captured by the SDC-based dataset is higher than that reported by UNCTAD, for certain years and certain categories the coverage is lower. Thus, whereas--compared with UNCTAD--SDC data report higher M&A FDI inflows into OECD countries (see below for notes on country groupings) for all years except 2000­2002, the OECD outflow volumes reported are routinely lower than those reported by UNCTAD. Country Groupings and Data Overview OECD membership was the principal determinant used to distinguish between devel- oped and emerging countries. Although Mexico and the Republic of Korea are now both OECD members, for the purposes of this analysis both were included in the emerging-countries grouping rather than the OECD grouping. Offshore financial centers and Mauritius were excluded from the analysis. 138 | A A D I T YA M AT T O O A N D A R V I N D S U B R A M A N I A N Notes The authors are grateful to Brad Jensen, Dani Rodrik, Tony Venables, Daniel Xie, and, espe- cially, our discussant Beata Javorcik for valuable comments; to Marko Klasnja and Janak Mayer for outstanding research assistance; and to Francis Ng for his generous statistical help. Janak Mayer prepared the annex to this paper. The research for this paper is supported in part by the governments of Norway, Sweden, and the United Kingdom through the Multidonor Trust Fund for Trade and Development. 1. Data for M&A and greenfield investments are from different sources, as described in the annex. 2. In subsequent work, we plan to increase the sample to cover most countries. 3. There are two definitions of sophisticated products. The first covers exports that lie above the median value of PRODY (described in the text), calculated for 1990. The second cov- ers exports that lie in the top 25th percentile of PRODY values (highly sophisticated prod- ucts, or HSPs). For each definition, we compute the weighted average per capita GDP of the exporting countries, with the weights being the share of each country in the total exports of sophisticated products. 4. Of course, this development could simply reflect the fact that richer countries, which are more likely to demand sophisticated goods, have grown faster than poorer countries. But during this period, the non-OECD countries in our sample grew substantially faster than the OECD countries. 5. It is, in principle, possible to combine OECD data and the IMF Balance of Payments Sta- tistics to obtain an estimate of the share of exports of skilled services by developing coun- tries directed toward OECD countries. However, significant inconsistencies in the data from those two sources prevent meaningful comparison. 6. The finding of an upward shift holds true when we estimate the relationship (a) without keeping the sample constant across time periods; (b) after controlling for a country's area, population, and remoteness from the world's center of gravity; and (c) using alternative measures of the uphillness of flows. Also, when we estimated the Preston relationships in a formal panel context, we found the coefficient on the 2005 dummy to be positive and statistically significant. 7. As a referee pointed out, in principle, it may not be necessary to make such a drastic distinction between sophisticated and unsophisticated goods. We could arrange goods along a continuum from less to more sophisticated on the basis of their PRODY values. A continuous measure of uphill exports of sophisticated goods could then be given by a weighted average of the product of the PRODY value of an export and the income level of the destination country, where the weight is exports of the product to a particular des- tination as a share of total exports (that is, the sum of exports of all products to all desti- nations). In notational form, such a measure would be xij aa PiYj i j a a xij i j where Pi is the PRODY value of product i and Yj is the per capita income level of the des- tination country j. This measure is analogous to an interaction between EXPY and the average income level of all exports. It turns out to be highly correlated (0.93) with EXPY because there is relatively little variation in the average income level of all exports. We therefore use our uphill measure, which has a binary definition of sophistication and is less correlated with EXPY. C R I S S C R O S S I N G G L O B A L I Z AT I O N | 139 8. The equality of these components provides additional econometric justification for com- bining them, as we have done in UPHILL1. 9. Population and area, which were decent instruments for our UPHILL1 measure, were poor instruments for the UPHILL2 measure, precluding the possibility of instrumental variable (IV) estimations. 10. In this panel, we retained Ireland and China because they made no difference to the results. 11. The results are available from the authors on request. References Amin, Mohammad, and Aaditya Mattoo. 2006. "Do Institutions Matter More for Services?" Policy Research Working Paper 4032, World Bank, Washington, DC. Bernanke, Ben S. 2006. "Global Economic Integration: What's New and What's Not?" Fed- eral Reserve Board, Washington, DC. http://www.federalreserve.gov/boarddocs/speeches/ 2006/20060825/default.htm. Bernard, Andrew B., J. Bradford Jensen, Stephen J. Redding, and Peter K. Schott. 2007. "Firms in International Trade." Journal of Economic Perspectives 21 (3): 105­30. Berthelon, Matias, and Caroline Freund. 2008. "On the Conservation of Distance in Interna- tional Trade." Journal of International Economics 75 (2): 310­20. Borensztein, Eduardo, José De Gregorio, and Jong-Wha Lee. 1998. "How Does Foreign Direct Investment Affect Growth?" Journal of International Economics 45 (June): 115­35. Burgess, Robin, and Anthony J. Venables. 2004. "Toward a Microeconomics of Growth." Policy Research Working Paper 3257, World Bank, Washington, DC. Caballero, Ricardo J., Emmanuel Farhi, and Pierre-Olivier Gourinchas. 2008. "An Equilibrium Model of `Global Imbalances' and Low Interest Rates." American Economic Review 98 (1, March): 358­93. Coe, David T., Elhanan Helpman, and Alexander W. Hoffmaister. 1997. "North-South R&D Spillovers." Economic Journal 107: 134­49. De Loecker, Jan. 2007. "Do Exports Generate Higher Productivity? Evidence from Slovenia." Journal of International Economics 73 (9, September): 69­98. Feenstra, Robert, and Hiau Looi Kee. 2004. "Export Variety and Country Productivity." Policy Research Working Paper 3412, World Bank, Washington, DC. Haskel, Jonathan E., Sonia C. Pereira, and Matthew J. Slaughter. 2002. "Does Inward Foreign Direct Investment Boost the Productivity of Domestic Firms?" NBER Working Paper 8724, National Bureau of Economic Research, Cambridge, MA. Hausmann, Ricardo, Jason Hwang, and Dani Rodrik. 2007. "What You Export Matters." Journal of Economic Growth 12 (1, March): 1­25. Helpman, Elhanan, and Paul R. Krugman. 1985. Market Structure and Foreign Trade: Increasing Returns, Imperfect Competition, and the International Economy. Cambridge, MA: MIT Press. Helpman, Elhanan, Marc J. Melitz, and Stephen R. Yeaple. 2004. "Export versus FDI with Heterogeneous Firms." American Economic Review 94 (1, March): 300­16. Hoekman, Bernard M., and Beata Smarzynska Javorcik, eds. 2006. Global Integration and Technology Transfer. Houndsmills, Basingstoke, Hampshire, U.K.: Palgrave Macmillan; Washington, DC: World Bank. 140 | A A D I T YA M AT T O O A N D A R V I N D S U B R A M A N I A N Hummels, David, and Peter Klenow. 2005. "The Variety and Quality of a Nation's Exports." American Economic Review 95 (3, June): 704­23. Javorcik, Beata S. 2004. "Does Foreign Direct Investment Increase the Productivity of Domes- tic Firms? In Search of Spillovers through Backward Linkages." American Economic Review 94 (3): 605­27. Kochhar, Kalpana, Utsav Kumar, Raghuram Rajan, Arvind Subramanian, and Ioannis Tokat- lidis. 2006. "India's Pattern of Development: What Happened, What Follows?" Journal of Monetary Economics 53 (5, July): 981­1019. Kremer, Michael, Nazmul Chaudhury, F. Halsey Rogers, Karthik Muralidharan, and Jeffrey Hammer. 2006. "Teacher Absence in India: A Snapshot." Journal of the European Economic Association 3 (2­3): 658­67. Krugman, Paul R. 2008. "Trade and Wages Reconsidered." Brookings Papers on Economic Activity 2008 (1): 103­54. Lumenga-Neso, Olivier, Marcelo Olarreaga, and Maurice Schiff. 2001. "On `Indirect' Trade- Related Research and Development Spillovers." Policy Research Working Paper 2580, World Bank, Washington, DC. Melitz, Marc. 2003. "The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry Productivity." Econometrica 71 (November): 1695­725. Prasad, Eswar, Raghuram G. Rajan, and Arvind Subramanian. 2007. "Foreign Capital and Economic Growth." Brookings Papers on Economic Activity 2007 (1): 153­230. Ramamurti, Ravi, and Jitendra V. Singh. 2009. "Indian Multinationals: Generic Internation- alization Strategies." In Emerging Multinationals in Emerging Markets, ed. Ravi Rama- murti and Jitendra V. Singh. Cambridge, U.K.: Cambridge University Press. Samuelson, Paul A. 2004. "Where Ricardo and Mill Rebut and Confirm Arguments of Main- stream Economists Supporting Globalization." Journal of Economic Perspectives 18 (3): 135­46. Schott, Peter K. 2005. "Across-Product versus Within-Product Specialization in International Trade." Quarterly Journal of Economics 119 (2, May): 647­78. ------. 2007. "The Relative Sophistication of Chinese Exports." Economic Policy 53: 5­49. Sequeira, Tiago Neves. 2003. "High-Tech Human Capital: Do the Richest Countries Invest the Most?" B. E. Journal of Macroeconomics, Topics in Macroeconomics 3 (1). http://www. bepress.com/bejm/topics/vol3/iss1/art13/. Subramanian, Arvind. 2007. "Precocious India." Business Standard (New Delhi), August 14. http://www.petersoninstitute.org/publications/opeds/oped.cfm?ResearchID=792. Comment on "Crisscrossing Globalization: The Phenomenon of Uphill Skill Flows," by Aaditya Mattoo and Arvind Subramanian BEATA SMARZYNSKA JAVORCIK Aaditya Mattoo and Arvind Subramanian have written an enjoyable and thought- provoking paper that challenges the reader to question some basic economic theories and conventional wisdom. In the first part of their study, the authors document a new phenomenon in which skills embodied in goods, services, or capital (in the form of entrepreneurial and man- agerial skills associated with foreign direct investment) flow from poor to rich coun- tries. They call this phenomenon uphill flows. The existence of uphill flows may come as a surprise, as it appears to contradict one of the basic theorems of interna- tional trade, which predicts that developing countries, where unskilled labor is abun- dant and skilled labor is scarce relative to the industrial world, will specialize in unskilled labor­intensive exports. The second part of the paper argues that there is a positive link between uphill export flows and economic growth in exporting countries. In other words, the authors contend that economic growth is affected not just by the product composi- tion of exports (as claimed by Hausmann, Hwang, and Rodrik 2007), but also by the type of destination market. The paper is likely to stimulate lively discussion on the link between the quality of exports and export markets and economic growth. Let me start the debate by posing a few questions under the following headings: (a) Are uphill flows real, or are they a statistical illusion? (b) Do uphill flows really contradict economic theory? (c) Uphill exports and growth: so what? Are Uphill Flows Real, or Are They a Statistical Illusion? There are several reasons why uphill export flows may not be "real." The first is that high skill intensity of developing country exports may simply reflect reliance of Beata Smarzynska Javorcik is a reader in economics at the University of Oxford and a research affiliate with the Centre for Economic Policy Research, London. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank 141 142 | B E ATA S M A R Z Y N S K A J AV O R C I K producers on skill-intensive imported inputs. For instance, China may import elec- tronic components and assemble them into computers that are then exported to the United States. Although the authors would classify such exports as uphill flows, in reality they would contain little input of Chinese skilled labor. This possibility can be easily checked because China's trade statistics explicitly designate "processing imports" as imports of intermediate inputs to be used to produce products solely for export and define "processing exports" as exports that use these imported inputs. Trade figures confirm the high import content of skilled labor­intensive exports from China. For instance, in 2004 processing exports constituted 96 percent of Chinese exports of office and computing machinery. In communications equipment, the cor- responding figure was 86 percent, and in medical, precision, and optical instruments, it was 76 percent (Dean and Lovely 2008). The second reason why uphill exports may create a misleading impression is that multinational enterprises from industrial economies are responsible for a large share of export flows from developing to developed countries. In China foreign investment enterprises accounted for nearly a third of industrial output produced in 2001. In the same year, manufacture of electronic components ranked second among Chinese sec- tors in terms of entry of foreign investment enterprises (Amiti and Javorcik 2008). Uphill exports associated with multinational enterprises may to a large extent embody knowledge produced in the developed-country headquarters of these enter- prises, rather than intangible assets created in the exporting country. The third difficulty posed by the definition of uphill flows stems from the large dif- ferences in regional characteristics observed within developing countries, which put into question the wisdom of treating a country like China as a unit of analysis. The coastal regions of China may be more similar to developed than developing economies, so should exports from these regions really be classified as uphill flows? Do Uphill Flows Really Contradict Economic Theory? The recent increase in foreign direct investment (FDI) from developing to developed countries is part of the uphill flows documented by the authors. On the surface, exports of capital from capital-scarce developing countries to capital-abundant industrial countries appear to contradict economic theory. But there are three simple reasons why this contradiction may not stand up to scrutiny. First, the apparent contradiction disappears when we reinterpret FDI undertaken by developing country firms as imports of intangible assets rather than as exports of cap- ital. For instance, Tata's acquisition of Jaguar may be viewed as a purchase of technol- ogy and a brand name. Lenovo's acquisition of IBM production facilities may be con- sidered a purchase of trademarks (ThinkPad), reputation, and distribution channels. Such "knowledge-sourcing" FDI, undertaken in order to import intangible assets to (relatively knowledge-scarce) developing countries, is consistent with economic theory. Second, export of capital may be a necessary complement to export of unskilled labor­intensive or natural resource­intensive products. For example, when a large C O M M E N T O N M AT T O O A N D S U B R A M A N I A N | 143 Indian outsourcing company, WIPRO, acquired American and European consulting firms, it purchased a client base in those markets in the hope of leveraging its exports of information technology and back-office services from India. Similarly, Gazprom's acquisition of stakes in oil distribution systems seems to be a natural step comple- menting its oil exports. Third, many highly publicized acquisitions undertaken in industrial countries by developing-country interests have been made by sovereign wealth funds and are not necessarily driven solely by market forces. (Political considerations may play a role.) Our economic theories simply do not apply to these cases. Uphill Exports and Growth: So What? The positive relationship between economic growth and uphill exports documented by the authors makes it really tempting to advise policy makers, "Export sophisti- cated goods to rich countries, and your country will grow faster." Yet doing so is akin to saying, "Practice bodybuilding, and you will become the next governor of California." Arnold Schwarzenegger was indeed a bodybuilder, but his highly suc- cessful Hollywood career and marriage into the Kennedy family may have been equally (if not more) important to his bid for political office. Even if we believe in a causal link from uphill export flows to economic growth, what is a "rich-country export good"? (Sophisticated exports are defined as products that tend to be exported by rich countries.) Even if a poor country exports a good classified in the same product category as a good exported by a rich country, the two can be very different. For instance, men's cotton shirts imported by the United States from Japan are roughly 30 times as expensive as the identically classified variety orig- inating in the Philippines (Schott 2004). Thus, ideally we would like to isolate char- acteristics of export goods that stimulate growth. Are such products characterized by long quality ladders? Speed of innovation? Or skill content? Furthermore, in order to provide meaningful policy advice, we need to understand the mechanism through which a certain type of export can lead to faster economic growth. After all, saying "All you need to do to become the next governor of California is to win the election" is not terribly helpful. One possibility is that the ability to serve developed-country markets creates an incentive for developing-country producers to increase the quality of their export products. This leads to additional investment in technology and physical assets, buildup of knowledge, and knowledge spillovers. Another possibility is that sophisticated buyers in developed countries transfer knowl- edge to developing-country producers. Patterns observed in Mexico after the North American Free Trade Agreement (NAFTA) came into effect are consistent with the above scenarios (Iacovone and Javorcik 2008). The data suggest that entry of Mexican producers into export mar- kets is preceded by increased investment in physical assets. There is also evidence sug- gesting that high-quality products are more likely to be exported (where quality is measured as the ratio of the domestic unit value obtained by a given producer for a 144 | B E ATA S M A R Z Y N S K A J AV O R C I K given product in a given time period to the average unit value of the same product observed in the same time period). An important implication of the paper by Mattoo and Subramanian is the need for further work aiming to understand how (if at all) learning from exporting takes place. The link between uphill flows and economic growth documented in their study suggests that such learning may occur only for some types of products and export markets--which could explain why the literature on this topic has produced conflicting results. References Amiti, Mary, and Beata S. Javorcik. 2008. "Trade Costs and Location of Foreign Firms in China." Journal of Development Economics 85 (1­2): 129­49. Dean, Judith, and Mary Lovely. 2008. "Trade Growth, Production Fragmentation, and China's Environment." NBER Working Paper 13860, National Bureau of Economic Research, Cambridge, MA. Hausmann, Ricardo, Jason Hwang, and Dani Rodrik. 2007. "What You Export Matters." Journal of Economic Growth 12 (1): 1­25. Iacovone, Leonardo, and Beata S. Javorcik. 2008. "Shipping Good Tequila Out: Investment, Domestic Unit Value and Entry of Multi-Product Plants into Export Markets." University of Oxford, Oxford, U.K. Schott, Peter. 2004. "Across-Product versus Within-Product Specialization in International Trade." Quarterly Journal of Economics 119 (2, May): 647­78. Migration, Remittances, and the Transition from Foreign Aid The Aid-Migration Trade-off JEAN-PAUL AZAM AND RUXANDA BERLINSCHI This paper highlights an empirically significant trade-off between the aid flows deliv- ered by donor countries and the inflows of migrants that they receive from develop- ing countries. It draws the implications for aid policy from a simple game-theoretic model, after reviewing the recent literature on the effects and motivations of foreign aid to developing countries. The paper is part of the recent effort by economists, goaded by the dead end in which the "aid-ineffectiveness" literature had cornered itself, to discover the hidden agenda behind foreign aid. Are migrants a blessing or a curse? In rich countries, the threat of an invasion by poor migrants from the South is evoked time and again, especially before important elec- tions. It seems that a sizable constituency exists for exerting pressure on governments with a view to inducing them to erect legal barriers against immigration. Some north- ern European countries that were once very liberal in this respect have recently wit- nessed the emergence of a National Front­type movement, with a fairly aggressive attitude toward immigration. This phenomenon can be observed even in countries where unemployment is negligible, as in the Netherlands. At the same time, there are voices cautioning that keeping migrants at bay is nothing like a free good. Borjas (1995), for example, analyzes the benefits accruing to the receiving coun- try in a competitive general equilibrium framework. His model indicates that natives benefit from immigration because of the production complementarities that exist between immigrant workers and other factors of production and that these benefits are larger when immigrants' assets are sufficiently different from the stock of native productive inputs. A different argument is used by Ortega (2000) in a dynamic labor Jean-Paul Azam is director of the Atelier de Recherche Quantitative Appliquée au Développement Économique (ARQADE), Toulouse School of Economics, University of Toulouse. He is a researcher with the Institut d'Economie Industrielle (IDEI) at the University of Toulouse and a senior member at the Institut Universitaire de France. Ruxanda Berlinschi is a lecturer at the Toulouse School of Economics. Comments by participants at the ABCDE conference are gratefully acknowledged. In particular, the authors thank the discussant, Melvin Ayogu. Jennifer Hunt and Devesh Kapur also offered helpful comments. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank. 147 148 | J E A N - PA U L A Z A M A N D R U X A N D A B E R L I N S C H I market model with multiple equilibria. In this model, immigration tilts the bargain- ing strength in favor of firms, with a positive impact on employment and wages. Other benefits have been identified outside the labor market. Gubert (2003) pre- sents a striking calculation: if France accepted just 60,000 more Malian migrants, and if the new migrants had the same propensity to send remittances home as those cur- rently living in France, total remittances would be equivalent to the aid that France is currently sending to Mali. This is a negligible number for a receiving country whose population is more than a thousand times larger than the number cited. The added migration would also help trim public expenditures, as many fewer police would be needed for tracking illegal migrants, and the French aid administration could also be cut sizably were the aid flow to be reduced correspondingly. Remittances are the key benefit that developing countries receive from the outflows of migrants that they send to rich countries every year. Klein and Harford (2005) demonstrate that remittances are now one of the main sources of external finance for developing countries, and one that is growing steadily, with a fairly smooth time profile. Remittances have become at least as important as foreign aid for many developing countries. From this kind of calculation, it is clear that the opportunity cost of fighting immi- gration in rich countries is sizable. Anti-immigration constituencies must therefore perceive considerable detrimental effects on their countries to convince them to accept these costs. Yet quantitative research has found that the externalities perceived by these constituencies are difficult to confirm statistically. This paper attempts to discover the main determinants of the inflow of migrants into rich countries, with a view toward identifying whether there are policy tools, apart from visa control, that governments of the North can (and do) use to curb immigration. The analysis is best seen as part of a research program that seeks to dis- cover the true agenda behind foreign aid, which the so-called aid-ineffectiveness lit- erature has shown to be different from the proclaimed goal of boosting growth and fighting poverty in the recipient country. The next sections review studies on the effects and determinants of migration, and the aid-ineffectiveness debate is briefly discussed. In the subsequent sections a simple game-theoretic model is sketched, and several testable predictions are derived. An empirical analysis then shows that aid does indeed belong to the toolbox used by rich- country governments to control immigration. The final section contains conclusions. Impacts of Immigration Flows The effects of immigrant flows on host countries have been widely studied in the eco- nomics literature. Most of the research on this topic has looked at the impact on the labor market and, in particular, on the wages and employment rates of the receiving country's natives. The simplest economic model of labor market equilibrium suggests that immigration is liable to create a pecuniary externality, as an increase in the labor supply resulting from the inflow of immigrants will lead to lower wages or, in the presence of wage rigidities, to higher unemployment. In reality, this negative effect may be mitigated by adjustments in the labor market; for example, firms may move to regions where labor is becoming cheaper, thus increasing labor demand there, or T H E A I D - M I G R AT I O N T R A D E - O F F | 149 natives may move away from the regions where the migrants have arrived. Moreover, in choosing their destination, migrants take into account expected future wages and possible demand shocks not observed by the econometrician (Borjas 2003). For these reasons, measuring the effect of migrants on wages is a difficult empirical exercise, and there is no general consensus on this question. Using data from U.S. decennial censuses for 1960­90 and from the 1998­2001 current population surveys, Borjas (2003) finds that immigration has a considerable negative impact on the wages of native workers. Card (2001) shows that immigra- tion flows in the late 1980s in U.S. cities with large immigrant populations reduced the relative employment rates of low-skilled natives by up to 1 percent and their relative wages by no more than 3 percent. Friedberg and Hunt (1995) review the existing theoretical and empirical literature on the impact of immigrants on host countries' wages and growth and conclude that the impact on natives' wages is very small. Longhi, Nijkamp, and Poot (2005), on the basis of their meta-analysis of 18 empirical studies of this type, conclude that there is a robust negative and statis- tically significant, but small, impact of immigrants on natives' wages and that this impact is larger in Europe than in the United States. Other types of externality have been discussed in the context of immigration. Besides its effects on the economy, immigration also has demographic and political effects on the host countries. Given that immigrant populations are generally younger than the natives and have higher fertility rates, immigration may offer a way of decreasing the age-dependency ratio in industrial countries. Turning to the political point of view, some countries may worry that immigrants threaten their national identity and their ethnic and cultural stability. The creation by French President Nicolas Sarkozy of a Ministry of Immigration, Integration and National Identity is a response to this type of anxiety. There are also fears of infil- tration by potential terrorists or drug traffickers (Neumayer 2006). These political concerns seem to play a role at least as important as the economic impacts described above in determining decisions by immigration authorities. Neumayer (2006) shows that the poorer, the less democratic, and the more exposed to armed political conflict a country is, the more its citizens are likely to be subject to visa restrictions. The same is true for nationals of countries that were the origins of terrorist attacks. It thus seems that migrants from the poorest countries are less welcome than migrants from rich countries. Since the poorest countries are also the most important recipients of foreign aid, it is natural to ask whether foreign aid is used to reduce immigration from them. Other arguments against immigration are based on its effect on the source coun- tries. One of the most important consequences of out-migration for the country of origin is the flow of remittances received by migrants' families and friends. Remit- tances are undoubtedly an essential means of reducing poverty and insuring the pop- ulation against risks. In many developing countries, remittances are a larger and more stable source of finance than official development assistance. But remittances do not have only positive consequences. As Kapur (2004) notes, in some cases they have been an important source of funding for terrorism and civil wars. In Somalia, for example, a large proportion of the remittances supported arms purchases for rural guerrillas; in Armenia diaspora remittances boosted tough nationalist regimes 150 | J E A N - PA U L A Z A M A N D R U X A N D A B E R L I N S C H I and complicated efforts to solve regional conflicts; and the regime of the Democratic People's Republic of Korea has been strengthened through access to scarce foreign currency resources. Kapur also cites the creation of a culture of dependency among, and lower labor market participation by, the population that did not migrate, as well as the risk of a "Dutch disease" phenomenon if remittances are spent largely on nontradable goods such as housing and land.1 Some empirical findings shed light on the relative impacts of economic and other variables on public opinion in receiving countries, and on the influence of public opin- ion on immigration policies. Scheve and Slaughter (2001), using a direct measure of the U.S. population's preferences concerning immigration, obtained from the 1992 National Election Studies, show that less-skilled workers are significantly more likely to prefer limitation of immigrant inflows into the United States. Mayda (2006) employs individual-level survey datasets, as well as aggregate data on international migration, to study attitudes toward immigrants and how these attitudes influence migration policies. She finds that skilled individuals are more likely to be pro- immigration in countries where the relative skill composition of natives vis-ŕ-vis immi- grants is high. Other influences on attitudes toward migration are concern about the impact of immigration on crime rates, individual perceptions of the cultural effect of foreigners, racist feelings, and the size of inflows of asylum seekers. Mayda observes that countries with higher per capita gross domestic product (GDP) are, on average, less open to immigration, after allowing for the influence of individual-level variables. A study by O'Rourke and Sinnott (2006), using a cross-country dataset to investigate the determinants of individual attitudes toward immigration, indicates that these attitudes reflect economic interests, as well as nationalist sentiment. The authors show that among labor market participants, the highly skilled are less opposed to immigra- tion than the low skilled; this effect is larger in richer countries than in poorer ones and in countries with greater equality than in countries with more inequality. Among those who are not in the labor force, noneconomic factors are much more important than economic considerations in determining attitudes toward migration. Facchini and Mayda (2008) use a sample of 34 countries which were included in the 1995 and 2003 rounds of the International Social Survey Programme to show that voters' negative opinions toward migration explain the restrictive migration policies in place in most destination countries. They demonstrate that countries in which the median voter is more opposed to migration tend to implement more restrictive poli- cies. Thus, it seems that besides the impact of immigrants on the economy, citizens of rich countries worry about immigrants' being a threat to security, national identity, and ethnic and cultural stability. The Determinants of Immigration Flows Many arguments have been advanced explaining, without necessarily justifying, why governments of the global North want to curb immigration into their countries. The key question is whether they actually do anything about it: are there policy handles T H E A I D - M I G R AT I O N T R A D E - O F F | 151 that they use to reduce inflows? Visa quotas allow only an imperfect control of immigration flows. Many industrial countries have family reunification laws that lead to chain immigration, and many countries have signed asylum and refugee pro- tection treaties that oblige them to accept some of these uprooted people. According to data from the United Nations High Commissioner for Refugees (UNHCR), about 560,000 asylum applications were received by 28 industrial countries in 2000, and about 1 million asylum seekers were awaiting a decision. Moreover, visa quotas do not reduce the number of illegal entrants, as discussed forcefully by de Haas (2006). Hatton and Williamson (2002) note that about 300,000 illegal immigrants enter the United States every year, and 400,000 to 500,000 enter Western Europe. Illegal aliens are estimated to add 10 or 15 percent to the foreign-born stock in countries of the Organisation for Economic Co-operation and Development (OECD). Thus, the visa instrument alone is not a sufficient solution if the desired immigration level is lower than the observed one or the one expected to prevail in the future if no other meas- ure is adopted. The rich empirical literature aimed at discovering the determinants of immigration flows has produced a good crop of convergent findings. For example, Mayda (2007), using a panel of bilateral migration flows to 14 OECD countries by country of origin between 1980 and 1995, finds that income improvements in the destination countries, as well as the share of the young population in the country of origin, have positive and significant effects on emigration rates and that distance between the countries and migration quotas have negative effects. Similar results are found by Jennissen (2003), who studied the economic determinants of net immigration in Western Europe for the period 1960­98, using net migration flows as a dependent variable. He finds that destination GDP per capita, existing migrant stock, and the educational level of the population have positive effects and that unemployment rates have a negative effect on net migration flows. Hatton and Williamson (2002) present a quantitative assessment of the economic and demographic fundamentals that drive world migration across historical periods and around the world, using data on aver- age net immigration rates over five-year periods from 1970­75 to 1995­2000 for 80 countries. The authors find that the share of the population age 15­29 in the receiving country has a negative effect on its immigration rates and that the immi- grant stock has a positive effect on net immigration. They also document that a rise in domestic income relative to the world and to the region both increase a country's net immigration. Lucas (2005) examines the causes and consequences of migration from lower-income countries. Neumayer (2005), using a panel on the annual num- ber of asylum seekers in Western European countries by country of origin between 1982 and 1999, finds that human rights abuses, political violence, and state failure are important determinants of asylum migration and that democracy has a significant but nonlinear effect. Economic conditions in the countries of origin are also an important determinant of the number of asylum seekers coming to Western Europe. Neumayer suggests that generous development assistance and the opening of pro- tected European markets to imports from the sending countries could ease migration pressure--a view that de Haas (2006) criticizes forcefully. 152 | J E A N - PA U L A Z A M A N D R U X A N D A B E R L I N S C H I Surprisingly few quantitative studies exist on the link between aid and migration. From the above-cited literature, income differentials appear to be one of the main determinants of the supply of immigrants. Because foreign aid is a transfer that reduces, at least at the margin, these differentials, it is natural to ask whether foreign aid reduces immigration. Castles (2003) characterizes the migration policies pursued by the United Kingdom and the European Union as generally unsuccessful and claims that reducing North-South inequality is a key to effective migration management. Morrison (1982) argues that the most promising way for development assistance to influence migration in the short-to-medium run is through employment-generating activities and, in the long run, by reducing population growth and improving income distribution. In fact, for countries with very low incomes, foreign aid could actually increase emigration rates in the short run because of a possible migration-hump phe- nomenon. Rotte and Vogler (2000), using a panel of data on international migration to Germany from 86 countries between 1981 and 1995, study the influences of eco- nomic, demographic, and political factors on inflows to Germany and find no sig- nificant effect from aid. Berthélemy, Beuran, and Maurel (2009) estimate the two- way impact of aid and migration using cross-country data and find a positive impact of aid on migration in a simultaneous-equation system. Searching for the Hidden Agenda behind Foreign Aid The academic literature on foreign aid has been at times quite paradoxical. It caught the public's attention under the generic name of the "aid-ineffectiveness" literature; Easterly (2006) surveyed it in an influential review aimed at a broad audience. The aid-ineffectiveness literature shows quite consistently that foreign aid is not very suc- cessful at boosting growth and reducing poverty in recipient countries. The contribu- tors to this body of work end up expressing severe criticism of the international com- munity, which appears unable to pursue its proclaimed objective. The World Bank's recent slogan, "Our Dream: A World Free of Poverty," seems bound to remain just a dream. This conclusion appears to challenge the standard methodology of economics at a fundamental level. How is it that the international community has consistently spent zillions of dollars in foreign aid for nearly six decades without being "effec- tive"? Are there no error-correction mechanisms that can put an end to this "massive waste"? But the apparent paradox only concerns a small share of the academic liter- ature on aid and is the result of some hasty interpretation of the findings. The Aid-Ineffectiveness Puzzle The root cause of the turmoil is that some economists have taken at face value the declared objectives of foreign aid. The stated objective has always been to boost eco- nomic growth in the recipient country. For a long time, aid was focused on filling the "saving gap"--the allegedly insufficient national saving flow that was supposed to afflict poor countries. Collier (2007) suggests that the change in emphasis which T H E A I D - M I G R AT I O N T R A D E - O F F | 153 occurred in the 1980s and 1990s, from economic growth to poverty alleviation, was the result of a public relations campaign aimed at harnessing electoral support in favor of foreign aid from all sides of the political spectrum in rich countries. Aca- demic economists, however, soon started to blow the whistle, demonstrating that there was not much empirical support for the view that foreign aid was promoting growth in poor countries or reducing significantly the incidence of poverty. A much-cited paper by Boone (1996) triggered a wave of debate on aid ineffec- tiveness by showing that no significant impact of aid on growth could be found in cross-country regressions. In a highly influential paper, Burnside and Dollar (2000) argue forcefully that in analyzing the effect of aid, due account must be taken of het- erogeneity among recipient countries. The authors favor an index of the quality of macroeconomic policies as their heterogeneity parameter because their findings sug- gest that aid boosts growth when it is given to countries that have a sound macro- economic policy framework. Similarly, Svensson (1999) presents cross-country regression findings showing that aid is more effective in affecting growth in more democratic countries. In the same vein, Kosack and Tobin (2006) find that foreign aid and democracy have a positive impact on economic growth and human development, provided that there is a minimum level of human capital in the recipient country. Nevertheless, the dominant diagnosis is that, in general, with some noteworthy exceptions, aid is not boosting growth. Some authors blame the failure of aid on the misconceived approach of conditionality (see, for example, Collier 1997). A number of theorists propose clever schemes for fixing aid (see Svensson 2000, 2003; Azam and Laffont 2003). Another influential response has been to claim that aid has not been effective because there has not been enough of it--what was needed was a "big push" to lift people out of the "poverty trap." This view was forcefully supported by Sachs (2005); Collier (2007) espoused a more subtle variant. These findings and the response that they triggered raise a more fundamental methodological issue: do they mean that aid is ineffective, or does the true agenda of aid differ from the much-publicized goals of fostering growth and alleviating poverty? The proper methodology of economics is based on revealed-preference theory: instead of trying to assess the effectiveness of foreign aid by looking at the extent to which it achieves its stated objective, we should try to infer its true agenda from its actual achievements. When people spend zillions of dollars over decades, they must certainly have achieved a measure of success that justifies the continuation of this expenditure flow. Some economists have tried to discover the hidden agenda of for- eign aid by looking at the determinants of its allocation across countries. Their results suggest that the impact of foreign aid on growth and development is probably not the crucial determinant of its allocation. For example, Burnside and Dollar (2000) find that the quality of the macroeconomic policies pursued by a given country does not make the country more likely to receive more aid, although it does makes aid more "effective." Similarly, Svensson (1999) presents cross-country regression analy- sis showing that although aid is more effective at promoting growth in more demo- cratic countries, those countries are not more favored as aid recipients. This finding suggests that aid allocation is governed by other considerations, hinting again that there is a hidden agenda aside from the generous drive to alleviate poverty. 154 | J E A N - PA U L A Z A M A N D R U X A N D A B E R L I N S C H I Donors' Revealed Preferences The political dimension of aid allocation is further analyzed by Alesina and Dollar (2000), who find that the colonial past and strategic alliances are the main determi- nants of the amount of aid received by poor countries. They also show, however, that in the time-series dimension, democratization is often followed by increased aid, although there is no significant static effect of democracy. By contrast, Berthélemy and Tichit (2004), in a panel data analysis covering 137 aid recipients and 22 bilat- eral donors during the period 1980­99, find a significant positive impact of the Free- dom House index of civil liberty and political rights. This finding is confirmed in a later study using a different estimation method (Berthélemy 2006). The latter two studies bring out quite strongly that in allocating aid, most bilateral donors seem to be guided by their self-interest and, in particular, by their commercial relationships. Fleck and Kilby (2006a) show that commercial concerns play an important part in determining the allocation of U.S. bilateral aid across countries, particularly when the president and/or Congress are conservative. The results reported by Fleck and Kilby (2006b) suggest that the validity of such a diagnosis can be extended to the case of the World Bank, whose aid allocation behavior is significantly influenced by U.S. trade and political interests. One may wonder, however, whether trade flows are per- fectly exogenous, at least as far as bilateral donors are concerned. Although most donors have formally ruled out tied aid, toward the end of the 1960 to 1997 sample period considered by Fleck and Kilby (2006a), some implicit and subtle ways of tying aid probably continued in operation. Moreover, aid helps finance the trade deficit of developing countries, and this certainly increases imports from industrial countries, which are also the main donors. This effect is even more likely toward the end of the period of analysis, as trade liberalization was a prominent feature of the reform pro- grams supported by foreign aid under the influence of the Bretton Woods institu- tions. Hence, some reverse causation between aid and trade may be present, chan- neled by various mechanisms, so that the above findings might be misleading. Chauvet (2002) looks at the relationship between aid allocation across countries and various kinds of "sociopolitical instabilities"--referring to events that reflect political troubles in the recipient countries. She distinguishes three types: (a) elite instability, including coups d'état, revolutions, and major government crises; (b) violent instability, including political assassinations, guerrilla warfare, and civil wars; and (c) social instability, such as strikes, demonstrations, and riots. She shows that these three types of event have different effects on the allocation of aid, depending also on the kind of aid. Instabilities of types (a) and (b) have a positive impact, suggesting that aid flows are directed to governments that are under political threat, while type (c) has a negative effect, showing that aid shies away from threats directed more specifically at the economy. These results suggest that donors give aid to recipient governments in response to political motivations, with a kind of conservative bent toward providing support to incumbent governments. Economic issues such as growth and poverty alleviation seem to play only a secondary role, in that govern- ments facing greater "social instability"--the likely response of some constituencies to economic hardship--are somehow punished by getting less aid money. This whole T H E A I D - M I G R AT I O N T R A D E - O F F | 155 line of empirical research tries to infer from the determinants of aid allocation across countries what donors are really trying to achieve, but it fails to test directly for the impact of aid on the presumed objectives. Alesina and Weder (2002) use a slightly different empirical strategy, looking directly at the effect of aid on some potential objectives of the donors. They show that the level of corruption plaguing the recipient government does not significantly affect the allocation of aid across countries but that there is a significant effect in the other direction. Their results suggest that an increase in aid this year increases the level of corruption next year--what they call the "voracity effect." They thus con- clude that donors do not care at all about corruption in the recipient country. Simi- larly, Azam and Delacroix (2006) and Azam and Thelen (2008) look directly at the effects of aid on some potential objectives of the donors while taking due account of reverse causation. Using such a structural econometric approach, they show that aid is effective at fighting terrorism and that donors allocate aid across countries with a view to pursuing this objective. Our paper represents a further attempt at identifying a donor objective, by testing whether aid is actually used to reduce migration from poor countries. Implications of the Aid-Migration Trade-off A very simple model is sufficient for capturing the main issues raised by the potential trade-off between aid and migration when rich countries wish to use aid policy to reduce migration inflows. It is most likely that (assuming the aid is effective in the first place) there are some spillovers, insofar as the aid given by one donor might reduce simultaneously migration outflows from the recipient country in the direction of both the donor country and other destinations. This means that some free riding is bound to occur unless donors coordinate their actions. The model discussed next illustrates this point. The Model Assume that there are three countries in the world: two donor countries, labeled 1 and 2, whose level of affluence potentially attracts migrants, and a developing coun- try, whose flows of migrants to each donor country are designated n1 and n2. The donors have the possibility of giving aid to the poor country, with a view to reduc- ing the flow of migrants that they receive from it. Two main mechanisms can explain why aid can have a negative impact on the migration flow. First, the aid can help cre- ate an improved economic situation in the recipient country by supporting produc- tive investments and creating jobs. Second, the aid can provide an inducement to the recipient government to try to deter out-migration, if the assistance is conditional on the adoption of policy measures aimed at reducing outflows. For example, financial incentives can be created in favor of returning migrants, thus reducing the net out- flow, other things being equal, or migration-prone groups can be targeted with spe- cific actions. In Mali and Senegal, for example, the Soninke ethnic group is the most 156 | J E A N - PA U L A Z A M A N D R U X A N D A B E R L I N S C H I migration prone because of the well-established diaspora that they can rely on (see Azam and Gubert 2006). A codevelopment project has been implemented with French aid money with a view to reducing migration by members of this group by developing attractive programs in their region of origin. Let a1 and a2 denote aid flows from countries 1 and 2, respectively. Assume that the inflow of migrants in donor country 1 is governed by the following function: n1 f (a1, a2, ). (1) We assume that the impacts of the two aid flows on n1 are negative, reflecting the aid- migration trade-off that we want to analyze. The negative impact of a country's own aid flow is quite obvious, as discussed above, but the cross-effect deserves additional comment. If either aid flow has a positive effect on the level of economic activity and the creation of jobs that might reduce the attractiveness of migration for nationals from the developing country, then it cannot be assumed that this will affect only the outflow directed at each donor country separately; there is necessarily some spillover on the outflow to the other country. In the limit, it could be argued that only the total aid flow, a1 a2, matters for the outflows of migrants, if the two aid flows have the same impact on the recipient economy. The more general specification embedded in equation (1), however, allows for some finer targeting by donor countries, which might devise policies that mainly affect the migration flow heading in their own indi- vidual directions. This could be done, for example, by targeting a specific ethnic group that is connected to an important diaspora in one of the donor countries. The parameter captures the set of the other variables that are liable in either country to affect the outflow of migrants. By permuting the subscripts 1 and 2, we can readily generate the equivalent function to equation (1) for donor country 2. Assume then that country 1 is prepared to incur the cost of providing aid if the aid shows some effectiveness in reducing the migration flow in its direction. This is cap- tured by assuming that country 1 seeks to minimize the following loss function: mina1,n1 L(a1, n1, ), (2) which is increasing and convex in its first two arguments. This captures the facts that aid entails a cost for the donating country, by using up some fiscal revenues, and that, for whatever reason, that country's government tends to feel that its country is attracting too many migrants. The shift parameter captures the contextual vari- ables that are liable to affect the government's feelings about immigration, such as elections or other political determinants. A similar function is assumed to govern the choices made by country 2. Nash-Equilibrium Aid and Migration Flows If the two countries determine their aid policies without any coordination between them, the aid flows and the migration flows will be determined by the Nash equilib- rium of the game. This is the standard equilibrium concept in noncooperative game theory, which assumes that each player takes the other one's equilibrium choice as given. T H E A I D - M I G R AT I O N T R A D E - O F F | 157 FIGURE 1. Nash-Equilibrium and Optimum Aid Flows n1 N N n1 N Ń f (a1, a2 , ) C a* 1 f (a1, a* , ) 2 a1 N a1 a* 1 Source: Authors' elaboration. Figure 1 describes how country 1 determines its best-response function, a1(a2, , ), by minimizing equation (2), subject to the trade-off embedded in equation (1), while taking a2, , and as given. The convex curve represents the aid-migration trade-off (equation 1), assuming that the aid flow chosen by donor 2 is at its Nash-equilibrium value, a N. The assumed convexity of the curve captures the idea that aid has a 2 decreasing marginal impact on the inflow of migrants so that even a very high aid flow would not reduce their number to zero. Then, donor 1 will choose its best- response aid flow, a N, at the point where an indifference curve for the loss function 1 (2), represented by the concave curve, is tangent to the aid-migration trade-off. The resulting point, labeled N in figure 1, is the Nash-equilibrium joint choice of aid flow aN and migration inflow n N made by country 1, given the equilibrium aid flow aN 1 1 2 chosen by country 2. A similar diagram could obviously be drawn for country 2. Case for Coordinating Aid It is easily shown that such a Nash-equilibrium point is inefficient from the point of view of the donor countries. It means that too little aid is being donated by the two donor countries because of a free-rider problem. The spillover effects of aid on migration analyzed above are likely to dilute the incentives of each donor to extend aid in order to reduce immigration. In the Nash equilibrium, each player takes the equilibrium choice of the other as given. The two players, however, could improve on this outcome by coordinating their aid decisions in order to take the spillover effects into account. The intuition for this result can be grasped by looking at the dashed lines in figure 1. Point C represents such a coordinated equilibrium outcome, 158 | J E A N - PA U L A Z A M A N D R U X A N D A B E R L I N S C H I as can be demonstrated by the following argument. Notice that if donor 2 increases its aid flow relative to aN , the aid-migration trade-off facing donor country 1 moves 2 downward, to a position illustrated by the dashed convex curve corresponding to a* a N. This downward shift reflects the spillover effect of country 2's aid flow in 2 2 reducing the inflow of migrants into country 1, for a given level of aid donated by country 1. Then, in the coordinated equilibrium, donor 1 will reciprocate the increased aid given by donor 2 at a point such as C, where a* a N. Point C is located 1 1 on a lower indifference curve than point N--thus corresponding to a lower value of the loss function (2)--and so donor 1 is better off in this coordinated equilibrium point than in the Nash equilibrium. This occurs even though donor 1 spends more money on aid, because it receives a lower inflow of migrants in return. A similar diagram could obviously be drawn for donor country 2. Figure 1 also suggests that such a coordinated equilibrium requires a highly cred- ible ability by the players to commit irreversibly in order to overcome the temptation to renege ex post. Once player 2 has engaged a* so that the aid-migration trade-off 2 has shifted downward to the dashed line f (a1, a*, ), player 1 is tempted to reduce its 2 own contribution by moving leftward along the trade-off in order to reach an even ~ lower indifference curve of its loss function--for example, point N. Anticipating this, player 2 might then be deterred from increasing its own aid flow in the first place. This is the essence of the free-rider problem, which was popularized in the form of the well-known "prisoner's dilemma." Both donor countries need to have a credible way of tying their own hands if a coordinated outcome is to come about. As a mat- ter of fact, we can observe in the real world that the donor community is expending considerable effort to make its pledged contributions credible, using methods rang- ing from the international definition of the Millennium Development Goals to the creation of powerful aid-dependent constituencies in their own countries (perhaps by tying aid to the advantage of some powerful firms or by creating an overstaffed aid administration). Nevertheless, unless we are prepared to assume that donor countries are coordinating perfectly their aid policies regarding the reduction of migration inflows, this free-riding problem suggests that the aid flows that we observe in the real world are probably below their optimal values. The foregoing short theoretical analysis of the implications of the aid-migration trade-off rests heavily on the assumptions that such a trade-off does exist in the real world and that there are some spillovers such that the aid given by one country is likely to affect the inflow of migrants entering another country. The empirical exer- cises offered in the next section aim at testing whether these two assumptions are supported by the data. Empirical Results A quick look at figure 2 does not seem very promising for the aid-migration trade- off hypothesis. This figure traces the yearly flow of immigrants into, and the total official development assistance (ODA) disbursements by, the members of the OECD's T H E A I D - M I G R AT I O N T R A D E - O F F | 159 FIGURE 2. Disbursements of Official Development Assistance (ODA) and Entry of Immigrants, DAC Countries 14 log (entry of immigrants) 12 10 8 4 6 8 10 log (ODA disbursements) Source: OECD.Stat. Note: DAC, Development Assistance Committee of the Organisation for Economic Co-operation and Development (OECD). Development Assistance Committee (DAC) between 1995 and 2005. There appears to be a positive correlation between the number of immigrants coming to a donor country and the amount of foreign aid that the latter disburses. The following sec- tion shows that this first impression is seriously misleading. Search for a Structural Equation The positive correlation shown in figure 2 does not represent any meaningful behav- ioral relationship between aid and migration because it fails to control for many rel- evant variables. Nevertheless, Berthélemy, Beuran, and Maurel (2009) find a similar positive relationship between the two in an equation that controls for numerous vari- ables. They cautiously explain it by referring to "policy coherence," arguing that donor countries are actively combining their aid and migration policies. Our results outlined below suggest instead that these authors face a specification problem. The model presented above implies that aid disbursements and migration inflows are jointly determined in equilibrium, as Berthélemy, Beuran, and Maurel suggest, but with different predictions. By changing the other determinants of the migration inflow that we have captured by the parameter , we can generate some comparative-static predictions that are com- patible with figure 2. Imagine that such an exogenous change shifts the aid-migration 160 | J E A N - PA U L A Z A M A N D R U X A N D A B E R L I N S C H I trade-off upward. Then it is most likely that the equilibrium points N or C will move to the northeast, that is, upward and to the right, indicating that both the aid disbursed and the migration inflow increase simultaneously. The reason for such shifts is that if more migrants are forthcoming for a given aid flow, the donor country will respond both by increasing its aid flows somewhat and by allowing a bit more migration because the marginal impact of aid on migration is decreasing. The latter effect entails an increase in the marginal cost of reducing immigration through increased foreign aid. This mental experiment suggests two things that a correct empirical analysis should take into account: (a) it is crucial to include the correct control variables in the migra- tion equation in order to identify correctly the aid-migration trade-off, and (b) the aid flow itself is probably endogenous, and this endogeneity should also be controlled for. Because it is well known that most econometric methods for controlling for endogene- ity entail a potential loss of efficiency, we first present the results without taking this problem into account. We then test whether this estimation procedure gives rise to a significant endogeneity bias in a second stage. This two-step approach allows us to perform two tests of interest with one equation--that is, to test (a) whether aid has a significant negative impact on the inflow of migrants, and (b) whether donors are actively using foreign aid as a policy response aimed at reducing the migration inflows that they face. Because of obvious availability problems, we are working with data on the num- ber of legal migrants when what we are really interested in is the total number of migrants. The latter is what our aid-migration trade-off is likely to govern. But because we are applying panel-data techniques, using country fixed effects, we can hope to learn a lot about total migrants from our econometric analysis. The following argument explains why. Assume that the number of legal migrants is a random fraction of their total num- ber, which reflects, among other things, the immigration-restriction policy enforced by the destination country. Then, because we are working with the logarithm of the number of legal migrants, the mean value of that random fraction feeds into the country fixed effect, while the deviations relative to that mean are feeding into the residuals. Formally, if nL is the number of legal migrants and n is the total number of migrants, we can assume that: nL (G, )n, (3) where G is a set of variables that captures the immigration-restriction policy stance of the government and 0 (G, ) 1 is the random share of legal migrants, depending on the exogenous shock variable . Then, by taking the logarithm of equation (3), we get Log nL Log (G, ) Log n. (4) Therefore, unless the restriction policy pursued by each government has changed drastically over our relatively short sample period, which spans 1995­2003, it should be well controlled for by the country fixed effects. This assumes that, with respect to immigration-restriction policy, there is more variation across countries than within T H E A I D - M I G R AT I O N T R A D E - O F F | 161 TABLE 1. Regression Results on Flows of Legal Migrants from Low-Income and Lower-Middle-Income Countries Variable (1) (2) (3) (4) Unemployment rate 0.30*** 0.18*** 0.30*** 0.14** (0.09) (0.08) (0.09) (0.08) Social expenditures (percent of GDP) 0.32*** 0.30*** 0.32*** 0.34*** (0.09) (0.09) (0.09) (0.10) Log of per capita GDP 0.54 9.53*** 0.54 14.80*** (1.42) (2.82) (1.42) (4.97) Log of stock of foreign population 0.19 0.57 0.17 0.91* (0.57) (0.50) (0.59) (0.47) Log of official development assistance 0.46 3.68*** 0.43 5.15*** (ODA) disbursements (0.32) (1.15) (0.28) (1.66) Log of multilateral disbursements -- -- 0.10 1.43 (0.27) (0.90) Endogeneity bias, ODA -- 4.47*** -- 5.90*** (1.26) (1.73) Endogeneity bias, multilateral -- -- -- 1.50 disbursements (0.96) Number of observations 118 117 118 116 F-test 9.50 9.84 7.87 7.64 Source: Authors' calculations. Note: GDP, gross domestic product. The dependent variable is the log of the inflow of migrants from low-income and lower-middle-income countries. All explanatory variables are characteristics of the destination country, and they are lagged once. The data on flow of migrants, social expenditures, stock of foreign population, ODA disbursements, and multilateral aid disbursements are taken from OECD.Stat. The data on GDP per capita and unemployment rates are taken from the World Development Indicators. The sample consists of the 22 donor countries of the Development Assistance Committee (DAC) of the Organisation for Economic Co-operation and Development (OECD) for 1995­2003, with the exceptions of New Zealand, for which the stock of foreign population is missing, and Australia and Canada, for which this information is available for only one point in time. Country fixed effects have been used but are not reported, to save space. Numbers in parentheses are robust standard errors. The instruments used for ODA and multilateral disbursements are the log of public expenditures on order and security, from OECD.Stat, and the percentage of right-wing members in parliament in the destination country, from "Parties and Elections," http://www.parties-and-elections.de, and Université de Sherbrooke, "World Perspective," http://per- spective.usherbrooke.ca/bilan/BMEncyclopedie/BMEncycloListePays.jsp. The log of government revenues (from OECD.Stat) is added as an instrument for equation (4). Their reduced-form impacts on the aid variables are presented in annex table A.1. *Significant at the 10 percent level. **Significant at the 5 percent level. ***Significant at the 1 percent level. each country over time. Then our equations explaining the logarithm of the number of legal migrants should, in fact, tell us a great deal about the total number of migrants entering each country. Table 1 presents the results of four regression equations that explain the inflow of legal migrants from low-income and lower-middle-income countries into donor countries. This restriction is meant to capture the idea that donor countries are not 162 | J E A N - PA U L A Z A M A N D R U X A N D A B E R L I N S C H I viewing the inflows of migrants from other rich countries in the same way as they do inflows from poorer countries. We also performed all the estimations with the total inflows, yielding mostly similar but sometimes significantly different results. In par- ticular, the income tax rate is significant for migrants from rich countries, but it does not matter for migrants from poorer countries. In columns (1) and (3) of table 1, no attempt is made to control for endogeneity; that is done in columns (2) and (4). The method used for performing this control is based on the standard Hausman test and is further discussed below. All the explana- tory variables are lagged once. This procedure potentially helps mitigate any remain- ing endogeneity problems, in particular those that might affect the various control variables, and it also provides some information about any potential time lag in the response of migration flows to changes in incentives. Four control variables are included, of which three consistently prove highly sig- nificant. The unemployment rate is highly significant, reflecting the deterrent effect of a depressed labor market in the host country. When the probability of finding a job is low in the destination country, migrants seem to postpone their travel or even to cancel it. The social expenditures policy pursued by the target country is an impor- tant attraction factor, and it is significant in all the columns. Industrial countries that spend more on social items such as health and education are obviously more attrac- tive to migrants than countries with a more conservative policy stance. Then, we find a strange result for per capita GDP in columns (1) and (3), where it seems to have a negligible impact on the inflow of migrants. This counterintuitive result suggests the presence of an estimation problem. Fortunately, this effect is not robust to the cor- rection of the endogeneity bias affecting the impact of aid, as this coefficient becomes positive and significant in columns (2) and (4). It thus seems that the impacts of the business cycle and national income on immigration are fully captured by the unem- ployment rate and GDP per capita. Finally, the existing stock of foreign population already residing in the country of destination is only significant at the 10 percent level in equation (4). This variable is meant to reflect the well-known network effects that play a key role in the migration process in many studies. For example, Azam and Gubert (2006) demonstrate that such an effect can explain an ethnic bias in migration. They show that two ethnic groups living in the same region of the Senegal River valley, in western Mali, and thus facing the same economic conditions, have very different migration patterns. The authors use historical evidence to document that the group with a long history of out- migration is sending a much higher fraction of its population abroad than the group without such a migration history. The established diaspora from the first group serves as a bridgehead that reduces the costs of migration for the prospective new migrants from that same group by helping them find jobs and accommodations and by pro- viding the informal credit and insurance services that migrants' networks are known for delivering to their members. Our findings reported above suggest that such an effect is not very strong at the country level, once the impact of foreign aid is taken into account. The stock of foreign population in the destination country is probably too coarse a measure to capture this effect. T H E A I D - M I G R AT I O N T R A D E - O F F | 163 Testing for the Impact of Aid The test variables capture the aid disbursed by the donor countries. To gain some insight into the relative effectiveness of various aid flows, we use the log of ODA disbursements by the destination country and test whether multilateral aid dis- bursements have a differential impact by introducing that variable in addition to ODA. Arguably, the multilateral aid variable captures a much better coordinated aid policy than ODA, which includes considerable bilateral aid. This interpretation reflects the idea that the principal donors have a say in the way the World Bank and other multilateral donor agencies determine multilateral aid disbursements and that they also have a clear opportunity to coordinate their decisions regarding these dis- bursements at board meetings or in the corridors. The model presented above sug- gests that coordinated aid flows could have some multiplier effects, as they imply a quid pro quo by other donors. It turns out that ODA disbursements represent the most significant aid variable coming out of our regressions (2) and (4), with the predicted negative sign. This is consistent with our theoretical framework, which suggests that foreign aid is probably an effective tool for reducing the inflow of migrants into rich countries. The model, however, also raises the question whether coordinated aid is a more powerful tool against immigration than uncoordinated aid. General ODA, which includes both bilateral aid and the contributions chan- neled through the Bretton Woods entities and other multilateral institutions, could be less effective than multilateral aid taken separately. Our findings, however, do not support the differential impact hypothesis. This suggests that, in fact, donor countries somehow manage to coordinate their bilateral and multilateral aid flows equally well. The technique applied in columns (2) and (4) for controlling for endogeneity is derived from the standard Hausman test. Two auxiliary reduced-form equations are estimated for log of ODA and log of multilateral aid, which are assumed endogenous in the theoretical framework presented above, using the log of public expenditures on order and security, the percentage of right-wing members in parliament, and the log of government revenues (in equation 4 only) as instruments, in addition to the four exogenous variables in table 1. These variables are regarded as the contextual variables captured by in equation (2). The first-stage reduced-form equations are presented in annex table A.1. The residuals from these equations are then included, in addition to the aid variables themselves, in columns (2) and (4), and their estimated coefficients provide estimates of the endogeneity biases for each variable. Inclusion of these residuals in the equations provides an additional benefit, as it corrects the esti- mated coefficients of the aid variables themselves for the endogeneity bias that affects them in the uncontrolled equation. This justifies the discussion presented above of the estimated coefficients of these aid variables. For the tests performed at columns (2) and (4), the residuals are obviously lagged. The corresponding estimates of the endogeneity biases are presented in table 1 under that entry. Moreover, this procedure yields the correct estimates for the coefficients of the variables themselves, as mentioned above. In column (2), where ODA only is 164 | J E A N - PA U L A Z A M A N D R U X A N D A B E R L I N S C H I included, we find that the latter is strongly significant, and the exogeneity assumption is rejected at the 1 percent level. The correct interpretation of this test as an endo- geneity test is not immediate. The specification of the immigration function in columns (1) through (4) assumes that the inflow of migrants responds to incentives with a one- year lag. The endogeneity test performed in columns (2) and (4) assumes that the government in the donor country adjusts its aid flow in year t 1 on the basis of its forecast that a component of the random shock will affect the immigration flow in year t; this forecast is presumed to be based on some information that is not available to the econometrician. For example, the government may be using a lead indicator based on the number of visa applications in year t 1 that will only show up as actu- al migration in the subsequent year, and this is a piece of information that we have not been able to include in our estimated equations. Similarly, the government of the host country might be aware of sociological or institutional changes affecting a resident dias- pora that are likely to affect the latter's ability to attract new migrants, but this is something the econometrician does not know. The reduced-form equation for aid reflects in its residuals this anticipation by the government in year t 1. The latter is then necessarily correlated with the random shock occurring in year t, by construc- tion, if our behavioral assumption correctly captures the way the donor government is forming its expectations. The two aid variables are included in column (4). Only ODA is found to be sig- nificant and endogenous. Hence, ODA might not be worse than multilateral aid after all, suggesting that donors have found various methods for obtaining the required coordination for their other aid flows. After all, many of them have been in this line of business for about six decades, so that the aid game might safely be approximated by an infinite-horizon repeated game. It is known that this kind of setting is likely to foster cooperation between the players. Our findings thus suggest that donor countries are doing a good job of equalizing the marginal impact of each kind of aid flow, so that aggregation of flows into a single ODA variable is legitimate for econometric purposes. In the foregoing econometric exercise, two key results seem robust: (a) foreign aid has a significant negative impact on migration inflows into donor countries, and (b) donors are actively using aid as a policy tool for reducing immigration. The third result that we tested, concerning the effectiveness of aid flows coordinated through multilateral institutions in relation to that of other aid flows, leads us to reject the view that bilateral aid is less effective. Conclusion This paper has investigated the assumption that donor countries employ foreign aid partly as a tool for controlling inflows of migrants. A brief theoretical analysis was used to bring out the main predictions that can be derived from such an assumption. T H E A I D - M I G R AT I O N T R A D E - O F F | 165 The model helped us identify the potential free-rider problem raised by the assumption and suggested that donors must find a coordinated equilibrium if they want to opti- mize the impact of their aid as a means of reducing immigration. Such a coordinated equilibrium requires donors to find a way of tying their own hands in order to make their commitment not to renege ex post on their pledged disbursements credible. We suggested that in the real world, donors are, in fact, using various mechanisms to cre- ate this credible commitment, ranging from the international definition of the Millen- nium Development Goals to the formation of powerful aid-dependent constituencies in their own countries. Techniques for developing the latter in the real world include, among other methods, the tying of aid to benefit powerful firms and the creation of an overstaffed aid administration. In addition, donors have created international aid institutions, such as the World Bank, whose job is precisely to coordinate at least some of the aid flows. The empirical tests performed using a panel of data from most DAC member countries show that our assumed aid-migration trade-off is indeed supported by the data. The empirical approach used to produce these findings is based on two require- ments brought out by the theoretical analysis. First, it is important to include in the estimations various control variables, which are likely to affect both immigration flows and aid disbursement flows. Second, due account must be taken of the fact that governments choose jointly the level of foreign aid that they deliver and the in- migration that they permit, so that the former must be regarded as endogenous in the econometric analysis. The findings of our econometric exercises provide robust sup- port for these two predictions. We further tested whether the amount of aid disbursed through a coordination mechanism, which we have proxied by multilateral aid dis- bursements, is any more effective than the other aid flows, here captured by ODA disbursements. Our results suggest that total ODA is not performing any worse than its multilateral aid component, but this might reflect econometric problems, as these two variables are strongly correlated with one another. Our econometric exercises fail to support the view that there is a significant free-rider problem with bilateral aid flows and hence that there is significant underprovision of aid. Nevertheless, our tests of this assumption do not seem very powerful, and further investigation of this issue is warranted. In particular, a finer disaggregation of aid flows might be required to perform a convincing analysis of the free-rider problem. This points the way for future research. Note 1. A large inflow of foreign currency drives up a country's exchange rate, which handicaps the sale of other exports and impairs the ability of domestic products to compete with imports. "Dutch disease" takes its name from the supposed effects of North Sea natural gas discoveries on the Netherlands economy. 166 | J E A N - PA U L A Z A M A N D R U X A N D A B E R L I N S C H I ANNEX TABLE A.1. First-Stage Reduced-Form Equations Log of ODA Log of ODA disbursements disbursements Log of multilateral for eq. (2) for eq. (4) disbursements Log of public expenditure on 0.21*** 0.21*** 0.09 order and security (0.07) (0.08) (0.11) Percentage of right-wing 0.003 0.002 0.01*** members in parliament (0.002) (0.002) (0.003) Log of government revenues 0.06 0.33* (0.14) (0.19) Unemployment rate 0.03 0.03 0.01 (0.02) (0.02) (0.02) Social expenditures 0.009 0.01 0.02 (percent of GDP) (0.013) (0.01) (0.02) Log of per capita GDP 1.87*** 1.95*** 1.10*** (0.36) (0.38) (0.42) Log of stock of foreign 0.04 0.05 0.23** population (0.08) (0.09) (0.59) N 159 156 156 F-test 48.59 39.86 17.31 Source: Authors' calculations. Note: GDP, gross domestic product; ODA, official development assistance. *Significant at the 10 percent level. **Significant at the 5 percent level. ***Significant at the 1 percent level. References Alesina, Alberto, and David Dollar. 2000. "Who Gives Aid to Whom and Why?" Journal of Economic Growth 5 (March): 33­63. Alesina, Alberto, and Beatrice Weder. 2002. "Do Corrupt Governments Receive Less Foreign Aid?" American Economic Review: Papers and Proceedings 92 (4): 1126­37. Azam, Jean-Paul, and Alexandra Delacroix. 2006. "Aid and the Delegated Fight against Terrorism." Review of Development Economics 10 (2): 330­44. Azam, Jean-Paul, and Flore Gubert. 2006. "Migrants' Remittances and the Household in Africa: A Review of Evidence." Journal of African Economies 15 (Suppl. 2): 426­62. Azam, Jean-Paul, and Jean-Jacques Laffont. 2003. "Contracting for Aid." Journal of Devel- opment Economics 70 (1): 25­58. Azam, Jean-Paul, and Véronique Thelen. 2008. "The Roles of Foreign Aid and Education in the War on Terror." Public Choice 135 (3­4): 375­97. Berthélemy, Jean-Claude. 2006. "Bilateral Donors' Interest vs. Recipients' Development Motives in Aid Allocation: Do All Donors Behave the Same?" Review of Development Economics 10 (2, May): 179­94. T H E A I D - M I G R AT I O N T R A D E - O F F | 167 Berthélemy, Jean-Claude, and Ariane Tichit. 2004. "Bilateral Donors' Aid Decisions--A Three-Dimensional Panel Analysis." International Review of Economics and Finance 13 (3): 253­74. Berthélemy, Jean-Claude, Monica Beuran, and Mathilde Maurel. 2009. "Aid and Migration: Substitutes or Complements?" World Development (in press). http://dx.doi.org/ 10.1016/j.worlddev.2009.02.002. Boone, Peter. 1996. "Politics and the Effectiveness of Foreign Aid." European Economic Review 40: 289­329. Borjas, George J. 1995. "The Economic Benefits from Immigration." Journal of Economic Perspectives 9 (2): 3­22. ------. 2003. "The Labor Demand Curve Is Downward Sloping: Reexamining the Impact of Immigration on the Labor Market." Quarterly Journal of Economics 118 (4): 1335­74. Burnside, Craig, and David Dollar. 2000. "Aid, Policies, and Growth." American Economic Review 90 (4): 847­68. Card, David. 2001. "Immigrant Inflows, Native Outflows, and the Local Labor Market Impacts of Higher Immigration." Journal of Labor Economics 19 (1): 22­64. Castles, Stephen. 2003. "Why Migration Policies Fail." Ethnic and Racial Studies 27 (2): 205­27. Chauvet, Lisa. 2002. "Socio-Political Instability and the Allocation of International Aid by Donors." European Journal of Political Economy 19: 33­59. Collier, Paul. 1997. "The Failure of Conditionality." In Perspectives on Aid and Development, ed. Catherine Gwinn and Joan M. Nelson. Policy Essay 22. Washington, DC: Overseas Development Council. ------. 2007. The Bottom Billion: Why the Poorest Countries Are Failing and What Can Be Done about It. Oxford: Oxford University Press. de Haas, Hein. 2006. "Turning the Tide? Why `Development Instead of Migration' Policies Are Bound to Fail." IMI Working Paper 2, International Migration Institute, Oxford. Easterly, William. 2006. The White Man's Burden. Why the West's Efforts to Aid the Rest Have Done So Much Ill and So Little Good. New York: Penguin Press. Facchini, Giovanni, and Anna Maria Mayda. 2008. "From Individual Attitudes towards Migrants to Migration Policy Outcomes: Theory and Evidence." Economic Policy 23 (56): 651­713. Fleck, Robert K., and Christopher Kilby. 2006a. "How Do Political Changes Influence US Bilateral Aid Allocation? Evidence from Panel Data." Review of Development Economics 10 (2): 210­23. ------. 2006b. "World Bank Independence: A Model and Statistical Analysis of US Influence." Review of Development Economics 10 (2): 224­40. Friedberg, Rachel M., and Jennifer Hunt. 1995. "The Impact of Immigrants on Host Country Wages, Employment and Growth." Journal of Economic Perspectives 9 (2): 23­44. Gubert, Flore. 2003. "Ces immigrés qui font vivre le Mali." Libération 10 (February 19). Hatton, Timothy J., and Jeffrey G. Williamson. 2002. "What Fundamentals Drive World Migration?" NBER Working Paper 9159, National Bureau of Economic Research, Cambridge, MA. Jennissen, Roel. 2003. "Economic Determinants of Net International Migration in Western Europe." European Journal of Population 19 (2, June): 171­98. Kapur, Devesh. 2004. "Remittances: The New Development Mantra?" G-24 Discussion Paper Series 29, United Nations Conference on Trade and Development (UNCTAD), New York. 168 | J E A N - PA U L A Z A M A N D R U X A N D A B E R L I N S C H I Klein, Michael, and Tim Harford. 2005. The Market for Aid. Washington, DC: International Finance Corporation. Kosack, Stephen, and Jennifer Tobin. 2006. "Funding Self-Sustaining Development: The Role of Aid, FDI and Government in Economic Success." International Organization 60 (1): 205­43. Longhi, Simonetta, Peter Nijkamp, and Jacques Poot. 2005. "A Meta Analytic Assessment of the Effect of Immigration on Wages." Journal of Economic Surveys 19 (3): 451­77. Lucas, Robert E. B. 2005. International Migration and Economic Development: Lessons from Low-Income Countries. Cheltenham, UK: Edward Elgar Publishing. Mayda, Anna Maria. 2006. "Who Is against Immigration? A Cross-Country Investigation of Individual Attitudes toward Immigrants." Review of Economics and Statistics 88 (3): 510­30. ------. 2007. "International Migration: A Panel Data Analysis of the Determinants of Bilateral Flows." CEPR Discussion Paper 6289, Centre for Economic Policy Research, Oxford. Morrison, Thomas K. 1982. "The Relationship of U.S. Aid, Trade and Investment to Migra- tion Pressures in Major Sending Countries." International Migration Review 16 (1): 4­26. Neumayer, Eric. 2005. "Bogus Refugees? The Determinants of Asylum Migration to Western Europe." International Studies Quarterly 49 (3): 389­410. ------. 2006. "Unequal Access to Foreign Spaces: How States Use Visa Restrictions to Regulate Mobility in a Globalized World." Transactions of the Institute of British Geographers 31 (1): 72­84. O'Rourke, K. H., and Richard Sinnott. 2006. "The Determinants of Individual Attitudes towards Immigration." European Journal of Political Economy 22 (4): 838­61. Ortega, Javier. 2000. "Pareto-Improving Immigration in an Economy with Equilibrium Unemployment." Economic Journal 110: 92­112. Rotte, Ralph, and Michael Vogler. 2000. "The Effects of Development on Migration: Theoretical Issues and New Empirical Evidence." Journal of Population Economics 13 (3): 485­508. Sachs, Jeffrey D. 2005. The End of Poverty: How We Can Make It Happen in Our Lifetime. London: Penguin Books. Scheve, Kenneth F., and Matthew J. Slaughter. 2001. "Labor Market Competition and Indi- vidual Preferences over Immigration Policy." Review of Economics and Statistics 83 (1): 133­45. Svensson, Jakob. 1999. "Aid, Growth and Democracy." Economics and Politics 11 (3): 275­97. ------. 2000. "When Is Foreign Aid Policy Credible? Aid Dependence and Conditionality." Journal of Development Economics 61 (1): 61­84. ------. 2003. "Why Conditional Aid Does Not Work and What Can Be Done about It?" Journal of Development Economics 70 (2): 381­402. Comment on "The Aid-Migration Trade-off," by Jean-Paul Azam and Ruxanda Berlinschi MELVIN D. AYOGU Jean-Paul Azam and Ruxanda Berlinschi have highlighted an inconvenient but important concern about foreign aid to developing countries. Starting from the assumption that aid is driven by the purely altruistic motive of improving the lot of the less fortunate, the literature on the impact of aid to needy countries suggests that such assistance has been largely ineffective--yet this does not seem to have dampened the flow of aid over time. An awkward question arises: if aid is known to be ineffec- tive but continues to be delivered, either the donors are irrational, or there are other, not yet revealed, reasons for giving. Azam and Berlinschi, on the grounds of the revealed-preference principle and empirical data, argue that part of the hidden aid agenda is to reduce the number of immigrants entering donor countries from devel- oping countries. Although the dataset covers legal migration only and does not explicitly classify immigrants by income level, nonetheless the discussion suggests that the target groups to be discouraged are the poor and the unskilled, desperate for work. The paper thus can be viewed as part of the strand of aid literature that seeks to rescue the "aid-ineffectiveness" literature from its current embarrassment. The Aid-Migration Theses The authors set out to test three propositions. The first is that donor countries use foreign aid partly to control immigration from recipient countries. This is the hidden agenda hypothesis. By controlling for the influence of drivers of aid flows other than immigration and for drivers of immigration other than aid flows, the authors attempt to isolate the effect of aid flows on migration flows. The second test is motivated by the insight afforded by the theoretical framework elaborated in the paper. According to this model, although donor countries that Melvin D. Ayogu is professor of economics and dean, Faculty of Commerce, University of Cape Town, South Africa. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank. 169 170 | M E LV I N D . AY O G U extend bilateral aid to poor countries secretly nurture the desire to stem migration flows from the beneficiaries of their financial assistance, their actions generate an inferior outcome for themselves. The paper suggests that donors should better coor- dinate their efforts in order to achieve what they really desire. But it would be polit- ically incorrect for donors to collude openly to curb the flow of migrants from the very countries that they seek to assist, particularly given that migration is a mecha- nism for poverty reduction, and the active espousal of such an agenda in a consor- tium would certainly be incompatible with the rhetoric of globalization. As a result of these implied discomforts, rich countries seek other, more subtle ways to achieve coordination at a supranational level. The third proposition concerns the relative effectiveness of the various multilateral institutions as coordinating mechanisms for the secret desires of donor countries. Empirical Evidence The paper finds that aid matters significantly in curbing migration flows. The clever construction of the test for this proposition enables the authors to implicitly identify donors' true objectives. Thus, by showing the presence of reverse causation between flows of assistance and of migrants, they establish that the level of immigration affects "giving." In addition, the paper finds empirical confirmation for the second proposition: that coordinated aid is relatively more effective in curbing immigration. On the third proposition, the evidence is inconclusive, and there is as yet no established theory that marshals the relative strengths and challenges of competing alternative multilateral institutions and helps predict in rank order those that are most effective as coordinating mechanisms. (Easterly and Pfutze 2008 recently exploited the ques- tion about the ranking of aid agencies, using a set of criteria viewed as crucial for efficient aid delivery.) Conclusion The theme of this year's conference is rightly about people and politics, which, pre- sumably, is what globalization is all about. Years ago, in 2000, Dani Rodrik asked whether trade can be global while politics remains local. Jurisdictional boundaries localize economic activities and define borders that then define trade as either national or international. Underpinning Rodrik's insightful analysis are concerns about likely future trends in the range of immigration policies toward people from poor countries and people with different religious and ideological persuasions. U.S. immigration policy of the recent past was an issue among Latino voters in the 2008 U.S. presi- dential election and, in fact, became one of the few issues on which candidates Barack Obama and John McCain were in agreement. In Europe concern is voiced about how immigration policy affects the career prospects of foreign soccer players in European leagues and what would be an appropriate policy stance on this issue. France, South Africa, and the United Kingdom have all experienced their versions of xenophobia in COMMENT ON AZAM AND BERLINSCHI | 171 recent memory. In the light of the times in which we live, the issues raised by Azam and Berlinschi cannot be ignored. Their results suggest the need for more studies at the micro level, with a focus on institutions. The authors do seem to have given us an example of aid that works. If foreign aid can generate a desired policy outcome when the agenda is hidden, why not when the objective is open? How can foreign aid be effective in furthering the hidden agenda (revealed preference) but ineffective in promoting declared objectives? And, how much can we rely on evidence from eight years of data in a history of compassionate or not-so-compassionate giving that spans more than three decades? Perhaps the real message here is that transparency in the aid business is overdue. Easterly and Pfutze (2008) emphasized the same message. Transparency can engen- der the correct attribution or decomposition of aid into its various goals (hidden and covert), opening a prospect for transforming the entire aid-ineffectiveness puzzle into an understandable picture. Finally, I have to acknowledge with a great deal of satisfaction that, as a by-product, Azam's and Berlinschi's paper provides support for the push by recipient countries for better donor coordination with a view to improving absorptive capacity (see, for instance, Birdsall, Williamson, and Deese 2002; de Renzio 2005; Ayogu 2006). References Ayogu, Melvin. 2006. "Can Africa Absorb More Aid?" In Aid, Debt Relief and Development in Africa: African Development Report 2006, ed. African Development Bank, 25­40. New York: Oxford University Press. Birdsall, Nancy, John Williamson, and Brian Deese. 2002. Delivering on Debt Relief: From IMF Gold to a New Aid Architecture. Washington, DC: Center for Global Development and Institute for International Economics, http://www.iie.com. de Renzio, Paolo. 2005. "Scaling Up versus Absorptive Capacity: Challenges and Opportuni- ties for Reaching the MDGs in Africa." ODI Briefing Paper, London. Easterly, William, and Tobias Pfutze. 2008. "Where Does the Money Go? Best and Worst Practices in Foreign Aid." Journal of Economic Perspectives 22 (2): 29­52. Rodrik, Dani. 2000. "How Far Will International Economic Integration Go?" Journal of Economic Perspectives 14 (1): 177­86. Are Remittances More Effective Than Aid for Improving Child Health? An Empirical Assessment Using Inter- and Intracountry Data LISA CHAUVET, FLORE GUBERT, AND SANDRINE MESPLÉ-SOMPS This paper analyzes the respective impacts of aid and remittances on human development as measured by infant and child mortality rates. Panel data on a set of 109 developing countries and cross-country quintile-level data on a sample of 47 developing countries are alternatively used. In addition to assessing the extent to which health aid and remittances contribute to reducing child health disparities between countries, the paper addresses two other questions: What is the net effect of migration, after accounting for the brain drain of health workers? What is the effec- tive impact of aid and remittances on intracountry child health disparities? Our results tend to show that remittances significantly improve child health and that the impact of health aid is nonlinear, suggesting that aid to the health sector is more effec- tive in the poorest countries. By contrast, medical brain drain, as measured by the expatriation rate of physicians, is found to have a harmful impact on health out- comes. The net impact of migration on human development is therefore weakened. Finally, remittances seem to be much more effective in improving health outcomes for children belonging to the richest households, whereas neither pro-poor nor antipoor effects are found for health aid. Poverty reduction is increasingly put forward as the main objective of official devel- opment assistance (ODA) to developing countries. National leaders and the interna- tional community have pledged to meet by 2015 a series of poverty reduction targets known as the Millennium Development Goals (MDGs).1 The pursuit of these goals calls for dramatic increases in infrastructure finance and in the provision of basic services to the population of the developing world that ODA alone cannot achieve. The Monterrey consensus, which emerged from the United Nations International Conference on Financing for Development in that city in 2002, highlighted the need Lisa Chauvet, Flore Gubert, and Sandrine Mesplé-Somps are researchers with the Institute of Research for Development (IRD), DIAL, Paris. Flore Gubert is also associate professor at the Paris School of Economics. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank. 173 174 | L I S A C H A U V E T, F L O R E G U B E R T, A N D S A N D R I N E M E S P L É - S O M P S to find new sources of financing, and the idea that more private funds should be invested in developing countries has received strong support since then. Given this context and an ever-increasing volume of flows from migrants, inter- national migrant remittances have attracted considerable attention in recent years. According to the latest World Bank estimates (see Ratha et al. 2007), recorded remittances to developing countries reached US$240 billion in 2007. The actual magnitude is even larger when transfers through informal channels are taken into account. In 36 out of 153 developing countries, remittances are larger than all cap- ital flows, public and private, and voices have already been raised here and there to call for progressive replacement of official aid by remittances. Little is known, however, about the respective effectiveness of aid and remit- tances in alleviating poverty. Despite a burgeoning literature examining the impact of ODA on aggregate welfare, there exists, to our knowledge, almost no studies analyzing to what extent aid and remittances may be substitutes or how they are related to inequality and poverty reduction. Exceptions include the work of Chauvet and Mesplé-Somps (2007), who analyze the distributive impact of trade flows, foreign direct investment (FDI), official aid, and migrants' remittances using Branco Milanovic's World Income Distribution database (Milanovic 2005). The authors find that FDI increases intracountry disparities and that remittances tend to decrease them. They also find that trade and aid have a nonlinear relationship with income distribution. The objective of our paper is to fill this knowledge gap by analyzing the respective impacts of aid and remittances on human development as measured by infant and child mortality rates. To what extent do aid and remittances help reduce child health disparities between countries? What are their respective impacts on child health dis- parities within countries? How do remittances compare with aid when migration costs (in the form of "brain drain") are accounted for? We choose basic indicators of human welfare instead of a monetary measure of poverty for three reasons. First, comparable cross-country data on monetary pover- ty over time are extremely scarce. Second, child health figures prominently among the MDGs. Donors have committed themselves to reducing by two-thirds the mortality rate among children under age 5 (goal 4), and to this end, they have devoted an increasing share of official aid to the health sector. There is, however, very little empirical evidence on the effect of increased aid flows on health outcomes in recipi- ent countries. Whether donors are right to prioritize the health sector in the intra- country allocation of aid is thus an unanswered question that needs to be addressed. Third, the relationship between migration and health is increasingly emphasized in the microeconomic literature, and donor agencies regularly report the success of most of their projects and programs in the health sector. It is therefore interesting to inves- tigate whether successful health interventions from the donors' side or the migrants' side at the micro level translate into improved health outcomes at the macro level and whether Paul Mosley's micro-macro paradox also applies to the health sector (Mosley 1987). We follow Mishra and Newhouse (2007) and use aid allocated to the health sec- tor, instead of aggregate aid, in our empirical analyses. Our implicit assumption is A R E R E M I T TA N C E S M O R E E F F E C T I V E T H A N A I D ? | 175 that not all types of aid can reasonably be expected to affect health outcomes and that narrowing the aid variable should help us better measure the impact (if any) of official development assistance on basic indicators of human development. Our empirical strategy relies on two econometric exercises. We first examine the respective impacts of aid, remittances, and medical brain drain on child health indicators, using panel data on a sample of 109 developing countries. We explore whether aid and remittances contribute to improving health outcomes and whether the brain drain of health workers vitiates the positive impact of remittances. This first econometric exercise raises substantial methodological issues such as measure- ment errors and endogeneity of our core explanatory variables, which we try to address. Keeping in mind the inherent weaknesses of this macro approach, our econometric results indicate that both remittances and health aid significantly reduce infant and child mortality rates but that the effect of health aid is nonlinear, suggesting that aid to the health sector is likely to be more effective in the poorest countries. Medical brain drain, as measured by the expatriation rate of physicians, is found to have a harmful impact on health outcomes. The net impact of migration on human development is thus diminished. We then assess the respective effectiveness of aid and remittances in lessening health disparities within countries, using cross-country quintile-level data. The results of this second econometric exercise show that remittances are effective in reducing infant and child mortality rates, but only for the richest quintiles. This finding suggests that although remittances contribute to better mean health out- comes in recipient countries, they tend to increase intracountry health inequality. The impact of health aid, by contrast, is hardly ever significant in our within-country regressions. The next section contains a review of the macroeconomic literature on the impacts of aid and remittances on poverty and inequality. The data, method of estimation, and results of the cross-country and intracountry analyses are presented in the sub- sequent two sections, followed by concluding remarks and a discussion of the policy implications of the findings. Effects of Aid and Remittances on Poverty and Inequality: A Review of the Literature From the early 1960s to the mid-1990s, the literature investigating the macroeco- nomic impact of aid focused on the link between aid and growth. The emerging pic- ture from this literature is that aid can enhance growth but that this result is very fragile and is highly dependent on the choice of data, sample composition, and esti- mation methods (Roodman 2007). The adoption of the United Nations Millennium Declaration in 2000 and the obligation accepted by donors to financially support developing countries' efforts to achieve the eight Millennium Development Goals by 2015 have progressively shifted the focus from the aid-growth nexus to the relation- ship between aid flows and welfare or poverty indicators. This relationship is exam- ined next, after which the effects of remittances are explored. 176 | L I S A C H A U V E T, F L O R E G U B E R T, A N D S A N D R I N E M E S P L É - S O M P S Poverty and Inequality Impacts of Aid The few existing studies on the impacts of aid on poverty and inequality have adopted standard cross-country growth regression approaches, replacing growth with an indicator of welfare or poverty as the dependent variable (see, for example, Boone 1996; Mosley, Hudson, and Verschoor 2004; Gomanee, Girma, and Morrissey 2005; Gomanee et al. 2005). Because comparable cross-country data on poverty over time are extremely scarce, most studies have concentrated on the effectiveness of aid in improving human development indicators such as the infant mortality rate, the under-five mortality rate, life expectancy, and primary schooling. In a famous paper, Boone (1996) finds no evidence that aid succeeds in improving human development indicators in recipient countries. Although aid could theoretically reduce infant mortality either through an increase in private consumption or through greater provision of public services to the poor, his results suggest that it increases the size of recipient governments but has no impact on basic measures of human develop- ment indicators. Pushing his analysis further, Boone investigates whether his result varies depending on the political regime. He finds some evidence that liberal political regimes, all else being equal, have lower infant mortality rates, which may reflect greater willingness of these systems to deliver public services to the poor. Boone's paper has been much criticized, on two grounds. In the first place, some authors have argued that the welfare impact of aid is not direct but operates through its effect on the amount of government spending allocated to social areas. Boone's regressions would thus be inappropriately specified. Mosley, Hudson, and Verschoor (2004) estimate a system of three equations, with poverty, aid, and pro-poor expen- diture as their dependent variables. They find that aid is associated with higher levels of pro-poor spending and that such spending is associated with lower poverty head- counts. Aid is also found to increase health spending, which in turn reduces infant mortality. Gomanee, Girma, and Morrissey (2005) reach the same type of conclusion, using quantile regressions. In a companion paper, however, Gomanee et al. (2005) find evidence that aid improves welfare indicators and that this effect works pre- dominantly through direct impacts. The second criticism is that although it may be true that aggregate aid has no impact on health, particular types of aid, including health aid, are effective in improving human development indicators (see, for example, Masud and Yontcheva 2005; Michaelowa and Weber 2007; Mishra and Newhouse 2007; Dreher, Nunnenkamp, and Thiele 2008). Mishra and Newhouse (2007), in particular, rely on a large dataset covering 118 countries between 1973 and 2004 to measure the effect of health aid on infant mortality. They estimate both ordinary least squares (OLS) regressions and a sys- tem of moment equations using the generalized method of moments (GMM) and find that increased health aid is associated with a statistically significant reduction in infant mortality. The estimated effect of health aid is small, however; since doubling health aid within a country would reduce infant mortality in the next five-year period by only 2 percent. In addition, the authors do not find any significant impact of overall aid. Masud and Yontcheva (2005) use data on assistance from nongovernmental organizations (NGOs) and on bilateral aid to assess the effectiveness of these financial A R E R E M I T TA N C E S M O R E E F F E C T I V E T H A N A I D ? | 177 flows on two social indicators, infant mortality and adult illiteracy. Their underly- ing assumption is that NGOs intervene at the grassroots level and may be more effective in alleviating poverty than other types of assistance. Using an unbalanced panel of 58 countries from 1990 to 2001, they find that health expenditure per capita reduces infant mortality, as does greater NGO aid per capita. By contrast, they do not find any significant impact of total bilateral aid on infant mortality. The authors then list a number of reasons why NGO aid might work better than bilat- eral aid in reducing infant mortality. First, NGO aid would be allocated more toward countries with high infant mortality rates, while bilateral aid would favor countries with lower infant mortality. Second, NGOs would have more direct links to the poor and vulnerable, which would make them more efficient. Third, in line with Boone (1996), aid transiting through recipient governments could be diverted for the benefit of wealthy elites. Pushing their analysis further, the authors find no evidence of a positive impact of NGO or bilateral aid on the share of spending on health care in total expenditure. The few existing studies examining the links between aid and aggregate welfare as measured by human development indicators do not permit clear conclusions. Some papers find no impact at all; others find evidence that aid decreases infant mortality rates, directly, or indirectly, through higher levels of pro-poor spending. This lack of consensus in the macroeconomic literature is surprising, given the number of suc- cessful health interventions financially supported by international assistance (Levine and the What Works Working Group 2004). Poverty and Inequality Impacts of Remittances Despite the increasing size of remittances, empirical macroeconomic evidence on the impacts of these financial flows on poverty and inequality is even scarcer than that related to aid. Here again, the scarcity of evidence stems mainly from the lack of reli- able and comparable cross-country data on several of the relevant variables, such as emigration rate by country and amounts remitted, and from the absence of the long series that are required if use is to be made of the latest macroeconometric tools. Con- sequently, the empirical literature is confined largely to a few case studies of villages or countries based on microeconometric data (see, for example, Leliveld 1997; Lachaud 1999; Adams 2004, 2006). At the cross-national level, to our knowledge, only four recent studies have looked at the poverty impact of remittances: Adams and Page (2005); World Economic Out- look (IMF 2005); Gupta, Pattillo, and Wagh (2007); and Acosta et al. (2008). Despite strong microeconomic evidence for a positive impact of remittances on education and health (see, for example, Kanaiaupuni and Donato 1999; Cox-Edwards and Ureta 2003; Hildebrandt and McKenzie 2005; Mansuri 2007), no one has ever investigated the impact of remittances on human development indicators at a macroeconomic level. Adams and Page (2005) use a panel of 71 low-income and middle-income coun- tries for which data on migration, remittances, poverty, and inequality are available and test whether countries that produce more international migration or receive 178 | L I S A C H A U V E T, F L O R E G U B E R T, A N D S A N D R I N E M E S P L É - S O M P S more international remittances have less poverty. After instrumenting for the potential endogeneity of remittances, they find that a 10 percent increase in per capita official remittances leads to a 3.5 percent decline in the share of people living in poverty. Using a broader sample of 101 countries, IMF (2005) provides further evidence that remittances have an effect on poverty. The effect, however, is rather small; on average, a 2.5 percentage point increase in the ratio of remittances to gross domestic product (GDP) is associated with a less than 0.5 percentage point decrease in poverty. As argued by the authors, this (disappointing) result could stem from the fact that average income and inequality, along with remittances, are included as regressors. Since these variables are themselves likely to be influ- enced by remittances, the true impact of remittances on poverty could actually be larger. Using a sample of 76 countries in which Sub-Saharan Africa is substantially rep- resented, Gupta, Pattillo, and Wagh (2007) adopt the same methodology as that of Adams and Page (2005) and model poverty as a function of mean income, some measures of income distribution, and remittances. Their findings indicate that a 10 percent rise in remittances is associated with a decrease of about 1 percent in the incidence of poverty. In the case of Sub-Saharan Africa, however, their results suggest that the impact of poverty on migration and remittances is greater than the impact of remittances on poverty. Finally, Acosta et al. (2008) use both cross-country and household survey data to assess the impact of remittances on growth, poverty, and inequality in Latin America. Their cross-country estimates suggest that remittances have a positive and statistically significant effect on growth, on average, but that they tend to increase the level of income inequality. For the average Latin American country, however, the effect is dif- ferent; an increase in remittances tends to be associated with lower levels of inequal- ity. Turning to the authors' microeconometric analyses, their findings suggest that the effects of remittances on poverty and inequality vary strongly across Latin American countries, depending on whether recipients are concentrated at the bottom or at the top of the distribution of nonremittance income. In what follows, our aim is to provide additional insights into the question of whether aid and remittances, as sources of external financing, are effective in improv- ing child health outcomes. Impact of Aid and Remittances on Health Outcomes: A Cross-Country Analysis This section assesses the impact of health aid and remittances on child health out- comes, using panel data on a sample of 109 countries from 1987 to 2004. (For a list of countries in the sample, see annex table A.1.) After a brief presentation of the empirical strategy and a description of the data, the results of our baseline model are discussed. The analysis is then pushed further by testing for nonlinearities in the aid-health relationship and investigating the effect of the medical brain drain on health outcomes. A R E R E M I T TA N C E S M O R E E F F E C T I V E T H A N A I D ? | 179 Model and Data To explore the relative impacts of aid and migration on child health indicators, we estimate a model of the following form: lnHealthi,t i t ln Xi,(t 1,t 4) lnRemittancesi,(t 1,t 4) ln Health aidi,(t 1,t 4) i,t . (1) Healthi,t is either the under-five mortality rate or the infant mortality rate from World Development Indicators (World Bank 2006). The under-five mortality rate is the probability (per 1,000 live births in a given year) that a newborn baby will die before reaching age 5, if subject to current age-specific mortality rates. The infant mortality rate is the number of infants dying before reaching age 1, per 1,000 live births. Since our dependent variables are bounded, we use a logarithmic transformation. To ease interpretation of our results and account for potential nonlinearities, all our independent variables are also log-linearized. We measure Health aidi,t using aid commitments to the health sector as defined by the Country Reporting System (CRS) of the Organisation for Economic Co-operation and Development (OECD).2 The main advantage of the CRS is that its data on aid commitments are highly disaggregated by purpose (sector). Its main disadvantage is that the data are only reliable for recent years--as reflected in figure 1 by the large underreporting of aid in the CRS data compared with data from the Development FIGURE 1. Total Aid, Remittances, and Health Aid, 1986­2004 1.40E+11 1.00E+10 9.00E+09 health aid, constant 2000 U.S. dollars 1.20E+11 8.00E+09 constant 2000 U.S. dollars total aid and remittances, 1.00E+11 7.00E+09 6.00E+09 8.00E+10 5.00E+09 6.00E+10 4.00E+09 4.00E+10 3.00E+09 2.00E+09 2.00E+10 1.00E+09 0.00E+00 0.00E+00 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 total aid commitments, DAC total aid commitments, CRS total remittances health aid commitments, CRS Source: Development Assistance Committee (DAC) of the Organisation for Economic Co-operation and Development (OECD); Country Reporting System (CRS), OECD; World Development Indicators 2006 (World Bank 2006). 180 | L I S A C H A U V E T, F L O R E G U B E R T, A N D S A N D R I N E M E S P L É - S O M P S Assistance Committee (DAC) of the OECD. As noted by Mishra and Newhouse (2007), the extent of underreporting varies by sector, donor, and time period. Missing data are therefore omitted from the sample rather than treated as zero health aid. We restrict our sample to relatively recent observations, starting in the mid-1980s.3 Figure 1 shows that the share of aid commitments to the health sector has gradually increased since the 1980s, when it was about 2 percent of total commitments; it is now about 5 per- cent. This increase in health aid reflects the switch in donors' priorities, notably, from aid for infrastructure to aid for social sectors, which reflects the adoption of the Millennium Development Goals. The CRS also provides a disaggregation of disbursements by sector. Unfortunately, disbursements are even more underreported than commitments. As an alternative vari- able for Health aidi,t, we proxy aid disbursements in the health sector by weighting total net disbursements with the share of health commitments in total commitments. Health aidi,t is expressed in per capita constant terms, using the DAC deflator. The term Remittancesi,t is defined as current transfers by migrants who are employed for more than a year in another country in which they are considered residents (World Bank 2006). We use the same deflator for remittances as for aid in order to transform this variable into per capita constant terms. As shown in figure 1, workers' remittances have been increasing in both absolute and relative terms since the mid-1980s; then, they accounted for only about 60 percent of total aid commitments, but since 2000 they have represented more than 200 percent of aid commitments, reflecting the growing importance of this financing for developing economies. The growth in remittances is partly attributable to the rise in the number of international migrants worldwide, but it also indicates that in recent years people have been shifting from informal to formal channels for sending funds. This potentially important source of measurement error is addressed in our econometric analysis. First, we include a time trend in our list of regressors in order to capture the increasing trend in remittances. Second, we control for unobservable heterogeneity among countries, hoping to account for some omitted variables that could explain simultaneously the increasing trend in remittances and the decreasing trend in child and infant mortality. Finally, this latter issue is also tack- led through the instrumentation of remittances (see the next subsection). Following the existing literature on cross-country determinants of child health out- comes, equation (1) controls for a set of relevant socioeconomic variables, Xi,t. Begin- ning with the work of Ravallion (1993) and Pritchett and Summers (1996), a consen- sus has emerged concerning the negative relationship between child mortality and national income. Female education, measured either by educational attainment or by illiteracy rates, has also been shown to be negatively correlated with child mortality (Filmer and Pritchett 1999; Anand and Bärnighausen 2004; Fay et al. 2005; McGuire 2006; Ravallion 2007). We express income per capita in purchasing power parity (PPP) constant terms (World Bank 2006) and measure female education by average years of schooling of the female population age 15 and older (Barro and Lee 2000). Anand and Bärnighausen (2004) show that the density of human resources in the health sector is significantly correlated with child health indicators. We proxy human resources for health with the number of physicians (per 1,000 inhabitants), from Docquier and Bhargava (2007). A R E R E M I T TA N C E S M O R E E F F E C T I V E T H A N A I D ? | 181 Other cross-country determinants of child health have been identified in the liter- ature, such as the size of the population (Mishra and Newhouse 2007), the share of urban population (Fay et al. 2005; Masud and Yontcheva 2005; Ravallion 2007), inequality indicators (Filmer and Pritchett 1999; Fay et al. 2005; McGuire 2006; Ravallion 2007), and poverty rates (Anand and Bärnighausen 2004), but none were significant in our analysis. Two other variables--ethnic fragmentation and whether the country is predominantly Muslim--were also significantly correlated with infant mortality in Filmer and Pritchett (1999) and McGuire (2006). Both are time invari- ant and could not be introduced in our fixed-effects analysis. Finally, there has been an intense debate concerning the impact of public spending on health outcomes. Because our core independent variable is health aid and the impact of health aid goes through the route of public spending, we exclude the pub- lic spending variable from our analysis. Another reason is that when public spending is introduced into the regressions, we lose half of the countries in the sample. Equation (1) is estimated on a panel of 109 developing countries, among them 39 Sub-Saharan countries, from 1987 to 2004. (See annex table A.1 for the country list.) Child health data are for every four or five years (1990, 1995, 2000, and 2004). The right-hand-side variables are averaged over three years, from t ­ 1 to t ­ 4, and are measured in logarithms. This is true for all variables except education because the Barro and Lee (2000) database on education is for every five years and is available only up to 2000. We therefore use the 2000 level of education to explain 2004 health outcomes, and so on. We control for unobservable heterogeneity with country fixed effects, . We also include time dummies t. Endogeneity of Aid, Remittances, and Income There are two potential sources of endogeneity of aid and remittances to child health indicators. First, aid and remittances are given purposively, and both donors and migrants are likely to take into account the child health situation when allocating their flows. Even if aid is determined at the macro level and remittances are determined at the micro level, both are likely to reflect, to some extent, the chances of survival of children. Second, there could be some omitted variables that affect aid, remittances, and child health. For example, natural disasters are likely to induce both a deteriora- tion of child health indicators and increased inflows of aid and remittances. We therefore instrument health aid and workers' remittances.4 As instruments for health aid, we use a set of variables that capture historical and cultural relationships between developing countries and donor or destination countries. These variables are more likely to be exogenous to child health than any characteristics of recipient or origin countries. Specifically, we use the total aid budget of the five main donors weighted by the cultural distance between receiving and donor or destination coun- tries (measured by whether they have the same religion) and by the geographic dis- tance (distance to Washington, Brussels, and Tokyo).5 As an instrument for health aid, we use health aid lagged twice. Workers' remittances are instrumented using the ratio of broad money supply (M2) to GDP because countries that are more financially 182 | L I S A C H A U V E T, F L O R E G U B E R T, A N D S A N D R I N E M E S P L É - S O M P S developed have been found to receive larger remittances. Income per capita is also endogenous to health indicators (Pritchett and Summers 1996; Filmer and Pritchett 1999). It is instrumented using twice-lagged income per capita. We also suspected education to be endogenous to health indicators. We tested this hypothesis, and it turned out that the exogeneity of education could not be rejected by our test. This result is partly explained by the fact that education in t 5 (or t 4) is used to explain health outcomes in t. The excludability and relevance of our instruments being legitimate concerns here, tests for their validity (Sargan test of overidentification, test of underidentification, test of weak instruments, partial R-squared) were systematically performed.6 Estimation of the Baseline Model Our estimation of the baseline model proceeds in three steps. Equation (1) is first estimated with simple OLS. We then introduce country fixed effects to take into account unobservable heterogeneity in our sample. Finally aid, remittances, and income are instrumented using two-stage least squares (2SLS), including country fixed effects and time fixed effects.7 Instrumentation equations are provided in annex table A.2. Regressions (1) through (6) in table 1 present the estimations of the baseline model when the dependent variable is either the under-five mortality rate or the infant mor- tality rate. Income per capita is highly significant and tends to reduce child mortality. The impact is quite strong: the coefficients of income in regressions (3) and (6) suggest that a 1 percent increase in income reduces child mortality by around 0.59 percent and infant mortality by about 0.50 percent. The coefficients of income per capita are interestingly close to the coefficients found by Pritchett and Summers (1996) in their instrumental variables (IV) estimation of infant mortality (around 0.3), using a different set of instruments. They are even closer for the fixed-effect estimations (0.31 in Pritchett and Summers 1996). Surprisingly, the number of physicians is not significant in table 1 except in OLS esti- mations. When significant, it is negative, suggesting that a larger number of doctors implies lower child and infant mortality rates. Anand and Bärnighausen (2004) find a strong impact of doctor and nurse density on various health indicators, which in their case is more robust than in our regressions. Only in OLS estimations does female edu- cation have a significant impact on child and infant mortality rates. In table 1 the neg- ative impact of the time dummies (1990 is the omitted time dummy) reflects the decreasing trend in child and infant mortality rates over the last two decades. Finally, aid and remittances both have a negative coefficient in regressions (1) through (6), but, contrary to Mishra and Newhouse (2007), we find no significant impact of health aid at this stage of our empirical analysis. By contrast, remittances are found to be strongly significant in most regressions, with the expected sign. When instrumented, the coefficient of remittances is multiplied more than fourfold: a 1 per- cent increase in remittances decreases child mortality by 0.12 percent and infant mortality by 0.10 percent. A R E R E M I T TA N C E S M O R E E F F E C T I V E T H A N A I D ? | 183 TABLE 1. Impact of Health Aid and Remittances on Child and Infant Mortality Rates, Baseline Model Child mortality rate Infant mortality rate OLS Within 2SLS OLS Within 2SLS Variable (1) (2) (3) (4) (5) (6) GDP per capitaa 0.553 0.263 0.595 0.482 0.218 0.500 (6.64)*** (3.03)*** (2.39)** (6.49)*** (2.78)*** (2.26)** Number of physicians 0.157 0.032 0.048 0.107 0.022 0.050 per 1,000 inhabitants (3.93)*** (0.82) (0.89) (2.88)*** (0.62) (1.04) Female educational 0.156 0.034 0.009 0.151 0.049 0.004 attainment (1.97)* (0.61) (0.09) (2.15)** (0.95) (0.05) Dummy for missing 0.181 0.008 0.217 0.170 0.038 0.168 education variable (1.82)* (0.15) (1.28) (2.08)** (0.81) (1.08) Remittances per capitaa 0.054 0.031 0.122 0.045 0.023 0.104 (2.37)** (2.37)** (2.97)*** (2.16)** (2.09)** (2.76)*** Health aid per capitaa 0.012 0.012 0.008 0.009 0.011 0.000 (0.60) (1.26) (0.31) (0.44) (1.29) (0.01) Year = 1995 0.058 0.102 0.068 0.048 0.094 0.081 (1.74)* (4.07)*** (1.29) (1.64) (4.16)*** (1.74)* Year = 2000 0.088 0.198 0.032 0.085 0.189 0.037 (2.14)** (6.92)*** (1.27) (2.27)** (7.34)*** (1.68)* Year = 2004 0.139 0.274 0.139 0.265 (2.94)*** (7.98)*** (3.24)*** (8.68)*** Constant 8.704 6.360 7.872 5.669 (13.88)*** (9.30)*** (13.93)*** (9.19)*** Fixed effects No Yes Yes No Yes Yes Number of observations 358 358 237 358 358 237 Number of countries 109 109 86 109 86 R2 0.75 0.57 0.72 0.57 Sargan (p-value) 0.31 0.27 Underidentification test 0.03 0.03 (p-value) Income instrumentation 0.000 0.000 F-statistic (p-value) Aid instrumentation 0.100 0.100 F-statistic (p-value) Remittance instrumentation 0.000 0.000 F-statistic (p-value) Note: 2SLS, two-stage least squares; GDP, gross domestic product; OLS, ordinary least squares. Numbers in parenthe- ses are robust t-statistics. All variables except the education variable are averages over three-year periods, from t 1 to t 4, measured in logs. In equations (1) and (4), standard errors are clustered by country. a. Instrumented regressors in equations (3) and (5). Instruments include twice-lagged GDP per capita; twice-lagged aid; and instruments for aid and remittances in the tradition of Tavares (2003), that is, total aid budgets of the five largest donors (the United States, Japan, France, the United Kingdom, and Germany) in constant dollars, weighted by a cultural distance variable (same religion) and a geographic distance variable. The ratio of broad money supply (M2) to GDP is also included as an instrument for remittances. Tests for excludability of the instruments are available on request. * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. 184 | L I S A C H A U V E T, F L O R E G U B E R T, A N D S A N D R I N E M E S P L É - S O M P S Nonlinearities in the Aid-Health Relationship As a next step in our analysis, we explore in greater detail the relationship between health aid and child health indicators. So far, we find no significant impact of health aid commitments. A relative consensus, however, has emerged in the literature: that aggregate aid disbursements affect macroeconomic outcomes such as economic growth in a nonlinear way.8 Similarly, the impact of health aid on health outcomes may be nonlinear. The nonlinearity may be attributable to constrained absorptive capacity. Constrained absorptive capacity in the health sector may be proxied through an interaction of health aid with income per capita; health aid would be relatively more effective in richer countries because of their greater capacity to absorb aid. To explore this kind of nonlinearity in the health-aid relationship, we estimate an equation of the following form: lnHealthi,t i t ln Xi,(t 1,t 4) ln Remittancesi,(t 1,t 4) 1 ln Health aidi,(t 1,t 4) 2 ln Health aidi,(t 1,t 4). ln Incomei,(t 1,t 4) i,t , (2) where lnHealth aid.lnIncome is an interaction variable of aid with income per capita. It is instrumented using the same set of instruments as those for health aid and income per capita. The results are presented in columns (1) and (2) of table 2. The absorptive capacity hypothesis is not supported by our results. The impact of health aid is nonlinear, but the nonlinearity suggests that aid to the health sector is more effective in poorer coun- tries. The threshold in income per capita corresponding to a switch to harmful aid is around US$4,100 per capita (in PPP). Figure 2 depicts the effect of aid on child mor- tality rates below and above this income threshold, respectively. The threshold is quite high and implies that most African countries belong to the decreasing part of the rela- tionship between health aid and health outcomes. Aid increases the child mortality rate in 8 of the 35 Sub-Saharan African countries in our sample: Botswana, Cape Verde, Gabon, Mauritius, Namibia, the Seychelles, South Africa, and Swaziland. In the remaining 27 African countries, aid tends to improve child health indicators. It is worth noting that our baseline specification implies that we capture the direct effects of aid and remittances on child health indicators. Another channel through which aid and remittances could affect health outcomes is their impact on GDP per capita. Since GDP per capita is included among our set of regressors, this indirect impact is not taken into account. Assuming that both remittances and aid tend to improve income, we therefore probably underestimate the impact of these sources of financing on child health indicators. An alternative way of testing the constrained absorptive capacity hypothesis is to introduce the square of health aid into the regression. A quadratic relationship between health aid and health indicators would reflect marginal decreasing returns to aid: after a given threshold of aid received, an additional dollar of aid is less effec- tive because the country no longer has the capacity to absorb it. Aid squared is never significant when introduced into any of the regressions.9 Moreover, its sign is negative, as is that of health aid. The absence of a quadratic relationship between health aid and health outcomes confirms our previous finding of no absorptive capacity constraints A R E R E M I T TA N C E S M O R E E F F E C T I V E T H A N A I D ? | 185 TABLE 2. Nonlinearity in the Health-Aid Relationship, Two-Stage Least Squares (2SLS) with Fixed Effects Child mortality rate Infant mortality rate (1) (2) GDP per capitaa 0.355 0.264 (1.41) (1.16) Number of physicians per 1,000 inhabitants 0.023 0.026 (0.40) (0.49) Female educational attainment 0.063 0.050 (0.53) (0.45) Dummy for missing education variable 0.305 0.255 (1.39) (1.26) Remittances per capitaa 0.115 0.097 (2.72)*** (2.46)** Health aid per capitaa 0.839 0.815 (1.87)* (1.99)** Health aid per capita income per capitaa 0.100 0.098 (1.85)* (1.98)** Year = 1995 0.071 0.084 (1.24) (1.62) Year = 2000 0.044 0.049 (1.69)* (2.09)** Fixed effects Yes Yes Number of observations 237 237 Number of countries 86 86 Sargan (p-value) 0.31 0.25 Underidentification test (p-value) 0.04 0.04 Income instrumentation F-statistic (p-value) 0.00 0.00 Aid instrumentation F-statistic (p-value) 0.10/0.12 0.10/0.12 Remittances instrumentation F-statistic (p-value) 0.00 0.00 Note: GDP, gross domestic product. Numbers in parentheses are robust t-statistics. All variables except the educa- tion variable are averages over three-year periods, from t 1 to t 4, measured in logs. a. Instrumented regressors. Instruments include twice-lagged GDP per capita; twice-lagged aid; and instruments for aid and remittances in the tradition of Tavares (2003), that is, total aid budgets of the five largest donors (the United States, Japan, France, the United Kingdom, and Germany) in constant dollars, weighted by a cultural distance vari- able (same religion) and a geographic distance variable. The ratio of broad money supply (M2) to GDP is also includ- ed as an instrument for remittances. Tests for excludability of the instruments are available on request. * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. on aid to the health sector. Health aid seems to be more effective where the prospects for improvements in health indicators are higher, that is, in poorer countries. Finally, we explore whether the results using health aid disbursements are similar to those with aid commitments.10 Regressions (1) and (3) of table 3 reproduce the baseline model; health aid disbursements are not significantly different from zero. In regressions (2) and (4), health disbursements interacted with income are significant, 186 | L I S A C H A U V E T, F L O R E G U B E R T, A N D S A N D R I N E M E S P L É - S O M P S FIGURE 2. Impact of Health Aid on Child Mortality Income per capita (PPP) less than US$4,100 4.9 4.8 4.7 log (child mortality) 4.6 4.5 4.4 4.3 4.2 4.1 4.0 ­4 ­3 ­2 ­1 0 1 2 3 4 log (health aid per capita) Income per capita (PPP) greater than US$4,100 4.4 4.2 4.0 log (child mortality) 3.8 3.6 3.4 3.2 3.0 ­8 ­6 ­4 ­2 0 2 4 log (health aid per capita) Source: Authors' calculations. confirming our previous finding. The resulting income threshold, US$4,000, is close to the one corresponding to regressions that include commitments. Medical Brain Drain Our baseline model suggests that migrants' remittances help improve health out- comes in developing countries. We now turn to analysis of the counterpart of workers' remittances--the impact of the brain drain induced by migration on health outcomes A R E R E M I T TA N C E S M O R E E F F E C T I V E T H A N A I D ? | 187 TABLE 3. Impact of Health Aid Disbursements on Health Indicators, Two-Stage Least Squares (2SLS) with Fixed Effects Child mortality rate Infant mortality rate (1) (2) (3) (4) GDP per capitaa 0.619 0.264 0.516 0.176 (2.47)** (1.00) (2.33)** (0.73) Number of physicians per 1,000 inhabitants 0.044 0.039 0.046 0.041 (0.77) (0.61) (0.90) (0.70) Female educational attainment 0.020 0.134 0.005 0.114 (0.19) (1.02) (0.05) (0.94) Dummy for missing education variable 0.248 0.445 0.194 0.383 (1.36) (1.80)* (1.16) (1.68)* Remittances per capitaa 0.124 0.123 0.106 0.104 (2.95)*** (2.70)*** (2.74)*** (2.46)** Health aid disbursements per capitaa 0.028 0.983 0.019 0.937 (0.96) (2.25)** (0.73) (2.28)** Health aid disbursements per capita 0.118 0.113 income per capitaa (2.16)** (2.20)** Year = 1995 0.061 0.075 0.075 0.088 (1.20) (1.38) (1.64) (1.79)* Year = 2000 0.029 0.047 0.035 0.051 (1.17) (1.79)* (1.55) (2.16)** Fixed effects Yes Yes Yes Yes Number of observations 233 233 233 233 Number of countries 86 86 86 86 Sargan (p-value) 0.29 0.47 0.26 0.36 Underidentification test (p-value) 0.03 0.03 0.03 0.03 Income instrumentation F-statistic (p-value) 0.00 0.00 0.00 0.00 Aid instrumentation F-statistic (p-value) 0.08 0.08/0.11 0.08/0.11 0.08/0.11 Remittances instrumentation F-statistic (p-value) 0.00 0.00 0.00 0.00 Note: GDP, gross domestic product. Numbers in parentheses are robust t-statistics. All variables except the educa- tion variable are averages over three-year periods, from t 1 to t 4, measured in logs. a. Instrumented regressors. Instruments include twice-lagged GDP per capita; twice-lagged aid; and instruments for aid and remittances in the tradition of Tavares (2003), that is, total aid budgets of the five largest donors (the United States, Japan, France, the United Kingdom, and Germany) in constant dollars, weighted by a cultural distance variable (same religion) and a geographic distance variable. The ratio of broad money supply (M2) to GDP is also included as an instrument for remittances. Tests for excludability of the instruments are available on request. * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. in developing countries. More specifically, we explore the impact of the medical brain drain. Docquier and Bhargava (2007) provide a rich dataset containing information on the expatriation rate of physicians.11 We introduce this latter variable into our model and estimate an equation of the following form: 188 | L I S A C H A U V E T, F L O R E G U B E R T, A N D S A N D R I N E M E S P L É - S O M P S lnHealthi,t i t ln Xi,(t 1,t 4) lnRemittancesi,(t 1,t 4) MedicalBrainDraini,(t 1,t 4) l lnHealth aidi,(t 1,t 4) 2 ln Health aidi,(t 1,t 4). lnIncomei,(t 1,t 4) i,t , (3) where MedicalBrainDraini,(t 1,t 4) is the expatriation rate of physicians averaged over a three-year subperiod and transformed in logarithms. Health outcomes and medical brain drain may be correlated with omitted variables such as the quality of health infrastructure. We therefore instrument this variable using the same set of instruments as for aid and remittances. Regressions (1) and (2) of table 4 present the results when medical brain drain is introduced into the analysis. The coefficient of medical brain drain is highly signifi- cant and has the expected positive sign: a 1 percent increase in the rate of expatria- tion of physicians increases child and infant mortality rates by around 0.5 percent. The expatriation of human resources in the health sector has a direct, harmful effect on health outcomes in developing countries.12 Interestingly, the medical brain drain does not really affect the impact of health aid on health outcomes. The threshold of income for which the relationship between aid and child health switches from negative to positive remains similar to that found in table 2, between US$4,700 and US$5,000, and the slope does not change greatly: from 0.815 it goes to about 1, suggesting that the health-improving impact of aid is not altered when the medical brain drain is taken into account. Intracountry Empirical Assessment In this section, we investigate the intracountry impact of aid and remittances on child health indicators by analyzing to what extent these transfers are targeted to the poor- est (or are not). The discussion begins with a description of the data and the empiri- cal strategy and ends with comments on our main findings. Model and Data We use the World Bank's comprehensive Health, Nutrition, and Population (HNP) database, in which development indicators from Demographic and Health Surveys (DHSs) are compiled by asset quintiles within countries (Gwatkin et al. 2007). Asset quintiles are computed using the first principal component in an analysis of the cor- relations between various consumer durables and other household characteristics, following a method proposed by Filmer and Pritchett (2001). Few studies have used the HNP database to analyze the determinants of child health outcomes. To our knowledge, the first is Fay et al. (2005). Using a sample of 39 countries and a country random-effect model, the authors assert that apart from traditional variables--such as GDP per capita, assets, education, and direct health interventions--better access to basic infrastructure services has an important impact on infant and child mortality and on the incidence of stunting. Ravallion (2007) questions the robustness of their results and criticizes their empirical strategy on three points. First, the model of Fay et al. (2005) is a linear model, but a logarithmic functional form would have been more appropriate, given that the dependent A R E R E M I T TA N C E S M O R E E F F E C T I V E T H A N A I D ? | 189 TABLE 4. Medical Brain Drain and Health Outcomes, Two-Stage Least Squares (2SLS) with Fixed Effects Child mortality rate Infant mortality rate (1) (2) GDP per capitaa 0.486 0.389 (1.77)* (1.56) Number of physicians per 1,000 inhabitants 0.379 0.365 (2.50)** (2.67)*** Female educational attainment 0.205 0.185 (0.99) (0.98) Dummy for missing education variable 0.536 0.476 (1.49) (1.45) Remittances per capitaa 0.134 (0.114 (2.54)** (2.35)** Health aid per capitaa 1.067 1.033 (2.02)** (2.14)** Health aid per capita income per capitaa 0.125 0.122 (1.99)** (2.12)** Medical brain drain (MBD)a 0.504 0.481 (2.61)*** (2.75)*** Year = 1995 0.093 0.105 (1.33) (1.64) Year = 2000 0.086 0.089 (2.39)** (2.75)*** Fixed effects Yes Yes Number of observations 237 237 Number of countries 86 86 Sargan (p-value) 0.99 0.99 Underidentification test (p-value) 0.07 0.07 Income instrumentation F-statistic (p-value) 0.00 0.00 Aid instrumentation F-statistic (p-value) 0.10/0.12 0.10/0.12 Remittances instrumentation F-statistic (p-value) 0.00 0.00 MBD instrumentation F-statistic (p-value) 0.01 0.01 Note: GDP, gross domestic product. Numbers in parentheses are robust t-statistics. All variables except the educa- tion variable are averages over three-year periods, from t 1 to t 4, measured in logs. a. Instrumented regressors. Instruments include twice-lagged GDP per capita; twice-lagged aid; and instruments for aid and remittances in the tradition of Tavares (2003), that is, total aid budgets of the five largest donors (the United States, Japan, France, the United Kingdom, and Germany) in constant dollars, weighted by a cultural dis- tance variable (same religion) and a geographic distance variable. The ratio of broad money supply (M2) to GDP is also included as an instrument for remittances. Tests for excludability of the instruments are available on request. * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. 190 | L I S A C H A U V E T, F L O R E G U B E R T, A N D S A N D R I N E M E S P L É - S O M P S variables are bounded. Second, by estimating a random-effect model, the authors implicitly assume that their country fixed effects are not correlated with the regres- sors. This is a strong assumption because many sources of latent heterogeneity across countries are suspected. Finally, there may be a strong presumption of bias arising from the omission of within-country differences in women's schooling. Using exact- ly the same data but estimating a fixed-effect model that includes female education and variables in log-linear form, Ravallion (2007) fails to detect any significant impact of infrastructure access on child health outcomes. His findings suggest a sig- nificant effect of access to health care and female educational attainment on child health.13 The study by Fielding, McGillivray, and Torres (2008) employs the same data. It examines, using a system of simultaneous equations, the relationships between four MDG-related variables (health, educational status, access to water, and access to sanitation) and aid; the authors also explore the impact of aid on these vari- ables. They find that although aid is effective overall, the poorest subgroups within each country are typically not the primary beneficiaries of the inflows. In what follows, we use an updated HNP database in which some countries have multiple-year data. (See annex table A.3 for a listing.) This temporal dimension of the panel makes it possible to assess the impact of country-specific variables that vary over time, such as GDP per capita, aid, and remittances, in a model that includes country fixed effects. The dataset covers 47 developing countries, of which 25 are in Sub-Saharan Africa, with five asset quintiles for each country-year, yielding a total of 380 observations. Table 5 provides summary statistics on the main variables of interest. It suggests that there are strong within-country health disparities that are correlated with asset inequality. Households belonging to the poorest asset quintile have the highest mean infant and child mortality rates; child mortality is almost twice as high for the poor- est quintile as for the richest one. A similar gap can be observed in the female school completion rate, which varies from 29.15 for the poorest quintile to 76.34 for the richest. It is worth noting that the differences in mean health indicators between the poorest and richest quintiles are always smaller than the ranges across countries within each quintile. The intracountry model to be estimated is very similar to the cross-country model presented in the preceding section in the sense that control variables are roughly the same and are expressed in log-linear form. The baseline model may therefore be writ- ten as follows: ln Healthj,i,t i lnXi,(t 1,t 4) lnXj,i,t ln Health aidi,(t 1,t 4) 5 5 lnRemiti,(t 1,t 4) jqj j qj *ln Health aidi,(t 1,t 4) j 2 j 2 5 j qj *ln Remiti,(t 1,t 4) j,i,t , (4) j 2 where j is the quintile index and qj are quintile dummy variables.14 Vector Xi,(t 1,t 4) includes GDP per capita in PPP constant terms and the number of physicians per 1,000 inhabitants. These variables are averaged over three years, A R E R E M I T TA N C E S M O R E E F F E C T I V E T H A N A I D ? | 191 TABLE 5. Summary Statistics Variable Mean Standard deviation Minimum Maximum Full sample (380 observations) Infant mortalitya 72.13 33.75 11.90 187.70 Child mortalityb 113.80 67.00 14.20 354.90 Female educational attainmentc 50.44 31.94 0.50 99.80 Poorest quintile, measured by an asset index (76 observations) Infant mortality a 86.88 31.32 32.00 187.70 Child mortalityb 140.08 62.82 39.10 297.90 Female educational attainmentc 29.15 25.98 0.50 98.70 Second quintile (76 observations) Infant mortalitya 82.62 32.71 23.80 152.30 Child mortalityb 132.33 69.25 27.30 354.90 Female educational attainmentc 39.24 29.75 1.00 99.50 Third quintile (76 observations) Infant mortalitya 75.91 34.14 19.70 157.20 Child mortalityb 120.08 69.44 23.50 348.30 Female educational attainmentc 48.38 30.98 1.50 99.80 Fourth quintile (76 observations) Infant mortality a 65.64 32.17 11.90 142.00 Child mortalityb 102.63 64.63 14.20 314.90 Female educational attainmentc 59.09 29.71 4.80 99.60 Richest quintile (76 observations) Infant mortalitya 49.58 24.51 13.80 97.20 Child mortalityb 73.88 45.93 15.80 183.70 Female educational attainmentc 76.34 20.13 27.00 99.80 Source: World Bank Health, Nutrition, and Population database. a. Infant mortality is measured by the number of deaths of children under 12 months of age per 1,000 live births, based on experience during the 10 years before the survey. b. Child mortality refers to the number of deaths of children under age 5 per 1,000 live births, based on experience during the 10 years before the survey. c. Female educational attainment is measured by the percent of women age 15­49 years who have completed fifth grade. from t 1 to t 4, and are measured in logarithms. Vector Xj,i,t contains quintile- and time-specific female educational attainment expressed in log-linear form. To test whether the impact of health aid and remittances differs between quintiles, lnHealthaid and lnRemit are interacted with quintile dummies q2 to q5, the poorest quintile being the reference. We choose not to interact quintile dummies with the other control variables such as GDP per capita. in order to limit the number of instru- ments when the IV specification is used. Finally, we control for quintile fixed effects (5 2 j qj ), as well as for country fixed effects. j 192 | L I S A C H A U V E T, F L O R E G U B E R T, A N D S A N D R I N E M E S P L É - S O M P S As in the cross-country analysis, endogeneity of aid, remittances, and income is controlled for using an IV specification. The education variable has also been found to be endogenous to health indicators. This is probably because education is meas- ured by the contemporaneous school completion rate. Instruments include lagged GDP per capita, the ratio of broad money supply (M2) to GDP, lagged health aid per capita, and total aid budgets for France, Japan, the United Kingdom, and the United States in constant dollars. We also include among the instruments lagged GDP per capita and lagged health aid per capita interacted with quintiles q2 to q5. Estimation Results The intracountry impact of aid and remittances on child health is assessed using child and infant mortality rates. We proceed in two steps. First, the baseline model is esti- mated without including any interaction terms between health aid and remittances, on the one hand, and quintile dummies, on the other hand (table 6). We then add these interaction terms to our set of regressors (table 7). Even though controlling for endogeneity and countries' unobserved heterogeneity is likely to provide more reli- able results, as in regressions (3) and (6) of table 1, tables 6 and 7 also present the results of simple OLS and fixed-effects regressions. As suggested by table 6, the impact of our control variables is quite similar to that found using our cross-country specification. GDP per capita, for instance, tends to decrease infant and child mortality rates. The coefficient of this variable suggests that an increase of 1 percent in GDP per capita reduces child and infant mortality by about 0.6 percent. As in the previous specification, the number of physicians per 1,000 inhab- itants is found to have no significant effect on child health outcomes. Female education is found to have a negative impact on the child mortality rate but not on the infant mor- tality rate. This result is in accordance with our previous results but not with those of Ravallion (2007), who found a significant negative impact of female education what- ever child health indicators were chosen. This lack of robust impact may come about because the education variable we use is less precise than that employed by Ravallion; we use the percentage of women age 15­49 who have completed the fifth grade, where- as Ravallion (2007) uses the average number of years of female schooling. Turning to our variables of interest, estimation results suggest that remittances and health aid have no impact at all. Adding interaction terms substantially alters the pic- ture. As suggested by table 7, migrants' remittances are now significant, and their impact on child health outcomes is found to be stronger for the richest quintiles than for the poorest ones. Remittances and remittances interacted with quintile dummies are jointly significant in the child and infant mortality equations. The impact of remittances on health indicators for the poorest quintiles is nil (column 3), whereas it is stronger for the middle and upper classes, at about 0.11, 0.16, and 0.23 for quin- tiles 3, 4, and 5, respectively. Overall, this result suggests that remittances tend to increase health disparities within countries. By contrast, neither an antipoor nor a pro-poor effect is detected for health aid. This finding contrasts with that of Fielding, McGillivray, and Torres (2008), who estimate a system of simultaneous equations on several welfare measures, including A R E R E M I T TA N C E S M O R E E F F E C T I V E T H A N A I D ? | 193 TABLE 6. Intracountry Specification without Interaction Terms Child mortality rate Infant mortality rate OLS Within 2SLS OLS Within 2SLS (1) (2) (3) (4) (5) (6) GDP per capitaa 0.272 0.871 0.673 0.281 0.868 0.620 (3.41)*** (5.21)*** (3.16)*** (4.52)*** (5.04)*** (2.71)*** Number of physicians 0.157 0.111 0.016 0.081 0.065 0.034 per 1,000 inhabitants (4.55)*** (1.61) (0.21) (2.82)*** (0.92) (0.39) Female educational attainmenta 0.132 0.047 0.220 0.100 0.052 0.186 (3.26)*** (2.01)** (1.79)* (3.08)*** (2.13)** (1.59) Remittances per capitaa 0.031 0.036 0.075 0.022 0.022 0.076 (1.64) (1.30) (1.53) (1.34) (0.77) (1.56) Health aid per capitaa 0.027 0.053 0.048 0.023 0.047 0.045 (0.71) (1.72)* (0.90) (0.66) (1.48) (0.83) Constant 7.054 11.163 6.696 10.738 (11.63)*** (8.84)*** (14.07)*** (8.26)*** Fixed effects No Yes Yes No Yes Yes Quintile dummies Yes Yes Yes Yes Yes Yes Number of observations 380 380 370 380 380 370 R2 0.79 0.72 0.74 0.64 Number of countries 47 46 47 46 Underidentification test 0.01 0.01 (p-value) Sargan (p-value) 0.52 0.20 Income instrumentation 0.000 0.000 F-statistic (p-value) Female education instrumentation 0.044 0.044 F-statistic (p-value) Aid instrumentation F-statistic 0.000 0.000 (p-value) Remittance instrumentation 0.000 0.000 F-statistic (p-value) Note: 2SLS, two-stage least squares; GDP, gross domestic product; OLS, ordinary least squares. Numbers in paren- theses are robust t-statistics. GDP per capita, number of physicians per 1,000 inhabitants, health aid per capita, and remittances are averages over three-year periods, from t ­ 1 to t ­ 4, measured in logs. Female educational attain- ment is measured at the same period as the outcome variable by quintile and is in logs. a. Instrumented regressors in equations (3) and (6). Instruments include lagged GDP per capita; lagged health aid per capita; ratio of broad money supply (M2) to GDP; and total aid budgets of France, Japan, the United Kingdom, and the United States in constant dollars. Tests for excludability of the instruments are available on request. * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. 194 | L I S A C H A U V E T, F L O R E G U B E R T, A N D S A N D R I N E M E S P L É - S O M P S TABLE 7. Intracountry Specification with Interaction Terms Child mortality rate Infant mortality rate OLS Within 2SLS OLS Within 2SLS (1) (2) (3) (4) (5) (6) GDP per capitaa 0.271 0.867 0.673 0.279 0.865 0.620 (3.38)*** (5.63)*** (2.72)*** (4.46)*** (5.33)*** (2.39)** Number of physicians per 0.156 0.109 0.016 0.080 0.063 0.034 1,000 inhabitants (4.48)*** (1.72)* (0.15) (2.79)*** (0.95) (0.31) Female educational attainmenta 0.137 0.037 0.220 0.103 0.045 0.186 (3.53)*** (1.69)* (1.49) (3.29)*** (1.92)* (1.39) Remittances per capitaa 0.017 0.015 0.035 0.022 0.024 0.011 (0.73) (0.56) (0.49) (0.99) (0.83) (0.16) Remittances per capita 0.033 0.040 0.044 0.034 0.040 0.029 quintile 2a (3.37)*** (2.75)*** (1.00) (2.80)*** (2.64)*** (0.69) Remittances per capita 0.055 0.063 0.114 0.052 0.059 0.087 quintile 3a (3.64)*** (4.35)*** (3.15)*** (3.60)*** (3.84)*** (2.58)*** Remittances per capita 0.074 0.077 0.163 0.067 0.070 0.132 quintile 4a (4.25)*** (5.31)*** (3.99)*** (3.84)*** (4.55)*** (3.40)*** Remittances per capita 0.085 0.076 0.232 0.068 0.060 0.187 quintile 5a (3.61)*** (5.21)*** (2.98)*** (3.05)*** (3.93)*** (2.60)*** Health aid per capitaa 0.028 0.003 0.073 0.031 0.002 0.072 (0.69) (0.08) (0.85) (0.79) (0.07) (0.90) Health aid per capita quintile 2a 0.041 0.040 0.078 0.037 0.037 0.074 (3.18)*** (1.75)* (1.32) (2.24)** (1.50) (1.28) Health aid per capita quintile 3a 0.067 0.062 0.121 0.059 0.055 0.120 (3.11)*** (2.70)*** (2.20)** (2.62)** (2.26)** (2.29)** Health aid per capita quintile 4a 0.096 0.087 0.194 0.097 0.089 0.178 (3.35)*** (3.75)*** (2.94)*** (3.30)*** (3.63)*** (2.76)*** Health aid per capita quintile 5a 0.074 0.061 0.210 0.076 0.066 0.209 (1.95)* (2.64)*** (2.17)** (2.09)** (2.69)*** (2.30)** Constant 6.969 11.069 6.620 10.655 (11.43)*** (9.51)*** (13.82)*** (8.70)*** Fixed effects No Yes Yes No Yes Yes Quintile dummies Yes Yes Yes Yes Yes Yes Number of observations 380 380 370 380 380 370 R2 0.81 0.76 0.75 0.69 Number of countries 47 46 47 46 Joint significance of aid variables 0.038 0.000 0.073 0.061 0.005 0.100 Joint significance of remittances 0.000 0.000 0.000 0.003 0.000 0.007 variables Underidentification test (p-value) 0.14 0.14 Sargan (p-value) 0.66 0.31 A R E R E M I T TA N C E S M O R E E F F E C T I V E T H A N A I D ? | 195 TABLE 7. (continued) Child mortality rate Infant mortality rate OLS Within 2SLS OLS Within 2SLS (1) (2) (3) (4) (5) (6) Income instrumentation F-statistic 0.000 0.000 (p-value) Female education instrumentation 0.000 0.000 F-statistic (p-value) Aid instrumentation F-statistic 0.000 0.000 (p-value) Aid q2 instrumentation 0.046 0.046 F-statistic (p-value) Aid q3 instrumentation 0.046 0.046 F-statistic (p-value) Aid q4 instrumentation 0.046 0.046 F-statistic (p-value) Aid q5 instrumentation 0.046 0.046 F-statistic (p-value) Remittances instrumentation 0.000 0.000 F-statistic (p-value) Remittances q2 instrumentation 0.003 0.003 F-statistic (p-value) Remittances q3 instrumentation 0.003 0.003 F-statistic (p-value) Remittances q4 instrumentation 0.003 0.003 F-statistic (p-value) Remittances q5 instrumentation 0.003 0.003 F-statistic (p-value) Note: 2SLS, two-stage least squares; GDP, gross domestic product; OLS, ordinary least squares; q, quintile. Numbers in parentheses are robust t-statistics. GDP per capita, number of physicians per 1,000 inhabitants, health aid per capi- ta, and remittances are averages over three-year periods, from t ­ 1 to t ­ 4, measured in logs. Female educational attainment is measured at the same period as the outcome variable by quintile and is in logs. a. Instrumented regressors in equations (3) and (6). Instruments include lagged GDP per capita; lagged health aid per capita; ratio of broad money supply (M2) to GDP; total aid budgets of France, Japan, the United Kingdom, and the United States in constant dollars; and lagged GDP per capita and lagged health aid per capita, both crossed with quintiles q2 to q5. Tests for excludability of the instruments are available on request. * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. health outcomes, and find a significant negative impact of total aid (as a percent of GDP) on child mortality and increased effectiveness of aid for the richest quintile. As a final step, we check whether including medical brain drain among the set of regressors changes the baseline results (table 8). The expatriation rate of physicians does not seem to have a direct significant impact, but its inclusion among the set of regressors mitigates the impact of remittances: whatever the quintile, the impact of remittances on child and infant mortality becomes indeed very low. 196 | L I S A C H A U V E T, F L O R E G U B E R T, A N D S A N D R I N E M E S P L É - S O M P S TABLE 8. Intracountry Specification with Medical Brain Drain, Two-Stage Least Squares Estimations Child mortality rate Infant mortality rate (1) (2) GDP per capitaa 0.823 0.814 (3.41)*** (2.71)*** Number of physicians per 1,000 inhabitants 0.913 1.123 (1.55) (1.56) Female educational attainmenta 0.195 0.348 (0.65) (0.95) Medical brain draina 0.983 1.268 (1.50) (1.61) Remittances per capitaa 0.115 0.114 (1.52) (1.18) Remittances per capita quintile 2a 0.064 0.055 (1.81)* (1.35) Remittances per capita quintile 3a 0.126 0.102 (3.85)*** (2.67)*** Remittances per capita quintile 4a 0.121 0.077 (2.78)*** (1.53) Remittances per capita quintile 5a 0.077 0.012 (0.69) (0.09) Health aid per capitaa 0.116 0.127 (1.54) (1.44) Health aid per capita quintile 2a 0.053 0.041 (1.10) (0.68) Health aid per capita quintile 3a 0.061 0.043 (0.98) (0.56) Health aid per capita quintile 4a 0.096 0.052 (1.11) (0.49) Health aid per capita quintile 5a 0.068 0.026 (0.57) (0.18) Fixed effects Yes Yes Quintile dummies Yes Yes Number of observations 370 370 Number of countries 46 46 Joint significance of aid variables 0.557 0.804 Joint significance of remittances variables 0.001 0.080 Underidentification test (p-value) 0.67 0.67 Sargan (p-value) 0.87 0.66 Income instrumentation F-statistic (p-value) 0.000 0.000 Female education instrumentation F-statistic (p-value) 0.000 0.000 Aid instrumentation F-statistic (p-value) 0.000 0.000 Aid q2 instrumentation F-statistic (p-value) 0.046 0.046 Aid q3 instrumentation F-statistic (p-value) 0.046 0.046 Aid q4 instrumentation F-statistic (p-value) 0.046 0.046 Aid q5 instrumentation F-statistic (p-value) 0.046 0.046 A R E R E M I T TA N C E S M O R E E F F E C T I V E T H A N A I D ? | 197 TABLE 8. (continued) Child mortality rate Infant mortality rate (1) (2) Remittances instrumentation F-statistic (p-value) 0.003 0.003 Remittances q2 instrumentation F-statistic (p-value) 0.003 0.003 Remittances q3 instrumentation F-statistic (p-value) 0.003 0.003 Remittances q4 instrumentation F-statistic (p-value) 0.003 0.003 Remittances q5 instrumentation F-statistic (p-value) 0.003 0.003 Medical brain drain instrumentation F-statistic (p-value) 0.000 0.000 Note: GDP, gross domestic product; q, quintile. Numbers in parentheses are robust t-statistics. GDP per capita, num- ber of physicians per 1,000 inhabitants, health aid per capita, and remittances are averages over three-year periods, from t ­ 1 to t ­ 4, measured in logs. Female educational attainment is measured at the same period as the outcome variable by quintile and is in logs. a. Instrumented regressors. Instruments include lagged GDP per capita; lagged health aid per capita; ratio of broad money supply (M2) to GDP; total aid budgets of France, Japan, the United Kingdom, and the United States in con- stant dollars; and lagged GDP per capita and lagged health aid per capita, both crossed with quintiles q2 to q5. Tests for excludability of the instruments are available upon request. * Significant at the 10 percent level. *** Significant at the 1 percent level. Conclusion For several years it has been asserted that the achievement of the Millennium Devel- opment Goals by 2015 will require increased external financing coupled with improved targeting effectiveness in favor of the poorest population. In this context, international migrants' remittances have been increasingly put forward as a promis- ing source of external financing. Nevertheless, empirical assessments of the respective impact of aid and remittances on aggregate welfare, measured either by poverty in monetary terms or by human development indicators, are rather scarce. In this paper we chose to focus on two child health outcomes--under-five mortal- ity (MDG 4) and infant mortality--in order to examine the direct impact of aid to the health sector and of remittances on these human development indicators. Given our primary focus, we do not enter the debate on the relative importance of the direct and indirect (via government pro-poor expenditure) impacts of aid. To complete our diagnosis, we push our analysis further and investigate the net impact of migration-- that is, the effectiveness of migration, including the effect of the medical brain drain. We also examine the intracountry allocation of aid and migrants' remittances. Our results for health aid are in line with the literature that examines the welfare impact of aid using cross-country data in the sense that they suggest a nonrobust rela- tionship between aid and welfare. Although the impact of health aid is found to be sig- nificant in our cross-country regressions (but only when aid is interacted with income per capita), this result vanishes when cross-country quintile level data are used. By contrast, and for the first time, the trade-off between the gains from migration and its costs is underlined. As suggested by our paper, the net impact of migration is 198 | L I S A C H A U V E T, F L O R E G U B E R T, A N D S A N D R I N E M E S P L É - S O M P S rather weak when the negative effect of medical brain drain is taken into account. Moreover, remittances are found to be more beneficial for children belonging to the richest households. This result is in line with other microeconomic evidence suggesting that remittances may increase within-country inequality. It differs from the finding of Chauvet and Mesplé-Somps (2007) that remittances have a pro-poor impact. The small estimated impact of health aid and remittances net of brain drain costs might well explain why child mortality rates have not substantially improved for three decades, as asserted by Murray et al. (2007), despite the growing volume of health aid and migrants' remittances. This does not imply that official assistance is inefficient, nor does it mean that private remittances should substitute for aid. Rather, it means that further investigation into the microlevel determinants of child mortality rates is needed to improve our understanding of the bad performance on child health outcomes in most developing countries, and in Africa in particular. ANNEX TABLE A.1 Cross-Country Regression Sample (109 countries) Albania* Egypt, Arab Rep. Madagascar South Africa Algeria El Salvador Malawi Sri Lanka Argentina Equatorial Guinea* Malaysia St. Lucia Armenia* Eritrea* Mali St. Vincent Azerbaijan Ethiopia Mauritania Sudan Bangladesh Fiji Mauritius Swaziland Belize* Gabon Mexico Syrian Arab Republic* Benin Gambia, The Moldova* Tajikistan* Bolivia Georgia Mongolia Tanzania Bosnia and Herzegovina Ghana Morocco Thailand Botswana Grenada* Mozambique Togo Brazil Guatemala Namibia Tonga Burkina Faso Guinea Nepal Trinidad and Tobago* Cambodia Guinea-Bissau Nicaragua Tunisia Cameroon Guyana Niger Turkey Cape Verde Haiti Nigeria Uganda Central African Republic* Honduras Oman* Uruguay* Chad* India Pakistan Vanuatu Chile* Indonesia Panama Venezuela, R. B. de China Iran, Islamic Rep.* Papua New Guinea Vietnam* Colombia Jamaica Paraguay Yemen, Republic Comoros Jordan Peru Zimbabwe* Congo, Rep. Kazakhstan* Philippines Costa Rica Kenya Rwanda Côte d'Ivoire Kyrgyz Republic* Samoa Croatia* Lao PDR Senegal Dominica* Lebanon Seychelles Dominican Republic Lesotho Sierra Leone Ecuador Macedonia, FYR* Solomon Islands * Countries excluded from the 2SLS estimations. A R E R E M I T TA N C E S M O R E E F F E C T I V E T H A N A I D ? | 199 ANNEX TABLE A.2 Instrumentation Equations Health aid Remittances Variable GDP per capita per capita per capita Number of physicians per 1,000 inhabitants 0.027 0.665 0.510 (1.050) ( 2.040)** (1.670)* Female educational attainment 0.019 0.582 0.204 (0.310) (0.780) (0.260) Dummy for missing education variable 0.059 0.886 1.425 ( 0.560) (0.660) ( 1.130) Year = 1995 0.187 0.012 0.923 ( 2.180)** ( 0.010) ( 1.580) Year = 2000 0.138 0.487 1.178 ( 1.630) (0.600) ( 2.140)** Twice-lagged health aid per capita 0.004 0.244 0.074 (0.860) ( 2.160)** (1.680)* Twice-lagged GDP per capita 0.289 0.170 0.500 (3.630)*** ( 0.310) ( 1.320) M2/GDP 0.054 0.093 0.738 ( 1.050) (0.230) (2.170)** Total French aid budget same religion 0.175 0.970 3.606 (1.030) ( 0.740) (4.140)*** Total French aid budget distance 0.000 0.000 0.000 ( 1.080) (0.940) (0.750) Total U.K. aid budget same religion 0.294 0.272 3.825 ( 1.600) (0.160) ( 3.620)*** Total U.K. aid budget distance 0.000 0.000 0.000 ( 1.470) (0.940) (1.990)** Total German aid budget same religion 0.272 1.517 6.228 ( 1.040) ( 0.620) ( 4.090)*** Total U.S. aid budget same religion 0.151 0.228 1.825 (1.200) (0.150) (2.400)** Total U.S. aid budget distance 0.000 0.000 0.000 (0.620) (0.620) ( 3.400)*** Total Japanese aid budget same religion 0.146 4.828 6.449 ( 0.590) ( 1.550) (1.630) Total Japanese aid budget distance 0.000 0.000 0.000 (1.150) (0.360) (1.620) Fixed effects Yes Yes Yes Number of observations 237 237 237 F-statistic (p-value) 0.0002 0.0997 0.0000 Partial R2 of excluded instruments 0.3139 0.1760 0.2102 * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. 200 | L I S A C H A U V E T, F L O R E G U B E R T, A N D S A N D R I N E M E S P L É - S O M P S ANNEX TABLE A.3 Intracountry Regression Sample Country name Year Country name Year Armenia 2000 Kyrgyz Republic 1997 Bangladesh 1996, 1999 Madagascar 1997 Benin 1996, 2001 Malawi 1992 Bolivia 1998, 2003 Mali 1995, 2001 Brazil 1996 Mauritania 2000 Burkina Faso 1993, 1999, 2003 Morocco 1992, 2003 Cambodia 1997 Mozambique 1997, 2003 Cameroon 1991, 1998, 2004 Namibia 1992, 2000 Chad 1996, 2005 Nepal 1996 Colombia 1995, 2000, 2005 Nicaragua 1997 Comoros 1996 Niger 1998 Côte d'Ivoire 1994 Nigeria 1990, 2003 Dominican Republic 1996, 2002 Peru 1996 Ethiopia 2000 Philippines 1998 Gabon 2000 Rwanda 2000 Ghana 1993, 1998, 2003 South Africa 1998 Guatemala 1995, 1998 Tanzania 1996, 1999 Guinea 1999 Togo 1998 Haiti 1994, 2000 Turkey 1993, 1998 India 1992, 1998 Uganda 1995, 2000 Indonesia 1997 Vietnam 1997, 2002 Jordan 1997 Yemen, Republic 1997 Kazakhstan 1995 Zimbabwe 1994, 1999 Kenya 1993, 1998, 2003 Notes 1. The eight Millennium Development Goals are, in brief: (1) to halve extreme income poverty; (2) to achieve universal primary education; (3) to promote gender equality; (4) to reduce the under-five mortality rate by two-thirds; (5) to reduce the maternal mortality rate by three-quarters; (6) to reduce the incidence of AIDS; (7) to promote sustainable develop- ment and to halve the percentage of people without access to safe drinking water; and (8) to set up a global partnership for development involving more generous and more widespread official development assistance. 2. CRS data are available at the OECD Web site http://www.oecd.org/document/0/0,2340, en_2649_34447_37679488_1_1_1_1,00.html. 3. Unlike us, Mishra and Newhouse (2007) include in their sample data on health commit- ments covering the 1960s and 1970s. 4. Note that the introduction of country fixed effects contributes to solving, although imper- fectly, the omitted variable bias. A R E R E M I T TA N C E S M O R E E F F E C T I V E T H A N A I D ? | 201 5. Tavares (2003) and Rajan and Subramanian (2005a, 2005b) use similar instruments for aid and remittances. 6. Tests of overidentification and underidentification are reported in each table. Tests for weak instruments, excludability, and partial R-squared are available on request. 7. Following the literature on the determinants of health, and contrary to Mishra and New- house (2007), we do not estimate a system of moment equations using generalized method of moments (GMM) with a lagged dependent variable. The main reason is that the num- ber of time periods is too small. 8. The kind of nonlinearity is still debated. Some authors argue that the relationship is quad- ratic (Hansen and Tarp 2001; Lensink and White 2001). Others claim that the impact of aid depends on economic policy (Burnside and Dollar 2000), on vulnerability to external shocks (Guillaumont and Chauvet 2001), on export price shocks (Collier and Dehn 2001), or on whether the country is tropical (Dalgaard, Hansen, and Tarp 2004). 9. Results are available from the authors on request. 10. We reran all our regressions using aggregate aid disbursements instead of health aid dis- bursements, but the variable was never significant, suggesting that not all types of aid affect health outcomes. Results are available on request. 11. The expatriation rate is also provided disaggregated by destination country. 12. Note that our estimations may underestimate the impact because the medical brain drain variable provided by Docquier and Bhargava (2007) measures only emigration of physi- cians, not that of other medical personnel such as nurses and midwives. 13. In a recent paper, Fay et al. (2007) briefly reply to Ravallion's comments. 14. We are not able to include time dummies because years vary from one country to the other and we only have one year of observation for half the sample. References Acosta, Pablo A., Cesar A. Calderon, Pablo R. Fajnzylber, and Humberto Lopez. 2008. "What Is the Impact of International Remittances on Poverty and Inequality in Latin America?" World Development 36 (1): 89­114. Adams, Richard H. 2004. "Remittances and Poverty in Guatemala." Policy Research Working Paper 3418, World Bank, Washington, DC. ------. 2006. "Remittances and Poverty in Ghana." Policy Research Working Paper 3838, World Bank, Washington, DC. Adams, Richard H., and John Page. 2005. "Do International Migration and Remittances Reduce Poverty in Developing Countries?" World Development 33 (10): 1645­69. Anand, S., and Till Bärnighausen. 2004. "Human Resources and Health Outcomes: Cross- Country Econometric Study." Lancet 364: 1603­9. Barro, Robert J., and Jong-Wha Lee. 2000. "International Data on Educational Attainment: Updates and Implications." CID Working Paper 42, Center for International Develop- ment, Harvard University, Cambridge, MA. Boone, Peter. 1996. "Politics and the Effectiveness of Foreign Aid." European Economic Review 40: 289­329. Burnside, Craig, and David Dollar. 2000. "Aid, Policies, and Growth." American Economic Review 90 (4): 847­68. 202 | L I S A C H A U V E T, F L O R E G U B E R T, A N D S A N D R I N E M E S P L É - S O M P S Chauvet, Lisa, and Sandrine Mesplé-Somps. 2007. "Impact des financements internationaux sur les inégalités des pays en développement." Revue Economique 58: 735­44. Collier, Paul, and Jan Dehn. 2001. "Aid, Shocks, and Growth." Policy Research Working Paper 2688, World Bank, Washington, DC. Cox-Edwards, Alessandra, and Manuela Ureta. 2003. "International Migration, Remittances and Schooling. Evidence from El Salvador." Journal of Development Economics 72 (2): 429­61. Dalgaard, Carl-Johan, Henrik Hansen, and Finn Tarp. 2004. "On the Empirics of Foreign Aid and Growth." Economic Journal 114 (496): F191­F217. Docquier, Frédéric, and Alok Bhargava. 2007. "A New Panel Data Set on Physicians' Emigra- tion Rates (1991­2004)." Economics School of Louvain, Louvain-la-Neuve, Belgium. http://perso.uclouvain.be/frederic.docquier/filePDF/MBD1_Description.pdf. Dreher, Axel, Peter Nunnenkamp, and Rainer Thiele. 2008. "Does Aid for Education Educate Children? Evidence from Panel Data." World Bank Economic Review 22 (2): 291­314. Fay, Marianne, Danny Leipziger, Quentin Wodon, and Tito Yepes. 2005. "Achieving Child- Health-Related Millennium Development Goals: The Role of Infrastructure." World Development 33 (8): 1267­84. ------. 2007. "`Achieving Child-Health-Related Millennium Development Goals: The Role of Infrastructure'--A Reply." World Development 35 (5): 929­30. Fielding, David, Mark McGillivray, and Sebastian Torres. 2008. "Achieving Health, Wealth and Wisdom: Links Between Aid and the Millennium Development Goals." In Achieving the Millennium Development Goals, ed. Mark McGillivray. Studies in Devel- opment Economics and Policy. Houndsmill, Basingstoke, Hampshire, U.K.: Palgrave MacMillan. Filmer, Deon, and Lant Pritchett. 1999. "The Impact of Public Spending on Health: Does Money Matter?" Social Science and Medicine 49: 1309­23. ------. 2001. "Estimating Wealth Effects without Expenditure Data or Tears: An Application to Educational Enrolments in States of India." Demography 38 (1): 115­32. Gomanee, Karuna, Sourafel Girma, and Oliver Morrissey. 2005. "Aid, Public Spending and Human Welfare: Evidence from Quantile Regressions." Journal of International Develop- ment 17: 299­309. Gomanee, Karuna, Oliver Morrissey, Paul Mosley, and Arjan Verschoor. 2005. "Aid, Government Expenditure, and Aggregate Welfare." World Development 33 (3): 355­70. Guillaumont, Patrick, and Lisa Chauvet. 2001. "Aid and Performance: A Reassessment." Journal of Development Studies 37 (6): 66­92. Gupta, Sanjeev, Catherine Pattillo, and Smita Wagh. 2007. "Impact of Remittances on Poverty and Financial Development in Sub-Saharan Africa." IMF Working Paper 07/38, International Monetary Fund, Washington, DC. Gwatkin, Davidson R., Shea Rutsein, Kiersten Johnson, Eldwan Suliman, Adam Wagstaff, and Agbessi Amouzou. 2007. Socio-Economic Differences in Health, Nutrition, and Population within Developing Countries: An Overview. Produced by the World Bank in collaboration with the government of the Netherlands and the Swedish International Development Cooperation Agency, Washington, DC. Hansen, Henrik, and Finn Tarp. 2001. "Aid and Growth Regressions." Journal of Develop- ment Economics 64 (2): 547­70. Hildebrandt, N., and David McKenzie. 2005. "The Effects of Migration on Child Health in Mexico." Policy Research Working Paper 3573, World Bank, Washington, DC. A R E R E M I T TA N C E S M O R E E F F E C T I V E T H A N A I D ? | 203 IMF (International Monetary Fund). 2005. World Economic Outlook, April 2005: Global- ization and External Imbalances. Washington, DC: IMF. Kanaiaupuni, Shawn M., and Katharine M. Donato. 1999. "Migradollars and Mortality: The Effects of Migration on Infant Survival in Mexico." Demography 36: 339­53. Lachaud, Jean-Pierre. 1999. "Envois de fonds, inégalité et pauvreté au Burkina Faso." Revue Tiers Monde 160: 793­827. Leliveld, André. 1997. "The Effects of Restrictive South African Migrant Labor Policy on the Survival of Rural Households in Southern Africa: A Case Study from Rural Swaziland." World Development 25 (11): 1839­49. Lensink, Robert, and Howard White. 2001. "Are There Negative Returns to Aid?" Journal of Development Studies 37 (6): 42­65. Levine, Ruth, and the What Works Working Group. 2004. Millions Saved: Proven Successes in Global Health. Washington, DC: Center for Global Development. Mansuri, Ghazala. 2007. "Migration, School Attainment and Child Labor: Evidence from Rural Pakistan." Policy Research Working Paper 3945, World Bank, Washington, DC. Masud, Nadia, and Boriana Yontcheva. 2005. "Does Foreign Aid Reduce Poverty? Empirical Evidence from Nongovernmental and Bilateral Aid." IMF Working Paper 05/100, Inter- national Monetary Fund, Washington, DC. McGuire, James W. 2006. "Basic Health Care Provision and Under-5 Mortality: A Cross- National Study of Developing Countries." World Development 34 (3): 405­25. Michaelowa, Katharina, and Anke Weber. 2007. "Aid Effectiveness in Primary, Secondary, and Tertiary Education." Background paper for the Education for All (EFA) Global Monitor- ing Report 2008, Education for All by 2015: Will We Make It? United Nations Educa- tional, Cultural, and Scientific Organization (UNESCO), New York. http://unesdoc. unesco.org/images/0015/001555/155559e.pdf. Milanovic, Branco. 2005. "Can We Discern the Effect of Globalization on Income Distribu- tion? Evidence from Household Surveys." World Bank Economic Review 19: 21­44. Mishra, Prachi, and David L. Newhouse. 2007. "Health Aid and Infant Mortality." IMF Working Paper 07/100, International Monetary Fund, Washington, DC. Mosley, Paul. 1987. Overseas Aid: Its Defence and Reform. Wheatsheaf Books, Brighton, U.K. Mosley, Paul, John Hudson, and Arjan Verschoor. 2004. "Aid, Poverty Reduction and the `New Conditionality.'" Economic Journal 114: F217­F243. Murray, Christopher L., T. Laakso, K. Shibuya, K. Hill, and Alan D. Lopez. 2007. "Can We Achieve Millennium Development Goal 4? New Analysis of Country Trends and Forecasts of Under-5 Mortality to 2015." Lancet 370 (September 22): 1040­54. Pritchett, Lant, and Larry Summers. 1996. "Wealthier Is Healthier." Journal of Human Resources 31 (4): 841­68. Rajan, Raghuram G., and Arvind Subramanian. 2005a. "Aid and Growth: What Does the Cross-Country Evidence Really Show?" NBER Working Paper 11513, National Bureau of Economic Research, Cambridge, MA. ------. 2005b. "What Undermines Aid's Impact on Growth?" IMF Working Paper WP/05/126, International Monetary Fund, Washington, DC. Ratha, Dilip, Sanket Mohapatra, K. M. Vijayalakshmi, and Zhimei Xu. 2007. "Remittance Trends 2007." Migration and Development Brief 3, Development Prospects Group, Migration and Remittances Team, World Bank, Washington, DC. http://siteresources .worldbank.org/EXTDECPROSPECTS/Resources/476882-1157133580628/Briefing Note3.pdf. 204 | L I S A C H A U V E T, F L O R E G U B E R T, A N D S A N D R I N E M E S P L É - S O M P S Ravallion, Martin. 1993. "Human Development in Poor Countries: On the Role of Private Incomes and Public Services." Journal of Economic Perspectives 7 (1): 133­50. ------. 2007. "`Achieving Child-Health-Related Millennium Development Goals: The Role of Infrastructure'--A Comment." World Development 35 (5): 920­28. Roodman, David. 2007. "Macro Aid Effectiveness Research: A Guide for the Perplexed." Working Paper 135, Center for Global Development, Washington, DC. Tavares, José. 2003. "Does Foreign Aid Corrupt?" Economics Letters 79 (1, April): 99­106. World Bank. 2006. World Development Indicators 2006. Washington, DC: World Bank. Comment on "Are Remittances More Effective Than Aid for Improving Child Health? An Empirical Assessment Using Inter- and Intracountry Data," by Lisa Chauvet, Flore Gubert, and Sandrine Mesplé-Somps MELVIN D. AYOGU In their paper, Chauvet, Gubert, and Mesplé-Somps investigate whether health aid or remittances matter for child health and, in particular, whether these help to reduce infant mortality. In this respect, the paper qualifies as one more aid-ineffectiveness study. The deeper issues that the authors address, however, are related to those exam- ined in the paper by Jean-Paul Azam and Ruxanda Berlinschi, in this volume. Chau- vet, Gubert, and Mesplé-Somps take note of the call for a progressive substitution of remittances for official aid. If, indeed, remittances from migrant workers prove, in general, more effective than foreign aid in alleviating poverty, the obvious next step is to promote more migration flows from poor to rich countries. The policy advice would be, do not offer aid in lieu of migration; instead, allow more migration in return for less aid--at least, those types of aid for which remittances have been found to be a superior remedy. For this reason alone, and given that the paper by Azam and Berlinschi suggests that rich countries have a hidden agenda of trading more aid for less migration (the opposite tack), this line of inquiry should be enthusiastically wel- comed. The excitement of the topic, however, may have also led the authors to attempt too much with a dataset that is arguably dirty. (On the general state of aid data, see Easterly and Pfutze 2008, 30, 51.) What the Authors Attempt to Do Looking within and across countries, the authors investigate two key issues and attempt to tackle related interesting questions. The two main issues are (a) whether foreign aid targeted to the health care sector reduces infant (below age 1) and child (under age 5) mortality rates, and (b) whether remittances from migrant workers reduce child and infant mortality rates. Other questions concern the circumstances under which one form of intervention may be more effective than the other. If Melvin D. Ayogu is professor of economics and dean, Faculty of Commerce, University of Cape Town, South Africa. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank 205 206 | M E LV I N D . AY O G U remittances are beneficial, does it matter that they come at the expense of loss of skilled labor from migrant-sending countries? To examine this issue, the authors focus on physician expatriation and its impact. The premise is that if expatriation is harmful, the negative consequences could be set against the perceived benefits of remittances, even though not all the measured remittances accrue from expatriated physicians alone. (The latter observation would, if correct, result in an underestima- tion of the net effect of brain drain, if such leakages do in fact occur.) The authors also look at the impact of aid and remittances across income deciles within a country and examine the issue of absorptive capacity that has been frequently raised as a serious negative consequence of poor donor aid delivery practices. In rankings of donor best practices, excessive fragmentation and overhead costs are key factors in the rating criteria (Easterly and Pfutze 2008). What the Authors Find Using a panel of developing countries, the authors determine that remittances pro- mote child health care but that health aid matters only when the relationships among child health indicators, aid, remittances, and income are taken into account. Of course, this endogeneity effect runs deeper than can be addressed by tinkering with instrumental variables. Here the authors could be picking up the consequences of existing aid practices, perhaps the effect of aid conditionality. It has been argued that the persistence of conditionality is partly attributable to its usefulness as an instru- ment for the pursuit of donor multiple objectives, of which only a few may be, in fact, altruistic (see Ayogu 2006 for a discussion). Recipient countries understand this larger game. One dimension of the game is the Samaritan's dilemma elaborated in Svensson (2000), according to which a quandary for the do-gooder arises because recipients behave strategically; they have no incentives for implementing poverty reduction strategies when an increasing proportion of aid is conditioned on poverty. Overall, after all the econometric adjustments, the authors find that both types of inter- vention (foreign aid and remittances from abroad) improve health care outcomes. Their indirect test of absorptive capacity constraints was not so fruitful, in that it was not supported by the data. Brain drain of doctors was, however, found to be harmful to child health. Medical personnel and health aid are complementary; the lack of one depresses the other. Brain drain therefore reduces the effectiveness of aid as well as the net benefit of remittances. Remittances are found to be more effective than health aid in improving health outcomes for children from richer households. The finding of higher marginal productivity of remittances for higher-income groups may be picking up several issues, such as the fact that remittances are fungible, whereas targeted health aid is not. Remittances have the capacity to improve overall family welfare in a way that targeted health aid is unable to match. Among poor communities, remittances carry a positive feedback and a selection bias. Families that receive remittances are big fish in a small pond--even if their relatives residing over- seas are little fish in a humongous pond. Therefore, selectivity bias could be C O M M E N T O N C H A U V E T, G U B E R T, A N D M E S P L É - S O M P S | 207 confounding the interpretation of the finding that the revealed positive impact of remittances on child health discriminates with respect to income. Wealth is good for nutrition and other complementary ingredients for a healthy child, and the income flow from remittances may well add to household wealth over time. What to Make of the Findings? Although the authors are careful not to draw any hard recommendations from their study, they have nonetheless made clear that there are no implications here for a migration-aid trade-off. That is a prudent view; the whole question of brain drain, remittances, undocumented workers, and immigration policies in the con- text of a globalizing world has yet to be addressed adequately. For instance, a recent study (Clemens 2007) finds no evidence of the harmful effects of the exter- nal migration of health professionals. The more deleterious effect seems to be inter- nal migration across sectors by professionals anxious to avoid "leaving the brain in the drain." According to the Southern African Migration Project (Crush 2008), working conditions emerge as the single most important predisposing factor for health professional emigration. Particularly influential are (a) the inequitable dis- tribution of personnel across the public versus private and primary versus second- ary health sectors, and (b) the urban-rural divide. With regard to the expatriation of physicians, analysts argue that many of the public health issues surrounding infant mortality--such as immunization and hygiene, diarrhea and dehydration, access to clean water, and other basic public health care issues--do not actually make huge demands of a physician's expertise.1 Physicians do not appear to be as crucial as this study would seem to suggest. Drawing from the experiences of Médecins Sans Frontičres (Doctors without Borders) in South Africa, it could be argued that even if the expatriation of physicians is assumed to be less than benign, the core of the primary health care problem lies in governance and in government underinvestment in clinics (Steinberg 2008, 273), rather than in the likely impact of brain drain on child health. Our view on governance is corroborated by a recent finding by Rajkumar and Swaroop (2008). Finally there are some results that are difficult to reconcile, such as the finding that aid raises infant mortality in some African countries. Clearly, given the HIV/AIDS pandemic, it would have been advisable to control for HIV prevalence (see Deaton 2008). I am equally sympathetic regarding the data problems confronting the authors, including the problematic instrumental variables deployed. Nonetheless, aid practices such as timing of disbursements, conditionality, and the use of bilateral agencies are much more important than cultural distance and should have been used in constructing the proxy for health aid disbursements. Note 1. I am grateful to Dr. Max Price for this insight. 208 | M E LV I N D . AY O G U References Ayogu, Melvin. 2006. "Can Africa Absorb More Aid?" In Aid, Debt Relief and Development in Africa: African Development Report 2006, 25­ 40. New York: Oxford University Press. Clemens, Michael. 2007. "Do Visas Kill? Health Effects of African Health Professional Emi- gration." Working Paper 114, Center for Global Development, Washington, DC. Crush, Jonathan. 2008. "Southern African Migration Project SAMP. Migration Resources: Brain Drain Resources." Queen's University, Kingston, ON, Canada. http://www.queensu .ca/samp/migrationresources/braindrain/ (accessed June 8, 2008). Deaton, Angus. 2008. "Income, Health, and Well-Being around the World: Evidence from a Gallup World Poll." Journal of Economic Perspectives 22 (2): 53­72. Easterly, William, and Tobias Pfutze. 2008. "Where Does the Money Go? Best and Worst Prac- tices in Foreign Aid." Journal of Economic Perspectives 22 (2): 29­52. Rajkumar, Andrew Sunil, and Vinaya Swaroop. 2008. "Public Spending and Outcomes: Does Governance Matter?" Journal of Development Economics 86 (1): 96­111. Steinberg, Johnny. 2008. Three-Letter Plague: A Young Man's Journey through a Great Epi- demic. Johannesburg: Jonathan Ball. Svensson, Jakob. 2000. "When Is Foreign Aid Credible? Aid Dependence and Conditionality." Journal of Development Economics 61: 61­84. The Role of Emigration and Emigrant Networks in Labor Market Decisions of Nonmigrants JINU KOOLA AND ÇAGLAR ÖZDEN The Indian state of Kerala is an ideal place to explore a question that is prominent in the migration literature: what role does the existence of migrant networks have in the labor market participation decisions of nonmigrant household members? Two linked statewide representative surveys in 1998 and 2003 that collected individual informa- tion about each member of 10,000 households, including members who had migrated, are used for this purpose. The analysis of the labor market participation of young men reveals interesting patterns. In cross-sectional data, young men in households with migrant members are found to be less likely to be employed, indicating that migration discourages labor market participation by nonmigrants. When, however, panel data are analyzed and individuals are followed over time, those males under age 30 are found to be more likely to migrate in the second period, taking advantage of their migrant networks. This result goes counter to the claim that migration induces unem- ployment or withdrawal from the labor market among family members. Rather, it sug- gests that young men in migrant households have a higher expectation of emigration and that they are less likely to take a job in Kerala while they prepare to emigrate. Almost 10 percent of the labor force of Kerala State--close to 2 million people--lives and works in a Persian Gulf country. These numbers make Kerala one of the largest migrant-sending regions in the world, and an interesting place to study various aspects of emigration. This paper focuses on a paradox, created by migration, in the employ- ment patterns seen in Kerala. Emigration there increased by around 35 percentage points between 1998 and 2003, and the unemployment rate for young males increased to 17.4 percent. Given the high unemployment rate in the face of massive emigration, the question arises: why has the exit rate of Kerala's labor force not decreased unem- ployment among nonmigrants in the state? More specifically, what influence does emigration have on the labor supply decisions of nonmigrant household members? Jinu Koola and Çag lar Özden are with the Development Research Group of the World Bank. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank 209 210 | J I N U K O O L A A N D Ç A GL A R Ö Z D E N Massive communal emigration rates are likely to have diverse effects on young men as they enter the labor market. The cross-sectional data show that young men in households that include migrant members and that are part of significant migrant net- works are less likely to be employed. The implication is that the presence of migrants in the household or the community decreases labor market participation and supply among nonmigrant members, mainly through a higher income effect from remit- tances. This point has been raised many times in the literature as an undesirable con- sequence of migration. However, panel data that follow individuals over time indicate that these same young males are likely to migrate in the second period, taking advan- tage of the migration networks their families or communities have established. This finding goes counter to the previous result and indicates a more complex decision- making process by nonmigrant household members. More specifically, young men in migrant households or communities seem to have higher thresholds of entry into the Kerala labor market, since they have higher opportunity costs. And if they fail to obtain the right labor market outcomes, they simply use the network to migrate. A unique feature of this paper is that it looks at the roles of both family networks and social networks in migration and labor market decisions. The network theory of migration states that an individual's probability of emigration increases with the size of that person's network in his or her locality (see Massey et al. 1998). Kerala's migrants, however, do not depend on the emigration experience of all the people in their panchayat (local jurisdiction); rather, they rely on the people they know--family, friends, and acquaintances.1 The common characteristics of their networks are not necessarily geographic, but religious. Social contacts are determined by religion, and in Kerala, as in the rest of India, religion is a community identity that distinguishes social groups. Furthermore, the support and services provided by the network seem to be excludable and rivalrous, in that they are provided only to network members within the same geographic location. Using religion to define a community's network, as in this paper, confers several main advantages. First and foremost, religion exogenously identifies a well-defined and well-established social group, overcoming the problem of peer effects. Religion also addresses the issue of self-selection because members of a religious group do not "choose" their religion today.2 Another benefit of using religion and geography to identify a social group is that these variables allow for variation within localities and religious groups. Finally, religious networks provide social support and many other excludable services that are especially valuable in labor market and migration contexts. Most of the empirical studies that have explored the relationship between migra- tion and labor market participation by nonmigrants have focused on the role of remittances (see Hanson 2005; Amuedo-Dorantes and Pozo 2006; Kim 2007). Yet there is a gap in our understanding of how the expectation of emigration, facilitated by emigration networks and the emigration of other household members, affects the household's labor supply and demand. One of the main determinants of migration is the social and communal network that influences the decision to migrate and helps lower migration costs. Identification of the influence of the family and the community is a key issue in this paper. The economic literature that studies migrant networks faces several empirical difficulties. E M I G R AT I O N A N D E M I G R A N T N E T W O R K S I N L A B O R M A R K E T D E C I S I O N | 211 The main econometric challenge for network analysis is to identify the direction of the causal relationship between migrant networks and the migration decision, which is complicated by peer effects. Peer effects confound the interpretation of the network effect because shared community characteristics can influence the migration decision but may be unobservable (Boyd 1989; Massey, Goldring, and Durnad 1994). Manski (1993) terms this the reflection problem. Another concern for an econometric analysis is that access to the migrant network may be endogenous to the household's own emigration experience (Taylor 1987; Orrenius 1999; McKenzie and Rapoport 2005). One way to isolate the network effect is to limit our analysis to new migrants who are temporarily exogenous to the network (Orrenius 1999; Zhao 2001). Empirical studies of migration have attempted to address these serious econometric challenges of endogeneity in other ways, by using community fixed effects, individual- level fixed effects, or instrumental variables such as rainfall data and historical state- level migration rates (Orrenius 1999; Winters, de Janvry, and Sadoulet 2001; Munshi 2003; McKenzie and Rapoport 2005; Giles and Yoo 2007; Woodruff and Zenteno 2007). In this study, the variation afforded by religion and geography is used to help identify the relationship between migrant networks and migration. The results in this paper have important implications for policies regarding migra- tion and labor markets. A common view, as mentioned above, is that emigration reduces the labor market participation and supply of the family members left behind. They rely on the remittances sent by family members abroad and make the optimal leisure-labor trade-off. This is viewed as a negative outcome because lower labor supply indicates lower income levels. This paper, however, shows that the decision- making process is more complicated and intertemporal in nature. It demonstrates that although the presence of migrants in the household or the community increases the likelihood of absence from the labor market because of higher opportunity costs, the outside option turns out to be migration for the nonmigrants, not reliance on remit- tances indefinitely. These results imply that policies should be designed accordingly. In the next section, the institutional setting and context for the study are outlined. Subsequent sections present the data and provide basic summary statistics for the sample population, describe the conceptual framework and empirical methodology, and set out the estimation results. The final section contains conclusions. Linkages between Kerala and the Persian Gulf In 2003, 1.84 million people from Kerala, or close to 10 percent of the state's total labor force, were living and working in a Persian Gulf country (Zachariah and Rajan 2004).3 Migration is almost all temporary because of the rather restrictive immigra- tion policies enacted by the countries of the Gulf Cooperation Council (GCC), which prohibit permanent resettlement of foreign nationals (Massey et al. 1998). Economic linkages between Kerala and the Persian Gulf have a very long history, but the current migration began to surge in 1973 with the sudden increase in oil prices.4 The Gulf countries, which faced a shortage of qualified local labor, sought 212 | J I N U K O O L A A N D Ç A GL A R Ö Z D E N foreign labor to satisfy their construction boom and other labor market demands. Kerala's history of trade with the region and its large surplus labor force motivated migration (Zachariah, Mathew, and Rajan 2001; see also Weiner 1982; Massey et al. 1998). By 1998, migrants from Kerala represented over 50 percent of India's total emigration to the Gulf countries (Prakash 1998). The state's geographic position, at the southwestern tip of India, with 580 kilometers of coastline, gives it privileged access to the Gulf. Its population density and literacy rates are also very high by Indi- an standards.5 Thus, Kerala's unemployed but relatively well-educated labor force was attracted to the employment opportunities that the Gulf countries offered. In a country with a Hindu-majority population (80.5 percent), Kerala is unique in that it has sizable Muslim and Christian populations (India 2001). The shares of total population are Hindu, 56.1 percent; Muslim, 24.7 percent; and Christian, 19 per- cent. Sufficient ethnographic data exist to suggest that social networks in Kerala have been created and are maintained within religious circles (Osella and Osella 2000; Kurien 2002). A distinguishing characteristic of Kerala's Gulf migration is its religious dimension; although Muslims make up only one-fourth of the state's total population, they represent close to half of the total emigrant population.6 In other words, current emigrants made up 13 percent of the total Muslim population in 1998, as against only 3 percent of the Hindu population and 6 percent of the Christian population. This is the main fact used to identify migrant networks in the paper. Data This study uses data from two linked household surveys conducted by the Centre for Development Studies in Thiruvananthapuram, Kerala, and covering a statewide rep- resentative household sample: they are the Kerala Migration Survey (KMS) and the South Asia Migration Study (SMS). A principal advantage of this paper is that it exploits the panel nature of the survey data; half of the households canvassed in 1998 were resurveyed in 2003. The KMS was administered to the residents of 9,995 households selected from 200 panchayats in each of Kerala's 14 districts between March and December 1998 (Zachariah, Mathew, and Rajan 2001, 64). Fifty households were selected at random from each panchayat.7 The survey collected individual information about each member of the household, including members who had migrated, from the head of the house- hold at the time of the household visit. The SMS survey, conducted in 2003, was designed to replicate the KMS and used the same questionnaire and a similar sampling methodology. The primary data for the SMS were collected from 10,012 households in 225 panchayats (rather than 200 panchayats, as in 1998) in the state's 14 districts (Zachariah and Rajan 2004, 12). Emigrant Characteristics In the survey, current emigrants (EMIs) are categorized as those members of the household who had emigrated from Kerala and who had been living abroad at the E M I G R AT I O N A N D E M I G R A N T N E T W O R K S I N L A B O R M A R K E T D E C I S I O N S | 213 TABLE 1. Characteristics of Sample Populations, Kerala State, 1998 and 2003 Year and migrant Percent Mean age Mean years Percent Percent category Observations male (years) of education married unemployed 1998 EMI 2,099 91 26.7a 8.7a 49a 26a NMI 26,560 45 34.5 7.5 67 9 2003 EMI 2,940 83 25.2a 9.7a 41a 33a NMI 27,273 45 35.6 8.2 65 14 Source: Authors' calculations from the 1998 Kerala Migration Study (KMS) and the 2003 South Asia Migration Study (SMS), Centre for Development Studies, Thiruvananthapuram, Kerala, India. Note: EMI, current emigrant; NMI, nonmigrant. a. At the time of migration. time of the survey for at least one year. Emigrants accounted for 6 percent of the total labor force in 1998 and for 8 percent in 2003. Return emigrants (REMs) are those members of the household who have returned to Kerala after living abroad. By 2003 almost one-fifth of Kerala's labor force population, or close to 3 million people, had had some sort of migration experience--they were either current emigrants or return emigrants. In this study, a nonmigrant (NMI) is a household member who was nei- ther a current emigrant nor a return emigrant in the survey year. The average emigrant is male, is under age 30, and has at least one more year of education than his nonmigrant counterpart (table 1). In 1998 Muslims represented nearly one-half of the total emigrant population; Hindus accounted for about one-third and Christians, for one-fifth. One reason that Muslims are twice as likely as non-Muslims to emigrate is that the overwhelming majority of Kerala's emigrants go to the oil-rich Persian Gulf states. As table 2 shows, the shares of emigrants going to a Gulf country were 94 percent in 1998 and 88 per- cent in 2003. More than two-thirds of all emigrants to the Gulf went to either the United Arab Emirates (UAE) or Saudi Arabia. The choice of destination country also varies by religion. Almost half of all Mus- lim emigrants went to Saudi Arabia in 2003, but only one-third of Christian or Hindu emigrants did. A large share (14 percent) of Hindu emigrants went to Oman, and a significant percentage of Christian emigrants went to Kuwait (10 percent) and the United States (8 percent). Only 1 percent of Muslim emigrants went to a country other than the Arab states of the Persian Gulf. Waiting to Emigrate: Those Who Stay Behind In this study, the unemployed are defined as those who are seeking a job, those who are performing unpaid household work, or those for whom a job is not required. Unemployment is higher among younger cohorts: in 1998 it was 16.7 percent for those under age 30 but only 7.5 percent for those over age 30, and in 2003 the unemploy- ment rate of the younger cohort was double that of the older cohort (see table 3). 214 | J I N U K O O L A A N D Ç A GL A R Ö Z D E N TABLE 2. Emigration from Kerala State, by Host Country, 1998 and 2003 (percent) 1998 2003 Host country EMI REM EMI REM Saudi Arabia 38 41 27 35 United Arab Emirates 31 26 36 31 Oman 10 13 8 11 Kuwait, Republic of 5 5 6 5 Bahrain 5 6 6 6 Qatar 5 3 5 4 Total, Gulf states 94 93 88 92 United States 2 0.5 5 1 Southeast Asia 1 1 1 1 United Kingdom 0.1 0.1 1 0.5 Other 4 6 5 5 Total, other regions 6 7 12 8 Source: Authors' calculations from the 1998 Kerala Migration Study (KMS) and the 2003 South Asia Migration Study (SMS), Centre for Development Studies, Thiruvananthapuram, Kerala, India. Note: EMI, current emigrant; REM, return emigrant. Figures may not sum to totals because of rounding. TABLE 3. Unemployment Rate by Age and Sex, Kerala State, 1998 and 2003 (percent) 1998 2003 Age Male Female Total Male Female Total 15­19 14.3 7.0 10.7 13.3 10.6 12.0 20­24 24.5 18.0 21.3 30.4 26.9 28.7 25­29 21.2 14.1 17.8 24.0 26.6 25.3 30­34 13.7 7.3 10.7 17.5 17.4 17.3 35­39 8.7 3.6 6.3 15.0 13.3 14.2 40­44 6.1 2.9 4.6 12.3 7.0 9.7 45­49 4.8 2.7 3.8 10.3 4.5 7.4 50­54 7.5 5.6 6.6 8.5 6.2 7.4 55­59 18.3 11.3 14.5 13.8 7.9 10.8 15­60 14.2 8.8 11.6 17.4 15.0 16.22 Source: Authors' calculations from the 1998 Kerala Migration Study (KMS) and the 2003 South Asia Migration Study (SMS), Centre for Development Studies, Thiruvananthapuram, Kerala, India. As table 3 shows, almost a quarter of males between ages 20 and 25 in 1998 and almost a third of males in the same age cohort in 2003 were unemployed. Given that emigration from Kerala is almost 90 percent male and that the average age of emi- gration in both years was under age 30, one might ask why the exit rate of Kerala's labor force has not decreased unemployment in the state. E M I G R AT I O N A N D E M I G R A N T N E T W O R K S I N L A B O R M A R K E T D E C I S I O N S | 215 One likely explanation for the high unemployment rate among young males is that they are waiting for the opportunity to emigrate and are unwilling to take a job in Kerala in the meantime. As table 4 shows, the total unemployment rate is 40 percent higher in emigrant households. Furthermore, the unemployment rate seems to increase with the number of current emigrants from the household. Table 4 also shows that the unemployment gap between emigrant and nonmigrant households is at least 5 percentage points higher for the male population. The figures in table 5 show that the unemployment rate is highest for males between ages 15 and 29 in emigrant households. Table 6 presents the unemployment rate among males according to whether the community is above or below the mean network size (or migration prevalence ratio). TABLE 4. Unemployment Rate for Age Group 15­60, by Number of Emigrants from the Household, Kerala State, 1998 and 2003 (percent) Total population Male population Number of emigrants 1998 2003 1998 2003 0 10.4 14.5 12.0 14.63 1 14.6 20.0 20.0 24.44 2 19.9 24.0 28.5 29.6 Source: Authors' calculations from the 1998 Kerala Migration Study (KMS) and the 2003 South Asia Migration Study (SMS), Centre for Development Studies, Thiruvananthapuram, Kerala, India. TABLE 5. Unemployment Rate, Males, by Age Cohort and Household Emigration Status, Kerala State, 1998 and 2003 (percent) Age 15­29 Age 30­60 Household type 1998 2003 1998 2003 Nonmigrant household 14.5 23.0 9.6 13.5 Current emigrant household 28.2 26.5 17.7 27.2 Source: Authors' calculations from the 1998 Kerala Migration Study (KMS) and the 2003 South Asia Migration Study (SMS), Centre for Development Studies, Thiruvananthapuram, Kerala, India. TABLE 6. Unemployment Rate, Males Age 15­60, by Emigrant Network Size, Kerala State, 1998 and 2003 (percent) Community type 1998 2003 Below mean network size 12.8 16.3 Above mean network size 16.2 18.9 Source: Authors' calculations from the 1998 Kerala Migration Study (KMS) and the 2003 South Asia Migration Study (SMS), Centre for Development Studies, Thiruvananthapuram, Kerala, India. 216 | J I N U K O O L A A N D Ç A GL A R Ö Z D E N TABLE 7. Job-Seeking Rate, Males, by Age and by Emigration Status of Household, Kerala State, 1998 and 2003 (percent) Age 15­29 Age 30­60 Household type 1998 2003 1998 2003 Nonmigrant household 16.9 21.6 5.6 10.3 Current emigrant household 26.0 24.8 14.2 24.4 Source: Authors' calculations from the 1998 Kerala Migration Study (KMS) and the 2003 South Asia Migration Study (SMS), Centre for Development Studies, Thiruvananthapuram, Kerala, India. One might assume from these findings that unemployed young males in emigrant households are simply free riders without any intention of obtaining employment. To the contrary, 90 percent of emigrants in 1998 stated that their main reason for migra- tion was to seek employment. Indeed, the rate of job seeking is highest for young males in emigrant households. Table 7 shows that a quarter of males under age 30 in emigrant households was engaged in looking for a job at the time of the survey, as against 17 percent of young males in nonmigrant households in 1998 and 21 percent in 2003. The data do not specify whether job seekers are looking for jobs within or outside Kerala, but it is most probable, given these findings, that young males in emi- grant households are likely to be seeking jobs outside the state. Empirical Methodology The methods used to model the behavior of those who stay behind are described in this section. The panel dataset gives us the unique opportunity to examine the situa- tion in 2003 of those who were unemployed in 1998. Cross-Sectional Regressions Using 1998 and 2003 Datasets The hypothesis to be tested is that a household's migration experience and the size of the emigrant network increase the probability of unemployment among nonmigrant males. The following equation is estimated to predict whether individual i in panchayat p in time period t is employed, using a probit model: Pr(Unemployedi) b0 b1EMI_hhi b2EMI_Netwp b3EMI_Netwp2 Zi ei, (1) where Unemployed is an indicator variable equal to 1 if the individual is unemployed in year t and 0 otherwise; EMI_hh is a dummy variable equal to 1 if the household has at least one emigrant in year t; and EMI_Netw refers to what we term the geo- graphic emigrant network and represents the proportion of adult males above age 16 who are current emigrants in year t from panchayat p. E M I G R AT I O N A N D E M I G R A N T N E T W O R K S I N L A B O R M A R K E T D E C I S I O N S | 217 To test the hypothesis that males in emigrant households are more likely to be job seekers, the following equation is estimated, using a probit model: Pr(JobSeekeri) b0 b1EMI_hhi b2EMI_Netwp b3EMI_Netwp2 Zi ei, (2) where JobSeeker is an indicator variable equal to 1 if the individual is seeking a job in the survey year and 0 otherwise, and all other variables are defined as before. The sample of individuals included in these regressions is restricted to adult males between ages 15 and 60. Current emigrants and students are omitted from the sam- ple. Equations (1) and (2) are estimated separately for males ages 15­29 and 30­60. All standard errors are clustered by household, since individuals in the same house- hold may share similar characteristics, and all regressions are weighted. Panel Regressions The main question is whether the young men who were unemployed in 1998 are waiting for the opportunity to emigrate because they are seeking employment outside the state. To test this, the following regressions are estimated using panel data for the sample of males age 15­29 in 2003, excluding those who were migrants or students in 1998: Pr(Employ_Statusi) b0 b1Unemployedi b2EMI_hhi*Unemployedi b3EMI_hhi b4EMI_Netwp b5EMI_Netwp2 Zi ei. (3) The dependent variable in this regression, Employ_Status, takes on three values: 1 if the individual is unemployed in 2003; 2 if the individual is employed in 2003; or 3 if the individual is an emigrant in 2003. Unemployed is a dummy variable equal to 1 if the individual was unemployed in 1998. EMI_hh*Unemployed is the interaction term if the individual was both unemployed in 1998 and a member of an emigrant household in 1998. The following regression is then estimated for job seeking: Pr( Jobseeking_Statusi) b0 b1 Jobseekeri b2EMI_hhi*Jobseekeri b3EMI_hhi b4EMI_Netwp b5EMI_Netwp2 Zi ei, (4) where the dependent variable in this regression, Jobseeking_Status, takes on three values: 1 if the individual is a job seeker in 2003; 2 if the individual is employed in 2003; or 3 if the individual is an emigrant in 2003. Jobseeker is a dummy variable equal to 1 if the individual was a job seeker in 1998, and EMI_hh*Jobseeker is an interaction term if the individual was both a job seeker in 1998 and was living in an emigrant household in 1998. In panel regressions, EMI_hh refers to the individual's 1998 status. The network variables refer to the individual's network in 2003. Control variables, described next, refer to the individual's 1998 status, with an additional dummy variable equal to 1 if the individual was a return emigrant in either 1998 or 2003. 218 | J I N U K O O L A A N D Ç A GL A R Ö Z D E N Control Variables Control variables are included in all regressions. The individual characteristics include the individual's years of schooling and years of schooling squared; age and age squared; and a binary variable equal to 1 if the individual is married or unemployed. The household-level characteristics include (a) a binary variable equal to 1 if the house- hold has at least one return emigrant; (b) a binary variable equal to 1 if the household has at least one out-migrant; (c) a binary variable equal to 1 if there are multiple emi- grants in the household; (d) family size; and (e) the number of adult males (i.e., above age 18) in the household. Family characteristics (family size and the number of adult males in the household) include the migrant, since the aim is to capture the household's "pre-emigration" state. The data do not contain information on household income. To control for that household characteristic, an income proxy, the log of the total amount of land (wet and dry) owned by the household, is included. The panchayat-level variables included are average cost of migration for all migrants from the panchayat; average house quality of all households in the pan- chayat, rated on a scale of 1 to 5, with 1 being very poor and 5 being luxurious; and a binary variable equal to 1 if the panchayat is urban. The latter two variables may also proxy for the overall wealth and level of infrastructure in the panchayat.8 Empirical Results: Those Left Behind The empirical results from the cross-sectional regressions described in equations (1) and (2) are presented in annex tables A.1 and A.2. The results show that the likeli- hood of unemployment increases for nonmigrant men between ages 15 and 60 if they live in an emigrant household. The probability of unemployment increases with the size of the emigrant network in the 1998 cross-section. When equation (1) is estimated separately for males ages 15­29 and 30­60, the dis- incentive effect on employment of living in an emigrant household is found to be two times stronger for the younger cohort than for the older cohort of males in the 1998 cross-section (see table 4). Unemployment also increases significantly with the size of the emigrant network for both cross-sections. Males in emigrant households are also more likely to be job seekers in both years, as was shown in table 5. In addition, the probability of job seeking increases with the emigrant network in the 2003 cross-section. When equation (2) is estimated separate- ly by age cohort, the positive effect on job seeking of living in an emigrant household is found to be much stronger for younger males than for their older counterparts. One of the main advantages of this study is the panel structure of the data. The results of the panel analysis, presented in annex tables A.3 and A.4, support the hypothesis that unemployed males are waiting for emigration and are seeking jobs outside Kerala. The probability of emigration increases for young males who were living in an emigrant household in 1998 and were also unemployed or seeking jobs in 1998. As expected, the probability of emigration increases with the size of the emi- grant network. E M I G R AT I O N A N D E M I G R A N T N E T W O R K S I N L A B O R M A R K E T D E C I S I O N S | 219 Consistent with the theoretical literature on the determinants of unemployment and emigration, certain individual and household characteristics (years of schooling, age, marital status, employment status, family size, and number of adult males in the household) also influence the probabilities of being unemployed, seeking jobs, and emigrating. Conclusions The findings in this paper support the hypothesis that young males in Kerala are unemployed because they are waiting for the opportunity to emigrate and are look- ing for jobs outside the state. These results indicate that young men who have a higher expectation of emigration are less likely to take a job in Kerala while they prepare to emigrate. In other words, vast emigration from the state creates only "temporary" unemployment, not permanent unemployment in which nonmigrants simply rely on remittances without actively engaging in any productive activity. One auxiliary effect from this stock of unemployed and ambitious young people is that they add (in absolute terms) to the number of highly skilled, albeit unemployed, individuals in Kerala's labor force. Emigrants are, on average, more educated than nonmigrants, and the proportion of emigrants or return emigrants with a secondary education is higher than the proportion of those with secondary education in the general population. Aspiring emigrants need higher levels of education because new policies and an increasing demand for skilled and professional workers in the Gulf countries require it. Since 1999, the United Arab Emirates no longer accepts applica- tions for visas for unskilled workers from Bangladesh, India, or Pakistan (Zachariah and Rajan 2004). Not surprisingly, the proportion of the population with secondary or tertiary education increased by 3 percentage points between 1998 and 2003 (Zachariah and Rajan 2004). Thus, those waiting to emigrate may be a potential "brain gain" for Kerala. Still, it is a matter of concern that in 2001 almost three- fourths of the unemployed had secondary or tertiary education (Zachariah, Mathew, and Rajan 2001). According to Zachariah and Rajan (2004) the total unemployment rate in Kerala increased by 8 percentage points between 1998 and 2003, but the unem- ployment rate among those who had completed secondary schooling increased 15 per- centage points. It is vital that the state engage these highly skilled but unemployed young people in the labor market even while they wait to emigrate. 220 | J I N U K O O L A A N D Ç A GL A R Ö Z D E N ANNEX TABLE A.1. Probit Estimates of the Marginal Effect of Migration on Unemployment, Kerala State, 1998 and 2003 (dependent variable: unemployed persons) 1998 2003 Age 15­60 Age 15­29 Age 30­60 Age 15­60 Age 15­29 Age 30­60 (1) (2) (3) (4) (5) (6) EMI households 0.05 0.06 0.04 0.04 0.02 0.07 (0.02)*** (0.03)** (0.01)** (0.02)** (0.03) (0.02)*** EMI network 0.09 0.22 0.05 0.16 0.24 0.08 (0.02)*** (0.08)*** (0.02)*** (0.02)*** (0.12)** (0.02)*** EMI network2 0.34 0.67 0.18 0.02 0.13 0.01 (0.15)** (0.34)* (0.13) (0.15) (0.34) (0.14) REM 1.87 2.30 1.12 2.89 7.60 1.52 (0.98)* (2.16) (0.84) (0.84)*** (2.59)*** (0.66)** Education 0.01 0.01 0.003 0.01 0.00 0.01 (0.004)** (0.01) (0.003) (0.004)*** (0.02) (0.003)*** Education2 0.001 0.001 0.0002 0.001 0.002 0.001 (0.0002)*** (0.006)* (0.0002) (0.0002)*** (0.001)** (0.0002)*** Married 0.16 0.15 0.15 0.21 0.17 0.20 (0.014)*** (0.021)*** (0.020)*** (0.014)*** (0.028)*** (0.020)*** Age 0.03 0.03 0.03 0.02 0.19 0.01 (0.002)*** (0.023) (0.004)*** (0.002)*** (0.024)*** (0.004)*** Age2 0.0003 0.001 0.0004 0.0002 0.005 0.0002 (0.00002)*** (0.0005)** (0.00004)*** (0.00002)*** (0.001)*** (0.00005)*** Family size 0.01 0.02 0.002 0.01 0.01 0.002 (0.002)*** (0.005)*** (0.002) (0.002)*** (0.005)** (0.002) Adult males 0.02 0.03 0.004 0.02 0.03 0.003 (0.005)*** (0.01)** (0.003) (0.004)*** (0.01)*** (0.004) Head 0.04 0.11 0.03 0.07 0.23 0.04 (0.01)*** (0.05)** (0.009)*** (0.01)*** (0.03)*** (0.01)*** Head's education 0.002 0.004 0.003 0.01 0.02 0.007 (0.002) (0.005) (0.002) (0.003)*** (0.006)*** (0.003)** Land 0.02 0.04 0.01 (0.003)*** (0.007)*** (0.002)*** Urban 0.01 0.03 0.01 (0.009) (0.022) (0.007) Number of observations 13,958 5,350 8,608 13,997 4,951 9,046 2 841.96 341.55 433.26 1,094.92 293.96 563.40 2 Pseudo R 0.1851 0.1208 0.1832 0.2054 0.1338 0.2061 Source: Authors' calculations from the 1998 Kerala Migration Study (KMS) and the 2003 South Asia Migration Study (SMS), Centre for Development Studies, Thiruvananthapuram, Kerala, India. Note: EMI, current emigrant; REM, return emigrant. Numbers in parentheses are robust standard errors. * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. E M I G R AT I O N A N D E M I G R A N T N E T W O R K S I N L A B O R M A R K E T D E C I S I O N S | 221 ANNEX TABLE A.2. Probit Estimates of the Marginal Effect of Migration on Job Seeking, Kerala State, 1998 and 2003 (dependent variable: job seeker) 1998 2003 Age 15­60 Age 15­29 Age 30­60 Age 15­66 Age 15­30 Age 30­60 (1) (2) (3) (4) (5) (6) EMI household 0.03 0.09 0.01 0.02 0.03 0.02 (0.008)*** (0.03)*** (0.005) (0.01)* (0.03) (0.009)** EMI network 0.09 0.18 0.04 0.13 0.08 0.07 (0.02)*** (0.08)** (0.01)*** (0.03)*** (0.04)** (0.02)*** EMI network2 0.12 0.37 0.03 0.01 0.26 0.01 (0.07)* (0.28) (0.03) (0.10) (0.31) (0.07) REM 1.04 3.85 0.20 2.54 8.73 1.11 (0.46)** (1.77)** (0.21) (0.62)*** (2.67)*** (0.38)*** Education 0.01 0.05 0.00 0.01 0.06 0.00 (0.003)*** (0.02)*** (0.001)** (0.005)** (0.02)** (0.003) Education2 0.0001 0.0006 0.0001 0.0002 0.0006 0.00 (0.00001) (0.001) (0.00005) (0.0002) (0.001) (0.000) Married 0.07 0.12 0.05 0.15 0.19 0.12 (0.009)*** (0.02)*** (0.01)*** (0.01)*** (0.02)*** (0.02)*** Age 0.01 0.02 0.00002 0.01 0.18 0.00 (0.001)*** (0.02) (0.001) (0.002)*** (0.02)*** (0.003) Age2 0.00005 0.0007 0.0000002 0.0006 0.004 0.00001 (0.0001)*** (0.0004)* (0.00001) (0.00002)*** (0.001)*** (0.00003) Family size 0.002 0.01 0.0000004 0.003 0.01 0.00004 (0.001)** (0.004)** (0.001) (0.001)** (0.005) (0.001) Adult males 0.0001 0.004 0.0004 0.01 0.02 0.001 (0.002) (0.009) (0.001) (0.003)** (0.011)** (0.002) Head 0.04 0.09 0.01 0.05 0.19 0.03 (0.007)*** (0.04)** (0.004)*** (0.008)*** (0.03)*** (0.006)*** Head's education 0.002 0.002 0.002 0.01 0.02 0.003 (0.001) (0.004) (0.001)** (0.002)*** (0.006)*** (0.001)** Land 0.003 0.01 0.001 (0.001)** (0.005)* (0.001) Urban 0.01 0.02 0.0004 (0.004) (0.016) (0.002) Number of observations 13,958 5,350 8,608 13,997 4,951 9,046 2 683.50 317.27 376.47 944.77 388.32 474.34 2 Pseudo R 0.2798 0.1407 0.3476 0.2930 0.1632 0.3464 Source: Authors' calculations from the 1998 Kerala Migration Study (KMS) and the 2003 South Asia Migration Study (SMS), Centre for Development Studies, Thiruvananthapuram, Kerala, India. Note: EMI, current emigrant; REM, return emigrant. Numbers in parentheses are robust standard errors. * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. 222 | J I N U K O O L A A N D Ç A GL A R Ö Z D E N ANNEX TABLE A.3. Panel Data: Multinomial Logit Estimates of the Impact of Previous Unemployment and Migration on Current Unemployment and Migration for Males Age 15­29, Kerala State, 1998 and 2003 Unemployed External emigrant Unemployed External or internal in 2003 in 2003 in 2003 migrant in 2003 Dependent variable (1) (2) (3) (4) Unemployed (1998) 0.61 (0.63 0.75 0.23 (0.228)*** (0.401) (0.260)*** (0.352) EMI household (1998) 0.21 0.45 0.15 0.15 (0.350) (0.353) (0.353) (0.319) Unemployed*EMI household 0.22 1.61 0.11 1.29 (0.646) (0.635)** (0.743) (0.596)** EMI network 1.44 20.62 1.39 11.84 (4.182) (6.137)*** (4.420) (5.063)** EMI network2 19.44 45.42 20.07 23.95 (16.853) (19.047)** (17.720) (16.114) REM (1998 or 2003) 0.35 0.04 0.30 0.20 (0.626) (0.500) (0.658) (0.492) Education 0.27 0.71 0.25 0.58 (0.103)** (0.327)** (0.107)** (0.246)** Education2 0.02 0.03 0.02 0.02 (0.006)*** (0.017)* (0.007)** (0.013)* Married 0.44 1.14 0.40 1.19 (0.562) (0.394)*** (0.657) (0.350)*** Age 0.09 0.95 0.08 0.41 (0.089) (0.444)** (0.033)** (0.318) Age2 0.00 0.02 0.00 0.01 0.61 0.63 0.75 0.23 Number of observations 2,117 2,117 2,053 2,053 2 187.65 187.65 147.17 147.17 Pseudo R 2 0.1241 0.1241 0.1023 0.1023 Source: Authors' calculations from the 1998 Kerala Migration Study (KMS) and the 2003 South Asia Migration Study (SMS), Centre for Development Studies, Thiruvananthapuram, Kerala, India. Note: EMI, current emigrant; REM, return emigrant. Numbers in parentheses are robust standard errors. Columns (1) and (2) refer to the multinomial logit in which the dependent variable takes on three values: 1 if unemployed, 2 if employed, or 3 if external emigrant in 2003. Columns (3) and (4) refer to the multinomial logit regression in which the dependent variable takes on three values: 1 if unemployed, 2 if employed, or 3 if external emigrant in 2003. All estimates are relative to the base outcome of employed in 2003. * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. E M I G R AT I O N A N D E M I G R A N T N E T W O R K S I N L A B O R M A R K E T D E C I S I O N S | 223 ANNEX TABLE A.4. Panel Data: Multinomial Logit Estimates of the Impact of Previous Job Seeking and Migration on Current Job Seeking and Migration for Males Age 15­29, Kerala State, 1998 and 2003 Job seeker External emigrant Job seeker External or internal in 2003 in 2003 in 2003 migrant in 2003 Dependent variable (1) (2) (3) (4) Job seeker (1998) 0.38 0.37 0.32 0.14 (0.246) (0.423) (0.302) (0.381) EMI household (1998) 0.24 0.83 0.22 0.21 (0.378) (0.380)** (0.378) (0.313) Job seeker*EMI household 0.44 1.29 0.27 1.20 (0.708) (0.670)* (0.854) (0.636)* EMI network 0.43 18.03 0.35 11.83 (4.345) (6.085)*** (4.532) (5.064)** EMI network2 15.99 39.74 15.67 24.02 (17.594) (18.974)** (18.235) (16.115) REM (1998 or 2003) 0.15 0.84 0.15 0.25 (0.271) (0.367)** (0.692) (0.479) Education 0.09 0.68 0.06 0.58 (0.132) (0.329)** (0.139) (0.242)** Education2 0.01 0.03 0.01 0.02 (0.007) (0.017) (0.008) (0.012) Married 0.87 1.25 1.05 1.24 (0.608) (0.404)*** (0.755) (0.349)*** Age 0.10 0.94 0.10 0.36 (0.034)*** (0.440)** (0.033)*** (0.317) Age2 0.00 0.02 0.00 0.01 (0.001) (0.011)* (0.001) (0.008) Number of observations 2,095 2,095 2,031 2,031 2 176.01 176.01 136.66 136.66 Pseudo R 2 0.1294 0.1294 0.1021 0.1021 Source: Authors' calculations from the 1998 Kerala Migration Study (KMS) and the 2003 South Asia Migration Study (SMS), Centre for Development Studies, Thiruvananthapuram, Kerala, India. Note: EMI, current emigrant; REM, return emigrant. Numbers in parentheses are robust standard errors. Columns (1) and (2) refer to the multinomial logit in which the dependent variable takes on three values: 1 if jobseeker, 2 if employed, and 3 if external emigrant in 2003. Columns (3) and (4) refer to the multinomial logit regression in which the dependent variable takes on three values: 1 if job seeker, 2 if employed, or 3 if external emigrant in 2003. All esti- mates are relative to the base outcome of employed in 2003. * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. 224 | J I N U K O O L A A N D Ç A GL A R Ö Z D E N Notes 1. Panchayats are often referred to as villages, but they are also found in cities. 2. Hinduism, Christianity, and Islam have all been established in Kerala for many centuries. Early Hinduism can be linked to the decline of Buddhism and a process of "Aryanization" in Kerala during the fifth and sixth centuries AD. The development of Islam can be traced to contacts with Arab traders in the eighth century AD. And the beginnings of Christianity are traditionally credited to Saint Thomas the apostle, in the first century AD (see Kurien 2002). 3. This figure is similar to estimations by the State Planning Board of Kerala, which deter- mined the number of emigrants in 1997 to be 1.6 million people. The estimates presented in this section are the authors' own calculations from the Kerala Migration Study (KMS) and the South Asia Migration Study (SMS), conducted in 1998 and 2003, respectively. 4. In 1973 a conference among GCC countries led to an agreement to form an oil cartel, the Organization of the Petroleum Exporting Countries (OPEC), that drove up the price of oil. By 1974 the price of oil had quadrupled (Massey et al. 1998, 137). 5. Kerala's literacy rate, at 91 percent according to the 2001 census, is the highest among Indian states. 6. In 1998, 50 percent of Gulf emigrants were Muslim. 7. Because the number of households at the panchayat level in each sample was not adjust- ed to reflect the total number of households in the panchayat, observations are weighted by the percentage of the panchayat population represented by the sample households. 8. Panchayat fixed effects are not included in the probit models because inclusion of a set of panchayat dummies would induce an incidental parameters problem (see Greene 2002). An additional problem with including panchayat dummies is that the emigration preva- lence rates are all measured at the panchayat level, which would make these variables per- fectly collinear with the panchayat dummies. References Amuedo-Dorantes, Catalina, and Susan Pozo. 2006. "Migration, Remittances, and Male and Female Employment Patterns." American Economic Review 96 (2, May): 222­26. Boyd, M. 1989. "Family and Personal Networks in International Migration: Recent Develop- ments and New Agendas." International Migration Review 23 (3): 638­70. Giles, John, and Kyeongwon Yoo. 2007. "Precautionary Behavior, Migrant Networks, and Household Consumption Decisions: An Empirical Analysis Using Household Panel Data from Rural China." Review of Economics and Statistics 89 (3, August): 534­51. Greene, William H. 2002. Econometric Analysis. 5th ed. Englewood Cliffs, NJ: Prentice Hall. Hanson, Gordon H. 2005. Emigration, Remittances, and Labor Force Participation in Mexico. INTAL Working Paper 28, Institute for the Integration of Latin America and the Caribbean, for the Inter-American Development Bank, Washington, DC. http://www.iadb.org/INTAL/aplicaciones/uploads/publicaciones/i_INTALITD_WP_28_20 07_Hanson.pdf. India. 2001. "Data on Religion," Census of India 2001. New Delhi: Office of the Registrar General. http://www.censusindia.net/. E M I G R AT I O N A N D E M I G R A N T N E T W O R K S I N L A B O R M A R K E T D E C I S I O N S | 225 Kim, Namsuk. 2007. "The Impact of Remittances on Labor Supply: The Case of Jamaica." Policy Research Working Paper 4120, World Bank, Washington, DC. http://econ. worldbank.org/external/default/main?pagePK 64165259&theSitePK 469372&piPK 64165421&menuPK 64166093&entityID 000016406_20070126111038. Kurien, P. A. 2002. Kaleidoscopic Ethnicity: International Migration and the Reconstruction of Community Identities in India. New Brunswick, NJ: Rutgers University Press. Manski, Charles F. 1993. "Identification of Endogenous Social Effects: The Reflection Prob- lem." Review of Economic Studies 60 (3): 531­42. Massey, Douglas S., Luis Goldring, and Jorge Durnad. 1994. "Continuities in Transnational Migration: An Analysis of Nineteen Mexican Communities." American Journal of Sociology 99 (6, May): 1492­1533. Massey, Douglas S., Joaquin Arrango, Graeme Hugo, Ali Kouaouci, Adela Pellegrino, and J. Edward Taylor. 1998. Worlds in Motion: Understanding International Migration at the End of the Millennium. Oxford, U.K.: Clarendon Press. McKenzie, D. J., and H. Rapoport. 2005. "Network Effects and the Dynamics of Migration and Inequality: Theory and Evidence from Mexico." Working Paper 063, Bureau for Research in Economic Analysis of Development (BREAD), Cambridge, MA. Munshi, K. 2003. "Networks in the Modern Economy: Mexican Migrants in the US Labor Market." Quarterly Journal of Economics 118 (2): 549­99. Orrenius, Pia M. 1999. "The Role of Family Networks, Coyote Prices, and the Rural Economy in Migration from Western Mexico: 1965­1994." Working Paper 9910, Federal Reserve Bank of Dallas, Dallas, TX. Osella, F., and C. Osella. 2000. Social Mobility in Kerala. London: Pluto Press. Prakash, B. A. 1998. "Gulf Migration and Its Economic Impact: The Kerala Experience." Economic and Political Weekly 33 (50, December 12): 3209­13. Taylor, J. E. 1987. "Undocumented Mexico-U.S. Migration and the Returns to Households in Rural Mexico." American Journal of Agricultural Economics 69: 626­38. Weiner, M. 1982. "International Migration and Development: Indians in the Persian Gulf." Population and Development Review 8 (1): 1­36. Winters, P., A. de Janvry, and E. Sadoulet. 2001. "Family and Community Networks in Mexico-U.S. Migration." Journal of Human Resources 36 (1): 159­84. Woodruff, C., and R. Zenteno. 2007. "Migration Networks and Microenterprises in Mexico." Journal of Development Economics 82 (2): 509­28. Zachariah, K. C., and S. Irudaya Rajan. 2004. "Gulf Revisited: Economic Consequences of Emigration from Kerala; Emigration and Unemployment." Working Paper 363, Centre for Development Studies, Thiruvananthapuram, Kerala, India. http://cds.edu/download _ files/363.pdf. Zachariah, K. C., E. T. Mathew, and S. I. Rajan. 2001. "Social, Economic, and Demographic Consequences of Migration on Kerala." International Migration 39 (2): 43­71. Zhao, Y. H. 2001. "The Role of Migrant Networks in Labour Migration: The Case of China." Working Paper N20001012, China Economic Research Center, Beijing University, Beijing. Comment on "The Role of Emigration and Emigrant Networks in Labor Market Decisions of Nonmigrants," by Jinu Koola and Çaglar Özden JEAN-LUC DEMONSANT The paper by Koola and Özden offers new insights on the functioning of migrant net- works and their impact on labor outcomes for those who stay behind. The authors present interesting empirical findings on the case of migrants from Kerala State, India, to the Arab states of the Persian Gulf. Because it is rooted in the cultural speci- ficities of the case study and is open to sociological approaches, the paper is particu- larly rich and relevant for development economists eager to avoid cultural blindness in their own research. The role of networks in migration decisions offers many per- spectives for research, as their functioning is complex and it is difficult to identify universal patterns. Cultural Biases of Earlier Studies The existing microeconomics literature on migration networks is a good example of how a focus on one geographic area can flaw the vision of development economists, in that it is clearly dominated by the case of Mexico-U.S. migration. For example, in the references to the paper by Koola and Özden, of the 10 entries on migrant net- works that mention a specific destination country in the title, 8 are on Mexico. Of course, since the United States and Mexico share the longest border between a devel- oped and a developing country, this emphasis does not come out of the blue and is perhaps legitimate. Nevertheless, the concentration on a particular case has adverse consequences for research in the field. Because theoretical economists draw their inspiration from empirical case studies, the tendency to infer general behaviors from specific cases where the context yields different outcomes is rather common. Proba- bly more dramatic, though, is the application by policy makers of "best practices" to inappropriate situations. Jean-Luc Demonsant is assistant professor of economics at the University of Guanajuato, Mexico. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank 226 COMMENT ON KOOLA AND ÖZDEN | 227 Earlier studies, working from homogenous village societies, have focused on the "quantity" rather than on the "quality" of networks. The paper by Koola and Özden remedies this deficiency by uncovering new mechanisms in play in network forma- tion and functioning. Religion as the Cement of Migration Networks Departing from the existing literature on migration networks, the authors convinc- ingly show that religious networks are more efficient than geographic ones in increas- ing the probability of migration of Keralites to the Gulf. In other studies, networks are usually measured by assessing the proportion of migrants from a geographic com- munity to a given destination. By differentiating migrants according to their religion, Koola and Özden make an important point: geographic proximity is not sufficient to bond people, especially in heterogeneous societies. In Kerala, where three major reli- gions coexist, religion is a strong identity marker. Religious communities are a central vehicle of socialization everywhere, but par- ticularly in poor countries, where social safety nets are incomplete or nonexistent and households must cope with a tough environment of severe poverty and shocks to eco- nomic and physical well-being. In order to face such an adverse situation, the house- hold cannot be isolated; it needs friends and relatives to count on. As a result, the religious community is a place of exchange in the village, as it is in migration. Indeed, economists working on networks of Mexican migrants in the United States may well become inspired by this breakthrough, as Catholicism is losing its monopoly in Mexico in favor of growing non-Catholic Christian movements. With even families now divided between Catholics and other Christians, the power of these new churches to mobilize community spirit is already indisputable. How Do Networks Operate? Koola and Özden find that the migration network has no spillover effect; having a network in country A increases the probability of migrating to that destination, but not to others. They conclude that the network is purely informative and is not used to finance migration; rather, experienced migrants provide referrals and circulate information about job offers, labor market conditions, and the like. The authors' main argument is that if the network financed migration, flows would not be desti- nation specific. I must admit that I was not fully convinced by that reasoning. Probably the diffi- culty is a matter of vocabulary; if "financing migration" means "paying for the trip," then it is true that migration would not be destination specific. But there are other migration costs, linked with relocation. The authors note that networks can provide "social support," and they do mention housing. For me, these kinds of support are also "economic." There are many accounts of established migrants paying the rent while the new migrant is getting settled. What about providing a safety net in case of misfortune such as unemployment or disease? The institution of rotating saving and 228 | JEAN-LUC DEMONSANT credit associations (ROSCAs, or, in India, chit funds) has been well documented in poor rural economies but also in migrant communities, where immigrants are more likely to be at a disadvantage in the credit market. Srinivasan (1995) describes ROSCAs in Oxford run by South Asian women. There are more studies in the Unit- ed States (where data availability is not such an issue) than elsewhere, but it may well be the case that there are Keralite ROSCAs in the Arab states of the Persian Gulf. Hence, without directly financing the migration, the migrant network finances the settlement of the new migrant and provides him or her with informal social security. Because settled migrants will be more in a position to lend a hand, they will be the main providers and will therefore finance this informal insurance at the time the new- comers need it the most. In turn, the new migrants will finance the next newcomers and thus form a chain of reciprocity with subsequent waves of migrants. Religious Clustering of Minorities Another aspect of the authors' findings is that Hindu networks have the highest impacts on both the decision to migrate and the choice of location, followed by Christian and then Muslim networks. The authors suggest that Hindus are more reliant on their networks than are the other groups. They mention that Muslims are more religiously compatible in the Arab Gulf states and that Christians have an educational advantage. I would suggest two further explanations. A first, noneco- nomic, one is that non-Muslims are clearly minorities in the Arab Gulf states. Religious practices are made easier when population is concentrated. The building of churches and Hindu temples, for instance, requires a minimum number of believers. Furthermore, some claim that freedom of religion would not be totally guaranteed for non-Muslims in some areas of the Gulf. The channeling effect for non-Muslims would then also be explained by the fact that a larger group is better able to defend itself. The second, complementary, explanation stems from the network externalities. As depicted by the authors in their study on geographic network effects, the rela- tionship between the stock of migrants in a given place and the probability of migrating is nonlinear. There are, in fact, two opposite effects: the stock of migrants reduces the relocation cost, but it increases competition on the labor mar- ket and thus decreases the incentive for migration. When the stock of migrants is still low, the positive network externality dominates; when the stock is high, the negative wage effect dominates. The authors observe that although Muslims rep- resent only one-fourth of Kerala's population, they make up close to half of the total emigrant population in the Gulf states. The proportion is not nearly as high for Christians and Hindus, which would explain the difference in the effects: Mus- lim migration is at a stage at which the competition effect is very high, whereas the positive network externality still strongly dominates for both Hindus and Chris- tians. Only longitudinal and historical data, since the 1970s, would allow us to ascertain whether this is relevant. In the results presented, with the data at hand, it appears that "own religious network" effects diminished slightly between 1998 COMMENT ON KOOLA AND ÖZDEN | 229 and 2003, but the differences are not statistically significant, and five years is, anyway, too short a time to detect such a pattern. Extended Family Networks When one looks carefully at the regressions concerning the migration decision, it appears that for a Muslim, having relatives who are migrants is (at least) as impor- tant as having other Muslims from the same village on migration. This does not hold for Christians or Hindus. This result could be interpreted as an indication that the family is taking over from the religious network among Muslims. If there is more competition on the labor market, it becomes more costly to help new migrants because of the unemployment risk. When times are rough, family represents a safer resort. There is a large body of literature documenting how the extended family pro- vides all kinds of insurance and protection against external events; Cox and Fafchamps (2007) offer a complete and up-to-date review. Additional information on the functioning of Keralite households would have been required in order to investi- gate the role of family networks, and this is beyond the scope of the paper. Unemployment of Potential Migrants? The authors obtain noteworthy results on employment decisions of nonmigrants. In a first econometric study, they show a standard outcome: young people are more like- ly to be unemployed in migrant households. Using the panel structure of the data, they show that the young unemployed in migrant households in 1998 had a higher probability of being on migration abroad in 2003. There seems, however, to be a selection bias because of an attrition problem that could potentially dampen this result. Suppose a young man was unemployed in 1998 but got married and formed a new home by 2003. Recall that in the sample regressions, persons who were between ages 10 and 24 in 1998 were between ages 15 and 29 in 2003; indeed, since students were excluded, the young persons were likely to have been between ages 14 and 24 in 1998 and therefore between ages 19 and 29 in 2003. This seems to be a reasonable marriage age range. The bias would stem from the following issue: is this individual still present in the sample, since he cannot be traced in 2003, when he is no longer part of "the initial household"? In the data presentation the authors mention that it is the households that are resurveyed, not the household members. In reality, we do not know the sit- uation of this particular young man in 2003. Has he found a job? Has he migrated? Leaving the parents' household and finding a job are likely to be linked. If it is the case that this young man is no longer in the sample and that he has not migrated, then the link between "unemployed in 1998" and "on migration in 2003" is biased upward. In other words, it is not to be taken for granted that unemployed young per- sons in migrant households are actually potential external migrants. They may still be in Kerala and may simply have left the initial household to form a new one. 230 | JEAN-LUC DEMONSANT Perhaps this comment is not relevant, as the issue may have been taken into account in the analysis. Nevertheless, it underscores the fact that I found no mention of the definition of "households" or "household members" in the paper. Are house- holds mainly nuclear families, or extended ones? Concerns about change in house- hold composition between the two dates may be quite relevant and decisive in the former case. To conclude, research on migrant networks is a challenging field of investigation for econometricians because it is data consuming. A first-best option would be to have data on both sides: at home and for the destination country. Such a data col- lection process is extremely difficult to implement and is therefore expensive, but it is rewarding. I surmise that the authors would agree with me that it would definite- ly be of great help in better grasping the migration process. Facing a second-best sit- uation, as development economists are trained to do, the authors fared very well in convincingly highlighting the importance of religious networks. Finally, I have to say that this article is particularly useful and relevant for such a conference. Aimed at a large audience, from policy makers to professional and aca- demic economists, it shows that noneconomic factors often play an important part in individuals' choices. A priori, religion has little to do with migration decisions, or at least it does not seem to be of such importance. Development economists have a tendency to overlook those issues, though times may be changing; as a recent study by Rao and Walton (2004) shows, interdisciplinarity is in the air in academia, as at the World Bank. References Cox, Donald, and Marcel Fafchamps. 2007. "Extended Family and Kinship Networks: Economic Insights and Evolutionary Directions." Handbook of Development Economics, vol. 4, ed. T. Paul Schulz and John Strauss, 3711­84. Amsterdam: North-Holland. Rao, Vijayendra, and Michael Walton. 2004. Culture and Public Action. Stanford, CA: Stan- ford University Press. Srinivasan, Shaila. 1995. "ROSCAs among South Asians in Oxford." In Money-Go-Rounds: The Importance of Rotating Savings and Credit Associations for Women, ed. Shirley Ardener and Sandra Burman. Washington, DC: Berg Publishers. Higher Education and High-Technology Industry The Role of Higher Education in High-Technology Industrial Development: What Can International Experience Tell Us? SACHI HATAKENAKA An extraordinary level of policy attention is being accorded higher education today. The reason is clear: higher education is considered one of the key factors for the development of science-based high-technology industries. This paper shows that the various types of higher education institutions play significantly different roles, which can influence the shape and nature of high-tech industry. The study further proposes a classification framework for institutions to illuminate such a difference. Drawing on experience in a variety of economies (the United States; Japan; Finland; Taiwan, China; the Republic of Korea; Ireland; Israel; China; and India), three separate dimensions are proposed for differentiating institutional characteristics: responsiveness, fundamental science orientation, and selectivity. Finally, the paper examines how the three dimensions developed in specific country contexts and suggests that early interventions for developing responsiveness may be particularly important. Higher education commands an extraordinary level of policy attention today. The reason is clear: with the rise of knowledge economies and science-based high- technology industries, higher education is considered crucial to economic development. Experience shows that high-tech industry can emerge in highly dissimilar countries, including those with limited industrial bases or low levels of economic development such as China, India, and Ireland. It is not surprising that innovation policies to create and support high-tech industries are a priority in almost all member countries of the Organisation for Economic Co-operation and Development (OECD) and in many developing countries. Development of an appropriate higher education sector is becoming a critical component of such policies. Sachi Hatakenaka is an independent researcher. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank 233 234 | S A C H I H ATA K E N A K A Much has been written about the role of American universities in high-tech indus- tries and more generally in industrial development, and some studies compare the roles of U.S. institutions with those in Europe or Japan. Studies in other countries conclude that the contributions of well-trained engineers and scientists--and, by implication, higher education systems--are significant. The comparative literature on higher education describes similarities and differences among national systems, but there tends to be little emphasis on their roles in economic development. Lester (2005), in a comparative study of several countries, argues that different types of industrial transformation require different contributions by universities. There is also an emerging literature on university-industry relationships in developing countries (Chen and Kenney 2007; Ma 2007; Wu 2007; Yusuf and Nabeshima 2007; Kroll and Leifner 2008), and the role of universities in economic development is beginning to be discussed in a more specific way (Mazzoleni and Nelson 2007; Leifner and Schiller 2008). Mazzoleni and Nelson provide an in-depth analysis of the role of public research in economic catch-up and argue for the increasing importance of such research, partic- ularly in application-oriented sciences and engineering. They also provide a helpful overview of the nature of American research universities and their economic roles. Leifner and Schiller contribute to the debate by introducing the notion of "academic capabilities" to help with thinking about how universities contribute to the development of technological capacity. However, few studies examine how different types of higher education institutions may contribute differently to the development of high-tech indus- try and how institutional characteristics influence the nature of their contribution. The objective of the paper is to fill this gap in the literature by proposing a classi- fication framework for higher education institutions. More specifically, the paper seeks answers to three questions: 1. What are the characteristics of higher education institutions that define the nature of their contributions to high-tech industry? 2. What roles do these institutions play in the development of high-tech industry? 3. How do institutional characteristics develop? The ultimate goal is to draw lessons for the future for higher education systems, par- ticularly in developing countries. This paper is based on a review of experience in a range of countries. Wherever possible, country-specific literature on higher education systems and their develop- ment is drawn on. The main emphasis is on two high-tech industrial sectors of relatively recent date: information and communications technology (ICT), both hard- ware and software, and biotechnology. The point of this paper is not to fully contrast the paths taken or to examine all the contributing factors in the development of high-tech industry. That a variety of factors was behind the take-off of high-tech industry is accepted as a given. The paper focuses on the specific roles higher education has played within that context. It exam- ines differences in capabilities for high-tech industry among countries, but only to the extent that they are related to the roles of higher education systems. HIGHER EDUCATION IN HIGH-TECH INDUSTRIAL DEVELOPMENT | 235 In order to explore the variety of roles played by higher education at different stages of industrial development, four sets of economies are examined: · The United States, which has held a leading position in both sectors--ICT and biotechnology--from their very beginning · Japan and Finland, which entered the global competition in high-tech products on a foundation of significant existing industrial capability · A group of late developers, including the Republic of Korea and Taiwan, China, that were building industrial capability around the time these industries were born · A group that developed high-tech industry from a much less developed industrial base (China, India, Ireland, and Israel). The paper attempts to draw on comparative historical data, to the extent they exist, so that comparisons can be made not only geographically but also over time. Framework for Understanding Institutional Characteristics In this section, a framework for classifying institutions is proposed, using three dimensions that measure institutional characteristics: responsiveness, basic science orientation, and selectivity (figure 1). Responsiveness refers to institutions' ability and inclination to respond to changing practical and industrial needs. Basic science orientation measures the level of commitment to fundamental science. Selectivity has to do with the stringency of the criteria for recruiting students and staff and distin- guishes elite institutions from those that are open to people more widely. These three dimensions emerged in the course of the review of the various cases, but also from prior academic work. The first two dimensions, responsiveness and basic science orientation (shown in figure 2), owe much to the work of Stokes (1997), FIGURE 1. Three Dimensions of a Higher Education Institution basic science orientation selectivity responsiveness Source: Author's elaboration. 236 | S A C H I H ATA K E N A K A FIGURE 2. Institutional Characteristics: Responsiveness and Basic Science Orientation responsiveness to economic needs NO YES YES basic relevant research research basic science institutions institutions orientation teaching-focused practically NO institutions oriented institutions Source: Adapted from Stokes (1997). who proposed two similar dimensions to distinguish between four different types of research orientation. Figure 2 shows the framework to differentiate between four types of institutions using these two dimensions. The top right-hand cell is occupied by relevant research universities, which con- duct fundamental research, but with an eye to its relevance to society. These institu- tions are committed to fundamental science but are also concerned about being useful, and they strive to remain responsive to socioeconomic needs. Surprisingly, not many institutions fit this category. A primary example is a group of American research universities that in their development embraced the value of relevance (Rosenberg and Nelson 1994; Geiger 2004a; Mazzoleni and Nelson 2007). When these universities became the envy of the world because of their contributions in high- tech industry such as biotechnology and ICT, the world initially looked to all American universities as if they were all the same. Over time, it became clear that even within the American system there were different types of university and that only certain research universities were proactive in industrial innovation in general and in the development of high-tech industry in particular. Some Israeli universities also appear to belong to this cell, although perhaps with less emphasis on basic science and more on applied research. The top left-hand cell is occupied by basic research universities, which are driven principally by the core values of fundamental science, with little interest in or capac- ity for responding to external needs. This is where the classic ivory-tower universities with well-developed research capabilities belong; indeed, the great majority of research universities in the world have belonged to this cell--at least until recently, when economic relevance became a global catchphrase. These institutions value cutting-edge science, reward scientific discoveries, and encourage their academics to define actively the direction of scientific research. Scientific developments were to be independent of practical requirements, which were regarded as a "corrupting" influ- ence on science. Indeed, until the 1990s, this cell represented what most higher education institutions aspired to be. Many Japanese national universities developed to fit into this category in spite of their origins as practically oriented universities, as is discussed next. HIGHER EDUCATION IN HIGH-TECH INDUSTRIAL DEVELOPMENT | 237 The bottom right-hand cell contains practically oriented institutions that aspire to meet the economy's needs for practical skills and knowledge. They offer courses rel- evant to employers and often conduct consulting and application-oriented research with and for employers. They have little interest in, or capacity for, furtherance of generic scientific knowledge or fundamental research. Examples are diverse; they range from the grandes écoles in France, which were designed to provide elite pro- fessional education (although some have now begun to develop basic research capac- ity), and many polytechnic schools and their equivalents in other countries. Ireland and Finland reformed their higher education sectors by introducing this type of insti- tution. Historically, the Japanese national universities would have belonged to this cell at their founding, as they were established to provide key technological skills needed for national modernization, but their commitment to bring science into prac- tice place them higher in the quadrant, in comparison with traditional trade schools, which focus on practical skills. Some Chinese universities have roots in this cell, and others appear to have moved rightward toward this cell over the last 20 years, becoming much more economically relevant. Some institutions fit this cell naturally by virtue of their emphasis on an application-oriented subject mix; engineering insti- tutions such as the Indian Institutes of Technology are good examples. The bottom left-hand cell is occupied by teaching-focused institutions, which are neither practically oriented in their teaching nor research oriented. U.S. liberal arts col- leges are good examples because they are committed to offering broad curricula, with an emphasis on generic skills rather than vocational or professional content. They define their identity in terms of their commitment to teaching and conduct hardly any research. (Some teaching institutions, however, are more academically oriented.) Most universities in developing countries fall into the teaching-focused category by default simply because they do not have the resources to be active in research. In many such cases their teaching is defined according to academic discipline. The third dimension, selectivity, differentiates further the nature of the institutions in each of the four cells. The more selective the institutions are in screening incoming students, the more elitist they are. Grandes écoles are differentiated from the bulk of polytechnics in that they are among the most highly selective institutions, which most polytechnics are not. Selectivity is often a system-level characteristic; some higher edu- cation systems are explicitly selective, characterized by highly competitive entrance examinations to allocate talents across institutions, as in Japan and Korea. Other sys- tems are much less so and are based on the notion that all universities command equal status; this is the tradition in Germany and many other European countries. As will be discussed later, selectivity appears to be an important characteristic that strengthens the roles institutions may play. Roles of Different Types of Institution This section describes how different types of institution contributed to the develop- ment of high-tech industry, drawing on the experiences of the countries listed above. One important observation is that none of the institutional categories appears to be 238 | S A C H I H ATA K E N A K A necessary for the development of high-tech industry; rather, they play different roles. The footprints of university roles are sometimes visible in the shape and nature of the resulting industry, but at other times industry developed while incorporating univer- sity contributions in a way that became almost invisible. Relevant Research Universities It is well established that some American research universities contributed significantly to the birth and development of biotechnology and the computer industry. For instance, fundamental scientific discoveries in university laboratories and the resulting patents (such as Cohen's and Boyer's discovery of recombinant DNA technique) led to the development of biotechnology as a whole new industry with strong linkages with academic scientific research (Henderson, Orsenigo, and Pisano 1999). The early evo- lution of the biotechnology industry was characterized by the formation of new firms, many of which were academic spin-offs; usually, academics worked in conjunction with professional managers, backed by venture capital. The early academic inventors were labeled "star scientists" because they were key players both in basic science and in its applications. They had a special role mainly because of the significant amount of tacit knowledge associated with the original scientific discoveries (Zucker and Darby 1996; Zucker, Darby, and Brewer 1998; Zucker, Darby, and Armstrong 2002). The norms in American universities that permitted academics to become deeply engaged in commercialization--which had developed early, given their culture of striving to be useful--was crucial in this process (Henderson, Orsenigo, and Pisano 1999; Etzkowitz 2002). American universities were also important to the development of the computer hard- ware and software industries, although their visibility varied over time (Rosenberg and Nelson 1994; Bresnahan and Malerba 1999; Mowery 1999). For instance, universities were actively engaged in the early development of computers in the 1940s. (During this period, European universities such as those in Manchester, Cambridge, and Berlin were as active as their American counterparts.) They were also important in the development of microcomputers, as exemplified by spin-offs from the Massachusetts Institute of Technology, Stanford University, and the University of Texas. In Europe, only Cambridge University in the United Kingdom played a similar role in microcomputers. There is evidence that industry values the more conventional channels of contri- butions made by universities. One survey found that industry regarded general research findings, or instrumentation and techniques developed through research, as more important university contributions than prototypes (Cohen, Nelson, and Walsh 2002). The same survey found that publications, meetings and conferences, informal interaction, and consulting were more important than patents as mechanisms through which industry gained access to university knowledge. Furthermore, industry reported that university research was helpful in solving problems, rather than in inspiring and triggering new research. These results concern contributions made by all types of American universities, including basic research institutions. Nonetheless, it is important to note that what is valued is not so much application capabilities in universities as relevant research capability. HIGHER EDUCATION IN HIGH-TECH INDUSTRIAL DEVELOPMENT | 239 There is also evidence that the development of the software industry was greatly facilitated by the early establishment of computer science as a new discipline in uni- versities (Mowery 1999). Universities contributed not only through formation of key knowledge but also through organizing and delivering education programs to supply updated skills. Indeed, American universities' ability to create and legitimate com- puter science as a new field was unparalleled by European or Japanese universities (Mowery 1999). In Israel, similarly, universities supplied highly skilled human resources and served as a source of leading technologies not only in ICT but also in biotechnology and agriculture-related research (de Fontenay and Carmel 2004; Breznitz 2007). From early on, the government made significant public investments in application-oriented scientific research in universities, public research institutions, and the military. Israeli universities became well-established scientific centers early, and today Israel is high in global rankings for key metrics such as the number of publications per capita and relative citation indexes (tables 1 and 2). But they were application oriented, as well; university-industry relationships had long been well established (Breznitz 2004), and three universities ranked among the top six patent holders in the country (Trajtenberg 2001; Breznitz 2004). Israel became one of the first research and development (R&D) sites abroad for U.S. multinational companies, with Motorola establishing a research unit in 1963 (de Fonteney and Carmel 2004). The contribution of the universities tends to be overshadowed by that of the military in the narrative of Israel's TABLE 1. Science and Engineering Articles, by Economy, 1995­2005 (number of articles, except as otherwise specified) Average annual change (percent) Per million 1995­ 2000­ 1995­ Economy 1995 2000 2005 population 2000 2005 2005 China 9,061 18,479 41,596 31 15.3 17.6 16.5 Finland 4,077 4,844 4,811 914 3.5 ­0.1 1.7 India 9,370 10,276 14,608 13 1.9 7.3 4.5 Ireland 1,218 1,581 2,120 502 5.4 6.0 5.7 Israel 5,741 6,290 6,309 926 1.8 0.1 0.9 Japan 47,068 57,101 55,471 434 3.9 ­0.6 1.7 Korea, Rep. 3,803 9,572 16,396 341 20.3 11.4 15.7 Taiwan, China 4,759 7,190 10,841 474 8.6 8.6 8.6 United States 193,337 192,743 205,320 678 ­0.1 1.3 0.6 Source: Thomson Scientific, Science Citation Index (SCI) and Social Sciences Citation Index (SSCI), http://scientific.thomson.com/products/categories/citation/; ipIQ, Inc.; and National Science Foundation, Division of Science Resources Statistics, special tabulations, as published in Science and Engineering Indicators 2008; population data from World Bank database, accessed May 7, 2008; National Statistics, Taiwan, China, http://eng.stat.gov.tw/ lp.asp?ctNode=2265&CtUnit=1072&BaseDSD=36. Note: Article counts are for the set of journals covered by the SCI and SSCI. Articles are classified by year of publica- tion and are assigned to economies on the basis of the institutional addresses listed in the article. For articles where collaborating institutions are from multiple economies, each economy receives fractional credit on the basis of the proportion of the participating institutions. 240 | S A C H I H ATA K E N A K A TABLE 2. Relative Prominence in Scientific Literature, Selected Economies, 1995 and 2003 (index: 1.00 indicates that an economy's share of cited literature equals its world share of scientific literature) Rank Economy or group 1995 2003 1 Switzerland 1.189 1.152 2 United States 1.013 1.026 5 United Kingdom 0.830 0.864 8 Finland 0.755 0.826 12 Ireland 0.662 0.764 14 Israel 0.682 0.742 15 EU-15 0.681 0.737 19 Hong Kong, China 0.438 0.672 22 Japan 0.551 0.575 24 Singapore 0.422 0.509 26 Chile 0.343 0.478 28 Korea, Rep. 0.320 0.439 29 Thailand n.a. 0.432 30 Czech Republic n.a. 0.417 31 Argentina 0.304 0.411 32 Slovenia n.a. 0.409 33 South Africa 0.325 0.404 34 Taiwan, China 0.321 0.401 35 Mexico 0.349 0.385 36 Poland 0.295 0.360 37 Brazil 0.314 0.359 38 Slovak Republic n.a. 0.339 40 Bulgaria 0.166 0.325 41 China 0.218 0.293 42 India 0.182 0.284 43 Croatia 0.257 0.281 44 Turkey 0.214 0.278 45 Saudi Arabia 0.220 0.256 Source: Thomson ISI, Science Citation Index (SCI) and Social Sciences Citation Index (SSCI), http://www.isinet.com/ products/citation/; ipIQ, Inc.; National Science Foundation, Division of Science Resources Statistics, special tabula- tions, Science and Engineering Indicators 2006. Note: n.a., not applicable; EU-15 refers to the 15 countries of the European Union before the expansion in member- ship of May 1, 2004. The relative prominence of scientific literature is measured on the basis of the relative citation index of an economy, that is, the economy's share of cited literature adjusted for its share of published literature. The economy's citation of its own literature is excluded. An index of 1.00 indicates that the economy's share of cited liter- ature equals its world share of scientific literature. An index greater than 1.00 or less than 1.00 indicates that the econ- omy's literature is cited relatively more or less, respectively, than its share of scientific literature. Economies with a share of world publications in the cited field of less than 10 percent during the period are either excluded or are listed as n.a. In the case of cited articles with collaborating institutions from multiple economies, each economy receives fractional credit on the basis of the proportion of its participating institutions. high-tech industry, but they have contributed to the emergence of an entrepreneurial sector, both directly and indirectly, through spin-offs and through science parks, some of which had been established as early as the 1960s (Roper and Grimes 2005). It is no accident that the resulting high-tech industry in Israel was highly R&D- oriented and had cutting-edge technological content. This poses a sharp contrast to HIGHER EDUCATION IN HIGH-TECH INDUSTRIAL DEVELOPMENT | 241 other emerging economies. Whereas the software industry in India and Ireland largely grew by providing software services, Israel's software industry developed on the basis of niche software products with global applications. Israel's innovation capabilities, as gauged, for example, by its patenting performance, lags behind that of Japan and the United States but is comparable with that of Finland and Taiwan, China (Trajtenberg 2001). Israel's ICT industry displays a marked concentration on upstream R&D, compared with the Finnish ICT industry, which covers a wider value chain ranging from R&D and production to marketing and distribution, or with Ireland's, which has hardly any R&D (Roper and Grimes 2005). In summary, relevant research universities appear to contribute by · Conducting basic research in fields pertinent to developments in industry and dis- seminating such knowledge through a variety of activities such as patenting and licensing, academic spin-offs, consulting, and advising · Providing updated skills in areas relevant to industry, both in undergraduate education and in advanced research training for masters and PhDs. Basic Research Universities Basic research universities are characterized by an academic ethos of scientific auton- omy, and it would not be surprising to find that they are less proactive than relevant research universities. Nonetheless, they are capable of supporting the development of high-tech industry with high technological content, through both research and edu- cation. One example is postwar Japan, where a group of national and some private universities developed to become basic research universities. Their contributions to the development of high-tech industry have not been very visible, and Japanese pol- icy makers and industrialists were often critical of them because they were deemed to be less responsive than their American peers. There is, however, consistent evidence that Japanese universities' contributions are underrated and that they have played a different but "important" role as compared with their American peers (Yoshihara and Tamai 1999; Pechter and Kakinuma 1999; Kodama and Suzuki 2007). Some analysts have coined the term "receiver-active paradigm" to describe this technology transfer phenomenon; it was the technology-receiving industry, rather than the uni- versities, that actively defined how and when technology transfer took place (Kodama and Suzuki 2007). Indeed, Japanese companies have been extremely active in using universities to acquire new scientific knowledge. An unusual mechanism that brought university academics and industry together was industries' practice of sending their staff to uni- versities, not for any specific education programs but, more generally, to study, some- times for a whole year. This practice was often motivated by a desire to learn about fields of science in which industry had limited expertise (Hicks 1993; Darby and Zucker 1996). Accordingly, in biotechnology, "star scientists" worked with company employees in their university laboratories, resulting in higher levels of coauthorship than in the United States (Darby and Zucker 1996). Other researchers also find 242 | S A C H I H ATA K E N A K A that Japanese academics' relationships with companies are equivalent to those pre- vailing in the United Kingdom and the United States, at least as measured by coau- thorships (Pechter and Kakinuma 1999), but that companies had to be proactive in eliciting scientific and technological information from the universities (Kodama and Suzuki 2007). Japanese universities have been much less active in directly engaging in the com- mercialization of science. In the early development of biotechnology, Japanese uni- versities played a negligible role in forming new companies in Japan, in striking contrast to the visibility and dominance of academic spin-offs in the United Kingdom and the United States (Henderson, Orsenigo, and Pisano 1999). This was partly because of the unevenness of Japanese science, at least at the outset; there were liter- ally no researchers in Japan working in genetic engineering at the time of the scien- tific breakthroughs (Darby and Zucker 1996). But even when they did catch up in science, universities as organizations did not permit their civil servant "star scientists" to work in the same way in new companies (Darby and Zucker 1996). Similar passive, albeit important, roles are also evident in education. Japanese universities made a significant contribution by supplying large numbers of engi- neers and scientists at the undergraduate and master's levels, something that was essential for the birth and development of the electronics industry. A group of highly selective Japanese national universities with large science and engineering faculties was particularly important in producing high-caliber scientists and engi- neers. Companies, however, had to develop significant capability for imparting the necessary skills to graduates through extensive in-house training. A form of division of labor developed whereby universities selected students and provided general academic training, while companies carried out the necessary specialist training, even for R&D personnel. Moreover, it was not universities that proposed massive expansion in science and engineering programs; it was government that, in response to industrial demand, planned such an expansion through funding. Indeed, Japanese universi- ties were generally slow to respond to emerging needs for industrial skills in, for instance, new areas such as software-related engineering, semiconductors, and biotechnology (Baba, Takai, and Mizuta 1996; Darby and Zucker 1996; Mowery 1999; Kobayashi 2001). They were also slow in establishing relevant research training. Although master's programs in engineering had become commonplace by the 1980s, doctoral programs in science and engineering remained small and unattractive for industry (Clark 1995; Shimizu and Mori 2001; see also table 3). PhDs from universities were too narrowly specialized and too academically ori- ented for industry. Instead, industry recruited bright potential researchers with less academic training and trained them in-house, using the unique Japanese system in which doctorates were awarded by universities for dissertations that were written mainly in the companies, without detailed supervision by academics (Shimizu and Mori 2001). The companies took the extreme route of developing in-house train- ing to cultivate doctoral-level researchers. In summary, basic research universities can contribute to high-tech industry in developing scientific capacity and in providing scientific information, although in the HIGHER EDUCATION IN HIGH-TECH INDUSTRIAL DEVELOPMENT | 243 TABLE 3. Total Enrollment, Advanced Research (PhD) Programs, Selected Economies and Selected Years, 1970­2006 (numbers enrolled, except as otherwise specified) Percent of Economy 1970 1980 1990 2000 2003 2006 total, 2006 China -- -- -- 54,038 108,737 -- -- Finland -- -- -- 19,750 19,846 22,145 7.2 India -- -- -- 55,019 65,357 36,519 -- Ireland -- -- -- 2,904 3,816 5,146 2.8 Israel -- -- -- 6,647 7,944 9,715 3.1 Japan 13,243 18,211 28,354 59,007 68,245 75,028 1.8 Korea, Rep. -- -- -- 28,924 34,712 41,055 1.3 Taiwan, China 166 673 4,437 13,822 21,658 29,839 2.3 United States -- -- -- 293,002 306,889 388,685 2.2 Source: UNESCO database accessed May 7, 2008; Education Statistics, Ministry of Education, Science and Technology, Japan; Education Statistics, Ministry of Education, Taiwan, China. Note: --, not available. UNESCO, United Nations Educational, Scientific, and Cultural Organization. case of Japan this required proactive work on the part of industry. Such universities can also provide large numbers of well-educated scientists and engineers, even though the graduates' training may not be fine-tuned to specific industrial technological needs. The great selectivity of Japanese universities was helpful for industry, which actively used such selectivity for identifying high-caliber talents. Practically Oriented Institutions Practically oriented institutions are unique in that they offer specialized engineering and science skills that are directly relevant to industrial needs. In Finland and Ireland professional institutions were created as an alternative to conventional university education, which was seen as unresponsive to industrial needs. Institutional respon- siveness was particularly important when new disciplines such as computer science emerged, and it appeared to give countries a competitive edge, as was clear in the case of Ireland. In Ireland, the government-led expansion of technically trained manpower in the 1970s and 1980s was important in attracting multinational corporations (MNCs), including ICT enterprises. The starting point was a consensus that emerged early among policy makers that technical education was essential but that universities were overly academic and that a different type of institution was needed to provide the critically needed technical manpower (White 2001). Thirteen regional technical col- leges and two national institutes of higher education were established in the 1970s for this purpose, representing the bulk of expansion in the tertiary sector. They were established specifically to be responsive to economic needs, and they are known today as having well-established practices in, for instance, assessing industrial needs and obtaining industrial inputs in curricular content (Breznitz 2007). 244 | S A C H I H ATA K E N A K A Finland also made a decisive move to create a set of responsive institutions when it established 29 polytechnics in the 1990s. By 2000, nearly 60 percent of entrants to higher education were going to polytechnics. Although it is still early to assess the full impact of polytechnics in Finland, their early performance reviews have been positive, and there are indications that their expansion coincided with the timing of major industrial needs, particularly in telecommunications (OECD 2003). In one survey of high-tech industry in northern Finland, 47 percent of companies said that poly- technics were important, compared with 38 percent saying the same of a university with a known regional orientation, and with 10­20 percent for other universities in Finland (Juahiainen 2006). Both Korea and Taiwan, China, invested in application-oriented research capacity in government research institutes, which had a significant part in developing domestic technological capacity within high-tech industry. The fact that such capacity was built outside universities is perhaps not surprising, given that it is faster and easier to concentrate resources in dedicated institutions. Such decisions did, however, have con- sequences, particularly in the case of Korea, in delaying the development of research capacity in universities, with the notable exception of the Korea Advanced Institute of Science (KAIS), discussed below. The experience of Taiwan, China, with high-tech industry cannot be described without citing the story of the Electronic Research Service Organization (ERSO), a public research institution. ERSO is a branch of the Industrial Technology Research Institute (ITRI) and was at the heart of the economy's acquisition of technological know-how in semiconductors. ERSO orchestrated the first major technology trans- fer agreement, with RCA in the 1970s. It subsequently developed an integrated cir- cuit (IC) fabrication technology and established a spin-off company that in turn was important in the diffusion of technological know-how (Wade 1990; Saxenian 2004). The industry in Taiwan, China (unlike the cases of Japan and Korea), included many small-to-medium-size enterprises that lacked the capability to undertake significant R&D in-house, and public investment in applied research was essential for their technological upgrading (Wade 1990). The Korean experience of investing in application-oriented research capacity in government research institutes is somewhat less visible, but it has been important in building corporate research capacity. An example is the Korea Institute of Science and Technology (KIST), established in 1966 and staffed by Korean scientists and engineers brought back from abroad, one of the early efforts to reverse the brain drain. The institute was designed to undertake contract research with both industry and government and was particularly active in facilitating technology transfer from foreign firms to local firms (Mazzoleni and Nelson 2007). KIST provides an inter- esting and exceptional example because it later became amalgamated with the Korea Advanced Institute of Science (KAIS), a graduate-level educational institution estab- lished in the 1970s specifically to fill an important gap in Korean higher education by creating high-level skills of industrial relevance (Mazzoleni and Nelson 2007). It is thus possible, even though not common, for research capacity developed in gov- ernment research institutes to contribute directly to the evolution of higher education institutions. HIGHER EDUCATION IN HIGH-TECH INDUSTRIAL DEVELOPMENT | 245 Teaching-Focused Institutions In the early stage of development of high-tech industry, companies often recruited generic science and engineering graduates who were not necessarily trained in a spe- cialization relevant to the industry. In China, India, Japan, Korea, and Taiwan, China, these graduates mainly came from teaching-focused institutions because many institutions simply had not yet developed research capacity. All these governments emphasized science and engineering. Beyond this general statement, there is a signif- icant divergence in what happened in each case. In Japan the government, in response to growing industrial demand in the late 1950s through the early 1960s, aggressively promoted science and engineering under- graduate programs not only in national universities but also in private universities, which were largely teaching oriented (Nakayama 1995). Most private universities in Japan had limited resources for undertaking science and engineering programs or research. It was through special government subsidies, which provided up to half of undergraduate student costs and two-thirds of the associated investment costs in science and engineering, that private universities managed to establish programs in these disciplines (Nakayama 1995). This push no doubt helped Japan produce a large number of scientists and engineers; students in these fields constituted 27­28 percent of all undergraduates in the late 1960s through the 1980s (table 4). Korea and Taiwan, China, made earlier moves toward emphasizing science and engineering relative to their industrial development, which resulted in massive brain drains. In Korea more than a third of undergraduate enrollments were in science and engineering by the mid-1960s (table 4), and the proportion increased to over 40 percent in the subsequent period in spite of the great expansion of the higher education sector that was taking place. Korea today has one of the world's highest TABLE 4. Undergraduate Science and Engineering Enrollments, Japan and the Republic of Korea, Including Agriculture and Excluding Medicine, Selected Years, 1960­2005 Country and indicator 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 Japan Number enrolled 136,818 238,596 375,598 443,180 451,904 463,336 524,201 611,351 625,371 590,549 Percent of total enrollment 23 27 28 27 26 27 26 26 25 24 Korea, Rep. Number enrolled -- 37,099 59,264 74,410 166,137 337,624 419,891 523,002 -- -- Percent of total enrollment -- 35 34 36 41 36 40 44 -- -- Source: Japan, MEXT, 2006; Weidman and Park 2000. 246 | S A C H I H ATA K E N A K A tertiary enrollment ratios, at 91 percent (table 5), and one of the highest proportions of science and engineering students (table 6). In Taiwan, China, the proportion of sci- ence and engineering enrollments was about a quarter through the early 1970s and gradually increased from the mid-1970s through the 1990s to over 40 percent (table 7). In contrast, in 2005 India's engineering students made up only 6 percent of total ter- tiary enrollments, and even when scientists are added, the proportion is significantly lower than for other economies, at 20 percent (see table 6). It is not that all these countries had high proportions of science and engineering in higher education. TABLE 5. Tertiary Gross Enrollment Ratio, Selected Economies, Selected Years, 1970­2006 Economy 1970 1975 1980 1985 1990 1995 2000 2005 2006 China 0 1 2 3 3 5 8 -- 22 Finland 13 28 32 34 48 70 83 92 93 India 5 5 5 6 6 7 10 11 12 Ireland 12 17 18 22 31 40 49 58 59 Israel 18 23 29 33 36 41 50 58 58 Japan 18 26 31 28 31 42 47 55 57 Korea, Rep. 7 9 15 34 39 52 73 90 91 Taiwan, China -- -- -- -- -- 39 56 82 84 United States 47 55 56 60 72 81 69 82 82 Source: UNESCO database; World Bank database accessed May 7, 2008; Education Indicators, Ministry of Education, Taiwan, China. Note: UNESCO, United Nations Educational, Scientific, and Cultural Organization. The gross enrollment ratio is the total number of students in tertiary education, including graduate school, as a percentage of the population in the five-year age group following on from secondary school­leaving age. TABLE 6. Total Tertiary Enrollment in Science and Engineering, Selected Countries, 2005 Percent of total enrollment Science Science/ Enrollment and engineering Economy Science Engineering Total Science Engineering engineering ratio a China 1,443,129 5,536,123 18,352,821 8 30 38 0.3 Finland 35,468 80,827 305,996 12 26 38 0.4 Indiab 1,689,504 696,609 11,777,296 14 6 20 2.4 Ireland 22,851 19,233 186,561 12 10 23 1.2 Israel 29,967 56,812 310,937 10 18 28 0.5 Japan 118,704 668,526 4,038,302 3 17 19 0.2 Korea, Rep. 264,259 1,022,845 3,224,875 8 32 40 0.3 United States 1,537,243 1,154,971 17,272,044 9 7 16 1.3 Source: UNESCO database, accessed May 7, 2008; for China, Zhou (2005). Note: UNESCO, United Nations Educational, Scientific, and Cultural Organization. a. 2004. b. 2001. HIGHER EDUCATION IN HIGH-TECH INDUSTRIAL DEVELOPMENT | 247 TABLE 7. Graduates in Science and Engineering, Including Agriculture and Excluding Medicine, Taiwan, China, Selected Years, 1960­95 Item 1960 1965 1970 1975 1980 1985 1990 1995 Number of graduates 1,655 2,689 7,987 19,716 25,672 37,965 50,490 82,938 Percent of total graduates 27 25 24 33 36 42 44 46 Source: Lin 2004. One factor that appears to exert an influence going well beyond that of numbers is selectivity. China, India, Japan, Korea, and Taiwan, China, are alike in that they have highly selective institutions which recruit high-caliber students into engineering nationally and which have set high quality standards in engineering schools. Fur- thermore, these institutions invariably have had a significant share of their institu- tional profile in engineering, and in that sense their origins were closer to practically oriented institutions than to generic teaching-focused ones. At the time of the eco- nomic take-off, there was little domestic demand for graduates from these selective schools. This meant that they were cheap, high-caliber resources with training in sci- ence and technology that emerging industries could readily tap into. Roles of Higher Education Institutions Revisited In this section, six different kinds of involvement by higher education institutions are summarized, mainly to highlight the dependence of the nature of their participation in the economy on both institutional characteristics and the industrial context. Education Institutions have made diverse contributions in education. Generic teaching institu- tions and basic research universities produced generic scientists and engineers and left the companies to take the responsibility for developing specialized knowledge and skills. This role was particularly pronounced when the government intervened by preferentially expanding science and engineering programs. Practically oriented insti- tutions were more proactive in keeping up with changing needs for industrial skills and supplying specialized knowledge and skills as industry evolved. Relevant research universities were able to be similarly proactive at a higher level, in their research training. Research Academics in relevant research universities were most proactive in conducting research and in working directly with research applications, assisted by universities' organizational capacity to promote and support consulting, patenting and licensing, and spin-offs. Basic research­oriented universities were less active in furthering industries' commercialization goals but helped companies acquire better understand- ing of new scientific fields, provided advice, and conducted joint research, although 248 | S A C H I H ATA K E N A K A generally on the universities' own terms. Other institutions, such as government research institutions and practically oriented universities, helped industry acquire existing technologies through application-oriented research. Spin-offs University spin-offs appeared to play a visible role when there was a significant knowledge or capacity gap between universities and domestic industry. The gap could take very different forms, as demonstrated by the cases of biotechnology in the United States and ICT in China. In the United States the knowledge gap had to do with fundamental scientific dis- coveries and with related tacit knowledge. In China the nature of the knowledge gap was related to established technology, which was nonetheless new to China. The highly selective nature of Chinese universities also meant that they were better posi- tioned to assemble a group of high-caliber individuals to lead and work in new ven- tures than normal companies; universities in effect were able to exploit the wide capacity gap of available human resources. Significant examples of academic spin-offs were associated with universities with relevant research capabilities. Thus, in biotechnology in the United States, academics from research universities became directly involved in spin-offs, and there were sim- ilar stories in various fields in Israel. In Ireland, where, because of lack of government funding, universities did not develop research capacity until late (Sands 2005), aca- demic spin-offs were generally not common, but one computer science department did develop an early research capacity, with European funding, and became known for its academic spin-offs (Breznitz 2007). A very different experience was seen in China, where universities had become active in creating enterprises since the late 1980s, even when they had little research capability. Although these enterprises are sometimes described as spin-offs, they are significantly different from the normal practice in that they are owned and managed by universities (Eun, Lee, and Wu 2006). Some of these companies have been spec- tacularly successful; they include three of the most successful personal computer (PC) companies, Lenovo, Founder, and Tongfang, which were created by the Chinese Academy of Sciences, Beijing University, and Tsinghua University, respectively. About 40 university enterprises are already listed on stock markets in China and Hong Kong, China (Eun, Lee, and Wu 2006). Interestingly, the knowledge content of these spin-off companies does not often derive from significant scientific research; rather, the spin-off was the mechanism through which skilled personnel moved from universities to the commercial sector (Chen and Kenney 2007). These enterprises were a simple mechanism through which universities could contribute to industrial capabilities in an environment of very lim- ited industrial capability (Eun, Lee, and Wu 2006). They resembled in this respect Japanese university start-ups in the early phase of industrial development (Odagiri and Goto 1996), when academics could behave as arbitragers of Western technology and were in a good position to create companies, given the underdeveloped industrial context. HIGHER EDUCATION IN HIGH-TECH INDUSTRIAL DEVELOPMENT | 249 It is not clear how long this practice of university enterprises will continue in China. Both the government and many universities may be going through a rethink- ing process, as many enterprises have not been successful and managerial responsi- bilities are increasingly demanding (Ma 2007; Kroll and Leifner 2008). It is likely that China's university enterprise experience was a phenomenon dictated by the specific context of underdeveloped industry and a high concentration of talent in universities. Science Parks Taiwan, China, provides the most visible example of a highly successful science park, which had a critical part in the development of the semiconductor industry there. In 1980 the government established a science park in Hsinchu, close to two of the island's best technical universities, and relocated ERSO to the park (Saxenian 2004). There was a certain division of labor, in that applied and relevant basic research were undertaken by ERSO, while the universities largely appeared to supply graduates as manpower to the park. In China many high-tech zones, industrial parks, and science parks have been established close to universities by the central government and local authorities, as well as by the universities themselves. The nature of their dynamic evolution is only begin- ning to be documented, but the diversity in approach is already evident (Chen and Kenney 2007). The park founded by local authorities in Zhongguancun in Beijing, close to the Chinese Academy of Science as well as to Beijing and Tsinghua Universities, is considered one of the most successful in attracting multinational corporations' R&D investments. The park established by Tsinghua University on its campus has many such R&D units, as well as spin-offs from the university. An interesting variant is Shenzhen, whose high-tech industrial park has been used to attract established universities to create a virtual campus. Licensing Licensing is one function that seems to be linked strongly to relevant research capacity in universities, as well as to the level of development of intellectual property­related institutions in the country. In the case of the United States, the patenting practice evolved as one of many mechanisms for technology transfer from universities, predi- cated on huge government funding in relevant scientific research (Rosenberg and Nelson 1994; Mowery et al. 2004). An early development of professional support units such as technology-licensing offices was also important for effective technology trans- fer through licensing. It is increasingly clear, however, that for most U.S. universities, patenting and licensing constitute a net drain rather than a source of additional income (Thursby and Thursby 2007). The wider acceptance of the World Trade Organization (WTO) and other interna- tional regimes related to intellectual property will undoubtedly have consequences for the paths that future developing economies can take in catching up. Although it is important for developing countries and their universities to increase their 250 | S A C H I H ATA K E N A K A understanding of intellectual property, indiscriminate opening of technology transfer offices in universities with a view to patenting inventions may be unwarranted. Brain Drain and Brain Circulation Brain drain transformed into brain circulation appears to be more and more impor- tant to the introduction of high-tech industry in developing countries. This undoubt- edly reflects the increasingly global nature of higher education as well as labor markets in developed countries, many of which are actively recruiting professionals from developing countries in areas of key skills shortages such as ICT. Such a brain drain can turn into brain circulation once appropriate domestic conditions are created for industrial development, when "brains" that are enriched with international expe- rience can return to their home countries. Indeed, one feature of high-tech development in Taiwan, China, that distinguishes it from the cases of Japan and Korea, is the visible role played by the diaspora of engineers, particularly in Silicon Valley, and the transformation of the brain drain into brain circulation linked directly to the birth of high-tech industry (Saxenian 1999, 2004, 2006; Saxenian, Motoyama, and Quan 2002). Almost 80 percent of the diaspora professionals in Silicon Valley had gone to the United States to study and remained there for work (Saxenian, Motoyama, and Quan 2002). Early generations of the diaspora were instrumental in informing the home government about Silicon Valley and its unusual dynamism, inspiring an early policy of building a similar cluster (Saxenian 2004). Hsinchu Science Park represents a Taiwanese attempt to recreate Silicon Valley, and indeed the park later became an active recipient of the returning diaspora, with 40 percent of its companies started by returnees (Wade 1990; Saxenian 2004). It is well known that Taiwan, China, suffered a huge brain drain that started in the 1950s and lasted until the 1990s. Interestingly, the government nevertheless continued organizing programs of overseas study to complement the domestic education system. In 1975, 2,300 students were studying under government auspices; in 1986 the figure was 7,000 (Wade 1990). This was happening in the midst of the massive brain drain, which was estimated at 20 percent of all engineering graduates in the late 1970s (Hou and Gee 1993). Instead of ceasing to send students abroad, the government stepped up efforts to keep in touch with the diaspora community (Wade 1990). The benefits of this policy were finally felt in the 1990s, when the brains started to return. China and India, too, experienced brain drain to Silicon Valley. In the survey of Silicon Valley professionals, over half of Indians and nearly 80 percent of mainland Chinese there had originally come to the United States for their education (Saxenian, Motoyama, and Quan 2002). China was able to benefit doubly from diasporas, as the Hsinchu­Silicon Valley­Shanghai nexus developed through (broadly defined) overseas Chinese engineers, including those originally from Taiwan, China (Saxenian 2006). In the early 2000s the Chinese government stepped up efforts to recruit back the overseas Chinese. The dominance of Indian entrepreneurs in U.S. high-tech companies, and not only in Silicon Valley, is well documented (Saxenian, Motoyama, and Quan 2002; HIGHER EDUCATION IN HIGH-TECH INDUSTRIAL DEVELOPMENT | 251 Wadhwa et al. 2007). This highly entrepreneurial diaspora was important as a "reputational intermediary" in connecting Indian firms to client U.S. firms and in bringing American multinationals to India (Kapur 2002; Kapur and McHale 2005; Saxenian 2006). Even during the brain drain, selectivity appears to have played a role. The com- petitive entrance system for higher education meant that some engineering colleges and universities had elite status, and it was largely graduates from such institutions who became critical members of the diaspora and formed vibrant alumni networks (Saxenian 2004). Such networks were important for solving technical problems, with former teachers as well as former classmates serving as important sources of technical information. In China, India, and Taiwan, China, the brain drain was often triggered by prospects of graduate education abroad. In all three cases, graduates from the most select institutions actively pursued opportunities in graduate education abroad and succeeded in finding places, notably in good American research universities. It is easy to imagine that in their search for the best U.S. graduate schools, these graduates would have benefited from having been at elite home institutions, with active peer interactions providing information about studying abroad and with letters of recommendation from home institutions that enjoyed credibility. It is almost as though graduate schools in the United States became a natural extension of the domestic education system. It is also interesting to observe the differing approaches taken by governments. The government of Taiwan, China, offered fellowships for study abroad even at the height of the brain drain, apparently in the belief that the economy would gain in the end. A decade later the Chinese government was doing something similar. Both valued foreign education sufficiently to send graduates abroad in the first place, but both also worked hard to bring them back. It is as if, for both, foreign institutions, particularly in the United States, functioned as a critical component of the higher education system, providing the kind of education that their domestic institutions could not offer. India, which also suffered a severe brain drain, particularly of high- caliber graduates, was somewhat hostile to the idea of expatriate Indians until much later and was also slower in orchestrating the use of expatriate brains (Saxenian 1999, 2006). Evolution of the Universities Although all four types of institution discussed above had some economic role in the development of high-tech industry, the "responsive" institutions in the right-hand cells of figure 2 appear to have participated much more proactively and directly in helping high-tech industry emerge and evolve. Yet such institutions do not seem to emerge naturally. In all cases government action was critical in founding them and in influencing their missions and orientations. In this section the evolution of the vari- ous types of institution in specific countries is examined to aid understanding of the underlying dynamics. 252 | S A C H I H ATA K E N A K A Building Responsiveness In the United States the emergence of a group of responsive institutions was no acci- dent. A whole group of American universities, the land grant colleges, was founded in the late nineteenth century with the explicit objective of serving the practical needs of the community. The federal government supported their establishment through special grants and continued to sustain their growth by funding some of their service activities, such as agricultural extension. This group was a wide mix, ranging from broad institutions such as Cornell and Purdue Universities to institu- tions oriented toward industrial training, such as the Massachusetts Institute of Technology. That universities could and should be useful to society became a belief shared by government, the public, and universities broadly. The responsiveness of American universities in teaching was well established in those early days, and it was their responsiveness that generated new subjects such as electrical, chemical, and aeronautical engineering (Rosenberg and Nelson 1994). Before World War II, private foundations provided key support to some universi- ties, helping them undertake research that was relevant and useful to society (Geiger 2004b). But it was the spectacular success with the use of science to develop key mil- itary technology during World War II that brought about a new level of commitment to "useful scientific research" by government, universities, and the public alike. Post- war government research funding was massive but also mission oriented, coming from key agencies with application interests such as defense, health, and energy (Rosenberg and Nelson 1994; Geiger 2004a). The role of the Defense Advanced Research Projects Agency (DARPA) in funding application-oriented basic research was legendary--so much so that the U.S. National Academy of Sciences has recently recommended the creation of a DARPA-like agency in energy to ensure the contin- ued competitiveness of American science (NAS 2005). In Ireland and Finland universities moved toward the right-hand side of figure 2, becoming more responsive over the past decades. In both cases there had been concern that universities were detached from industry and were unable to meet changing labor market needs (White 2001; OECD 2003). Both governments invested in creating a large group of practically oriented institutions, as described above, but also pushed their universities to become more economically relevant. Investments to create sizable alternatives to universities exerted pressure on universities to become more sensitive to economic needs. In Ireland the government funded the establishment of new technical programs at universities and technical institutions that were considered relevant to industry. In Finland significant public investments were made in strategic areas of R&D, with much emphasis on collaboration between companies and universities (OECD 2003; Dahlman, Routti, and Yla-Anttila 2007; Srinivas and Viljamaa 2008). Today, Finland has a diverse university sector, with varying levels of university inclination to work with industry. Some universities with a technological orientation or regional development missions are much more active in working with industry and receive much larger shares of funding from the Finnish Funding Agency for Technol- ogy and Innovation (TEKES), which provides competitive grants for application- oriented research (Finland, Ministry of Education, 2005). HIGHER EDUCATION IN HIGH-TECH INDUSTRIAL DEVELOPMENT | 253 The Oulu University of Applied Sciences was important in the transformation of Oulu Province from a sleepy backwater to a vibrant high-tech center (Donnelly and Hyry 2004). Established in the 1950s with the specific mission of helping the region develop, the university has been proactive in its choice of disciplinary coverage and responsive in its educational program. An electrical engineering department was established early, well before industrial needs emerged, but the university also responded quickly to meet emerging skills shortages, for instance, in management education (Donnelly and Hyry 2004). The university's responsiveness led to the development of relevant research centers in which both basic and applied research were undertaken, often jointly with industry, and with some academic spin-offs. In China the government has, since the 1980s, been taking steps to reform the higher education system. In retrospect, the measures employed have been remarkably consistent in pushing key universities toward the American relevant research univer- sity model (Ma 2007). In the science and technology reform that began in the 1980s, the role of these fields in economic development was strongly emphasized. Public research institutes were to reorient the content of their research to meet economic needs, and universities were to develop research capacity relevant to society, for the first time. The official endorsement did not mean that government funding was forth- coming for research. Although the Chinese National Science Foundation was estab- lished in 1986 to provide competitive grants for basic research projects in public research institutions, as well as in universities, university budgets had been cut and were extremely tight. Universities had strong incentives to generate their own incomes through industrial contracting (Ma 2007). This was the context in which universities began to develop a certain responsiveness to industry, through contract research and consulting and by setting up their own enterprises. The government has also supported the emergence of elite research universities through a series of special programs, from the key university and key laboratory pro- grams in the 1980s to the more famous Projects 211 and 985 in the 1990s, which concentrated government funding on the top 100 and top 9 universities, respectively (Ma 2007). Together with the gradual development of competitive funding for research, these initiatives resulted in strong incentives for universities to be research oriented and compete globally to become world-class institutions (Ma 2007). It is no accident that one of the first global rankings of universities was designed by a Chinese university; Chinese institutions were developing such indicators to gauge their positions in the world. Moving away from Responsiveness The path taken by Japanese universities provides an interesting contrast to that in the United States. The starting points in the two countries were similar in that Japan's first university, the University of Tokyo, was developed in the late nineteenth century as a critical instrument for Japan's modernization--specifically, to teach useful scientific knowledge. At the time, the university's structure of having practically oriented facul- ties such as engineering, agriculture, medicine, and law, along with science and litera- ture, was highly unusual even in global terms (Bartholomew 1989). Other imperial 254 | S A C H I H ATA K E N A K A universities were established along the lines of the Tokyo University model, and collectively they helped introduce key technologies and knowledge essential for Japan's early industrialization. University academics and graduates were critical in the birth of many of Japan's technology-oriented enterprises, such as Toyota, Toshiba, and some pharmaceutical companies (Odagiri and Goto 1996). Indeed, Japanese versions of university start-ups are visible in many corporate histories. Large companies habit- ually sponsored university facilities, and academics actively served as technical consultants (Hashimoto 1999). Even the largest industrial corporations had limited research facilities or capabilities in those days, and university academics undertook application-oriented research that was relevant to the companies. After World War II, Japanese universities embarked on an evolutionary path toward becoming basic research universities and began to be somewhat removed from industrial development (Hatakenaka forthcoming). In a striking contrast to the United States, where the successful role of science in military technology had reinforced public confidence, Japanese academics moved away from the military- industrial complex (Hatakenaka 2004). When industry began to acquire foreign technology in the 1950s, universities had a minimal role in the process (Hashimoto 1999). In the 1960s, industries recruited their own researchers, often from universi- ties, to establish corporate laboratories (Nakayama 1991), marking the beginning of a new era in which corporate research became a dominant force in industrial inno- vation (Nakayama 1995). In the late 1960s, student uprisings led to further criticism of university-industry relationships, and a strong sentiment for limiting such rela- tionships prevailed (Hashimoto 1999; Hatakenaka 2004). Institutional mechanisms for working with industry remained underdeveloped under this culture of censorship; the bulk of scientists and engineers were civil servants in national universities, sub- ject to rules that forbade them to consult with industry, and universities lacked the autonomy to create their own rules. Although individual academics did work with industry, their modes of engagement were distinctly different from those of American universities. Relationships were formed principally by industries doing the legwork of coming to individual academ- ics, extracting information, and making every effort to learn from the universities, in a receiver-active technical transfer model (Kodama and Suzuki 2007). There was sur- prisingly little interaction between universities and companies concerning the content of education (Dore and Sako 1989), and universities made little organizational effort to work with industry, until recently. Rather, companies expected universities to be no more than a selection mechanism, and they developed in-house capabilities and prac- tices to train graduates in the needed specializations. An important consequence of this system was that Japanese universities did not learn to respond to emerging needs for industrial skills. This detachment increasingly became an issue for industry as it moved into advanced-technology fields, particularly when new industries such as computer software, semiconductors, and biotechnology developed in the 1980s (Baba, Takai, and Mizuta 1996; Darby and Zucker 1996; Kobayashi 2001). In a comparative analy- sis of Stanford University and the Tokyo Institute of Technology (TIT), curricular change was found to be infrequent and limited in scope at the TIT, whereas at Stanford updating took place organically and almost constantly (Harayama 2001; Hoshino, HIGHER EDUCATION IN HIGH-TECH INDUSTRIAL DEVELOPMENT | 255 Harayama, and Hatatani 2003). The difference appears to reflect different institu- tional philosophies; the TIT wished to emphasize basic core knowledge essential even in the changing environment, and Stanford had an explicit emphasis on remaining relevant to industrial needs. The universities' lack of responsiveness was particularly pronounced in research training. Although master's programs in engineering were commonplace by the 1980s, PhD programs in science and engineering remained small and highly aca- demic. Even when these PhD programs expanded in the 1990s, little effort was made to reform the curricular content, and 40 percent of the companies surveyed cited skills mismatch as a reason for not recruiting PhDs (Shimizu and Mori 2001). This was in some ways an inevitable and consistent development, given the corporate per- sonnel practice of recruiting the best and brightest early and training them in-house. Indeed, a form of doctoral training developed in which industrial researchers could submit dissertations to university academics to obtain degrees when their research work was undertaken within industry (Clark 1995). Developing a Basic Research Orientation Developing a research capacity of whatever nature is often a difficult step in the evo- lution of higher education systems. In Korea universities were not very visible in the transition that Korean electronics companies made to high-tech fields such as com- puters and semiconductors during the 1980s. The reason was the universities' inad- equately developed research capacity, a consequence of government underinvestment (Kim 1993, 2000). Korean universities were particularly underresourced in view of the rapid and massive expansion in undergraduate enrollments in the 1970s and 1980s (Kim 1993). The student-staff ratio rose from 23 in 1966 to 36 in 1985 (Kim 1993), and by 1985 nearly half of national and public university revenues came from tuition and fees (KCUE 1990). Even today, Korea's public per-student spending is strikingly low compared with that in other countries (tables 8 and 9). It was not until TABLE 8. Public Expenditure per Tertiary-Level Student as Percent of GDP per Capita, Selected Countries, Selected Years 1999­2005 Country 1999 2000 2004 2005 China 90 -- -- -- Finland 41 39 37 35 India -- 91 94 58 Ireland 29 31 24 25 Israel 33 32 26 -- Japan 15 18 21 19 Korea, Rep. 8 -- 9 -- United States 27 -- 23 23 Source: UNESCO database, accessed May 7, 2008. Note: --, not available. GDP, gross domestic product; UNESCO, United Nations Educational, Scientific, and Cultural Organization. 256 | S A C H I H ATA K E N A K A TABLE 9. Tertiary Education Expenditures as Share of GDP. Selected Countries, Selected Years 1999­2005 Country and type of institution 1999 2000 2004 2005 China Private -- -- -- -- Public 0.4 -- -- -- Total -- -- -- -- Finland Private -- -- 0.1 0.1 Public 1.7 1.7 1.7 1.7 Total -- 1.7 1.8 1.8 India Private -- -- 0.2 -- Public -- 0.9 1.0 -- Total -- 0.9 -- -- Ireland Private -- -- -- 0.2 Public 1.0 -- 0.9 0.9 Total -- -- -- 1.2 Israel Private -- 0.9 1.0 -- Public 1.2 1.1 1.0 -- Total -- 2.0 2.0 -- Japan Private 0.6 0.6 0.8 0.9 Public 0.5 0.5 0.5 0.5 Total 1.1 1.1 1.3 1.4 Korea, Rep. Private 1.7 -- 1.9 -- Public 0.4 -- 0.5 -- Total 2.2 -- 2.3 -- United States Private -- 1.7 2.0 2.0 Public 1.1 0.9 1.1 1.1 Total -- 2.6 3.1 3.0 Source: UNESCO database, accessed May 7, 2008. Note: --, not available. GDP, gross domestic product; UNESCO, United Nations Educational, Scientific, and Cultural Organization. Figures may not sum to totals because of rounding. the late 1990s, when government for the first time concentrated large amounts of resources in the best institutions for research and graduate education, that research performance began to improve in a significant way (Kim and Nam 2007). Scarcity of government funding was actually beneficial from the point of view of forcing universities to develop ties with industry. Some of the quoted statistics in Korea indicate a relatively high level of engagement with industry. For instance, HIGHER EDUCATION IN HIGH-TECH INDUSTRIAL DEVELOPMENT | 257 TABLE 10. Academic Research and Development Financed by Industry, Selected OECD Countries, China, and Russian Federation, Selected Years, 1981­2006 (percent) Korea, United United All Russian Year France Germany Japan Rep. Kingdom States OECD China Federation 1981 1.3 1.8 1.0 -- 2.8 4.4 2.9 -- -- 1986 2.0 5.8 1.7 -- 5.7 6.5 4.5 -- -- 1991 4.2 7.0 2.4 -- 7.8 6.8 6.0 -- -- 1996 3.2 9.2 2.4 50.5 6.7 7.1 6.9 -- 23.7 2001 3.1 12.2 2.3 14.3 6.2 6.5 6.4 -- 26.5 2002 2.9 11.8 2.8 13.9 5.8 5.8 6.2 -- 27.2 2003 2.7 12.6 2.9 13.6 5.6 5.3 6.0 35.9 27.9 2004 2.7 13.2 2.8 15.9 5.1 5.1 6.2 37.1 32.6 2005 -- -- -- 15.2 -- 5.2 -- 36.7 29.3 2006 -- -- -- -- -- 5.3 -- -- -- Source: National Science Foundation, Division of Science Resources Statistics, special tabulations, as published in Science and Engineering Indicators 2008. Note: --, not available. OECD, Organization for Economic Co-operation and Development. about half of university research was funded by industry in the early 1990s (Kim 2000), and one survey found that nearly half of the companies canvassed had worked with universities (Sohn and Kenney 2007). The industry share of research funding declined in the late 1990s to about 15 percent, presumably reflecting increasing levels of government funding in research (see table 10). Nevertheless, it remains higher than the 6 percent in OECD countries, indicating that industry engagement continues to be important, even if university contributions are not readily acknowledged, and in spite of the poor performance record in proactive measures such as academic spin- offs (Sohn and Kenney 2007). Some would argue that research orientation was slow to develop in Korean universities because of the historical division of labor whereby public research institutions rather than universities were responsible for research (Lee 2000; Sohn and Kenney 2007). Similar difficulties are encountered in India, where there is a well- established group of public research institutions whose status is higher than that of the universities (Jayaram 2007). The best researchers tend to be recruited into research institutions, leaving universities in a difficult position for developing research capabilities. The fact that the research institutions have no teaching respon- sibilities makes the situation worse in that the best researchers are not contributing to the training of the next generation (Jayaram 2007). There is no question that public research institutions can be spectacularly success- ful in the development of high-tech industry, as the experience with ERSO in Taiwan, China, shows. It is hard to imagine that similar results could have been obtained in so short a time if similar capacity had been established in a university. Still, the fact that relevant research capacity was concentrated outside universities probably repre- sents a lost opportunity for universities in Taiwan, China, as well. 258 | S A C H I H ATA K E N A K A Finland's experience shows that it is possible to link applied research capacity in public research institutions to universities in a productive way. When Finland made its bid to turn itself into a knowledge economy in the early 1990s, a key component of its policies was large public investment in collaborative R&D among universities, public research institutions, and companies, orchestrated by TEKES (Dahlman, Routti, and Yla-Anttila 2007). China promoted the research role of universities early on, despite the existence of public research institutions. In the science and technology reform starting in the 1980s, the Chinese government took action both to reorient government research institutions toward becoming economically relevant and to recognize, for the first time, the research role of universities in economic development. Concluding Remarks This paper has argued that institutions contribute to the development of high-tech industry in different ways. It has also proposed that their roles are largely determined by characteristics which can be defined on three dimensions: their responsiveness, their basic research orientation, and their selectivity. Enhancing university characteristics along any of these three dimensions can strengthen the nature of the contributions uni- versities make. The more responsive are the universities, the more companies can depend on universities in tailoring academia's research and education contributions to their specific needs. The more the institutions are oriented toward basic research, the better they are able to provide a sophisticated scientific base to enhance the high- technology content of industry. The more selective the universities, the less difficult it is for companies to identify appropriate talents. The paper also introduced four key types of institutions, using the first two dimen- sions, responsiveness and basic research orientation. The four are basic research uni- versities, relevant research universities, teaching-focused institutions, and practically oriented institutions. Interestingly, no single type of institution appeared to be essen- tial for starting up high-tech industry; every one of the four types of institution was instrumental in the establishment of high-tech industry in one setting or another. In fact, different university types appear to lead to different shapes and natures of high- tech industry. For instance, the production of high-quality scientists and engineers in large numbers by teaching-focused institutions was sufficient to ensure the availabil- ity of cheap, high-caliber labor in India, leading to key production advantages. Basic research universities help companies equip themselves with key scientific knowledge and well-trained employees in generic science and engineering. But suc- cessful companies have to be sufficiently proactive to extract appropriate scientific know-how from universities or to provide further specialized training to employees to ensure their competitiveness. Practically oriented institutions can provide application-oriented research assis- tance and specialist training, as needed by industry, and can help companies reduce the costs of in-house research or training. However, they are not able to assist com- panies in upgrading the scientific content of their work or to provide postgraduate HIGHER EDUCATION IN HIGH-TECH INDUSTRIAL DEVELOPMENT | 259 training in emerging fields of science. They can provide institutional support for start-ups, although these are unlikely to be characterized as science based. Relevant research universities are characterized by their ability to be more directly involved in commercialization of scientific knowledge. These universities can also pro- vide high-level postgraduate training in emerging fields to supply key skills for emerg- ing industry. They tend to develop and attract appropriate ecosystems (such as venture capital and professional support services) for science-based start-ups over time. Industrial demand for the contributions of universities is likely to become more stringent. Global competitive pressures will demand constant upgrading and redefi- nition of industry's competitive advantage. Universities will need to provide "higher- order" contributions, and to do so, they will need to move along at least one of the three dimensions. A good supply of cheap generic scientists and engineers may have been enough to start high-tech industry yesterday, but the growth of such industry will lead to higher wages, which in turn will demand redefinition of its competitive advantage. By enhancing responsiveness, basic research orientation, or selectivity, the power of university support to industry can be strengthened. The paper has also proposed that moving universities along any of the dimensions requires nontrivial effort. Governments can be an important influence in orienting and guiding institutional development, in subtle or not so subtle ways. Responsiveness does not appear to emerge naturally; in all the cases examined, some specific interventions had occurred to instill the culture of responsiveness, often at the time of founding, and often with government intervention. There is no simple recipe for encouraging institutions to become responsive. The demand to which insti- tutions must be sensitive is often local, intangible, and ill defined. Governments may require that institutions provide new educational programs or expand existing ones, but simply following a top-down requirement does not make an institution respon- sive. The Korean government's central requirement of maintaining a quota for engi- neering programs may not have encouraged responsiveness. Indeed, it is possible that the more frequently a government interferes through regulations or even through financial incentives, the less institutions learn to be creative and strategic about what they do. In Ireland and Finland new groups of institutions were established by gov- ernment from scratch in order to build responsiveness into the system. Peer pressures and academic norms tend to push universities to become nonre- sponsive over time--toward becoming basic research universities or academically ori- ented teaching institutions. On the whole, there appear to be more cases in which institutions developed responsiveness first and then evolved toward a fundamental science orientation than vice versa. (A good example is the Massachusetts Institute of Technology; see Etzkowitz 2002.) It is probably easier to develop in that order than the other way around. Today, economic responsiveness on the part of universities has become a global mantra. The last decade has seen an ever-increasing number of government policies seeking to push universities to develop such a dimension, and this has led to changes in academic values. For instance, it would be unusual to meet an academic in any OECD country who would contest the need for a university to be economically rel- evant. Notwithstanding this dynamic change, most institutions have a long way to 260 | S A C H I H ATA K E N A K A go toward developing a full set of internal and external mechanisms that are respon- sive and proactive. The Japanese national universities, for example, are as collaborative with industry as are U.S, ones, but they still lag substantively in proactive commer- cialization as measured by metrics such as licensing and spin-offs. Indeed, in order for universities to become truly proactive, a whole new set of relationships is likely to be needed, not only with industry but also with other players, ranging from venture capital to proactive research funding bodies. It takes a full ecosystem to enhance the responsiveness of universities. One advantage that developing countries have over old economies may be the potential to develop new institutions that together can make up such a complete system. Building a basic research orientation also requires serious commitment and invest- ment, usually from government. Investments are needed not only in human capital, in the form of advanced research training, but also in infrastructure such as libraries and laboratories, as well as for direct expenses associated with research. A research orientation cannot be developed overnight, but it is difficult to change once intro- duced; PhD-trained academics usually gravitate toward research rather than teach- ing, and a sense of elitism is often associated with being research oriented--the more basic, the more elite. Active intervention is usually needed to counter the academic drift away from becoming responsive. Selectivity is one dimension of higher education institutions that often emerges naturally as the demand for higher education expands and the competition for places increases. However, it needs to be underpinned by robust selection systems that are fair but also provide good learning incentives for future students. This is not easy, but it merits considerable attention, as selection systems can directly influence the repu- tational hierarchy of institutions, as well as students' learning incentives, for years to come. Some examinations inadvertently encourage students to become oriented toward test-taking tactics rather than learning, and some scoring systems give a spurious impression of accuracy. The fact that all three dimensions are important in enabling institutions to play key roles does not mean that all institutions must be the same. Because different dimensions bring different advantages, there is a strong case for a variety of institu- tions, as the experience in the United States and Finland shows. It is reasonable for a national system of higher education to seek to have as much diversity as possible in terms of types of institution. A historical legacy that the case countries share is some form of national champion of science and technology. The postwar United States was a special place in that pol- icy makers' endorsement of the employment of science for societal benefit led to the development of a whole host of federal funding institutions that supported funda- mental science, for different reasons (Geiger 2004a). Similarly, without the highest commitment from Jawaharlal Nehru, India's first postindependence leader, the country's institutes of technology would never have received the extraordinary technical assis- tance from the United States, the Soviet Union, and Germany that enabled them to develop their high standards. Without the commitment of Deng Xiaoping, the Chinese government would not have devoted so many resources to science and tech- nology, which in turn helped attract the best and the brightest to the field. Such visible HIGHER EDUCATION IN HIGH-TECH INDUSTRIAL DEVELOPMENT | 261 support of science and technology at the highest levels helped foster the commitment to science and technology within the society that led students more readily to choose to study these disciplines. National championship of this kind is not easy to maintain. Science is expensive, and it takes a long time for its impact to be felt. It is easy for commitment to science to give way to fiscal pressures, particularly since it is never clear how much is needed. It is also not easy to cultivate and then maintain the high level of interest within the society that induces students to choose science and engineering. Many OECD countries, including Japan, the United Kingdom, and the United States, are increas- ingly concerned about the level of interest in science among students. Even in India, there is growing concern that fewer students are choosing a science track. A final point for reflection is the role of governance arrangements at both the insti- tutional level and the national level. Quality assurance considerations often lead gov- ernment bodies to become overprescriptive about inputs, as well as being concerned about outputs. This is a significant dilemma in developing countries, where there is little tradition and culture concerning the quality of higher education. Attempts by governments to "control" quality on the basis of inputs can lead to institutions' becoming similar to each other and being passive in defining what they are. To make matters worse, some dimensions are easier to measure than others, which can then result in their being given undue prominence; publications and citations are estab- lished metrics for evaluating the productivity and quality of scientific research, but metrics for responsiveness are only just emerging. Moreover, tight regulatory control by government can be a serious impediment when institutions need to respond to changing needs. Yet the desire for the system to respond to the needs of both labor markets and future students, particularly in dynamically developing countries, often pushes governments to centralize key deci- sions about the size and direction of the higher education system. Institutional capacity to respond to complex change is unlikely to develop if insti- tutions do not have autonomy in making strategic decisions. Furthermore, institu- tions are unlikely to be able to make good strategic decisions without appropriate governance structures and internal processes. The key for the future is to ensure that there are governance arrangements at both the institutional and the system levels that will facilitate and even encourage further evolution of higher education. References Baba, Y., S. Takai, and Y. Mizuta. 1996. "The User Driven Evolution of the Japanese Software Industry: The Case of Customized Software for Mainframes." In The International Com- puter Software Industry: A Comparative Study of Industry Evolution and Structure, ed. D. C. Mowery. New York: Oxford University Press. Bartholomew, James R. 1989. The Formation of Science in Japan. New Haven, CT: Yale University Press. Bresnahan, Timothy F., and Franco Malerba. 1999. "Industrial Dynamics and the Evolution of Firms' and Nations' Competitive Capabilities in the World Computer Industry." In Sources of Industrial Leadership: Studies of Seven Industries, ed. David C. Mowery and Richard R. Nelson. Cambridge, U.K.: Cambridge University Press. 262 | S A C H I H ATA K E N A K A Breznitz, Dan. 2004. "The Israeli Software Industry." In From Underdogs to Tigers: The Rise and Growth of the Software Industry in Brazil, China, India, Ireland, and Israel, ed. Ashish Arora and Alfonso Gambardella. Oxford, U.K.: Oxford University Press. ------. 2007. Innovation and the State: Political Choice and Strategies for Growth in Israel, Taiwan and Ireland. New Haven, CT: Yale University Press. Chen, Kun, and Martin Kenney. 2007. "Universities/Research Institutes and Regional Inno- vation Systems: The Cases of Beijing and Schenzhen." World Development 35 (6): 1056­74. Clark, Burton R. 1995. Places of Inquiry: Research and Advanced Education in Modern Universities. Berkeley and Los Angeles: University of California Press. Cohen, W. M., R. R. Nelson, and J. P. Walsh. 2002. "Links and Impact: The Influence of Public Research on Industrial R&D." Management Science 48 (1): 1­23. Dahlman, Carl, Jorma Routti, and Pekka Yla-Anttila. 2007. Finland as a Knowledge Economy: Elements of Success and Lessons Learned. Washington, DC: World Bank Institute. Darby, Michael R., and Lynne G. Zucker. 1996. "Star Scientists, Institutions and the Entry of Japanese Biotechnology Enterprises." NBER Working Paper 5795, National Bureau of Economic Research, Cambridge, MA. de Fontaney, Catherine, and Erran Carmel. 2004. "Israel's Silicon Wadi: The Forces behind Clus- ter Formation." In Building High Tech Clusters: Silicon Valley and Beyond, ed. Timothy Bresnahan and Alfonso Gambardella. Cambridge, U.K.: Cambridge University Press. Donnelly, Tom, and Martti Hyry. 2004. "Urban and Regional High Technologies: The Case of Oulu." Local Economy 19 (2): 134­49. Dore, Ronald P., and Mari Sako. 1989. How the Japanese Learn to Work. London: Routledge. Etzkowitz, Henry. 2002. MIT and the Rise of Entrepreneurial Science. London: Routledge. Eun, Jong-Hak, Keun Lee, and Guisheng Wu. 2006. "Explaining University-Run Enterprises in China: A Theoretical Framework for University-Industry Relationship in Developing Countries and Its Application to China." Research Policy 35: 1329­46. Finland, Ministry of Education. 2005. "OECD Thematic Review of Tertiary Education: Country Background Report for Finland." Organisation for Economic Co-operation and Development, Paris. Geiger, Roger. 2004a. Research and Relevant Knowledge: American Research Universities since World War II. New Brunswick, NJ: Transaction Publishers. ------. 2004b. To Advance Knowledge: The Growth of American Research Universities 1900­1940. New Brunswick, NJ: Transaction Publishers. Harayama, Yuko. 2001. "Silicon Valley no sangyo hatten to Stanford daigaku no curriculum hensen" [Silicon Valley's Industrial Development and Curricular Change in Stanford Uni- versity]. In Daigaku kaikaku: kadai to souten, ed. Masahiko Aoki, Akihiko Sawa, and Michio Daito, 31­66. Tokyo: Toyo Keizai Shinposha. Hashimoto, Takehiko. 1999. "The Hesitant Relationship Reconsidered: University-Industry Cooperation in Postwar Japan." In Industrializing Knowledge: University-Industry Link- ages in Japan and the US, ed. Lewis Branscomb, Fumio Kodama, and Richard Florida. Cambridge, MA: MIT Press. Hatakenaka, Sachi. 2004. University-Industry Partnerships in MIT, Cambridge and Tokyo: Storytelling across Boundaries. New York: Routledge. ------. Forthcoming. "What's the Point of Universities? The Economic Role of Universities in Japan." Japan Forum. Henderson, Rebecca, Luigi Orsenigo, and Gary P. Pisano. 1999. "The Pharmaceutical Indus- try and the Revolution in Molecular Biology: Interactions among Scientific, Institutional and Organizational Change." In Sources of Industrial Leadership: Studies of Seven HIGHER EDUCATION IN HIGH-TECH INDUSTRIAL DEVELOPMENT | 263 Industries, ed. David Mowery and Richard Nelson. Cambridge, U.K.: Cambridge University Press. Hicks, Diana. 1993. "University-Industry Research Links in Japan." Policy Sciences 26: 361­95. Hoshino, Yu, Yuko Harayama, and Shigeo Hatatani. 2003. "Daigaku no gaibu henka heno taio sei" [Responsiveness of Universities to External Change]. In Sangaku Renkei, ed. Yuko Harayama. Tokyo: Toyo Keizai Shinposha. Hou, Chi-Ming, and San Gee. 1993. "National Systems Supporting Technical Advance in Industry: The Case of Taiwan." In National Innovation Systems: A Comparative Analysis, ed. Richard R. Nelson. New York: Oxford University Press. Japan, MEXT (Ministry of Education, Culture, Science, and Technology). 2006. Monbuka- gakutokeiyoran. Tokyo: Ministry of Finance Press. Jayaram, N. 2007. "Beyond Retailing Knowledge: Prospects of Research-Oriented Universities in India." In World Class Worldwide: Transforming Research Universities in Asia and Latin America, ed. Philip G. Altbach and Jorge Balán. Baltimore, MD: Johns Hopkins University Press. Juahiainen, Jussi S. 2006. "Multipolis: High-Technology Network in Northern Finland." European Planning Studies 14 (10): 1407­28. Kapur, Devesh. 2002. "The Causes and Consequences of India's IT Boom." India Review 1 (2): 91­110. Kapur, Devesh, and John McHale. 2005. "Sojourns and Software: International Mobile Human Capital and High Tech Industry Development in India, Ireland and Israel." In From Underdogs to Tigers: The Rise and Growth of the Software Industry in Brazil, China, India, Ireland, and Israel, ed. Ashish Arora and Alfonso Gambardella. Oxford, U.K.: Oxford University Press. KCUE (Korean Council for University Education). 1990. Korean Higher Education: Its Development, Aspects and Prospects. Seoul: KCUE. Kim, Ki-Seok, and Sunghee Nam. 2007. "The Making of a World-Class University at the Periphery." In World Class Worldwide: Transforming Research Universities in Asia and Latin America, ed. Philip G. Altbach and Jorge Balán. Baltimore, MD: Johns Hopkins University Press. Kim, Limsu. 1993. "National System of Industrial Innovation: Dynamics of Capability Build- ing in Korea." In National Innovation Systems: A Comparative Analysis, ed. Richard R. Nelson. New York: Oxford University Press. ------. 2000. "Korea's National Innovation System in Transition." In Technology, Learning And Innovation: Experiences of Newly Industrializing Economies, ed. Limsu Kim and Richard R. Nelson. Cambridge, U.K.: Cambridge University Press. Kobayashi, Shinichi. 2001. "Daigaku kyoiku no shokugyoteki relevance to daigaku no soshiki sekkei" [Relevance of University Education to Work and Organizational Design]. In Daigaku kaikaku: kadai to souten, ed. Masahiko Aoki, Akihiko Sawa, and Michio Daito. Tokyo: Toyo Keizai Shinposha. Kodama, Fumio, and Jun Suzuki. 2007. "How Japanese Companies Have Used Scientific Advances to Restructure Their Businesses: The Receiver-Active National System of Innovation." World Development 35 (6): 976­90. Kroll, Henning, and Ingo Leifner. 2008. "Spin-Off Enterprises as Means of Technology Com- mercialisation in a Transforming Economy--Evidence from Three Universities in China." Technovation 28: 293­313. Lee, Won-Young. 2000. "The Role of Science and Technology Policy in Korea's Industrial Development." In Technology, Learning and Innovation: Experiences of Newly Industri- alizing Economies, ed. Limsu Kim and Richard R. Nelson. Cambridge, U.K.: Cambridge University Press. 264 | S A C H I H ATA K E N A K A Leifner, Ingo, and Daniel Schiller. 2008. "Academic Capabilities in Developing Countries: A Conceptual Framework with Empirical Illustrations from Thailand." Research Policy 37: 276­93. Lester, Richard K. 2005. "Universities, Innovation, and the Competitiveness of Local Economies: Summary Report from the Local Innovation Systems Project--Phase I." MIT IPC Working Paper 05-010, Industrial Performance Center, Massachusetts Institute of Technology, Cambridge, MA. Lin, Tin-Chun. 2004. "The Role of Higher Education in Economic Development: An Empiri- cal Study of Taiwan Case." Journal of Asian Economics 15 (2): 355­71. Ma, Wanhua. 2007. "The Flagship University and China's Economic Reform." In World Class Worldwide: Transforming Research Universities in Asia and Latin America, ed. Philip G. Altbach and Jorge Balán. Baltimore, MD: Johns Hopkins University Press. Mazzoleni, Roberto, and Richard R. Nelson. 2007. "Public Research Institutions and Economic Catch Up." Research Policy 36 (10, December): 1512­28. Mowery, David C. 1999. "The Computer Software Industry." In Sources of Industrial Leadership: Studies of Seven Industries, ed. David C. Mowery and Richard R. Nelson. Cambridge, U.K.: Cambridge University Press. Mowery, David, Richard R. Nelson, Bhaven N. Sampat, and Arvids A. Ziedonis. 2004. Ivory Tower and Industrial Innovation: University-Industry Technology Transfer before and after the Bayh-Dole Act. Stanford, CA: Stanford University Press. Nakayama, Shigeru. 1991. Science, Technology and Society in Postwar Japan. Japanese Studies Series. London: Kegan Paul International. ------. 1995. Kagaku gijutsu no sengoshi. Tokyo: Iwanami Shoten. NAS (National Academy of Sciences). 2005. Rising above the Gathering Storm: Energizing and Employing America for a Brighter Economic Future. Washington, DC: National Academies Press. Odagiri, Hiroyuki, and Akira Goto. 1996. Technology and Industrial Development in Japan: Building Capabilities by Learning, Innovation and Public Policy. Oxford, U.K.: Oxford University Press. OECD (Organisation for Economic Co-operation and Development). 2003. Reviews of National Policies for Education: Polytechnic Education in Finland. Paris: OECD. Pechter, Kenneth, and Sumio Kakinuma. 1999. "Co-authorship Linkages between University Research and Japanese Industry." In Industrializing Knowledge: University-Industry Link- ages in Japan and the US, ed. Lewis Branscomb, Fumio Kodama, and Richard Florida, 102­27. Cambridge, MA: MIT Press. Roper, Stephen, and Seamus Grimes. 2005. "Wireless Valley, Silicon Wadi and Digital Island-- Helsinki, Tel Aviv and Dublin and the ICT Global Production Network." Geoforum 36: 297­313. Rosenberg, Nathan, and Richard Nelson. 1994. "American Universities and Technical Advance in Industry." Research Policy 23: 323­48. Sands, Anita. 2005. "The Irish Software Industry." In From Underdogs to Tigers: The Rise and Growth of the Software Industry in Brazil, China, India, Ireland, and Israel, ed. Ashish Arora and Alfonso Gambardella. Oxford, U.K.: Oxford University Press. Saxenian, AnnaLee. 1999. Silicon Valley's New Immigrant Entrepreneurs. San Francisco: Public Policy Institute of California. ------. 2004. "Taiwan's Hsinchu Region: Imitator and Partner for Silicon Valley." In Build- ing High-Tech Clusters: Silicon Valley and Beyond, ed. Timothy Bresnahan and Alfonso Gambardella. Cambridge, U.K.: Cambridge University Press. HIGHER EDUCATION IN HIGH-TECH INDUSTRIAL DEVELOPMENT | 265 ------. 2006. The New Argonauts: Regional Advantage in a Global Economy. Cambridge, MA: Harvard University Press. Saxenian, AnnaLee, Yasuyuki Motoyama, and Xiaohong Quan. 2002. Local and Global Net- works of Immigrant Professionals in Silicon Valley. San Francisco: Public Policy Institute of California. Shimizu, Kiyoyuki, and Kazuhiro Mori. 2001."Nihon no rikokei daigakuinkyoiku no bap- ponteki kaikaku" [Fundamental Reform for Graduate Schools in Japan]. In Daigaku kaikaku: kadai to souten, ed. Masahiko Aoki, Akihiro Sawa, and Michio Daito. Tokyo: Toyo Keizai Shinposha. Sohn, Dong-Won, and Martin Kenney. 2007. "Universities, Clusters, Innovation Systems: The Case of Seoul, Korea." World Development 35 (6): 991­1004. Srinivas, Smita, and Kimmo Viljamaa. 2008. "Emergence of Economic Institutions: Analysing the Third Role of Universities in Turku, Finland." Regional Studies 42 (3): 323­41. Stokes, Donald E. 1997. Pasteur's Quadrant: Basic Science and Technological Innovation. Washington, DC: Brookings Institution Press. Thursby, Jerry G., and Marie C. Thursby. 2007. "University Licensing." Oxford Review of Economic Policy 23 (4): 620­39. Trajtenberg, Manuel. 2001. "Innovation in Israel 1968­1997: A Comparative Analysis Using Patent Data." Research Policy 30: 363­89. Wade, Robert. 1990. Governing the Market: Economic Theory and the Role of Government in East Asian Industrialization. Princeton, NJ: Princeton University Press. Wadhwa, Vivek, AnnaLee Saxenian, Ben Rissing, and Gary Gereffi. 2007. "America's New Immigrant Entrepreneurs: Part I." Duke Science, Technology and Innovation Paper 23, Master of Engineering Management Program, Pratt School of Engineering, Duke Univer- sity, Durham, NC. Weidman, John C., and Namgi Park. 2000. Higher Education in Korea: Tradition and Adaptation. New York: Falmer Press. White, Tony. 2001. Investing in People: Higher Education in Ireland from 1960­2000. Dublin: Institute of Public Administration. Wu, Weiping. 2007. "Cultivating Research Universities and Industrial Linkages in China: The Case of Shanghai." World Development 35 (6): 1075­93. Yoshihara, Mariko, and Katsuya Tamai. 1999. "Lack of Incentives and Persisting Constraints: Factors Hindering Technology Transfer at Japanese Universities." In Industrializing Knowledge: University-Industry Linkages in Japan and the US, ed. Lewis Branscomb, Fumio Kodama, and Richard Florida, 348­64. Cambridge, MA: MIT Press. Yusuf, Shahid, and Kaoru Nabeshima. 2007. How Universities Promote Economic Growth. Washington, DC: World Bank. Zhou, Ji. 2005. Higher Education in China. Singapore: Thomson Learning. Zucker, Lynn G., and Michael R. Darby. 1996. "Star Scientists and Institutional Transforma- tion: Patterns of Invention and Innovation in the Formation of the Biotechnology Industry." Proceedings of the National Academy of Sciences 93 (23): 12709­116. Zucker, Lynne G., Michael R. Darby, and Jeff S. Armstrong. 2002. "Commercializing Knowl- edge: University Science, Knowledge Capture and Firm Performance in Biotechnology." Management Science 48 (1): 138­53. Zucker, Lynne G., Michael R. Darby, and Marilyn B. Brewer. 1998. "Intellectual Human Capital and the Birth of US Biotechnology Enterprises." American Economic Review 88 (1): 290­306. Comment on "The Role of Higher Education in High-Technology Industrial Development: What Can International Experience Tell Us?" by Sachi Hatakenaka ERIK SANDER Over the last 20 years, technology-based multinational corporations (MNCs) and small and medium-size enterprises (SMEs) have increasingly looked to universities to undertake basic research that spawns breakthrough products for the future. This is research that companies are either not equipped to undertake themselves or are unable to fund internally, given their focus on near-term product development that has an impact on the bottom line. Concurrently, high-technology industry across the globe continues to rely on universities to produce an employee population that can be drawn on to fill company pipelines with highly skilled professionals who can contribute to the company's profitability quickly and with less postrecruitment training. In this paper, Hatakenaka provides a strong foundation for discussing the diverse history of government-university-industry collaboration mechanisms across interna- tional borders and highlights the complex variables and key motivators affecting how these groups work together. Essentially, the particular characteristics of each univer- sity, company, and government entity are crucial to how these collaborations have grown (or have failed to grow substantially) in various settings. Whereas in some countries, such as Japan, industries have focused on the academic role of screening and providing foundational training for future long-term employees, in other countries, such as the United States, they have taken a broader viewpoint toward leveraging academic research and educational resources to build the company's tech- nology and employee pipelines. Viewed globally, government-university-industry collaborations can be seen as a jumble of disjointed education, research, and outreach programs. Nevertheless, Hatakenaka clearly identifies the salient variables and moti- vational factors that determine how these collaborations have been birthed and have grown in different countries. The themes of organizational characteristics and responsiveness arise several times in Hatakenaka's discussion of global government-university-industry collaboration Erik Sander is director of industry programs, University of Florida College of Engineering, Gainesville. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank 266 C O M M E N T O N H ATA K E N A K A | 267 models. It can be instructive to extend and reframe these points through a brief discussion of such collaborations from the perspective of cultural barriers and issues, not only across geographic borders, but also across the cultural borders of the gov- ernment, academic, and private sectors. Support for Higher Education One could argue that a key to understanding what drives culture and motivation within higher education is a grasp of the support mechanisms for an institute's research and education mission. In general, many U.S. universities are recognized as having built strong industrial relations through research and education programs, and so it is sensible to study the U.S. system as a baseline for understanding the broader global environment for the effect of universities on high-technology indus- tries across borders. As illustrated in figure 1 and discussed by Hatakenaka, U.S. university research has grown substantially since World War II, primarily in response to new government funding programs established to advance the technological and manufacturing capa- bilities of U.S. industry. Beginning with relatively minuscule funding levels in the 1950s, federal support of U.S. university research today spans myriad government agencies providing more than US$40 billion (in constant 2000 dollars) to assist vir- tually every field of scientific advancement. U.S. research-intensive universities have grown reliant on federal support, which typically funds more than 60 percent of a university's research base (NSF 2008). Government funding agencies have their individual funding profiles; for example, whereas the Defense Advanced Research Projects Agency (DARPA) tends to fund later stages of research and is, not surprisingly, oriented toward defense applications, the National Institutes of Health (NIH) emphasizes the discovery phase. Nevertheless, the greatest part of government support for university research is targeted toward basic research, which is focused on knowledge discovery rather than on specific, predeter- mined market applications. The government typically seeds next-generation discover- ies, some of which industry can ultimately take to market. This prevalence of govern- ment basic research funding dovetails well with university researchers' quest for basic knowledge. The result is that 75 percent of U.S. university research is defined as basic. There is a stark contrast with the profile of high-technology industry research, only 4 percent of which is categorized as basic (NSF 2008). High-technology industry is driven by market economics to direct its internal resources toward applied research and development; it is focused on meeting a specific market need through develop- ment of a product or service and its underlying technology. As Hatakenaka rightly notes, the U.S. government funds basic research at higher education institutions to provide the foundation for fundamental technology breakthroughs for next-generation corporate products--thus establishing a key role for U.S. universities in the overall advancement of technology today and in the future. Universities in several other countries draw a much greater proportion of their funding from industry. For instance, Hatakenaka observes that industry supports more than 268 | ERIK SANDER FIGURE 1. Funding Sources for U.S. University Research 45,000 40,000 millions of constant 2000 U.S. dollars 35,000 30,000 25,000 20,000 15,000 10,000 5,000 0 1965 1970 1975 1980 1985 1990 1995 2000 2005 other nonprofit industry university federal Source: NSF 2008. 30 percent of university research in China and Russia, although one could question whether this is because of greater industry involvement or less government support. Transitioning the Valley of Death Industry's own development of next-generation technology advances must ultimately span the continuum of basic research to applied research and development, although companies in different societies have made conscious decisions as to where they choose to focus in that spectrum. For example, Hatakenaka points out that industry in some countries, including China, has historically chosen to focus on development of new applications of existing technologies, whereas countries such as the United States have distributed resources across the entire spectrum. Whichever model of industrial product development is chosen, one common theme drives resource alloca- tion: a market pull for the technologies being developed. This market pull, entailing a focus on customer needs in research and in product and service design and develop- ment, can produce an efficient allocation of research and development (R&D) resources, but sometimes at the cost of discovery of next-generation technologies to underpin future product lines--essentially, the basic research element. Government-university-industry research collaborations provide an alternate means of technology development whereby the government funds next-generation, C O M M E N T O N H ATA K E N A K A | 269 FIGURE 2. Bridging the Technological "Valley of Death" technological push market pull funding/interest university research Valley of industry development government funding Death industry funding industry--product development university--research commercialization Source: Author's elaboration. high-risk basic research in universities. Often, however, the greatest challenge is to find a home for these technologies in industry. This transition point can be difficult to negotiate because the technology is sometimes too far along for further significant government funding and university development but is not sufficiently developed and risk mitigated to attract substantial industry interest in adoption or funding. Successful government-university-industry research collaborations require the bridging of what has been commonly termed the Valley of Death--the point in a technology's life cycle at which the research is transitioning from an academic basic research program to an applied industrial development program (figure 2). The chal- lenge usually revolves around (a) the university's further development, and therefore risk mitigation, of the technology, perhaps to the point of a demonstrable prototype or initial clinical trials, and (b) industry's willingness to adopt earlier-stage technolo- gies through funding and licensing. This analytical approach can also be applied to the educational mission of higher education if one considers the output of the collaborative process to be university students/company recruits. Although most students do not fall into a true Valley of Death in the same way that university technologies may completely fail to make the transition to the private sector, there is an analogous problem of how best to fill the educational gap between the basic science knowledge of academia and the more specialized knowledge of industry. Kapur and Crowley (2008) raise an important question: is the purpose of higher education to train people for a labor force, or to train a labor force that is in turn trainable by employers? Hatakenaka points out that 270 | ERIK SANDER industry addresses the issue of educational roles through a variety of models across national borders. For instance, Japanese industry has historically relied on universi- ties' stringent screening and selection processes for admission and basic preparation and has, for its part, focused on downstream training of employees who spend most of their careers with a single company. By contrast, U.S. technological MNCs have typically relied on universities to provide a greater share of future employees' training within the academic environment. These two models bridge the educational Valley of Death from different directions--academic and industrial. As in technology develop- ment, the salient point concerns which group takes the lead in refining the raw prod- uct (the student or new employee). Whether research or education is under discussion, efficient bridging of the gap between academia and industry can prove critical for achieving an economic impact through production of high-value employees, products, and services. Hatakenaka makes that clear. Bridging Cultural Barriers to Collaboration The value perceived by all parties to these potential collaborations would seem to sow the seeds for natural partnerships between government funding sources, institutes of higher education, and high-technology industry that seek to fill product pipelines and researcher ranks in the most efficient manner possible. The question then is, why do strong government-university-industry collaborations not form more naturally across international borders? Hatakenaka makes an excellent case that this process is more complex than would appear at first glance. She identifies a number of barriers that she categorizes mostly along the lines of institutional characteristics and responsiveness. A complementary perspective is that of a cultural spectrum spanning the wide range between institutes of higher education and MNCs, with technology SMEs often caught in the middle. Although it is dangerous to generalize, figure 3 illustrates some of the pri- mary cultural barriers that naturally exist between the academic framework of open FIGURE 3. Cultural Barriers to Collaboration between Universities and Industry university culture technology SME industry culture open--publishing closed--proprietary long-term focus shorter-term focus basic research applied R&D disciplinary departments multidisciplinary programs edge of the envelope risk management collaboration primarily internal R&D organized anarchy project management forcus on science forcus on monetary return Source: Author's elaboration. Note: SME, small and medium-size enterprise; R&D, research and development. C O M M E N T O N H ATA K E N A K A | 271 discourse and the more focused and proprietary company environment. It is readily seen that the cultural gaps between universities, small technology-based companies, and large industry players can be manifested in definitive ways that affect collaboration potential, such as timeframes (semesters versus weeks), focus (basic versus applied research), management (decentralized versus hierarchical and project driven), and bottom-line emphasis (discovery versus profits and shareholder value). The challenge, many times, is to overcome these cultural barriers and make full use of university, private sector, and government resources to maximize ultimate economic impact. There are multiple avenues for higher education-industry engagement, as described by Hatakenaka, but not all universities and not all companies and governments are equipped to take advantage of them. Hatakenaka describes this relationship building in terms of responsiveness to meeting industry needs and notes that it succeeds only if it serves the university, industry, and, ultimately, the government sources that provide sub- stantial support to the academic system. Box 1 surveys the numerous potential avenues of interaction between the industrial and academic sectors. The list is not meant to be BOX 1. Pathways for Interaction between Higher Education and High-Technology Industry Leveraging research assets Technology licensing Industry-sponsored research Joint research Facility usage agreements Technology donations Leveraging researcher assets Industrial advisory boards Faculty sabbaticals Faculty consulting Visiting industry researchers Leveraging educational assets Publications Alumni networks Distance education Lifelong learning Short courses, seminars, and conferences Leveraging student assets Proactive student recruitment Student internships/cooperative education Student scholarships Source: Author's elaboration. 272 | ERIK SANDER all-inclusive, but the point is that there are myriad ways in which higher education and high-technology industry can establish mutually beneficial relationships. As Hatakenaka discusses, across national borders, industry does not engage equal- ly with higher education institutions through all avenues, and not all academic units are equally equipped to respond fully to all high-technology companies. For instance, U.S. firms are more likely to have engaged in multiple avenues of collaboration, including technology licensing and industry-sponsored research; Japanese firms have historically not taken advantage of some of these university capabilities but have focused on others, such as student recruitment and in-house training. Even this situation is changing, as Japan is now entering an era of university spin-offs and technology commercialization catalyzed by government programs. Global Best Practices in Public-Private Partnerships The U.S. Science and Technology Policy Institute, with funding from the National Science Foundation (NSF), recently conducted a best practices study of government- university-industry partnerships worldwide to aid understanding of the various col- laboration mechanisms (Lal et al. 2007). A team of institute personnel and various domain experts studied and visited more than 40 public-private partnerships in nine countries. The study focused on five areas of inquiry: program-level vision; center- level organization; industry partnerships; international partnerships; and engineering education. In the research domain, the study confirmed Hatakenaka's broad model of responsiveness. Even within a given country--for example, Germany--the basic research focus of, say, the Max Planck Institutes stands out in contrast to the Fraunhofer Institutes' focus on applied research in close collaboration with industry partners. In the United Kingdom, the Warwick University Manufacturing Group, which is very industrially focused and does outstanding applied research, provides another excellent example of Hatakenaka's model of responsiveness to industrial R&D needs. An outstanding mixed model is Belgium's International Microelectronics Center (IMEC), which has established a robust model of working closely with a large population of MNCs and SMEs through a multitiered association model (see figure 4). In each of these examples, it is important to point out that the technology devel- opment spectrum discussed earlier is not isolated from the student development spec- trum. Students in application-oriented academic centers typically work closely with industrial researchers and are well prepared to make significant contributions as employees of the company or industry. What is untested is whether those students are better prepared in the longer term for a broader spectrum of careers. Conclusion Hatakenaka provides an excellent foundation for exploring the myriad relationships among government agencies, higher education institutions, and high-technology industry across national borders. Her parsing of higher education institutions and high-technology industry by their historical interactions is exemplary and provides a C O M M E N T O N H ATA K E N A K A | 273 FIGURE 4. Differences in Focus among Public-Private Partnerships, Selected Countries: Findings from a Survey ERC (U.S.) focus on basic and curiosity-driven research Global COE (Japan) SRC/ERC/NCRC (Korea, Rep. of) MPI (Germany) SAIT (Korea, Rep. of) WTCC (Japan) INQIE (Japan) Key Labs (China) IMEC (Belgium) CSET (Ireland) IMRC (U.K.) VUTC (U.K.) Fraunhofer (Germany) WMG (U.K.) FMTC (Belgium) IAE (Korea, Rep. of) NERCs (China) focus on technology transfer and industry needs Source: Lal et al. 2007. Note: The institutions shown in the figure are research partnerships in the countries named. The partnerships men- tioned in the text are the International Microelectronics Center (IMEC), the Max Planck Institutes (MPI), the Fraunhofer Institutes (Fraunhofer), and the Warwick University Manufacturing Group (WMG). framework for an expanded discussion of the role of higher education and high- technology industry cultures in establishing collaborative relationships. Inherent culture is a critical variable driving the influence of higher education on high-technology indus- try, and vice versa, and Hatakenaka rightly points out that collaborative mechanisms are varied and differ across borders. There is much that can be added to this discussion, and Hatakenaka has supplied an important first step in a field of study that has until now not gained the attention it deserves. Further studies can profoundly reshape these models of collaboration between higher education and high-technology industry to the benefit of the global economy, especially in the developing world. References Kapur, Devesh, and Megan Crowley. 2008. "Beyond the ABCs: Higher Education and Devel- oping Countries." Working Paper 139, Center for Global Development, Washington, DC. Lal, Bhavya, Craig Boardman, Nate Deshmukh Towery, Jamie Link, and Stephanie Shipp. 2007. Designing the Next Generation of NSF Engineering Research Centers: Insights from Worldwide Practice. Washington, DC: U.S. Science and Technology Policy Institute. NSF (National Science Foundation). 2008. "NSF Science and Engineering Indicators." NSF, Washington, DC. African Higher Education and Industry: What Are the Linkages? AKILAGPA SAWYERR AND BOUBAKAR BARRY Productive linkages between higher education and industry presuppose the existence of institutions able to prepare skilled technical personnel and to produce and trans- fer innovative, industry-relevant research and technology. The necessary counterpart is industry hungry for new knowledge and technology and prepared to invest in developing or acquiring it. Weaknesses on both sides are evident in Sub-Saharan Africa. Higher education has suffered from years of neglect and underfunding and is generally considered to be ori- ented away from science and technology. Although there are signs of a revival, over- all, higher education, particularly as regards science and technology research, is not up to the task of lifting African industry to a globally competitive level. On the other side, high-technology industry is largely absent in the region. This paper consequently focuses on small and medium-size industry, which is important for Africa's immediate future. But given the lack of steady pressure to acquire new knowledge and technology so as to remain competitive, there is little demand from this sector for industry-relevant research by knowledge institutions. Part of the solution to this impasse must lie in a policy framework and material interventions aimed at boosting research capacity and industrial development in priority areas while at the same time providing appropriate bridging mechanisms and supportive fiscal and other incentives. There is also scope for regional and continental collaboration to harvest economies of scale. Massive transformations in global communication, production and management in the second half of the twentieth century placed knowledge and its applications at the center of economic and social development in all parts of the world by Akilagpa Sawyerr is a former secretary-general of the Association of African Universities. Boubakar Barry is coordina- tor of the Research and Education Networking Unit, Association of African Universities. Ransford Bekoe, assistant proj- ect officer in the Office of the Secretary-General, Association of African Universities, provided substantial research sup- port in the preparation of this paper. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank 275 276 | A K I L A G PA S A W Y E R R A N D B O U B A K A R B A R RY integrating the generation and application of advanced knowledge into material and cultural production to an extent unknown in history. This evolution has given added significance to systems and institutions for knowledge production. The spectacular economic and social developments of recent decades, especially in the newly industrializing countries of Southeast Asia, have gone hand in hand with rapid and substantial expansion of higher education and skills training. Although the exact character of the nexus between knowledge production and economic and social development remains unclear, there is little doubt about its existence or about its reciprocal nature, with knowledge production and the con- ditions and operations of the productive sectors of economy and society influ- encing each other. To introduce our examination of one aspect of this general question, we first briefly review the key features of linkages between higher education and industry and the conditions for their success. This will set the background for the examination and assessment of the specific case of Africa. Channels and Conditions of Interaction Although there is no reason to expect all kinds of industry to interact with the knowl- edge production process in exactly the same way, even in the same country, the general pathways of interaction are clear. On the side of knowledge production institutions, the contribution to industry takes a variety of forms and moves through a variety of chan- nels (see Brennenraedts, Bekkers, and Verspagen 2006, 4).1 The main ones are · Preparation of well-educated technical graduates, ensuring a cadre of skilled and trainable personnel for industry, especially in the fields of science, engineering, and management · Pursuit of fundamental as well as applications-oriented research that produces industrially relevant new knowledge, including management science · Direct generalized transmission of research products through publications, patents, prototypes, and the like · Direct targeted transmission through conferences, workshops, meetings, net- works, and other formal and informal interactions · Indirect transmission in the course of collaborative ventures and exchanges of personnel · Transmission under commercial arrangements such as research contracts, licenses, and consultancies. With regard to the provision of skilled cadres, the outcome turns, in the first instance, on the number, variety, and quality of technical graduates produced by the education and training system and available for absorption into industry as managers, skilled and semiskilled technical workers, and researchers. Equally sig- nificant for successful linkage, although often ignored, is the capacity (or other- wise) of industry to absorb the products of the universities and training institutions and put them to productive use. A F R I C A N H I G H E R E D U C AT I O N A N D I N D U S T RY: W H AT A R E T H E L I N K A G E S ? | 277 For meaningful interaction around the production of knowledge and its injec- tion into industry, there needs to be (a) a basis for interaction, namely, relevant knowledge and awareness of the need for it; (b) a disposition for productive engagement; and (c) a supportive policy framework. Capable Knowledge Producers In the absence of knowledge institutions with both the capability to produce usable knowledge and the orientation and ability to transfer it to industry, the supply side will be absent or weak. Thus, the higher education and training system must have the capability and flexibility to produce and supply technical graduates on a suitable scale and in fields relevant to industry and its evolving needs. Such graduates will staff the managerial and technical levels of industry and, in the case of small and medium-size industry, even provide entrepreneurial leadership. To meet the special needs of high-technology industry, these graduates must include a substantial pro- portion of PhDs in mathematics and the various branches of engineering to ensure high-level appreciation of emerging technologies and the possibilities for innovation. Even more than in the supply and absorption of graduates, the linkage between higher education and industry is often perceived in the production of knowledge and its injection into production and management as a basis for technological and man- agerial innovation. A key condition for this is the capacity of the knowledge produc- tion centers, through cutting-edge as well as applications-oriented research, to pro- duce usable new knowledge and new applications. Institutions that lack high-quality research capability thus have no part in such interactions. Yet the capacity to produce the relevant knowledge is not sufficient; it must be combined with the appropriate orientation and the ability to transmit knowledge to potential users. Academic insti- tutions, through the publication of research findings, through conference presenta- tions, and through the taking out of patents, do much of this in the normal course of their work. The critical factor is to go beyond this and engage in more deliberate and targeted transmission to industry and other potential users of the research product or, indeed, to pursue coproduction of knowledge with users. Modes of targeted transmission include workshops, meetings, networks, and other formal and informal interactions; collaborative ventures and exchanges of personnel between industry and knowledge centers; and commercial arrangements such as research contracts, licenses, and consultancies (Brennenraedts, Bekkers, and Verspagen 2006, 15­16). To help discharge this extension function effectively, institutions frequently provide special incentives to their staffs and set up specific structures, such as consultancy or commercialization units. Receptive Industry For its part, industry must adopt a posture of active receptiveness for there to be suc- cessful interaction. Where industry feels no need for new knowledge or technology, or is unwilling or unable to invest time or resources in its acquisition, there will be 278 | A K I L A G PA S A W Y E R R A N D B O U B A K A R B A R RY no effective demand for locally produced knowledge and technology. There must be a perception of need, recognition that the proffered or potential knowledge products will meet that need, and preparedness to invest in the acquisition of such products. In the case of skilled technical and managerial cadres, this is usually not a problem-- certainly not for high-technology and medium-size enterprises, which need such cadres if they are to be minimally effective. In relation to the acquisition of knowl- edge, however, a distinction is usually drawn between, on the one hand, high- technology industries and enterprises engaged in the export of manufactures, which need such knowledge in order to raise their productivity and international competi- tiveness, and, on the other hand, small, low-technology enterprises, which do not per- ceive such a need. In the latter instance, the absence of a technology culture--which develops out of the pressure for constant updating and deepening of technology in order to survive competition--means that there is little disposition to invest in the acquisition of new knowledge for upgrading or adapting current technology beyond what is embodied in new machinery and equipment (World Bank 2009). Even in the case of high-technology producers and export-oriented manufacturers, whether and under what conditions they will engage with local knowledge produc- ers for needed knowledge depends on several factors. For local branches of transna- tional enterprises, the primary source of research and new technology is the parent company and its branches and affiliates, rendering marginal any local knowledge production capacity, a situation that often applies in Africa. In other instances, a good deal turns on the existence or otherwise of a technology culture, as well as the degree to which local knowledge producers are perceived as credible partners from whom to procure knowledge or with whom to collaborate in the coproduction of knowledge. As will be seen from the consideration of the case material, the lack of these preconditions constitutes a major barrier to higher education­industry linkages in Sub-Saharan Africa. An element that runs through all discussion of higher education­industry linkages is the importance of proactiveness on both sides. Universities and research centers must provide incentives and set up structures aimed at promoting and facilitating linkages with industry, even as industry must be actively receptive. There is evidence, however, that, even though university interest is essential for engagement, successful linkages have come into being and have survived principally through the commitment of industry and its investment of time and resources in specific linkage arrangements, whether by seeking collaboration, commissioning research, setting up an incubator or a research laboratory in a university or center, receiving student interns, or releas- ing staff as adjuncts to institutions. Indeed, it may be stated with some force that without vibrant industry driven by the pressure to increase productivity and lower costs in order to thrive in a fiercely competitive, usually export, market, the drive to innovate through the application of new knowledge and new technology remains weak, and recourse to knowledge institutions for support or collaboration is highly unlikely.2 In such conditions, the response to overtures from the knowledge institu- tions is bound to be lukewarm, at best. This must go a long way toward explaining why the many university-initiated measures aimed at promoting and facilitating linkages with industry have not achieved long-term success (see box 1 and the accom- panying text). A F R I C A N H I G H E R E D U C AT I O N A N D I N D U S T RY: W H AT A R E T H E L I N K A G E S ? | 279 BOX 1. Selected Examples of University-Industry Partnerships in Africa Makerere University, Uganda · The Uganda Gatsby Trust, based at Makerere's Faculty of Technology, is a not- for profit entity established in 1994 with seed funding from the Gatsby Chari- table Foundation (GCF) in the United Kingdom. The trust seeks to support manufacturing and value-adding businesses with growth potential. Its activities include training courses, business development services, student internships, technology development and transfer, and a business park. · The Makerere University Private Sector Forum serves as a platform for interac- tion between the university and the private sector, with the mission of encourag- ing support for university programs and promoting entrepreneurship training, joint research, curriculum review, technology innovation and transfer, and devel- opment of business clusters. University of Dar es Salaam, Tanzania The Institute of Production Innovation was founded in 1981 to carry out product innovation, transfer innovations to industry, and provide technical consultancy for enterprises. Kwame Nkrumah University of Science and Technology, Ghana The Technology Consultancy Centre was originally established in 1972 as a pro- duction unit of the University of Science and Technology in Kumasi. It serves as a conduit through which university research is made available to industry. Most of its clients are small and medium-size enterprises in the informal sector, and the technologies transferred are mainly in food processing, fabrication of small-scale machinery and parts, and ceramics manufacture and foundry works. Source: Tindimubona (2000, 18, 19); UNCTAD (2003, 57, 70). For further information on the Uganda Gatsby Trust, see http://gatsbyuganda.com/; for information on the Tech- nology Consultancy Centre, see http://www.knust.edu.gh/tcc/index.htm. Supportive Policy Framework A third general condition for the success of higher education­industry linkages is a sup- portive public policy framework. Experience shows that appropriate public policy inter- ventions are often necessary to strengthen both the supply and demand sides and to bridge the gap between knowledge producers and industry. Such interventions take a variety of forms: for example, creating the necessary competences on the supply side, mainly through support for graduate training and research in selected priority areas in universities and research centers; establishing official structures to facilitate collabora- tion and information flow; and offering direct incentives to stimulate industry demand and receptivity to local knowledge workers and products. In some instances, the state has collaborated directly, in a variety of ways, with universities and industry. 280 | A K I L A G PA S A W Y E R R A N D B O U B A K A R B A R RY The African Scene: Background Some clarification is necessary to help frame this review of the African situation. First, the discussion is essentially limited to Sub-Saharan Africa, excluding South Africa. This is because the histories and levels of development of the Maghreb coun- tries and South Africa are sufficiently different from the rest of Africa to make mean- ingful generalizations across the continent difficult for the purposes of our topic. On the other hand, even though the rest of Africa is by no means homogenous, there is enough in common to allow for reasonable generalizations. Second, an undertaking such as this is faced with the usual difficulties of lack of adequate and consistent national and intercountry data. The problem is compound- ed in this instance by the virtual absence of literature dealing with the actual forms and structures of such university-industry interaction as there is, in contrast to the burgeoning literature elsewhere. In the circumstances, we have been fortunate to come across a number of country case studies that provide us with useful, if limited, material on which to proceed. Third, given the near-absence of high-technology industry (see "Industry in Africa," below), the prevalence of small and middle-size industry, and the critical role of the latter in Africa's development, our presentation will deal primarily with such industry, concentrating on pathways between industry and knowledge production centers and the extent and potential of such linkages. Finally, although we recognize the contribution of university-based management consulting to organizational improvement and to productivity increases, the topic does not receive direct attention in our discussion. This is largely because it is so sparsely covered in the literature. Higher Education in Africa As indicated above, the key conditions for effective higher education­industry link- ages include (a) the capability of higher education to produce skilled technical and managerial cadres and usable knowledge and (b) the capacity and disposition of higher education institutions to transmit such knowledge to end users. To assess these conditions, it is necessary to review briefly the state of African higher education and its relationship with industry. Capability Most African higher education institutions, from their colonial or immediate post- colonial beginnings, have been oriented toward the liberal arts and basic sciences. This orientation, which suited the immediate postcolonial agenda of providing a cadre of leaders and workers to run the public and private sectors of the newly inde- pendent states, was, by and large, effective for the first decade or so of independence. Before the end of that period, however, a clamor arose to alter the mission of the almost exclusively public universities and training institutions to make them more A F R I C A N H I G H E R E D U C AT I O N A N D I N D U S T RY: W H AT A R E T H E L I N K A G E S ? | 281 development oriented. This campaign culminated in a call by Africa's higher educa- tion leaders for the transformation of the leading African universities into "develop- mental universities," as voiced at two major events: a United Nations­sponsored 1962 conference in Tananarive, and a workshop entitled "Creating the African University: Emerging Issues in the 1970s," organized by the Association of African Universities in Accra in 1972 (see Sawyerr 2004).3 In response to the changing global environment and the increasing centrality of knowledge to production and social development, the call for developmentalism became insistent, and steps were taken in many places to give it meaning. Universi- ties of science and technology and of agriculture were established; new emphasis was placed on polytechnics and technical training; and in the early 1990s technology parks began to be set up around universities and other higher education institutions. These attempts to reorient higher education in Africa have to date met with very limited success, partly because of the persistence of the ingrained orientation and practices of the leading universities but also because of lack of complementary trans- formations in economy and policy. What did change by the end of the 1980s was the profile of the typical higher education system, reshaped by an explosion of enrollments, an aging faculty, and reduced research and graduate study programs. The resulting deterioration was com- pounded by a combination of neglect and underfunding, as well as policy errors at both the system level and the institution level. By the late 1990s, most African uni- versities were in dire straits, even the region's premier institutions, which had, in their prime, vied with the best in the world. (For fuller discussion of these issues and rele- vant references, see Sawyerr 2004; World Bank 2009.) The situation is perhaps best captured in a set of paradoxes: · The number of universities expanded dramatically following the wave of political independence, from 52 in 1960 to over 400 in 2000, and it continues to climb steeply, especially on the private side. Yet many countries still have only one, two, or three universities, and there is little diversity in institutional forms and functions. · University enrollments rose fourfold between 1975 and 1985, almost tripled again during the next decade, and are still rising. Yet, with an average participation rate of barely 5 percent, pressure for expansion of enrollment is unrelenting. · Expenditure per student as a proportion of gross national product (GNP)--that is, in relation to national wealth--is far higher in Africa than anywhere else in the world. Yet, owing to a combination of difficult economic conditions in most coun- tries and the fact that most universities depend on public support, the provision of resources has not kept pace with enrollment. Another problem relevant to our discussion is the aging of the faculty in virtually all universities, as senior faculty retire without being replaced at the rate required to maintain the appropriate levels of leadership of research and graduate programs. This can be seen from the following selective figures: · Since 1990, according to the International Organization for Migration, Africa has been losing an estimated 20,000 professionals each year through migration, many of them academics. 282 | A K I L A G PA S A W Y E R R A N D B O U B A K A R B A R RY · In 2003 Makerere University in Uganda had an estimated vacancy rate of about 40 percent, as did the Ghanaian university system. · In Nigeria 400 professors (45 percent of the top-level professoriate) were expect- ed to reach the mandatory retirement age in 2008. · In South Africa 50 percent of the academic staff was over age 50 in 2004. · Only a third of the academic staff at Makerere University held doctoral degrees as of December 2006. · In 2008, for the first time, teaching positions at the University of Dar es Salaam were to be filled by staff with only a bachelor's degree--about 91 positions were involved. The overall effects include · Heavy teaching loads resulting from hugely expanded enrollments, unmatched by commensurate faculty increases or the benefit of modern teaching aids · Involvement of academic staff in nonacademic activities as a means of supplement- ing low official incomes, cutting down on time for staff development and research · The absence of mentoring by senior colleagues and the lack of a critical mass of other researchers, heightening the isolation of young researchers and reducing the scope for learning on the job. All these have obvious implications for the quality of teaching, learning, and research. There have, of course, been remarkable instances of coping and pockets of signif- icant revitalization. Particularly in the last few years, there has also been a general awakening at the institutional, national, and continental levels to the need for cor- rective action. Nevertheless, the general situation remains problematic. This is in part because the attempt at revival has coincided with the rise of the knowledge society and the imperative of global competitiveness. The consequence is that Africa's insti- tutions, unlike most others in the world, face a dual challenge: · The old task of producing experts for managing the essentially primary-production economy and developing-country social conditions, and · The new task of providing cutting-edge knowledge, advanced management, and innovation to develop and manage a modern society and economy in the context of global competitiveness. Although this picture of degeneration in the conditions for knowledge production and innovation may appear unduly bleak, it highlights the challenges, the better to enable us to appreciate the limited capability of the higher education system to respond effectively to the needs of modern industry. Turning directly to the matter of university-industry linkages, Sub-Saharan Africa has the lowest tertiary enrollment rate in the world, particularly in the scientific and technical fields (tables 1 and 2). The world average gross tertiary enrollment rate is 24 percent, and that of North America and Europe is 70 percent, but in Africa it is only 5 percent. In 1995 Sub- Saharan Africa had a tertiary-level enrollment rate of 0.28 percent in technical fields; the developing-country average was 0.82 percent, and that for developed countries A F R I C A N H I G H E R E D U C AT I O N A N D I N D U S T RY: W H AT A R E T H E L I N K A G E S ? | 283 TABLE 1. Tertiary Gross Enrollment Ratio (GER), 2004 Region GER (percent) Arab states 21 Central and Eastern Europe 54 Central Asia 25 East Asia and the Pacific 23 Latin America and the Caribbean 28 North America and Western Europe 70 South and West Asia 11 Sub-Saharan Africa 5 World (average) 24 Source: UNESCO 2006. was 4.06 percent (table 2). From another angle, Sub-Saharan Africa, with 12 percent of the developing world's population, accounts for only 3.1 percent of the technical tertiary enrollments and 1.7 percent of the engineering enrollments in developing countries. The total number of engineers enrolled in the whole of Sub-Saharan Africa represents about 12 percent of the corresponding figure for the Republic of Korea alone. In 2002 the three universities with substantial technology programs in Uganda had approximately 1,000 students enrolled in four-year engineering degree programs; in Australia, a country with a population one-third smaller than Uganda's, the figure was 53,000. Orientation In a comment that captures conditions throughout Sub-Saharan Africa, a report by the United Nations Conference on Trade and Development (UNCTAD) observes, "The skill base is very weak and the educational system is generally not geared to meeting the skill needs of industrial competitiveness" (UNCTAD 2003, 16). In fair- ness to higher education institutions, it must be acknowledged that most are fully aware of this situation and have made efforts to improve it, through internal meas- ures and by reaching out to industry. In many instances outreach to the productive sector has been undertaken both as an extension of the conventional objective of community service and as a means of generating income. Outreach efforts include incentives for staff to undertake industry-relevant research, accept research contracts, and provide consultancy and other services; structures to facilitate interaction with industry; technology-licensing offices to market inventions; and, in some cases, spin- off companies for the commercial exploitation of research outcomes and fuller inte- gration into the productive sector. Some notable initiatives are Makerere University's Uganda Gatsby Trust, the University of Dar es Salaam's Institute of Production Inno- vation, and the Technology Consultancy Centre of the Kwame Nkrumah University of Science and Technology (see box 1). Similar examples can be found just about any- where on the continent (World Bank 2009). 284 TABLE 2. Total Tertiary-Level Enrollment in Technical Subjects, 1995 Tertiary-level enrollment Technical subjects Total Percent of All technical Natural Mathematics/ Country group/country students population subjects sciences computing Engineering Developing countries 35,345,800 0.82 7,021,929 2,046,566 780,930 4,194,433 Sub-Saharan Africa 1,542,700 0.28 220,660 111,500 39,330 69,830 Ghana 9,600 0.06 2,100 1,200 200 700 Kenya 31,300 0.12 4,600 3,600 -- 1,000 Mauritius 5,500 0.49 500 100 100 300 South Africa 617,900 1.49 72,200 21,700 30,500 20,000 Tanzania 12,800 0.04 3,600 800 100 2,700 Uganda 27,600 0.14 2,600 800 300 1,500 Zimbabwe 45,600 0.41 9,700 2,200 800 6,700 Latin America 7,677,800 1.64 1,404,402 212,901 188,800 1,002,701 Asia 21,553,400 0.72 4,584,300 1,513,100 438,600 2,632,600 Transition economies 2,025,800 1.95 440,800 55,500 30,600 354,700 Developed countries 33,774,800 4.06 5,754,419 1,509,334 1,053,913 3,191,172 Source: UNCTAD 2003. Note: --, not available. A F R I C A N H I G H E R E D U C AT I O N A N D I N D U S T RY: W H AT A R E T H E L I N K A G E S ? | 285 In the end, these initiatives have proved neither substantial nor sustainable, for reasons to do partly with the histories and ways of work of the knowledge insti- tutions, which make it difficult to fully integrate such initiatives within the insti- tutions. Other factors are the state and orientation of industry, as will emerge from the discussion that follows, and the overall policy framework within which the interaction takes place, especially dependence of research projects on donor funding for survival. Turning to the institutions, the problematic features relate to curriculum coverage and enrollment patterns, the extent and nature of research activities, and attitudes toward involvement with industry and commerce. Curriculum and enrollment patterns. Enrollment is still heavily biased in favor of lib- eral arts courses, with not enough science students and even fewer in engineering and mathematics (see table 2). The roots of the problem lie largely outside the university, in that secondary schools do not turn out enough science-ready university entrants. Moreover, because the unit cost of producing science and technology (S&T) gradu- ates is so much higher than for liberal arts graduates, most public universities and virtually all the private universities, which account for a large part of the recent insti- tutional expansion, concentrate on non-S&T courses. The upshot is that the pool of science-trained graduates from which industry would draw its high-level technical workers is much shallower than in other regions of the world. In addition, the teaching and learning process tends not to emphasize problem-solving skills or exposure to real-life situations. There are notable exceptions, especially in the medical schools and in those institutions where problem-based learning, community work, and industrial attachment still form part of teaching and learning. The situation in most institutions which had these programs in the past is that they are no longer affordable, especially in view of bloated enrollments, overcrowded classrooms and laboratories, and grossly inadequate staffing. A particularly serious aspect of this situation is the absence or virtual collapse of graduate study in most universities. Although the issue affects all parts of the aca- demic enterprise and research culture, the relevant point here is that African coun- tries are turning out neither the numbers of PhDs in mathematics and engineering required for high-technology take-off nor enough of the holders of master's degrees needed to provide leadership in small and medium-size enterprises if African economies and societies are to be globally competitive. This is not to say that a shortage of science graduates with problem-solving skills or PhDs is currently handicapping local industry. There is evidence to the contrary, with the arrival of the "unemployed science graduate." Although the nub of the issue appears to lie in the absorptive capacity of local industry (which we address later), the point to note is that African institutions are currently in no position to provide the depth and range of basic science and high-level problem-solving skills needed to support industrial development of the sort and at the levels required to move African societies out of poverty and into a solid competitive posture such as is demanded by the times. Research. In the second place, there is not enough research, basic or applied, in African universities, and what does exist is of neither the type nor the quality to 286 | A K I L A G PA S A W Y E R R A N D B O U B A K A R B A R RY attract the interest of industry or to lift it to a higher level (World Bank 2009). The problem is partly attributable to the quality of human resources available at the uni- versities and the incentives for undertaking industry-relevant research. Many faculty members do not hold PhDs and--an especially serious shortcoming in technological fields--time for research is constricted by heavy teaching loads. Issues such as the development of new products and successful linkages with industry are rarely taken into account for purposes of promotion and career advancement, creating a major disincentive to engaging in research and development (R&D) activities that do not lead to scientific publication. A further issue concerns the relevance of such research as is undertaken. Because of the "publish or perish" syndrome, research agendas tend to focus on purely aca- demic and narrowly scientific objectives so as to ensure publication in refereed jour- nals, with little regard for developmental needs. This situation is compounded by undue dependence on donor funding for research, which tends not to be industry oriented. Not surprisingly, the results of such research are rarely relevant to local entrepreneurs and industries, which, as a consequence, feel no inclination to forge partnerships with local researchers or their institutions. Attitudes. In the third place, and linked to some of the factors identified above, there is little understanding in the university setting of the ways of work and the constraints of industry, of what industry considers important, or of what it takes to prepare for and work effectively with it. In the absence of such understanding, it is virtually impossible for the two sides to act proactively in planning and undertaking collabo- rative work (van Zyl, Amadi-Echendu, and Bothma 2007, 2). There are, of course, attempts to get around this problem by establishing structures, including special cen- ters, within the universities , for facilitating and promoting industrial outreach and interaction (see box 1). But by and large, this is not enough to overcome the drag of the factors mentioned above. Finally, there is a line of thinking, particularly within the leading public universi- ties, which holds that the role of universities is to develop knowledge through research, publication of research results, and the training of students, not to solve routine problems. In such circles, research commissioned by or targeted to industry is often regarded as "not real science." This attitude is linked to the insistence on pub- lication in peer-reviewed journals as the basis for academic career advancement. Industry in Africa The basic unit of technological activity is the industrial firm, which is responsible for the acquisition, development, adaptation, and use of technology (UNCTAD 2003, 1). The firm's activities, particularly in a competitive setting, stimulate innovation in tech- nology and its applications, which usually drives the higher education and research sector. Thus, the presence and activities of such firms are key to the technological and broader development of any society in the contemporary world. Experience shows that the factors driving technological development in such firms include competition and trade policies, governing rules and regulations, physical infrastructure, skills, and A F R I C A N H I G H E R E D U C AT I O N A N D I N D U S T RY: W H AT A R E T H E L I N K A G E S ? | 287 financing (UNCTAD 2003, 2). Among these, skills acquired through high-level and specialized training and knowledge generated from relevant research are major deter- minants of success in the new competition. Against this background, what can we say about industrial firms and the condition and structure of industry in general in Africa, in relation to technology develop- ment and knowledge acquisition? As previously noted, the relevant data and literature are thin. We therefore rely heavily on country case studies of Kenya, Ghana, Uganda, and Tanzania published by UNCTAD in 2003. Although these four cases cannot represent the total situation on the continent, the main findings are confirmed by other studies (see, for example, World Bank 2009) and are sufficiently robust to form the basis of comments on Africa outside South Africa and the Maghreb. By and large, African manufacturing industry, as portrayed in the case study coun- tries, is dominated by low-level processing of natural resources and the production of simple consumer goods for local consumption. Describing a typical situation, the UNCTAD study observed, [Ghana's] industry structure consists mainly of a large refining industry and aluminum smelting facility, with very limited manufacturing capacity. The modern sector, largely owned by foreign companies, is concentrated on food processing and on industrial inter- mediaries manufactures. . . . Most of the semi-modern enterprises are relatively small- scale African-owned operations running on simple machinery and on low-level techni- cal and managerial skills. They generally produce lower-quality goods geared to the domestic market. The informal sector is entirely African and operates on even simpler technologies. (UNCTAD 2003, 43) Thus, rather like Africa's systems of higher education, Africa's industry has remained substantially untransformed--in this case, from its colonial raw material production and export origins. At a time when world trade has shifted strongly from resource- based to technology-based commodities, Africa has the highest proportion of resource-based exports, and its share of high-technology products is the lowest. Pri- mary products account for over three-quarters of all exports, with two-thirds of export revenues coming from petroleum alone. The shares of engineering, food, and garment exports are relatively low, except for Mauritius (World Bank 2009). Further- more, there is hardly any high-technology (i.e., truly R&D intensive) industry such as wireless technology, advanced biotechnology, materials science, and so on. What industry there is consists essentially of light manufacturing, often involving simple and relatively self-contained engineering activities and agroprocessing, both largely reliant on technology embodied in simple equipment, and requiring limited skills. Not surprisingly, manufacturing value added in Sub-Saharan Africa, excluding South Africa, grew only 0.1 percent per year over the period 1990­97 and has stagnated at 0.4 percent as a share of global manufacturing value added, at a time when the developing-country share of global manufacturing exports had grown from 9.6 percent in 1980 to 26.9 percent in 2000 (UNCTAD 2003, 2, 6). Nevertheless, as long as these medium-technology enterprises remain the most sig- nificant forms of industry in Africa, their performance will be vital to the develop- ment and competitiveness of African economies. Improvements in their productivity and coverage through engagement with local and foreign knowledge and technology 288 | A K I L A G PA S A W Y E R R A N D B O U B A K A R B A R RY sources are therefore of the first importance. That is the perspective that shapes the discussion that follows. The key findings of the UNCTAD case studies can be summarized as follow: · In general, industrial structure is characterized by an emphasis on low-technology products, with little evidence of technology deepening or the adaptation of foreign technologies to generate improved or new products. · The passive importation of relatively simple technologies and their use at relatively low levels of technical efficiency, as well as the relative absence of local capital goods production and design engineering, mean that technological learning and diffusion are very limited. · Employment of technically qualified personnel such as trained engineers is low, and in-house training is restricted to creating the basic skills needed to operate equipment and simple systems.4 · In general, firms show little awareness of the importance of science and technology activities and of technological capacity for ensuring industrial competitiveness. This reflects the absence of what may be termed a "technology culture." As a con- sequence, firms are reluctant to pay for technology services, nor do they spend enough on R&D. · R&D spending in Sub-Saharan Africa was 0.28 percent of GNP in 1997, as against a developing country average of 0.39 and a world average of 0.92 (table 3). Indeed, formal R&D is confined to a very few large enterprises, while the bulk of R&D is conducted in public research institutions and universities, mainly with public funding. On average, in both Ghana and Kenya, for instance, the government funds 90 percent of what little R&D expenditure there is (UNCTAD 2003, 24, 49). · Links with technology institutions are restricted to necessary, basic activities, such as mandatory certification of products or materials testing. There is very little research contracting or use of institutions to search for and adapt new foreign technologies. TABLE 3. Research and Development (R&D), by Major Country Group, c. 1997 Scientists and engineers in research and development Per million Total R&D Country group or region inhabitants Numbers (percent of GNP) Industrial market economiesa 1,102 2,704,205 1.94 Developing economiesb 514 1,034,333 0.39 Sub-Saharan Africa, excluding South Africa 83 3,193 0.28 North Africa 423 29,675 0.40 Latin America and Caribbean 339 107,508 0.45 Asia, excluding Japan 783 893,957 0.72 World (79­84 countries) 1,304 4,684,700 0.92 Source: UNCTAD 2003. Note: GNP, gross national product. a. United States, Canada, Western Europe, Japan, Australia, and New Zealand. b. Includes the Middle Eastern oil-producing countries, Turkey, Israel, and the former socialist economies in Asia. A F R I C A N H I G H E R E D U C AT I O N A N D I N D U S T RY: W H AT A R E T H E L I N K A G E S ? | 289 Considering its percentage of the world's total population, it is estimated that Africa undertakes only about a tenth of the R&D it ought to be conducting. Where R&D does occur, the research output is often prematurely abandoned and is rarely trans- formed into usable technologies (Massaquoi 2001, 6). Ironically, the low volume of R&D and the low rate of transformation of research results into practical use lead to low returns to R&D investment in Africa, compared with those in the developed and emerging countries, which in turn, leads to low investments, in a downward spiral. This general absence of a technology culture is attributable, in part, to the lack of competition that would spur demand for better and more efficient technology, which in turn could stimulate science and technology institutions to create more advanced, useful, and affordable products. Most countries have extensive public infrastructure devoted to the production of industrially useful knowledge and technology and their transmission to industry through research centers, industrial promotion units, and funding arrangements. Yet the evidence from the case studies and other sources is that, by and large, firms in Africa are unaware of the technological services or capabilities available within the official technology infrastructure and universities or regard them as irrelevant or inef- ficient. There are thus few links with those institutions and such facilities and resources as they have (UNCTAD 2003, 34). For all these reasons--in particular, the generally weak skills base, the absence of a technology culture, and the resulting low inflows of high technology--most indus- tries lack the ability to absorb sophisticated technologies, which constitute the cut- ting edge of industrial dynamism and competitiveness. Furthermore, especially because of the low level of in-house technological activity, firms cannot always iden- tify their specific technological needs and so lack the necessary basis for seeking, acquiring, and absorbing essential knowledge and technology inputs through research contracting, collaborative ventures, and the like. This situation, together with the generally weak drive for technological innova- tion, constitute a formidable barrier to linkages with agencies that can conduct research for or in collaboration with industry or that have the means to facilitate such research. That, we would argue, constitutes the main constraint on the provision of technology support by research and development institutions. As succinctly put in the UNCTAD study, Where such [in-house] capabilities are lacking, even good technology institutions can do little. (UNCTAD 2003, 33) Without a more dynamic technology culture in private industry, an institution such as the Kenya Industrial Research and Development Institute cannot provide much more than mundane technical services. (UNCTAD 2003, 31) A final feature of note is that the principal industries in Africa are for the most part affiliates or subsidiaries of multinational companies established in developed coun- tries and operate mainly in the primary sector and in services. These affiliates and subsidiaries carry out very little research on the continent, almost all the needed research being undertaken at headquarters or other locations of the parent compa- nies. In consequence, their presence has little impact on domestic technology devel- opment (World Bank 2009). 290 | A K I L A G PA S A W Y E R R A N D B O U B A K A R B A R RY Concluding Comment It is difficult to bring such a wide-ranging discussion to a neat conclusion, and there is not much point in offering "recommendations." What we do instead is to suggest some initial ideas aimed at encouraging a systematic approach to what ought to be a burning national and continental issue: the mobilization of Africa's knowledge resources in support of its sustained development in the current global dispensation. Given the lack of data and of basic understanding on such a crucial issue, a first step has to be the establishment, preferably in each country and certainly in each region, of a well-resourced "observatory" involving industry, government, the uni- versities, and the technology institutions. The primary task of this body should be the thorough study and proper observation and visioning of this phenomenon, namely, that of mobilizing and applying Africa's knowledge resources, as it evolves over the coming decade. Experience elsewhere should be carefully examined and adapted to the changing African realities. An obvious second thrust is the revitalization and strengthening of Africa's uni- versities and the facilitation of their repositioning to provide the intellectual sinews of a new drive for African development on all fronts. The case for such a revitaliza- tion drive is now generally accepted, nationally, continentally, and among Africa's donor partners. The principal gain will be the combination of expanded access with quality and lifelong learning skills, along with the general spread of the scientific approach to all aspects of life. This will have immediate significance not only for the conditions of individual and social life but also for the quality of the labor force as a whole and, thus, for the productive sector, including industry. The improvement of general education should be complemented by a special pro- gram for the strengthening of staff quality, research, and graduate study in carefully selected priority areas aligned with specific national industrial development objec- tives. The determination of priority development areas should be approached as an issue of critical national policy, involving detailed technical and sociological studies and wide consultation. In some circumstances, a subregional approach will offer the best options. Necessary intermediary arrangements, including the provision of pub- licly supported laboratories and other facilities, should be made, both to support priority-area research and to promote linkages to targeted industrial activities. A crucial factor will be the stimulation of a technology culture in industry in general and especially in the targeted sectors. As appears from all that has been said above, without such a culture the demand side of the knowledge-industry interaction will remain so weak as to render the linkage anemic and unsustainable, providing no impe- tus to industrial innovation and competitiveness. Part of the answer will consist in the measured exposure of industry to internal and external competition. But note should also be taken of the lessons from the positive East Asian experience, as well as from Africa's not-so-positive experience: at this stage of maturity, domestic industry, exposed to unrestrained foreign competition, faces the danger of decimation, and, therefore, a policy of measured exposure must incorporate specific and targeted state support for local industry and enterprise if it is to elicit the appropriate responses (UNCTAD 2003, 4; A F R I C A N H I G H E R E D U C AT I O N A N D I N D U S T RY: W H AT A R E T H E L I N K A G E S ? | 291 World Bank 2009). What is needed, drawing on East Asian experience, is a regime of carefully considered and proactive state support and incentives, aimed at building up a strong technology culture in local industry, with special emphasis on priority areas, and at raising the capacity of industry to compete locally and internationally. State- supported apprenticeships for engineers and technologists, and encouragement of firms to develop active in-house and external staff training and upgrading, can make a major contribution. Industry should be facilitated increasingly to engage with and reinforce the local knowledge and technology infrastructure, in what was known in Brazil as the "Sabato Triangle," involving close collaboration among government, universities, and industry. To give this entire process the necessary weight and urgency, it must be champi- oned by the highest political levels. In other countries this has been achieved by fram- ing it as a special presidential or prime ministerial initiative. Something similar should be done in Africa. Although the primary locus of the effort must be national, the scale of the prob- lem, in terms of resources, skill pools, and markets, transcends national boundaries. In this connection, it is encouraging that the African Union is providing continental leadership in the revitalization of higher education. The new Second Decade for Edu- cation in Africa (2006­15), declared in 2007, includes, for the first time, a compo- nent on higher education, signaling the African Union's recognition of the vital role that higher education must play in the region's development agenda. In relation, specifically, to science and technology, this new awareness is symbolized by the Science and Technology Consolidated Plan of Action launched in 2005 by the African Union and the secretariat of the New Partnership for Africa's Development (NEPAD). This plan (NEPAD 2006) is aimed at developing and using science and technology for the socioeconomic transformation of the continent and its integration into the world economy. Although its implementation has been stalled by the failure, so far, of African leaders to reach a consensus on how to finance it, the plan is significant because, as with the declaration of the Second Decade of Education, it signals, finally, the engagement of Africa's political leadership. For its part, the African Development Bank has approved a Strategy for Higher Education, Science, and Technology aimed, among other things, at helping member countries link higher education to the productive sector. Supplemented by appropriate national-level policies and structures, such as those outlined above, and combined with well-resourced regional nodes of strength in selected areas, these efforts provide a starting point for productive collaboration between higher education and industry nationally and across the continent. Notes 1. We use the term "knowledge production institution," which is more general than "higher education" or "university," to emphasize that relevant knowledge production occurs out- side the formal university system, although the latter remains the principal locus for research and knowledge generation. 292 | A K I L A G PA S A W Y E R R A N D B O U B A K A R B A R RY 2. An exception must be made in the case of state-sponsored research or industrial drive, as for the arms industry. 3. The Tananarive conference was sponsored by the United Nations Educational, Cultural, and Scientific Organization (UNESCO) and the United Nations Economic Commission for Africa. 4. Although the point about low employment of technically qualified personnel holds for the typical medium-to-low-technology enterprise, the situation is significantly different in the larger firms and major public enterprises and utilities, where the top engineers and managers are drawn from local graduates, occasionally with foreign postgraduate qualifications. Bibliography African Ministerial Council on Technology. n.d. "Flagship R&D Programmes: Building Engi- neering Capacity for Manufacturing." http://www.nepadst.org/platforms/engcapacity.shtml. Ajayi, J. F. Ade, Lameck K. H. Goma, and G. Ampah Johnson. 1996. The African Experience with Higher Education. Accra: Association of African Universities; London: James Currey; Athens, OH: Ohio University Press. Breier, Mignonne, and Mahlubi Mabizela. 2008. "High-Skill Requirements in Advanced Man- ufacturing." In Human Resources Development Review: Education, Employment and Skills in South Africa, ed. Andre Kraak and Karen Press, 300­21. Cape Town: HSRC Press. Brennenraedts, Reginald, Rudi Bekkers, and Bart Verspagen. 2006. "The Different Channels of University-Industry Knowledge Transfer: Empirical Evidence from Biomedical Engi- neering." Working Paper 06.04, Eindhoven Centre for Innovation Studies, Eindhoven, Netherlands. Engelhard, Rutger. 2000. National Policy Dialogue on Research and Technology for Development. Introduction to Assessment Reports Prepared in Ghana, Senegal, Uganda and Vietnam. Maastricht, Netherlands: European Centre for Development Policy Management. Lorentzen, Jo Angelique Wildschut. 2008. "High-Skill Requirements in Advanced Manufac- turing." Human Resources Development Review: Education, Employment and Skills in South Africa, ed. Andre Kraak and Karen Press, 345­64. Cape Town: HSRC Press. Lwakabamba, Silas, and Francis Mbuza. 2004. "African Centre for Engineering and Technol- ogy Education." Global Journal of Engineering Education (Australia) 8 (1): 93­100. Massaquoi, J. C. M., ed. 2001. University-Industry Partnership for Cooperative Technology Development in Africa: Opportunities, Challenges, Strategies and Policy Issues. Nairobi: United Nations Educational, Scientific, and Cultural Organization (UNESCO). NEPAD (New Partnership for Africa's Development), Office of Science and Technology. 2006. Africa's Science and Technology Consolidated Plan of Action. Midrand, South Africa: NEPAD. Sawyerr, Akilagpa. 2004. "Challenges Facing African Universities: Selected Issues." African Studies Review 47 (1): 1­59. Scott, Alister, et al. 2001. The Economic Returns to Basic Research and the Benefits of University-Industry Relationships: A Literature Review and Update of Findings. Science and Technology Policy Research, Report for the U.K. Office of Science and Technology, London. Tindimubona, Alex R. 2000. "National Policy Dialogue on Research and Technology for Development in Uganda: An Assessment." European Centre for Development Policy Man- agement, Maastricht, Netherlands. A F R I C A N H I G H E R E D U C AT I O N A N D I N D U S T RY: W H AT A R E T H E L I N K A G E S ? | 293 UNCTAD (United Nations Conference on Trade and Development). 2003. Africa's Technolo- gy Gap: Case Studies on Kenya, Ghana, Uganda and Tanzania. UNCTAD/ITE/IPC/ Misc.13. New York: United Nations. UNESCO (United Nations Educational, Scientific, and Cultural Organization). 2006. "Glob- al Education Digest 2006: Comparing Education Statistics around the World." UNESCO Institute for Statistics, Montreal. van Zyl, A., J. Amadi-Echendu, and T. J. D. Bothma. 2007. "Nine Drivers of Knowledge Transfer between Universities and Industry R&D Partners in South Africa." South African Journal of Information Management 9 (1, March): 1­22. World Bank. 2009. Accelerating Catch-up: Tertiary Education for Growth in Sub-Saharan Africa. Directions in Development Series. Washington, DC: World Bank. Yesufu, T. M. 1973. Creating the African University: Emerging Issues in the 1970s. Ibadan, Nigeria: Association of African Universities; Oxford University Press. Comment on "African Higher Education and Industry: What Are the Linkages?" by Akilagpa Sawyerr and Boubakar Barry SHAHID YUSUF The paper by Sawyerr and Barry is a model of balanced exposition. It rightly starts by underscoring the axial role of knowledge for development and the increasing con- tribution knowledge producers can make to the pace and technological content of industrial change and to gains in productivity. The authors enumerate the pathways through which knowledge finds its way into production processes, but they stress, again rightly, that the outcomes depend on the capability of the producers to gener- ate a stock of usable knowledge and on the efforts to disseminate this knowledge to potential users. Equally, those who can benefit from the knowledge must be active in seeking it out, and well prepared and entrepreneurial in putting it to the best possi- ble use. Fumio Kodama has labeled firms with absorptive capacity as "receiver active" (Kodama and Suzuki 2007). The dynamism and growth of a knowledge economy depend on the initiative of both parties. To what degree this interaction between producers and users results in a virtuous economic spiral depends, accord- ing to Sawyerr and Barry, on the emergence of a technology culture that stimulates the continual deepening of knowledge, as well as unceasing efforts by industry to har- ness knowledge effectively for the sake of survival in a unsparingly competitive world economy. An economy that induces the continuous production, borrowing, adaptation, and utilization of knowledge has been slow to emerge in African countries, for a variety of reasons. African universities are struggling to rise to the challenge of graduating enough engineers and scientists with skills of the requisite quality and to conduct research that can be of practical relevance. African firms, by and large, evince little desire to upgrade their technologies and diversify into areas of higher technology by fully exploiting the available supply of skills and seeking out technology from all pos- sible sources, domestic and foreign. They are behaving passively, preferring to remain in low-technology, niche areas. As a consequence, African firms too often forgo the Shahid Yusuf is an economic adviser at the World Bank Institute. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank 294 C O M M E N T O N S A W Y E R R A N D B A R RY | 295 opportunities that could be grasped through an upgrading of their workforces, through technology catch-up, and through strategies that give greater primacy to in-house applied research and to innovation. Sawyerr and Barry conclude by calling for a determined political drive designed to strengthen the technology culture by, for example, enhancing market competition; to revitalize the universities; and to create an "observatory" composed of key stake- holders to help chart the way forward. The authors' diagnosis is balanced and well- rounded. But they never explain why African countries have been slow to evolve a technology culture, why competitive pressures impinging on African economies have been relatively weak, and why industry in African countries has remained low tech- nology. A stab at each of these questions is surely the first step in arriving at solu- tions, however tentative. Absent such analysis of persistent shortcomings, the authors' recommendations, although sensible, do not go beyond the generic. If one compares some of the Southeast Asian "tiger" economies with countries in Sub-Saharan Africa, it is apparent that Malaysia and Thailand started out from a resource base that was not different from that of several of the coastal economies in Africa. Neither of these countries had or even now possesses a technology culture. Neither has a strong tertiary education sector, and their universities are only now beginning to develop postgraduate and postdoctoral programs. Yet they have acquired substantial manufacturing capability and have absorbed the expertise to produce high-tech products. The question that arises is why Southeast Asian coun- tries have responded differently to global competition, have integrated with global value chains, and have built an internationally competitive industrial base even though the technology culture was no stronger there than in a number of African countries. There are several possible explanations that Sawyerr and Barry could have examined. It could be, for example, that the African economies were (and remain), on balance, more closed and less politically stable than their Southeast Asian com- parators and therefore attracted little foreign direct investment in manufacturing activities; that their market institutions were and are relatively weaker; that their pri- mary and secondary education systems did not impart the basic skills required of an industrial workforce; and that the region has lacked role models such as Japan and the Republic of Korea to generate a variety of spillover and neighborhood effects. And there are other reasons, as well. Southeast Asian countries are now attempting to build technological capabilities and deepen embryonic innovation systems, much like the African countries that are the focus of this paper. However, they have the advantage of starting from a solid manufacturing base, and one that has absorbed the advances in technology in the electronics, engineering, and transportation industries. They have also begun invest- ing substantially in tertiary education and research. Nevertheless, the deepening of domestic technological capabilities poses severe challenges, especially for latecomers struggling to climb onto the first rung of the global market for manufactures. The majority of countries in Sub-Saharan Africa lag far behind, and developing a competitive knowledge economy will take years of effort and investment. Sawyerr and Barry identify a broad objective, the technology culture, and they point to some 296 | SHAHID YUSUF of the weaknesses of African universities, which fail to emphasize problem-solving skills and are struggling to enlarge the pool of graduates with mathematics, science, and engineering skills. The authors assert that a relatively protected domestic market might explain the small appetite for technology among firms and the low investment in research and development. But all they offer by way of remedies is to call for active involvement by the state and a pan-African effort to develop science and technology skills. This is a small part of the answer that policy makers are seeking, not just in Sub-Saharan Africa but also in Asia and elsewhere. How to make government inter- ventions more effective is the more compelling question the paper fails to address. Nor does the paper acknowledge and tackle another equally important policy ques- tion: what kinds of specific actions by firms and other nonstate actors are needed to complement the efforts of the state to embed a technology culture and to make it the foundation of an economy that derives competitive strength from innovation? Deal- ing with these questions in any depth would go far beyond the authors' remit, but a start at providing answers beyond the commonplace, grounded in their vast knowl- edge of the subject, would have augmented the contribution of the paper to policy making. Reference Kodama, Fumio, and Jun Suzuki. 2007. "How Japanese Companies Have Used Scientific Advances to Restructure Their Businesses: The Receiver-Active National System of Inno- vation." World Development 35 (6): 976­90. An Arrested Virtuous Circle? Higher Education and High-Technology Industries in India RAKESH BASANT AND PARTHA MUKHOPADHYAY The authors provide a brief but comprehensive overview of linkages between higher education and the high-technology sector, followed by an analysis of the major link- ages in India. Links outside the labor market are found to be weak--a consequence of a regulatory structure that separates research from the university and discourages good faculty from joining the university. The effect is to erode the quality of the intel- lectual capital necessary to generate new knowledge. In the labor market, the authors find a robust link between higher education and high-tech industry, but despite a strong private sector supply response to the growth of high-tech industry, the quality leaves much to be desired. Poor university governance may be limiting both labor market and non­labor market linkages. Industry efforts to improve the quality of graduates are promising, but overreliance on industry may compromise workforce flexibility. Addressing governance failures in higher education is necessary in order to strengthen the links between higher education and high-tech industry. For late-industrializing countries, the creation of a high-technology industrial base is critical for development. As the links between technology and development evolve, the potential role of higher education in contributing to the creation of a high-tech industrial base is becoming a major policy focus.1 In this context, the links between high-tech industries and higher education become crucial. The available literature has not explored these linkages adequately, especially in the context of developing coun- tries. The aim of this paper is to explore empirically such linkages in India. Rakesh Basant is a professor at the Indian Institute of Management, Ahmedabad. Partha Mukhopadhyay is a senior research fellow at the Center for Policy Research, Delhi. This is an extensively revised version of the paper presented at the Annual World Bank Conference on Development Economics. Rakesh Basant would like to acknowledge the support of the Observer Research Foundation, New Delhi, for part of the work reported here. The authors are also grateful to Shobha Tekumalla and Rohan Mukherjee for excellent research assistance and to Pankaj Chandra and an anonymous referee for useful comments. Annual World Bank Conference on Development Economics 2009, Global © 2010 The International Bank for Reconstruction and Development / The World Bank 297 298 | R A K E S H B A S A N T A N D PA R T H A M U K H O PA D H YAY The next section contains a brief but comprehensive overview of linkages between higher education and the high-tech sector. This is followed by a survey of the structure of higher education in India. Empirical evidence on the major linkages in India is then examined in the context of the recent growth of high-tech industry there, and concrete examples of labor market channels in the information technology, financial services, and high-end manufacturing sectors are presented. The response of the higher educa- tion sector to developments in the labor market, and the differences in this respect between Indian states, are examined, and we argue that the response has been restricted by the nature of regulation in India. Some issues raised in earlier sections are then discussed in the context of other South Asian countries. The final sections identify policy lessons and present conclusions. Linkages between Higher Education and High-Technology Industry Three broad conceptual frameworks inform the analysis of the linkages between higher education and high-tech industry: the national innovation system (NIS); the triple-helix paradigm; and university-industry linkages (UILs). The concept of NIS was developed to explain the differences in the innovative per- formances of industrial countries. More recently, the framework has been applied to the analysis of the experiences of developing countries, especially the newly industrializing economies of East Asia. NIS characterizes a system of interacting agents--firms, univer- sities (including research institutions), and government agencies--that are involved in the development and commercialization of science and technology. These interactions are undertaken within national borders and encompass technical, commercial, legal, financial, and social transactions. (See Metcalfe and Ramlogan 2008 for a succinct sum- mary of the NIS literature.) Differences in innovative performance among nation-states are ascribed to differences in the way institutions combine and interact to generate, improve, and diffuse new technologies--that is, products, processes, and practices. The initial applications of this framework focused on nation-states, but NIS is increasingly being used to look at regional, local (cluster), and sectoral situations. It is often also employed to contextualize case studies of specific institutions--firms, universities, and state agencies. It has been suggested that the literature in this tradition has focused more on the "invention system" than on the "innovation system" and has thus accorded less importance to understanding the complementary economic processes needed to convert invention into innovation and then to diffuse the innovations (Metcalfe and Ramlogan 2008). The triple-helix paradigm extends the NIS framework. In triple helix I, "the state encompasses industry and academia and directs relations between them." In triple helix II, the three actors are in individual and separate domains "with strong borders dividing them and highly circumscribed relations" among them. Triple helix III gen- erates a "knowledge infrastructure . . . with each taking the role of the other and with hybrid organizations at the interface" (Etzkowitz and Leydesdorff 2000, 111). The framework thus develops a model of innovation that captures multiple reciprocal relationships among industry, university, and government which help capitalize knowledge at different stages of its development process (Etzkowitz 2003). In so AN ARRESTED VIRTUOUS CIRCLE? | 299 doing, it focuses on the internal transformation of each of the helixes and the influ- ence of one helix on the other. The recent literature on university-industry linkages appears at first glance to be both an extension and a combination of these two frameworks, as it focuses on two of the three components of NIS and two helixes of the triple-helix system. UIL studies, howev- er, are different, in three ways. First, UILs are typically analyzed in the context of geo- graphically bound clusters, related to the literature on regional or cluster-specific innovation systems. In adopting this focus, they tend to deemphasize the macro linkages between the education system and industry. Second, UIL studies focus on the variety of industry-academia linkages and their measurement. Third, these studies increasingly explore the complementary economic processes that are required in order to facilitate and even push the invention-innovation-diffusion process. This exploration has led to the examination of various policy options that bring universities closer to the market and facilitate commercialization of technology developed at the university, through licensing and start-ups. Thus, the emerging focus of this work is on understanding the factors that help "traditional universities" become "entrepreneurial universities."2 Broadly, all three streams of literature encompass the linkages between higher edu- cation and high-tech industries, with the UIL framework giving these interactions greater attention. Nevertheless, the linkages remain largely underexplored in the con- text of developing countries. In particular, labor market linkages have received inad- equate attention, whether in local or national contexts. This is surprising because the labor market connection between industry and academia remains the most promi- nent link even in developed countries, where universities contribute relatively little to patenting, licensing, and new enterprise creation except, probably, in life sciences (Hershberg, Nabeshima, and Yusuf 2007). In developing countries, such linkages are even weaker. Conversely, industries can influence higher education institutions through a vari- ety of channels. These relationships, again, have been little explored in developing countries, where the relative capacities of the private sector and the state differ from those in a developed country. The ability of institutions to respond positively to industry and create a virtuous feedback loop depends on the regulatory structure of higher education, especially the constraints on establishing new institutions, hiring faculty, setting salaries, and making changes in curriculum--that is, it depends on the degree of agency possessed by the institution. The Developing-Country Context It is useful to briefly describe the linkages between higher education and high-tech industries and how the nature of these interactions may differ in developing countries. Linkages from Higher Education Institutions to High-Technology Industries Consider first the links that operate from higher education institutions to high-tech industries. 300 | R A K E S H B A S A N T A N D PA R T H A M U K H O PA D H YAY 1. Higher education institutions provide a common and possibly neutral platform for discussions about the broader goals of innovation policy and a forum that permits relatively open interaction between industry and government.3 This feature may be valuable in developing countries, where the relationship between the bureau- cracy and industry is often antagonistic or clientelist, either of which precludes productive dialogue. Its function may be especially important when a country is moving from one type of industry-government relationship toward another, for example, during a liberalization. We do not explore this linkage here. 2. Another mundane link is through the institutions' supply of services such as test- ing, training, certification, and prototype development.4 This is especially impor- tant in developing countries, where firms are typically small and their number is relatively large. Developing skills in-house is more difficult for a small firm than for a large one; furthermore, the country's talent pool is relatively limited and is unable to meet the needs of the large number of firms. Higher education institu- tions such as universities can function as a common pool resource, aggregating the limited talent and making it available to all firms on a fee-for-service basis. This, again, is not an area we explore here. 3. The third link is through an ostensibly primary function of higher education, cre- ation of knowledge, through technology licensing, industry-institution research and development (R&D) projects, and the like. Given the relatively poor state of the education infrastructure in developing countries, the predominance of state funding for higher education, and the low levels of funding of R&D by industry, this is unlikely to be a strong channel. 4. A step beyond the creation of knowledge is its commercialization through the cre- ation of new enterprises that put to use technologies developed in the institutions by students or faculty. Often, formation of new enterprises is facilitated by science parks, incubators, and the like that are established around research-based aca- demic institutions. The growth of high-technology industries, however, requires inputs other than those from higher education institutions. For example, if a city is to become a financial center, human skills have to be combined with urban infrastructure. Similarly, limited availability of risk capital may strongly inhibit the growth of high-tech firms. 5. Finally, and most important, higher education institutions supply talent for high- tech industries. Students trained at these institutions become part of the labor pool, which may or may not be restricted to the region. This is not true just for service sectors commonly regarded as high-tech areas, such as information tech- nology (IT) and financial services; most types of manufacturing, too, need a cer- tain component of high-end skills. As the higher education sector evolves, it acquires the capacity to handle more complex technologies ("deepening"), and its influence extends to more industries ("broadening"). The focus of the UIL literature has been on the creation and commercialization of knowledge and the creation of enterprises. The links through the supply of labor have, in some sense, been considered too mundane. Yet, although the former types of linkage are possibly more important in countries where growth is driven by pushing AN ARRESTED VIRTUOUS CIRCLE? | 301 the technological frontier, the labor market link may be significant in developing countries, where the growth of high-tech industry may depend more on how quickly existing technologies can be exploited. In industries such as software, this ability may well be constrained by the supply of highly skilled labor, which will determine the cost and the pace of expansion. Linkages from High-Technology Industries to Higher Education Institutions The linkages that extend from high-tech industries to higher education can, again, be grouped under five broad types. 1. The first and most mundane is the supply of goods and services; for example, high-tech industry provides specialized equipment to institutions, as well as technicians to serv- ice the equipment. In developing countries, the servicing function is likely to be more important because much of the equipment may be imported and may also see much longer usage periods than in developed countries. We do not explore this link here. 2. In developing countries, where teachers are often in short supply, industry profes- sionals may serve as teaching staff. Of course, industry also affects the supply of teachers by offering an alternative career choice for students, which could influence both the quality and the compensation of the teaching pool. This is especially true for technical and professional education, which is relevant to high-tech industries. 3. Industry can influence the curriculum by working with government or higher edu- cation institutions to effect changes in line with its own requirements. To the extent that industry is more outward-oriented than higher education institutions in developing countries, this can be a source of new ideas, but it also carries a risk of making education narrower. 4. High-tech industries can be a source of funding for higher education. Because of the stronger links between educational inputs and success in industry, alumni in these industries may be more predisposed toward giving back to the institution. Industry may also fund research projects. Given the low levels of government funding, both in general and specifically for research, and the conditionalities that restrict the use of such funding, contributions from industry can be more important in developing than in developed countries. In research, especially, the added funding from indus- try can be a sizable proportion of the overall research budget in universities. 5. Finally, demand for labor on the part of industry can lead to demand for new insti- tutions and expansion of the capacity of existing institutions. Depending on the response, this can affect the structure of the higher education sector. Concentra- tion of industry in a few areas, as is more likely in developing countries, may sim- ilarly concentrate the structure of higher education. Higher Education in India India has several types of higher education institutions: universities, "deemed to be uni- versities," colleges, institutions of national importance, and postgraduate institutions 302 | R A K E S H B A S A N T A N D PA R T H A M U K H O PA D H YAY TABLE 1. Types of University-Level Institutions, by Mode of Establishment, India Type Central Private State Total University 20 7 209 236 Deemed-to-be university 36 59 8 103 Institute of national importance 13 0 0 13 Established by state act 0 0 5 5 Total 69 66 222 357 Source: University Grants Commission (UGC) Web site, http://www.ugc.ac.in. (table 1).5 Universities can be set up by an act of parliament or by the state legislature. In general, only universities, deemed-to-be universities, and institutions of national importance are authorized to grant degrees. Other postgraduate institutions and poly- technics that are recognized by specialized accreditation organizations such as the All India Council of Technical Education (AICTE) can grant postgraduate diplomas and diplomas. We next highlight some key features of the higher education system. Funding and Fee Regulation Institutions of national importance and public universities chartered by an act of par- liament or by state legislatures are funded by states, the University Grants Commis- sion (UGC), or both. Private universities established by state legislatures are funded privately. Central funding is directed largely to central institutions, and little of it goes to support higher education at the state level. Typically, the size of the fees that can be charged has to be justified by the institution before a quasi-judicial body in each state. Whereas the popular perception is that the share of funding from fees is low, many public universities earn a substantial portion of their operating income from fees; in engineering and technology institutions, fees range between 22,000 rupees and 72,000 rupees per year (Agarwal 2006, 177). CABE (2005) also provides evi- dence to this effect (see table 2). In addition, the growing number of student loans and the rising share of private colleges testify to the importance of student fees as a financing source, especially for technical and professional education. Uniform Compensation and Automatic Promotion of Faculty The UGC specifies a salary structure for universities that receive its funding. Com- pensation in the various universities is therefore nominally uniform except for minor differences in allowances. Most other publicly owned academic institutions bench- mark themselves to this salary structure, which implies that faculty at these institu- tions also face a uniform compensation structure.6 In addition, most institutions typically follow a pattern of time-bound automatic promotion of faculty. This is a departure from the past, when the number of senior positions was limited and recruitment to these positions was by open competition, making it possible for fac- ulty to be promoted early by applying for positions in other institutions. AN ARRESTED VIRTUOUS CIRCLE? | 303 TABLE 2. Distribution of Selected Public Universities by Share of Fees in University Income, Late 1990s 10 percent 10­30 percent 31­50 percent 50 percent Hyderabad Andhra Guru Jambeswar Mumbai Kalyani Bhavnagar Punjab Karnataka State Open Maharshi Dayanand Calcutta S. N. D. T. Women's Kuvempu Rabindra Bharati Dibrugarh Bangalore M. D. Saraswati Tamil Dr Harsingh Pune Visva Bharati Sri Venkateshwara YCM Open Anna Calicut Delhi Goa J. N. Vyas Kannur Osmania Karnataka Kumaon Mangalore Mysore Saurashtra IGNOU Source: CABE (2005). Affiliated Colleges Undergraduate teaching mostly takes place at colleges affiliated with a university and is based on a predetermined curriculum. Affiliated colleges may offer professional education, as well. Evaluation of students is carried out externally through a univer- sitywide examination. Autonomous colleges, accredited by the National Assessment and Accreditation Council (NAAC), also exist, but their number is not high relative to the total number of colleges. Usually, admission is based on performance on the qualifying school-leaving examination. Separation of Teaching and Research Currently, teaching and research are increasingly carried out at different locations because of the prevalence of specialized research institutions. The establishment of institutions in science and technology through the Council of Scientific and Industrial Research (CSIR) and the creation of social science research institutes through the Indian Council of Social Science Research (ICSSR) resulted in an exodus of research- oriented faculty from universities to these institutions, where research facilities were generally better. The consequent deterioration of the university system led to a self- reinforcing downward spiral that put the relationship between teaching and research further out of balance. One of the key adverse impacts of such a separation has been that undergraduate teaching primarily undertaken by affiliated colleges has become almost completely delinked from research (Yarnell 2006). 304 | R A K E S H B A S A N T A N D PA R T H A M U K H O PA D H YAY Specialized Institutes for Professional Education Much professional education is provided through specialized institutes at both the undergraduate and postgraduate levels. These professional institutions may be affili- ated with a university but may be independently or, often, concomitantly approved by the AICTE or other professional councils such as the Medical Council of India. They have flexibility in designing their curriculum, and student evaluation is per- formed in-house. Some of these entities are accredited by the National Accreditation Board (NAB). Admissions are based on highly selective nationwide or statewide competitive examinations. In some states, the increasing number of institutions has reduced the degree of selectivity overall, but the quality of the institutions varies considerably, and higher-quality institutions maintain a high level of selectivity. Fees in private educational institutions are currently regulated through a com- mittee headed by a former high court judge. This mechanism allows considerable variation across institutions. Some institutions, including a few that are not in the private sector, such as the Indian Institutes of Management (IIMs), are able to charge market-related fees for professional education, and these have been rising recently.7 Linkages between Higher Education and High-Technology Industries in India The high-tech industry in India has seen some growth, and the share of communica- tions and computer-related services has increased significantly in recent years (fig- ure 1). This growth may have led to an increase in demand for skilled workers; recent studies have indicated the presence of output-skill complementarities (Berman, Somanathan, and Tan 2005; Chamarbagwala and Sharma 2008). There is some evi- dence for growing wage premiums for skilled workers, driven by growth in demand for skills.8 Developments in science and technology, and changes in global production and R&D networks, are creating new opportunities for interaction between academic institutions and firms in India. The evidence reported in Basant and Chandra (2007) suggests that some academic institutions and other entities in the cities of Pune and Bangalore have utilized these opportunities to build linkages and interact with the city clusters in a rich variety of ways. Only a few institutions, however, have the relevant knowledge base needed to undertake high-end knowledge networking activ- ities with industrial entities. The survey data reveal a hierarchy of institutions accord- ing to the strengths of their linkages: those that undertake only teaching (linkage through the labor market); those that carry on research and teaching and provide services such as testing (linkages of all the types mentioned earlier); and those that focus on specialized research (linkages predominantly driven by knowledge genera- tion and dissemination). Given the existence of this hierarchy, reflecting low institu- tional capabilities, academic institutions rarely come together to advance the linkages collectively, but many are gearing up to participate in such linkages in a more sys- tematic manner. The experience of Pune and Bangalore broadly supports the idea AN ARRESTED VIRTUOUS CIRCLE? | 305 FIGURE 1. Share of GDP, Select High-Technology Sectors, India, 1999­2000 to 2007­8 6 5.7 5 4.9 4.2 4 3.6 3.3 percent 3.1 3 2.7 3.0 1.6 2.2 2.8 2 1.9 2.4 2.1 1.8 1.6 1.4 1 1.0 0 0 ­1 ­2 ­3 ­4 ­5 ­6 ­7 ­8 00 00 01 02 03 04 05 06 07 ­2 20 20 20 20 20 20 20 20 99 19 communication computer-related services Source: Authors' computation from India, Central Statistical Organisation, National Accounts Statistics, various issues. Note: GDP, gross domestic product. that institutions (if given some autonomy) and industry coevolve to take advantage of emerging opportunities. Linkages from Higher Education to High-Technology Industries Higher education can contribute to high-tech industry through the creation of knowl- edge (as embodied in patents and publications) and by facilitating the creation of enterprises. Each form of linkage is examined below. Creation of knowledge: Patenting and research output. Table 3 presents information from the U.S. patent database about patents issued to inventors in India, starting in 1976 (Gupta 2008). Certain broad trends are evident: 1. Patenting in India has increased over time, particularly in recent years. Indian universities, too, have increased their patenting over time in line with this trend, but their share still remains low and is in fact a little less than in 1976­80. 2. Public sector research institutions constitute most of the Indian assignees, and this share is steadily increasing--even more so than that of private firms, whose patenting is also growing. Of the public institution patents, over 80 percent is assigned to the CSIR, a network of publicly funded research laboratories around the country.9 These laboratories do not have educational functions, reflecting the sepa- ration of teaching and research in Indian higher education. 306 TABLE 3. Patents Awarded to Inventors in India, by Type of Assignee, 1976­2005 Indian assignee Foreign assignee University or University or Individual Private firm Public sector institution NGO Private firm State institution Number of patents 1976­80 16 11 3 3 0 50 0 5 1981­85 15 10 3 1 0 34 9 4 1986­90 27 10 16 1 0 36 16 15 1991­95 31 20 35 2 0 114 8 12 1996­2000 70 90 167 13 0 272 12 20 2001­5 82 267 677 45 5 831 2 31 Total 241 408 901 65 5 1,337 47 87 Share of patents of a given assignee over time (percent) 1976­80 6.6 2.7 0.3 4.6 0.0 3.7 0.0 5.7 1981­85 6.2 2.5 0.3 1.5 0.0 2.5 19.1 4.6 1986­90 11.2 2.5 1.8 1.5 0.0 2.7 34.0 17.2 1991­95 12.9 4.9 3.9 3.1 0.0 8.5 17.0 13.8 1996­2000 29.0 22.1 18.5 20.0 0.0 20.3 25.5 23.0 2001­5 34.0 65.4 75.1 69.2 100.0 62.2 4.3 35.6 Share of assignees as percent of total patents within a period 1976­80 18.2 12.5 3.4 3.4 0.0 56.8 0.0 5.7 1981­85 19.7 13.2 3.9 1.3 0.0 44.7 11.8 5.3 1986­90 22.3 8.3 13.2 0.8 0.0 29.8 13.2 12.4 1991­95 14.0 9.0 15.8 0.9 0.0 51.4 3.6 5.4 1996­2000 10.9 14.0 25.9 2.0 0.0 42.2 1.9 3.1 2001­5 4.2 13.8 34.9 2.3 0.3 42.8 0.1 1.6 Total 7.8 13.2 29.1 2.1 0.2 43.3 1.5 2.8 Source: Gupta 2008. Note: NGO, nongovernmental organization. AN ARRESTED VIRTUOUS CIRCLE? | 307 3. The growing number of patents issued to foreign firms (although it is a declin- ing share, owing to the rise in patenting by domestic firms and public research labo- ratories) indicates that the extent of spillover from research outsourcing may be weak. This is an indirect effect of the structure of Indian higher education, in that for- eign firms established research bases in India to utilize Indian researchers produced by the country's higher education system. Gupta (2008) provides an interesting insight. Of the top 10 Indian public and pri- vate firms (excluding CSIR), which account for over half the patents, only one is in the semiconductor and IT area; the rest are in the pharmaceutical and chemical sec- tors. Of the top 10 foreign firms, which account for more than 40 percent of the total assignees, only 3 (including a foreign subsidiary of an Indian firm) are in pharma- ceuticals and chemicals. Five are IT firms, and two, GE and Unilever, are diversified. It would thus appear that the research areas of domestic and foreign firms do not overlap significantly. Athreye and Prevezer (2008) and Athreye and Puranam (2008) provide more detailed evidence on this point; Indian involvement is much higher in chemistry-oriented sectors than in IT (table 4). The data also suggest that patenting activity has not matched industry growth in telecommunications and IT (table 4). TABLE 4. Patent Characteristics by Patent Subclassification Share of Indian Average Average Share of inventors backward forward Indian entities Patent subclassification (percent)a citationb citationc (percent)d Agriculture, food chemistry 92.6 1.4 3.7 86.7 Chemical engineering 86.8 3.1 10.4 82.0 Metallurgy 86.4 4.3 8.1 80.5 Organic chemistry 87.4 1.9 5.4 73.2 Chemical and petrol industry 87.6 2.4 7.2 72.2 Biotechnology 82.8 0.6 2.5 71.4 Pharmaceuticals, cosmetics 85.7 1.9 7.4 70.6 Materials processing 79.0 36.4 49.8 66.7 Macromolecular chemistry, polymers 71.2 2.1 10.5 47.2 Control technology 71.4 7.2 15.2 36.9 Optics 58.9 19.1 33.5 36.4 Medical technology 69.2 18.8 18.4 31.3 Surface technology 63.8 13.5 28.9 21.4 Semiconductor devices 54.5 8.0 9.9 12.2 Electrical machinery 67.0 5.5 28.0 7.1 Information technology 67.5 3.3 12.8 6.3 Telecommunications 73.3 4.4 11.0 4.5 Audio-visual technology 68.6 4.4 21.1 3.9 Source: Athreye and Puranam 2008. a. The share of Indian inventors is calculated as the proportion of the number of Indian assignees in a patent to the total number of assignees. b. Backward citations refers to the number of patents cited by the patent in question. c. Forward citations refers to the number of times other patents cite the patent in question. d. Indian entities include public sector institutions, private firms, universities and institutions, and private individuals. 308 | R A K E S H B A S A N T A N D PA R T H A M U K H O PA D H YAY This raises the question of the capacity of domestic industry to benefit from spillovers, since it diminishes the likelihood that the domestic firm is receiver active in the sense described by Kodama, Kano, and Suzuki (2006). Thus, the direct gener- ation of knowledge can largely be attributed to public research labs instead of to higher education institutions, and the indirect generation of knowledge, too, seems to be in areas where linkages to domestic industry appear low. In the case of the phar- maceutical and biotechnology industry, however--consistent with Athreye and Pre- vezer (2008), Athreye and Puranam (2008), and Gupta (2008)--there seems to be an indication that linkages are beginning to form (see box 1), matching the pattern in developed countries, where such linkages are concentrated in the life sciences sector. Creation of knowledge: Publications. Even in basic research, the capacity of uni- versities is limited and skewed. About 80 percent of the doctorates awarded in engineering were from 20 universities, and about two-thirds of the science doctor- ates were from 30 universities. Only about 20 universities have a fellow in one of the three science academies (Agarwal 2006). Even in social sciences, where infra- structural constraints are presumably less binding, just 26.2 percent of the 454 arti- cles published in a leading domestic social science journal between 1998 and 2000 were from universities. Of that share, 13.9 percent came from three universities in Delhi and Mumbai, and the Indian Institutes of Management originated another 4.4 percent; that is, 70 percent of the university total was from these few institu- tions. Specialized research institutes and organizations accounted for more than universities, with 27.3 percent of all the articles (Chatterjee et al. 2002). Of the total publications from the 15 Indian universities most active in publishing in the areas of health and biotechnology in international peer-reviewed journals during the period 1991­2002, 20.5 percent came from one university, the Indian Institute of Science (Kumar et al. 2004). As is shown below, that institution is also the dom- inant player in the linkages between academia and industry, especially in the fields of pharmaceuticals and biotechnology. Research has moved out of Indian universities and other academic institutions over the years. For many years now, the public sector research institutions have been the main centers of research activity, and universities have largely become teaching institutions. Creation of enterprises. Apart from working in high-tech industries as employees, persons with higher education may set up enterprises in high-tech sectors. Enterprise creation as a part of UILs is still at a nascent stage in India, and relationships outside the labor market are limited to research support through consulting and other research projects. But the focus on enterprise creation is causing considerable excite- ment among research-oriented science, technology, and management institutions. Conventional incubators are proliferating in India today. Virtually all the well- known technology institutions have one, and some management institutions are experimenting with the concept. Basant and Chandra (2006) analyze the emerging enterprise-creation role of academic institutions in India and the incubation models used by them. It is too early to assess the impact of these enterprise creation efforts; most have begun only within the last five years. No outstanding success has come to AN ARRESTED VIRTUOUS CIRCLE? | 309 BOX 1. University-Industry Linkages in the Biotechnology and Pharmaceutical Sectors The Indian biotechnology and pharmaceutical industries have seen significant growth in recent years. There is ample evidence to show that links with academia have also grown, helping to create "training opportunities for in-house staff, improve access to research facilities and expensive equipment, expand clinical trial capabilities and provide access to government-sponsored research funds" (Frew et al. 2007, 408). The variety of linkages is large; partnerships may be focused on product development, drug delivery platforms, drug candidate screening, medical diagnostics, clinical trials, or data management. A few educational institutions such as the Indian Institute of Science have created new enterprises on the basis of tech- nologies developed within the institution. Many partnerships are developing prod- ucts to address India's local health needs, especially in the area of vaccines. (For descriptions of these partnerships, see, for example, Maria, Ruet, and Zerah 2003; Kumar et al. 2004; Yarnell 2006; Frew et al. 2007.) A striking finding from a (nonexhaustive) survey of company linkages with uni- versities and public research institutes is that relatively few institutions and firms are involved. The institutions most active in these alliances are the Indian Institute of Science (IISc), the International Centre for Genetic Engineering and Biotechnol- ogy (ICGEB), the National Institute of Immunology (NII), the All Indian Institute of Medical Sciences (AIIMS), the Indian Institute of Chemical Technology (IICT) in Hyderabad, and the Centre for Cellular and Molecular Biology (CCMB). Only one of these, the IISc, is a university; participation by other universities, which include the Jawaharlal Nehru University and the Universities of Delhi, Madurai, and Osmania, is fairly limited. Shanta Biotech, Bharat Biotech, Bharat Biotech Interna- tional, Nicholas Piramal, Dr Reddy's Laboratories, and Ranbaxy are among the few firms that are active in these partnerships. It thus appears that the linkages are still in their infancy, and it will take some time before they deepen enough to make a significant impact on technology development and diffusion. Participation of only a select set of mainly specialized research institutions highlights the quality constraints that prevent most academic institutions from contributing to these alliances and also points to the adverse effect that the separation of research and teaching may have had on the ability of Indian universities to effectively participate in such collaborations. light, but many incubators can boast of moderate success. Although precise numbers are not available, many companies from technology institutions have survived the rigors of market competition after graduating from the incubators. At the moment, the most important contribution of such efforts has been to highlight the possibility of creating enterprises based on technological innovation in educational institutions. Even moderate success generates a sharper focus on technology-based entrepreneur- ship as a career option. 310 | R A K E S H B A S A N T A N D PA R T H A M U K H O PA D H YAY A small survey of 11 incubators (35 originally received questionnaires) reveals that 9 are publicly funded; 1 incubator is supported by alumni and another by a venture capital institution. Nine of these incubators are associated with educational institu- tions, but most of the incubatees are from outside the educational institution, and in eight of the incubators, all the incubatees but one are external to the institution. This is a form of UIL that is perhaps more akin to enterprise nurturing than creation. (It can be argued that this activity is part of the supply of services rather than enterprise creation.) Many incubatees have sought intellectual property (IP) protection (espe- cially in the incubator funded by a venture capital institution), but many have not. Although in this sample most of the incubatees are "outsiders," a large proportion of the incubators in the Indian Institutes of Technology (IITs) and other nationally known institutes of technology incubate their own faculty and students--an activity that comes closer to enterprise creation in the conventional sense of the term. (See Basant and Chandra 2006 for details.) An alternative model of enterprise creation through focused agenda-based research--as is being pursued by the TeNet group at IIT Madras (http://www.tenet. res.in)--appears to offer an interesting alternative to the conventional model. How- ever, it requires considerable motivation on the part of the research group and the ability to address the trade-off between publication and enterprise creation. Some R&D institutions have just started to grapple with this dilemma, and the idea of enterprise creation might further sharpen the trade-off. Linkages from High-Technology Industry to Higher Education As mentioned earlier, industry can support higher education by providing teaching talent; pushing for curriculum reform; taking measures to improve the labor supply through assessment, certification, and registration of professionals; and directly fund- ing institutions. We look at these linkages in the Indian context. Supply of teachers. The number of higher education institutions has shown strong growth. This is especially true of professional institutions, most of them in the pri- vate sector. The number of teachers has not kept pace with this growth, as the num- ber of PhDs produced indicates. Consequently, many of these institutions are staffed by part-time faculty or faculty whose experience has been more in industry than in academia. Although it is mandatory for institutions to disclose the qualifications of faculty, this information is not collated.10 A quick survey of institution Web sites reveals, however, that a number of faculty members lack PhD degrees and that faculty size is small. In fact, even at the National Institutes of Technology, a large proportion of faculty is without PhDs.11 The supply of faculty members from industry, both part-timers and those embark- ing on a second career, is therefore a significant source of linkage. Although this practice can strengthen industry-academia links, it is possible that, given the differences in com- pensation, many of those who are choosing to enter academia from industry are not doing so out of choice, and their links with their former organizations may be weaker than would be initially evident. AN ARRESTED VIRTUOUS CIRCLE? | 311 Modifications and enhancements to curriculum. The IT industry has been in the forefront of demands for changes in the educational curriculum in India as a result of the industry perception that the quality of existing graduates is low. According to the National Association of Software and Service Companies (NASSCOM), only about 25 percent of technical graduates and 10 to 15 percent of general college grad- uates are currently suitable for employment in the offshore IT and business product outsourcing (BPO) industries, respectively (NASSCOM 2005).12 NASSCOM has signed memorandums of understanding with the UGC and the AICTE to strengthen professional education in line with the IT industry's require- ments for skilled professionals. A key component is a mentorship program between a higher education institution and a firm. Some examples of these mentoring rela- tionships are those of Zensar with VIT; Pune XANSA with Jammu University; Pixtel Technologies with the ISB Engineering College, Ghaziabad; and Pixtel Technologies with the Galgotia College of Engineering, Greater Noida. In addition, companies such as ITC InfoTech, Accenture, SUN, MindTree, Microsoft, and Patni are under- taking faculty training programs. NASSCOM is also piloting a "finishing school" for engineering graduates who are still seeking employment. One pilot was conducted during May and June 2007 for a period of eight weeks in eight institutions, including IIT Roorkee and seven National Institutes of Technology (NITs): Khozikode, Durgapur, Kurushetra, Jaipur, Surathkal, Thiruchirapalli, and Warangal. The "finishing school" addresses both technical and soft skills development and is staffed by trained faculty and practicing IT and IT-enabled services (ITES) industry consultants. The students receive an opportunity to reinforce key basic engineering skills and, in addition, acquire industry-specific knowledge, soft skills, and management and employment skills (NASSCOM 2007). Some NASSCOM members have intensively promoted training and supplemen- tary education for their recruits and employees. Wipro Technologies has had an Academy of Software Excellence for over 10 years; the Infosys Campus Connect program (see box 2) is about 5 years old. The extent of remediation efforts being undertaken by the IT industry is an indication of the inappropriate training of grad- uating engineers from higher education institutions, at least from the point of view of the IT sector.13 These industry initiatives are in addition to a private training industry that began to develop along with the growth of the IT sector and has now diversified into training for back-office operations, retail employment, and the like. The courses do not have a certification other than the brand name of the institute.14 The respon- sibility for quality assurance therefore rests with the student. Some pharmaceutical firms have also developed linkages with universities, for example, through continuing-education PhD programs with universities for their lab researchers.15 These linkages, however, are restricted to a few institutions. Bhat- tacharya and Arora (2007) find that of seven select universities, only three (IIT Delhi, Jawaharlal Nehru University, and Delhi University) interacted actively with industry in curriculum design across departments. In two (Pune and Jadavpur) such interac- tion was linked to a few departments, and Hyderabad University and the Indian Insti- tute of Science reported no such linkages. These limited relationships may become 312 | R A K E S H B A S A N T A N D PA R T H A M U K H O PA D H YAY BOX 2. The Infosys Campus Connect Program Infosys, a large Indian IT services provider with over 100,000 employees, annu- ally recruits more than 10,000 professionals. In May 2004 the company launched Campus Connect with 60 colleges, as an "industry-academia collabo- ration program to align engineering student competencies with industry needs" (see Infosys 2007). The program has grown rapidly, and as of March 2008, it extended to 500 colleges. It operates at present in nine cities in India and is now expanding to China, Malaysia, and Mexico. In its short history, Campus Connect has already trained 25,000 students and enhanced the skills of 2,000 faculty members. The core of Campus Connect is the Foundation Program, a 130-classroom- hour proprietary educational supplement for a batch size of 60­75 students that is integrated with the college's academic schedule and may include industrial vis- its to Infosys development centers. The course material is provided by Infosys and is based on resources it uses for its induction programs, including assign- ments, case studies, and a student project bank that simulates live project situa- tions adapted for academic environments. All other facilities are provided by the participating college. A companion soft skills program, delivered by alliance partners, is intended to develop students' proficiency in communications, team- work, corporate work culture, and so on. In contrast to the foundation program, there is a charge for the soft skills program. The college is given incentives on the basis of the performance of the number of graduates joining Infosys and is expected to pass on the cash benefits received from Infosys to faculty members and other individuals according to college- specific norms. Campus Connect also engages in such activities as facilitating faculty sabbaticals and sponsoring presentations of papers at conferences. By aligning the skill needs of IT services with the college curriculum, Campus Connect reduces learning time and training cost after employment. Infosys does not guarantee that graduates of Campus Connect will be offered employment, nor is it incumbent on graduates to accept an offer from Infosys, should it choose to make one. Infosys thus banks on its reputation as a superior employer and its large annual recruitment to ensure that a sufficient number of Campus Connect graduates accepts Infosys offers to make its investment in the program worthwhile. weaker if PhD programs begin to exit the university and move to specialized insti- tutions, as incipient trends indicate. National laboratories rather than universities are becoming the locus of PhD research; for example, the National Chemical Laboratory may well be the largest producer of PhDs in the chemical sciences (see Yarnell 2006). Industry participation in the curriculum is now spreading to other areas, foremost of which is vocational training, where the private sector already has a strong AN ARRESTED VIRTUOUS CIRCLE? | 313 presence. India has many industrial training institutes (ITIs) that offer a variety of courses, including long-term courses ranging from two to three years, especially for the manufacturing sector. As part of an ongoing reform program, many industries and industry associations are now becoming involved in the management of these institutions and in improving the relevance to industry of their instruction.16 Although the system can be commended for allowing the growth of such initia- tives by benignly neglecting to regulate them, their expansion does raise questions about the ability of the formal higher education system to respond to market demand, especially with respect to professional courses.17 Improvement in labor quality. In addition to modifications to curriculum, the IT industry, through NASSCOM, has undertaken other initiatives to improve the qual- ity of the labor supply, especially for the ITES sector. The degree of selectivity in the basic educational qualification for the ITES sector, which comprises university edu- cation in all disciplines, is less than that of engineering, which is the base qualifica- tion in the IT sector. The improvement initiatives can be grouped under the heads of testing and verification. The NASSCOM Assessment of Competence (NAC) is designed to address possi- ble talent shortages by creating a robust and continuous pipeline of talent for the ITES industry through a standard assessment and certification program. The assess- ment program tests the aptitude of a candidate on seven skill sets.18 NASSCOM has also partnered with other organizations to develop customized certification programs for frontline managers in this sector, for example, certified BPO quality analyst (CBQA) and certified BPO team leader, level 1 (CBTL-1). A test for the IT sector, NAC-Tech has also been developed. The need for such tests reflects the limited effec- tiveness of quality regulation with respect to educational institutions. Labor quality is also affected by fraudulent reporting of qualifications. To facili- tate more accurate verification of credentials, NASSCOM has fostered a National Skills Registry (NSR) to ensure that there is a verified database of the skill sets and talents of the human resources within the ITES industry. The NSR contains personal, academic, and employment details for IT professionals (ITPs), as entered by the indi- viduals themselves, with associated background check information. It covers ITP employees and prospective employees in the IT and ITES industries. Every ITP regis- tered in the NSR is identified uniquely by fingerprints, and the database includes the ITP's photograph. As of September 2009, 79 major companies and over 485,551 professionals have registered (see www.nationalskillsregistry.com). Funding of higher education institutions. Although the government provides most of the funding for public institutions, private funding has made noticeable contribu- tions to some. Much of this funding, however, is limited to marquee institutions such as the IITs and IIMs. On occasion, Indian industry has contributed to institutions abroad. It has been argued (for example, by Kapur and Mehta 2007) that a bur- geoning trend toward private contributions may have been nipped in the bud by the government's desire to control the flows by insisting that they be channeled through a central fund, the Bharat Shiksha Kosh (India Education Fund), whose use would be determined by the central Ministry of Education. There is, however, considerable 314 | R A K E S H B A S A N T A N D PA R T H A M U K H O PA D H YAY private funding of higher education through the establishment of privately owned educational institutions, which now constitute the majority of institutions in profes- sional and technical education. Relative Importance of Different Linkages The creation of knowledge and the creation of enterprises are, at this time, relatively weak channels of interaction between industry and academia. The universities produce few patents, and even the generation of publications is limited and is con- centrated in a few universities. As for enterprise creation, the role of educational insti- tutions can better be described as nurturing rath