Report No. 39221-UG Uganda Moving Beyond Recovery: Investment and Behavior Change, For Growth Country Economic Memorandum Volume II: Overview September 2007 Poverty Reduction and Economic Management Unit Africa Asia Region Document of the World Bank ODA Overseas Development Assistance UETL Uganda Electrification Transmission Ltd OECD Organization for Economic Co-operation UNHS Uganda National Household Survey and Development UIA Uganda Investment Authority OIBM Opportunity International Bank of Malawi UNIDO United Nations Industrial Development PA Poverty Assessment Organization PAF Poverty Action Fund UNMP UN Millennium Project PEAP Poverty Eradication Action Plan UPE Universal Primary Education PER Public Expenditure Review UPDF Uganda People's Defense Forces PIP Public Investment Plan URC Uganda Railways Corporation PMA Plan for Modernization of Agriculture USAID United States Agency for International PPI Private Participation in Infrastructure Development PPP Public Private Partnerships UTB Uganda Tourist Board PSD Private Sector Development UTL Uganda Telecommunications Limited RAFU Roads Agency Formation Unit VAT Value Added Tax RAI Rural Access Index WFP World Food Program RPED Regional Program for Enterprise Development RTA Regional Trade Arrangement ROO Rules of Origin SACCO Savings and Credit Cooperative SADC South African Development Community SIDA Swedish International Development Agency SME Small and Medium Enterprise SOE State Owned Enterprise SPS Sanitary Phyto-sanitary SSA Sub-saharan Africa TFP Total Factor Productivity TOT Terms of Trade TTF Transport and Trade Facilitation UBI Uganda Business Inquiry UBOS Uganda Bureau of Statistics UCBL Uganda Commercial Bank Limited UCC Uganda Communications Commission UCDA Uganda Coffee Development Authority UCSCU Uganda Credit and Savings Cooperative Union UEB Uganda Electricity Board UEGCL Uganda Electricity Generation Company Limited UEPB Uganda Export Promotion Board Vice President: Obiageli Katryn Ezekwesili Country Director: John Murray McIntire Sector Director: Sudhir Shetty Sector Manager: Kathie Krumm Task Team Leader: Dino Merotto TABLE OF CONTENTS 1. SOURCES OF GROWTH ..............................................................................................................1 A. SUMMARY...........................................................................................................................................1 B. RECENT ECONOMIC HISTORY: FROM POST CONFLICT, TO MATURE REFORMER...............................2 C. IS AN IMPRESSIVE GROWTH SPURT BY WORLD STANDARDS SLOWING DOWN? ...............................5 D. ACCOUNTING FOR GROWTH ­ MACRO ANALYSIS............................................................................16 E. ACCOUNTING FOR GROWTH: EVIDENCE FROM FARMS AND FIRMS .................................................28 2. AGRICULTURE SECTOR PERFORMANCE..........................................................................47 A. INTRODUCTION .................................................................................................................................47 B. AGRICULTURE AND PRO-POOR GROWTH..........................................................................................47 C. AGRICULTURE SECTOR.....................................................................................................................48 D. DRIVERS OF PAST AGRICULTURAL GROWTH....................................................................................53 E. UNDERSTANDING GROWTH CONSTRAINTS.......................................................................................69 F. SUMMARY AND THE WAY FORWARD...............................................................................................93 3. GROWTH DIAGNOSTIC............................................................................................................96 A. GROWTH DIAGNOSTIC: BACKGROUND ­ THEORETICAL FRAMEWORK.............................................96 B. INITIAL RESULTS ­ TOWARDS IDENTIFYING BINDING CONSTRAINTS TO PRIVATE INVESTMENT AND GROWTH IN UGANDA .....................................................................................................................100 4. PATTERNS OF GROWTH AND PUBLIC SPENDING IN UGANDA: ALTERNATIVE SCENARIOS FOR 2003-2020...................................................................................................................128 A. SUMMARY AND CONCLUSIONS.......................................................................................................128 B. INTRODUCTION ...............................................................................................................................131 C. BASE SCENARIO..............................................................................................................................131 D. ALTERNATIVE PATTERNS FOR SECTORAL PRODUCTIVITY GROWTH........................136 E. GOVERNMENT SPENDING ALLOCATION AND EFFICIENCY .............................................................140 F. FOREIGN AID AND GOVERNMENT SPENDING.................................................................................146 5. EXPORTS, TRADE POLICY AND GROWTH.......................................................................148 A. ECONOMY-WIDE ISSUES..................................................................................................................149 B. SUB-SECTOR SPECIFIC ISSUES.........................................................................................................154 C. MARKET ACCESS ISSUES ................................................................................................................159 D. CONCLUSIONS AND STRATEGIC ISSUES...........................................................................................160 6. INFRASTRUCTURE FOR GROWTH.....................................................................................162 A. INFRASTRUCTURE SERVICES IN UGANDA: A SNAPSHOT.................................................................162 B. INFRASTRUCTURE IS IMPORTANT TO GROWTH: UGANDA AND THE INTERNATIONAL EXPERIENCE 172 C. DEVELOPMENT PRIORITIES AND SPENDING IN INFRASTRUCTURE ..................................................175 D. PUBLIC SPENDING PATTERNS AND MECHANISMS...........................................................................177 F. ALTERNATIVE MEANS FOR INFRASTRUCTURE FINANCING.............................................................200 7. FINANCE FOR GROWTH ­ MAKING UGANDA'S FINANCIAL SYSTEM WORK HARDER FOR ECONOMIC DEVELOPMENT...................................................................................209 A. INTRODUCTION ...............................................................................................................................209 B. FINANCIAL DEEPENING AS A PREREQUISITE FOR SUSTAINABLE GROWTH .......................................209 C. IS UGANDA SAVINGS OR INTERMEDIATION CONSTRAINED?............................................................210 D. WHERE ARE UGANDA'S INTERMEDIATION CONSTRAINTS?.............................................................211 E. COST, EFFICIENCY AND COMPETITIVENESS OF FINANCIAL INTERMEDIATION .................................212 F. OUTREACH OF THE BANKING SYSTEM.............................................................................................214 G. RURAL FINANCE ­ ISSUES AND SOLUTIONS....................................................................................217 H. SUPPORTING THE DEVELOPMENT OF THE MARKET FOR TERM FINANCE ..........................................221 8. MANAGING THE MACROECONOMIC IMPACT OF AID BY INVESTING IN INFRASTRUCTURE.................................................................................................................................223 A. SUMMARY.......................................................................................................................................223 B. BACKGROUND.................................................................................................................................223 C. A WORD ON AID INTENSITY & OTHER FLOWS...............................................................................226 ANNEXES ...................................................................................................................................................240 NOTES ON SUB-SECTOR SPECIFIC STRATEGIES .........................................................................................240 NOTES ON SUB-SECTOR SPECIFIC INSTITUTIONAL ASPECTS.....................................................................241 ANNEXES FROM CHAPTER 2 ......................................................................................................................245 ANNEXES FROM CHAPTER 4 ......................................................................................................................257 ANNEXES FROM CHAPTER 8 ......................................................................................................................259 BIBLIOGRAPHY.......................................................................................................................................268 LIST OF FIGURES Figure 1-1: Savings, Investment, Trade and $US GDP per Capita PPP Since 1970 ......................................3 Figure 1-2: Exchange Rate Depreciation Kick-Starts Growth Acceleration ..................................................4 Figure 1-3: Impact of Dependency Ratio on Per Capita GDP ........................................................................7 Figure 1-4: Uganda's Terms of Trade Fell by Much More Than Other African Countries ...........................8 Figure 1-5: World Commodity Prices 1999-2005...........................................................................................8 Figure 1-6: Uganda's Growth: Actual Versus Terms of Trade Adjusted Growth..........................................9 Figure 1-7: Evolution of Sectoral shares of Gross Domestic Product (1981/82-2003/04)...........................10 Figure 1-8: Structure of Exports, 1994/5-2004/5 ....................................................................................14 Figure 1-9: Growth in GDP, GDE and GDY and the Terms of Trade over 1990/91-2004/05.....................15 Figure 1-10: Comparison of Growth Performance and TFP Trend in Uganda (1960-2000)........................19 Figure 1-11: Annual Growth in the Labor Force (1960-2006) .....................................................................21 Figure 1-12: Educational Attainment of Workers (1995-2000) by Cohort...................................................22 Figure 1-13: Population Pyramid of Uganda in 2002 ...................................................................................23 Figure 1-14: Labor Force and Dependency Ratio (1980-2006)....................................................................23 Figure 1-15: Comparison Human and Physical Capital Growth since 1965 ................................................24 Figure 1-16: Formal Sector Firms: Start-Ups by Age...................................................................................37 Figure 1-17: Size Distribution of Firms Started Since 1989.........................................................................37 Figure 1-18: Age and Average Firm Size: Whole Economy ........................................................................38 Figure 1-19: Manufacturing Firms Age and Average Firm Size: Selected Districts....................................39 Figure 1-20: Regional Pattern of Employment by Firm Size........................................................................42 Figure 1-21: Sector Pattern of Employment in 2001/02 ...............................................................................43 Figure 1-22: Sector Distribution of Value Added in 2001/02.......................................................................43 Figure 2-1: Agricultural Growth Remain the Engine of Growth for Poverty Reduction in Low Income Countries.........................................................................................................................................................47 Figure 2-2: Composition of Agricultural GDP..............................................................................................49 Figure 2-3: Agricultural Growth Volatility and Poverty Estimates ..............................................................52 Figure 2-4: Change in Growth Volatility ......................................................................................................52 Figure 2-5: Agricultural Performance: Trends.............................................................................................54 Figure 2-6: Yields and Prices of Food Crops and Industrial Crops in Uganda(1990-2004).........................55 Figure 2-7: Structure of Major Crops by Value (% of total gross crop value) .............................................56 Figure 2-8: Land Use.....................................................................................................................................61 Figure 2-9: Cereal Yields (Average 1999-2004)...........................................................................................62 Figure 2-10: Composition of Crops Grown per Plot......................................................................................63 Figure 2-11: Sole Crop Verus Intercrop........................................................................................................64 Figure 2-12: Land and Labor Productivity....................................................................................................64 Figure 2-13: Uganda's Land Productivity as Compared to other Countries.................................................65 Figure 2-14: Agricultural Value Added per Worker.....................................................................................66 Figure 2-15: Agricultural and Food Crop Monetization ...............................................................................67 Figure 2-16: Shift in Area Allocated to Higher Value Crops........................................................................68 Figure 2-17: Change in Exports ....................................................................................................................69 Figure 2-18: Fertilizer Use (100 gms per Ha) ...............................................................................................77 Figure 2-19: Household Per Capita Expenditures.........................................................................................81 Figure 2-20: Trends in Price Indices .............................................................................................................85 Figure 2-21: Average Monthly Wholesale Price Differential to Kampala (Jan 2001 ­ Dec. 2005).............87 Figure 2-22: CV of Monthly Wholesale Price Differential to Kampala (Jan. 2001 ­ Dec. 2005) ...............88 Figure 2-23: Average Wholesale Marketing Margins for Beans & Real Petroleum Pr................................88 Figure 2-24: Price Margins for Beans ...........................................................................................................90 Figure 2-25: Returns to Intra-Year Storage: Average across Markets (Percent Difference between Min and Max Prices).....................................................................................................................................................90 Figure 2-26: Market Integration ....................................................................................................................92 Figure 3-1: Growth Diagnostics ....................................................................................................................98 Figure 3-2: Enrollments in Primary Education Have Increased..................................................................114 Figure 3-3: The Higher Returns to Higher Education Depict a Skills Premium.........................................115 Figure 3-4: Supply of Workers with Primary and some Secondary Education has Increased Markedly...116 Figure 3-5: The Higher Returns to Higher Education: Evidence of a Skills Premium?.............................117 Figure 3-6: Nominal Prices for Traditional Fuels 1990-2005.....................................................................123 Figure 3-7: Freight Transport Rates of Selected Countries (in percent).....................................................124 Figure 3-8: Returns to Capital in Manufacturing Sector versus Infrastruture availability by District .......125 Figure 4-1: Poverty ­ Human Development Trade-Offs..............................................................................142 Figure 6-1: Details of Load Shedding (2001-05) ........................................................................................170 Figure 6-2: Uganda, Public Investment Composition (as of 2003).............................................................183 Figure 6-3: Intra-Infrastructure Sector Breakdown.....................................................................................184 Figure 6-4: Sectoral Breakdown of Infrastructure Spending funded by Donors (2004).............................185 Figure 6-5: Budgetary Spending on Water and Roads................................................................................186 Figure 6-6: Breakdown of Commercial Bank Credit to the Private Sector ­ by Sectors............................201 Figure 7-1: Bill Rate and Portfolio Flows...................................................................................................211 Figure 8-1: Uganda's real GDP growth and Aid inflows; 1966-2004..........................................................224 Figure 8-2: Aid Intensity Indicators .............................................................................................................227 Figure 8-3: Structure of foreign exchange inflows: 1994q1-2004q4...........................................................228 Figure 8-4: Exports and real effective exchange rate movement, 1994/5-2004/5 .......................................229 Figure 8-5: Aid, Exports and IRER movement, 1994/5-2004/5. .................................................................230 Figure 8-6: Composition of Public Expenditure ..........................................................................................231 Figure 8-7: Volatility In Aid and Tax Revenues..........................................................................................234 LIST OF TABLES Table 1-1: Uganda's Recent Growth Performance in International Context...................................................6 Table 1-2: Estimation of Uganda's GDP Growth Trend..................................................................................7 Table 1-3: Sectoral Composition and Growth of GDP (1990-2005) Shows Slower Structural Transformation ...............................................................................................................................................11 Table 1-4: Uganda's Investment Growth (1997/98=100)........................................................................13 Table 1-5: Growth Accounting Decomposition of Uganda's Growth:..........................................................18 Table 1-6: Selected Human Capital Development/Accumulation Indicators (average2000-2003)...............24 Table 1-7: Agricultural Value Added Growth Rates (1990-2004)................................................................31 Table 1-8: Growth rate of GDP and Change in Shares of Production and Labor Force by Sector: .............32 Table 1-9: Change in Average per Capita Consumption by Sector of Household Head (in constant, 1997/98 prices) .............................................................................................................................................................32 Table 1-10: Household-Based Enterprises by Age .......................................................................................33 Table 1-11: Ratios of Value Added (VA) to Capital and Worker.................................................................33 Table 1-12: Average Consumption by Sector of Household Head...............................................................35 Table 1-13: Monthly Wages of Paid Workers In ISS 2002/03 .....................................................................36 Table 1-14: Percentage of Firms, Number of Employed and Percentage of Employed in Labor Force by Firm Size, 2001/02 .........................................................................................................................................40 Table 1-15: Comparable Numbers for Shares of Formal Employment and Numbers Employed * .............41 Table 1-16: Formal Employment and Number of Firms in Formal Sector by Region and in the Capital, 1989 and 2000/01 ...........................................................................................................................................41 Table 1-17: Sector Shares of Formal Employment­ For Registered Firms Established Before and Since 1989 ................................................................................................................................................................44 Table 1-18: Formal Sector - Real Output, Labor, Capital, Profits and Inputs per Worker 1989 vs 2001 by Sector..............................................................................................................................................................45 Table 2-1: GDP Growth and Sector Contributions .......................................................................................49 Table 2-2: Agricultural Value-Added ...........................................................................................................50 Table 2-3: Sub-Sectoral Contributions to Agricultural Growth....................................................................51 Table 2-4: Principal Crop Growth Estimates from 1989/90-1996/97...........................................................57 Table 2-5: Principal Crop Growth Estimates from 1997/98-2003/04...........................................................58 Table 2-6: Output Growth Decomposition for Individual Crops..................................................................58 Table2-7: Value Growth Decomposition for Individual Crops.....................................................................59 Table 2-8: Comparison of Existing Yields to Research Trials.......................................................................64 Table 2-9: Average Production.......................................................................................................................67 Table 2-10: Averages by Value of Production per Acre Groupings.............................................................71 Table 2-11: Household Use of Improved Agricultural Inputs in 1999/00 ....................................................78 Table 2-12: Results of Engle Curve Estimation............................................................................................83 Table 3-1: Selected Indicators of Financial Intermediation in 2004 across Countries ...............................100 Table 3-2: Sources and Uses of Credit ­ Rural Households.......................................................................101 Table 3-3: Credit Market Participation and Firm Size (by no. of Employees .............................................102 Table 3-4: Formal Credit Market Participation by Firm Size (Percentages of Firms)................................103 Table 3-5: Why did Firms not apply for Loans? By Firm Size (Percentage of Firms)...............................103 Table 3-6: Credit Constraints by Firm Size.................................................................................................103 Table 3-7: Cross Country Comparison (Independent of Firm Size) ...........................................................104 Table 3-8: Investor Perceptions of Likely Direction of Investment in 2003 ..............................................106 Table 3-9: Top 10 Taxed Manufactures (VAT Plus Import Duty as a share of supply).............................107 Table 3-10: Cross- Country Comparison of Selective Micro Risks............................................................108 Table 3-11: Firms' Evaluation of Some Selective Constraints....................................................................109 Table 3-12: Cross- Country Comparison of Cost of Doing Business Surveys...........................................110 Table 3-13: Share of Discovered (non-traditional) Products in Total Exports (%) ....................................111 Table 3-14: Share of Infrastructure Charges in Total Cost, by Sector (2000/01 List)................................118 Table 3-15: Net Versus Gross TFP, Adjusted Prices ..................................................................................119 Table 3-16: Household Fuel Use.................................................................................................................120 Table 3-17: Costs of Running a Generator Versus Obtaining Electricity from the Public Grid ................121 Table 3-18: Costs of Running a Generator versus obtaining Electricity from the Public Grid in 2005.....121 Table 3-19: Trend of Petroleum Prices for a Slecetion of Countries and Regions.....................................122 Table 4-1: Structure of Uganda's Economy in 2003 and 2020 (base simulation).......................................135 Table 4-2: Definitions of Simulations with Accelerated TFP Growth.........................................................136 Table 4-3: Summary at Macro/Aggregate Level ­ BASE Versus TFP Simulations ...................................137 Table 4-4: Sectoral Growth in Production and Trade ­ BASE Versus TFP Simulations............................138 Table 4-5: Labor Employment an Unemployment ­ BASE Versus TFP Simulations................................139 Table 4-6: Definitions of Policy, Government Efficiency and Aid Simulations ........................................141 Table 4-7: Summary of Results ­ BASE and Policy Simulations ...............................................................144 Table 4-8: Sectoral Growth in Production and Trade ­ BASE and Policy simulations ..............................145 Table 4-9: Labor employment and unemployment: BASE versus policy simulations...............................145 Table 6-1: Uganda and International Peers Infrastructure Basic Access Indicators...................................163 Table 6-2 : Change in Household Access to Basic Infrastructure Services................................................164 Table 6-3: Uganda, Share of Roads in Good/Fair Conditions, 1988 and 2002...........................................165 Table 6-4: Uganda and International Peers: Infrastructure Basic Qualitative indicators............................171 Table 6-5: Infrastructure Spending per Capita (Selected Countries) ..........................................................177 Table 6-6: Uganda Overall Infrastructure ...................................................................................................177 Table 6-7: A quick take of infrastructure delivering, spending, policy making and regulatory functions (figures as 2003) ...........................................................................................................................................179 Table 6-8: On- and Off-Budget Investment and Maintenance, Sectoral Breakdown (2003) .....................182 Table 6-9: Infrastructure and Non-Infrastructure in the National Budget....................................................183 Table 6-10: On-Budget Spending: Development and Recurrent versus Capital and O&M.......................187 Table 6-11: Capital versus O&M Spending per Infrastructure Sub-Sector .................................................188 Table 6-12: Central government versus local government spening on Infrastructure................................189 Table 6-13: Indebtedness of Infrastructure Parastatals ...............................................................................190 Table 6-14: Uganda, Infrastructure Parastatals ­ Annual Investment, 2000-04 (1997- USh `000).............191 Table6-15: Operating Cost Recovery of Uganda Parastatals......................................................................191 Table 6-16: Implicit Subsidies to Parastatals and Private Infrastructure Entities.......................................192 Table 6-17: Uganda, Parastatals' Spending Composition Capital vs. O&M, Sectoral Breakdown (2003) 194 Table 6-18: Uganda, Parastatals' Assets Stock, Real Prices (2002-04) ......................................................194 Table 6-19: Overview of Extra-Budgetary Funds in Infrastructure............................................................195 Table 6-20: Uganda Projects with Private Participation (Annual Commitments)......................................197 Table 6-21: Risk sharing Arrangements in Uganda's Infrastructure PPPs.................................................198 Table 6-22: Capital Market Indicators 2005 ...............................................................................................202 Table 6-23: Capital Market Issuance Costs.................................................................................................205 Table 7-1: Financial Intermediation across Countries, 2004 ......................................................................210 Table 7-2: Outreach of the Banking System in International Comparison.................................................215 Table 7-3: Use of Banking Services in International Comparison..............................................................215 Table 8-1: Correlation coefficients (Aid and IRER)....................................................................................230 Table 8-2: Share of Government Spending (Functional Classification) 1998/9-2004/5 ............................231 Table 8-3: Import Share of on-Budget Public Expenditure ........................................................................232 LIST OF BOXES Box 2-1: Some Facts about Coffee and the Ugandan Economy...................................................................51 Box 6-1: Kampala Roads.............................................................................................................................166 Box 6-2: Key Budgeting Principles according to PEAP.............................................................................180 Box 6-3: Overview of Uganda's budget Process ........................................................................................180 Box 6-4: Road Fund, Some International Experience.................................................................................196 Box 6-5: International Experience with PPP Units.....................................................................................200 Box 6-6: Policies to Improve the Availability of Infrastructure Term Finance..........................................208 Box 7-1: Policies to Improve Efficiency and Competitiveness of Financial Intermediation .....................217 Box 7-2: Pilot Weather-Based Insurance in Malawi....................................................................................219 Box 7-3: Price risk insurance in Tanzania....................................................................................................219 Box 7-4: Policies to improve access to finance in rural areas.....................................................................220 1. SOURCES OF GROWTH A. SUMMARY 1.1 The first section of this retrospective briefly traces Uganda's economic history from Independence in 1965, describes the current status and unique characteristics of Uganda's economy, and sets out how the economy has evolved since peace and stability were restored in the late 1980s, and reforms began. Section 2 compares Uganda's growth performance to other countries in the region, and to other regions of the world, and then begins a detailed decomposition of growth; by production (sector) and by expenditure [C+I+G+(X-M)]. The section then assesses the underlying factors behind the apparent recent slowdown in growth, then draws implications of the slow down for policy makers. Section 3 presents the results of growth accounting exercises for Uganda, using the augmented Solow-decomposition1, and applies an econometric analysis to try to account for the role of policy reforms in explaining the residual2. The section then looks more qualitatively at the short-term and transformational factors which explain the large residual value for TFP in the decompositions, considering the respective roles of product discovery and technical change, post-conflict `rebound' effects on capacity utilization, structural transformation, terms of trade, the reallocation of labor, entry and exit of firms, and ratios of land to capital and labor. Section 4 looks for evidence of changes in productivity in firms and farms. 1.2 The key conclusions are positive: 1. Uganda's remarkable run of growth has continued in the face of shocks. 2. Trend growth has not slowed down significantly since 1999. The rebound phase of growth is over, but after deducting adverse terms of trade and re-basing the National Accounts, underlying growth since 1999 is close to that in the 1990s. 3. The economic reform program is working. This underscores the importance of maintaining the existing policy framework of macro stability competition in markets, and level playing field for investors. 4. The quality of growth seems to be improving. Many of the features of a dynamic economy are visible in Uganda's growth experience: · farmers are switching to new higher value crops, · labor is moving into new occupations, · new products are being discovered, and · new exports are emerging. · the technology component of imports and exports is improving. 5. Human capital is improving from a very low base. Primary and secondary enrollment, and recently even university enrollment, has increased sharply. The seeds for an 1See Collins and Bosworth (1996) "Economic Growth In East Asia: Accumulation versus Assimilation", Brookings Papers on Economic Activity 1996(2). 2Following Easterly & Levine (2001). 1 improvement in human capital have been laid. The labor force is set to expand dramatically in terms of size and skills. 6. Now the physical capital stock needs to improve to keep pace with labor growth. Within capital formation, the shares of machinery and equipment and public infrastructure need to expand. 1.3 Some challenges are outlined too: 1. High population growth and a high dependency ratio especially amongst poor families is a concern; this could limit domestic savings and investment and so provide a brake on national welfare unless Uganda can hasten a demographic transition. 2. Much of the growth in output and employment has been in informal and micro enterprises. 76 percent of new firms on the Uganda Business Registry in 2002 which were established between 1999 and 2002 employed less than 5 employees in 2002. 3. Much of the employment growth and new firm entry has been in services. Retail, restaurants, security services, health and education and more recently telecommunications sectors, have seen rapid job growth. 4. Employment in manufacturing and large-scale commercial agriculture has not grown so fast. 5. The Eastern and Northern Regions are trailing behind. B. RECENT ECONOMIC HISTORY: FROM POST CONFLICT, TO MATURE REFORMER 1.4 In the last 20 years, Uganda has undergone an incredible transformation from a failed state to one of the world's fastest growing economies. With the passage of time it is all too easy to forget just how far the country has come. From 1971 to 1986 Uganda's economy was blighted by the succession of economic tyranny under Idi Amin, extreme political instability under Milton Obote, and ultimately the descent into an asset-destroying civil war. During this tumultuous period, 7 per cent of the population was displaced. Exports, which peaked at 30 percent of GDP prior to Independence in 1965, fell to just 8 percent in 1987. Openness (imports and exports as a share of real current price GDP) fell from 50 percent at Independence in 1966, to just 10 percent in 19773. Real GDP per capita for the average Ugandan fell by a quarter (Figure 1-1). By 1987 real GDP was declining by 2.7 percent annually, and inflation was in 3 digit levels. The currency was heavily overvalued, both the current and capital accounts were controlled, as were interest rates and commodity export prices. All sectors of the economy were operating below capacity. Gross domestic investment halved from its peak of 16 percent of market price GDP in 1971, to average 7.8 percent in the period until 1986. Gross domestic savings fell from a post-Independence peak of 17 percent of GDP in 1970, to negative 0.4 percent of GDP in 1980, averaging just 6 percent for the period 1972-86 despite a significant boom in coffee prices in the late 1970s. 3Penn World Tables (PWT 6.1) 2 Figure 1-1: Savings, Investment, Trade and $US GDP per Capita PPP Since 1970 60.00 1200 50.00 1000 Total Trade as % GDP 40.00 800 Gross domestic 670 30.00 680 investment GDP 600 % Gross Domestic 20.00 Savings 400 10.00 GDP Per Capita In US Dollars 200 0.00 60 63 66 69 72 751978 19 81 84 87 90 93 96 99 19 19 19 19 19 19 19 19 19 19 19 19 -10.00 0 Source: Penn World Table 6.1 1.5 Physical, human and social capital was depleted, and firms and farms were run down. Roads were not maintained, and water and power infrastructure was depreciated. Social capital was broken by the conflict and intimidation of people by Amin's army. Trading virtually stopped, and people retreated into subsistence farming. Thousands of Asian families who largely made up the business class were expelled, losing or transferring their wealth in capital flight as they fled. A large stock of private land and productive assets were seized by the state, and given to cronies. These firms were mismanaged, and the appropriated land was inefficiently distributed for political favor. Many homes, factories and offices were destroyed in the conflict. Capital flight was dramatic, despite the controls: by 1986, as much as 60 percent of Uganda's private wealth was held abroad. 1.6 By 1986 Uganda was a heavily under-capitalized, closed subsistence economy with failed public services and severe price distortions. Its structure was adversely affected by civil disorder. The already low share of manufacturing had halved (from just 8.8 percent to 4.4 percent of GDP) while that of subsistence agriculture rose (from 21 to 36 percent), as people retreated into self-sufficiency. Consequently the service sub-sectors also declined (banking, transport, construction etc) fell from 33.6 percent to just 19.6 percent as demand for them from agriculture and industry shrank. The public sector was in disarray. Whereas data for the period are poor, it is estimated that tax revenues had fallen to just 5 percent of GDP, and the black market premium on the exchange rate was several hundred percent, peaking at about 1,000 percent in 1978. The extent of capital depletion has been estimated by Easterly et al. (GIST growth project), and in global calculations by Mahajan (World Bank)4. According to their estimates, Uganda's real physical capital stock in 2000 was still at the level of 1975. With the 4PREMEP database (June 2003), using Nehru-Dhareshwar database and SIMA, and assuming 4% depreciation. 3 labor force rising by around 3 percent annually throughout, this means the capital to labor ratio in the economy would have more than halved in this 20 year period. 1.7 Between 1986 and 1999, Uganda experienced a remarkable rebound in economic growth and poverty reduction5. Much of this was a rebound to past levels of economic activity: by 1996 the economy only just recovered to its nominal 1971 $US per capita GDP. Some of the rebound was down to good luck6; international coffee prices were buoyant in the early 1990s, resulting in an improvement in Uganda's terms of trade by 100 percent between 1992 and 1995. Some was due to the resumption of trading. Some was due to the demand-stimulus of large scale donor-financed rehabilitation in the post-war period. Figure 1-2: Exchange Rate Depreciation Kick-Starts Growth Acceleration 180 10% 160 8% 140 6% 120 x 100 4% de In 80 2% Real exchange rate (PPP) 60 0% 40 3year per capita GDP growth(TOT -2% 20 adjusted) 0 -4% 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 20 Year 1.8 The early rebound was accelerated by good policies. These other factors clearly played an important initial role. Indeed, most post-conflict countries tend to rebound with peace. But most post-conflict rebounds typically run their course within 6 years (Collier & Hoeffler (2001)). And most growth episodes, however they were started, tend to run out within 8 years or so, having run-up against a binding constraint (Rodrik and Velasco (2003) "Growth Accelerations"). Uganda's growth spurt is still running, for more than twice as long as would be regarded as "normal". There can be little doubt that the rebound was first accelerated, and then kept alive, by the authorities' careful sequencing and determined implementation of the most far- reaching stabilization and structural reform program in Africa, and one of the most comprehensive reform efforts in the world7. It appears that Uganda's officials got a lot of things right ­ by first anticipating, and then removing the right policy barriers to growth. So what exactly did they do8? 5 This experience is thoroughly documented in Reinikka and Collier (2001), and in IMF (1994) and is covered in the first background paper in CEM volume 2. 6Consistent with the findings in Easterly, Kremer, Pritchett and Summers (1993) who find that favorable terms of trade are as important as good policies in medium-term growth performance over decades.. 7A comprehensive catalogue of Uganda's reforms was prepared as a background paper and is available on request. 8Appendix 1 presents a comprehensive catalogue of reform measures supported under the ERP. 4 1.9 Policy reform started with a focus on price stabilization and liberalization. Within one year of coming to power, the NRM Government had secured peace across all but the Northern Region. In May 1987 they initiated the first phase of reforms under an "Economic Recovery Program" (ERP), starting with currency reform, devaluations, liberalization of domestic prices, and eventual conversion to floating exchange rate regime by 1993, under phase 2. Figure 1.29 suggests the early decision to depreciate the real exchange rate was closely associated with the growth acceleration which ensued in the following 8-10 years. 1.10 Structural reforms deepened as the reform program matured, and kept providing a growth impetus up to 1999. Following the early first generation of reforms, Uganda's authorities embarked upon a sequenced package of structural reform policies and investments designed to free up markets and create price incentives, stimulate private investment, and encourage competition. Marketing boards were abolished and the financial sector was liberalized. A privatization package focused initially on banks and then on public enterprises, and eventually in 1998 on utilities, including telecoms and electricity sectors. Substantial aid inflows and return of flight capital rewarded the early reforms, followed by an increase in FDI, particularly following the immensely successful telecommunications privatization. Throughout, macro economic stability was maintained. 1.11 From 1997, the focus of reforms switched to poverty reduction. A substantial scale up of aid flows followed the introduction of Universal Primary Education and establishment of sector investment programs. Reform emphasis switched somewhat to improving public service delivery, tightening public expenditure and procurement systems, and building broader systems of accountability, law and order. 1.12 The next section describes Uganda's remarkable growth acceleration. It tries to account for it, and the recent slowdown. The section then ends by putting this recent growth acceleration into the broader perspective of Uganda's relative status today. C. IS AN IMPRESSIVE GROWTH SPURT BY WORLD STANDARDS SLOWING DOWN? Aggregate GDP Growth 1.13 Uganda's per capita GDP growth of 3.5 percent per annum in the 1990s was amongst the fastest in the world10 (see Table 1-1). Annual real growth averaged 6.9 percent, well above that the sub-Saharan African average of 2.0 percent, and also close to that of East Asia and Pacific region of 7.6 percent (Table 1-1). This performance was all the more impressive considering the conflict in the North. 9From Hausmann, Pritchett and Rodrik (2004), "Growth Accelerations". 10Source: World Bank (2005) "Development Lessons of the 1990s" estimates per capita income growth for Uganda at 3.2% per annum. The top 20 fastest growing countries of 152 for which data are available, were: China(8.6%), Vietnam(5.7%), Korea(5.0%), Lebanon(4.4%), Chile(4.3%), Mozambique(4.3%), Mauritius(4.1%), Sudan(3.7%), Malaysia(3.7%), Dominican Republic(3.7%), Loa PDR(3.6%), India(3.6%), Thailand(3.4%), Bhutan(3.4%), Uganda(3.2%), Sri Lanka(3.1%), Poland(3.1%), Bangladesh(3.0%), Tunisia(29%), Iran(2.7%) 5 Table 1-1: Uganda's Recent Growth Performance in International Context Country/region/group Annual GDP growth rates(%) Per capita income growth(%) 1990-99 2000-03 1990-99 2000-03 a. East African region Kenya 2.14 0.82 -0.43 -1.22 Uganda/1 6.86 5.52 3.46 1.80 Tanzania 3.14 5.91 0.22 4.41 b. Fastest growing African countries: 5.38 4.46 Mozambique 5.67 7.30 8.10 5.11 Mauritius 5.38 4.58 4.20 3.47 Botswana 4.70 5.35 1.74 4.17 Tunisia 5.08 4.63 c. Other regions/groupings Sub-saharan Africa region 1.96 3.27 -0.64 0.98 East Asia & Pacific 7.63 6.80 6.24 5.85 South Asia 5.38 5.16 3.37 3.37 Low Income Countries (LICs) 4.36 4.87 2.18 2.95 Source: Uganda Bureau of Statistics for Uganda and World Development Indicators for other countries and regions Note: 1/ Uganda averages 2000-05 1.14 Trend growth over the last 2 decades is estimated at 5.7 percent per year. Trend GDP growth drawn from a univariate fit also appears to show a rise in 1992, as economic reforms gained pace11. This is shown by the positive coefficient for the dummy used to differentiate the post liberalization/reform era in Table 1-2. 1.15 Since 1999 real GDP growth slowed down to 5.5 percent, compared to 6.9 percent in the 1990s. This was still higher than the 3.3 percent recorded for the Sub-Saharan region; but Uganda's exceptionally high population growth coupled with a high dependency ratio, means it translated into per capita income growth of just 1.8 percent. This growth rate is below the LIC average since 2000. 11The long term trend is estimated from the function log(Ymkt) = b1 + b2*Trend + et, where b1 is a constant and b2 is the trend growth of GDP, but also included a dummy to capture the post liberalization era starting in 1992. 6 Table 1-2: Estimation of Uganda's GDP Growth Trend Dependent Variable: Log of GDP at market prices Sample: 1982 2004 Variable Coefficient t-Statistic Prob. C 14.86169 493.9339 0.0000 @TREND 0.057474 17.20766 0.0000 @TREND*DUM 0.003296 1.628607 0.1229 AR(1) 1.075015 7.903529 0.0000 AR(2) -0.461320 -4.413977 0.0004 R-squared 0.998669 Adjusted R-squared 0.998336 Durbin-Watson stat 2.025162 Source: World Bank Staff calculation using data from UBOS 1.16 At this rate, it would take Uganda 40 years to double average income per head, compared to just 20 years with the per capita growth rates of the 1990s. The effect of the dependency ratio is most noticeable when comparing real GDP per worker with real GDP per capita (Figure 1-3). Figure 1-3: Impact of Dependency Ratio on Per Capita GDP 2500 2000 1500 1000 500 0 0 4 195 195 58 2 6 0 4 8 2 86 0 4 8 19 196 196 197 197 197 198 19 199 199 199 GDP per Worker(US $PPP 1996) GDP per Capita(US $ PPP 1996) 1.17 Much of the recent slowdown in growth is explained by adverse terms of trade. Apart from a period in the mid-1990s when coffee prices boomed, Uganda's terms of trade have been trending downwards throughout the reform period, and have in fact fallen by more than any other African country during this time (Figure 1-4). In fact, the terms of trade deteriorated by about 40 percent between 1998/99 and 2003/04 (before improving slightly in 20004/05), with nearly all of Uganda's main exports - traditional and non-traditional - suffering slumps in international prices (Figure 1-5). This coincided with escalating prices for petroleum just as the intensity of petroleum use in Uganda was increasing. As an example, world prices for Robusta coffee in August 2004 were just 36 percent of the level in May of 1997, whilst petroleum prices 7 were doubled. Recently this differential between export and import unit values has got even wider, as petroleum prices have hit near record highs. Figure 1-4: Uganda's Terms of Trade Fell by Much More Than Other African Countries 250.0 Uganda 1986 Africa's Terms of Trade 200.0 African Oil Producers 150.0 % 100.0 1999 1993 50.0 2003 0.0 0 2 6 8 2 4 0 198 198 1984 198 198 1990 199Year199 1996 1998 200 02 20 2004 Figure 1-5: World Commodity Prices 1999-2005 400.0 ) Petroleum 100 350.0 Coffee (Robusta) = Cotton 300.0 Tea Fish 250.0 AVERAGE Poly. (Coffee (Robusta)) 200.0 Poly. (Petroleum) ANNUAL 150.0 1999( 100.0 DEX 50.0 NI - p c t l c t ay Nov Apr Se Feb l Ju De May Oc Mar Aug Jan Jun 4 Ju De M Oc 1 2 Nov3 Apr 3 Sep 4 Feb 04 05 1997 1998 1998 1999 19991999 2000 20002001 200 2002 2002 00 2 200 200 200 200 20 200520 8 1.18 Adjusted for terms of trade, Uganda's recent annual real GDP growth would have been maintained at the same level as the 1990s (Figure 1-6)12. The terms of trade decline propelled a faster deceleration in Gross Domestic Expenditure (GDE), compared to GDP. GDE decelerated from 8.1 percent for the period of the 1990s, to 5.8 percent from 1998/9 to 2003/04. Over most of the 1990s, thanks to favorable commodity prices, annual growth rates of GDE were higher than those of GDP. Figure 1-6: Uganda's Growth: Actual Versus Terms of Trade Adjusted Growth 8.0% 1989/90-1998/99 1999/2000-2004/05 6.9% 7.0% 6.2% 6.2% 6.0% 5.5% 5.0% 4.0% 3.0% 2.0% 1.0% 0.0% TOT Adjusted GDP growth Actual GDP at market price 1.19 Uganda's economy also weathered a banking crisis and a severe drought, but currently faces a power crisis. In 1998, the State-owned Uganda Commercial Bank was closed, restructured, and sold. Lending by Stanbic, the new owner, only resumed in 2004, leaving a credit crunch at the start of the century. Shortly after, in 2002/03, Uganda suffered a severe drought, which affected food crop output and drove up food prices and created a severe power shortage on account of low levels of the Lake Victoria. Coming off the back of severe terms of trade decline, it is remarkable that Uganda's robust growth performance ­ at close to trend ­ continued. Aggregate Supply 1.20 There was a rapid transformation in sector production between 1990 and 1999. Industry was the fastest growing sector, led by manufacturing and construction, which together accounted for 1.8 percentage points of average annual GDP growth of 6.3 percent. Services were the principal driver of overall growth in value added however, providing 2.7 percentage points to GDP growth, with transport & communications, hotels and restaurants and general government recording strong growth. Agriculture grew at a steady rate of 3.9 percent per annum between 1990 and 1999, contributing 1.8 percentage points to growth. 12The TOT adjusted GDP growth is estimated by assuming that TOTs mainly affect exports as imports are more price inelastic. Adjusted GDP growth therefore estimates what the contribution of exports would have been had the terms of trade not changed (= change in export price/import unit price*GDP share of exports). 9 1.21 According to the National Accounts, structural transformation has slowed down since 1999. The contribution of all sectors to GDP growth fell, except for construction. Growth in agriculture decelerated to 3.3 percent over the period 1999/00-2004/05. The sector's contribution to annual GDP growth of 5.5 percent was just 1.3 percentage points, half a percentage point lower than in the 1990s. Industry decelerated to 7.0 percent, contributing 1.6 percentage points to GDP growth. Within industry, manufacturing growth decelerated from 12.3 percent over 1990s to 5.6 percent over the period 1999/00-2004/05, and contributed just 0.5 percentage points. The services sector remained the biggest source of GDP growth, decelerating only modestly, to 6.8 percent. Its contribution to GDP was 2.6 percentage points. The slow down in transformation has led some commentators to suggest that either Uganda has reached the limits of its manufacturing recovery, or that industry needs incentives in order for firms to invest. Figure 1-7: Evolution of Sectoral shares of Gross Domestic Product (1981/82-2003/04) 100% 90% Gas, Electricity, Services 80% mining and water 70% quarrying 60% Construction % 50% Manufacturing 40% Agriculture 30% Slower 20% Minimal Rapid Transformation Transformation 10% Transformation 0% /82 /84 /86 /88 /90 /92 /94 /96 /98 /00 /02 /04 81 83 85 87 89 91 93 95 97 99 01 03 19 19 19 19 19 19 19 19 19 19 20 20 Source: Data from Uganda Bureaux of Statistics 1.22 This apparent slowdown in structural transformation should however be interpreted with some caution, until GDP accounts are updated. The national accounts weights are still based on 1997/98 data and hence could be underestimating the contribution of fast growing sub-sectors to GDP. For example, a simple re-weighting of the index of industrial production for the period 1997/98 to 2004/05 would put average growth at 8.8 percent, higher than the 6.4 percent, derived from the index based on 1997/98 weights. National accounts based in 1997/98 are also probably under-recording value added growth in the fast growing telecommunications and banking sectors, and perhaps also understating education and hotels and restaurants. 10 Table 1-3: Sectoral Composition and Growth of GDP (1990-2005) Shows Slower Structural Transformation Growth rates (%) Sectoral contribution Sectoral shares in to GDP growth (%) nominal GDP (%) 1990-99 2000-05 1990-99 2000-05 1990-99 2000-05 Supply of GDP/1 6.3 5.5 6.3 5.5 100.0 100.0 1. Agriculture 3.9 3.3 1.8 1.3 46.4 33.8 2. Industry 10.0 7.0 1.8 1.6 18.5 24.6 Manufacturing 12.3 5.6 0.8 0.5 7.3 9.4 Construction 8.3 8.3 0.8 0.9 9.6 13.0 Gas, electricity, water 7.6 6.5 0.1 0.1 1.2 1.4 Mining and quarrying 34.1 8.5 0.1 0.1 0.4 0.8 3. Services 7.8 6.8 2.7 2.6 35.1 41.6 Demand of GDP/2 6.9 5.5 6.9 5.5 100.0 100.0 1. Domestic Absorption 7.5 5.3 7.8 5.8 112.4 113.6 Consumption 7.3 5.1 6.4 4.7 96.0 93.2 · Public 7.6 5.1 0.9 0.6 12.1 14.4 · Private 7.3 5.1 5.5 4.1 83.9 78.8 Investment 9.0 6.6 1.3 1.0 16.5 20.5 · Public 4.0 -1.2 0.1 -0.1 5.5 5.3 · Private 12.7 9.4 1.2 1.1 10.9 14.5 · Change in stocks 21.2 18.5 0.0 0.1 0.0 0.4 2. External Absorption 5.5 0.4 0.3 0.0 -12.4 -13.6 Exports 12.7 7.2 1.0 0.8 10.4 12.0 Imports 7.4 4.3 1.3 0.8 22.9 25.6 3.Statistical Discrepancy -0.6 -0.3 ... ... Memo: Monetary GDP 7.9 6.1 5.7 4.8 72.7 80.6 Non-monetary GDP 2.3 3.3 0.6 0.7 27.3 19.4 Source: Uganda Bureau of Statistics Note: 1/ GDP at factor cost 2/ GDP at market price, but Sectoral Shares are based on Total Expenditure on GDP (i.e. exclude the statistical discrepancy) Aggregate Demand 1.23 The economy is still consumption-driven, but is becoming less so. In the past 5 years, consumption has accounted for about 85 percent of growth in aggregate demand (private consumption was 75 percent of total growth in GDE from 2000-05). This compares with 93 percent for the 1990s (of which 80 percent was private). This slow down in private consumption growth ­ most of which occurred after 1996/97, probably implies a return to more normal levels following the post-conflict rebound, and following the end of terms of trade gains in the mid- 1990s13. Given the level of poverty in Uganda, the high share of consumption in income, and the relatively closed economy, there can be quite large multiplier effects to increased incomes. For example, preliminary analysis using the draft Social Accounting Matrix compiled under this study seems to suggest that in Uganda higher incomes from a shift in employment; eg the rise in 13 See Collier, P. and Gunning, J and Associates, (1999) "Trade Shocks and Developing Countries" for a summary of evidence on how economies respond to terms of trade shocks. 11 manufacturing employment in the formal sector, generate gains for the rural economy too, by increasing the demand for higher-value (income elastic) foods such as matooke, rice, poultry and milk. It is possible that growth in Uganda has `fed off itself' since the restoration of formal enterprises. The injection of additional purchasing power from teachers' wages under the education sector expansion since 1998, may have had a similar positive effect. 1.24 Private investment and exports ­ though small - continue to be the fastest growing components of aggregate demand, and will increasingly need to be the drivers of Uganda's growth. Both remain relatively small, but both have been growing. Since 2000, private investment has accounted for 20 percent of annual growth, compared to 17 percent in the recovery and reconstruction period of the 1990s. Exports accounted for 15 percent of growth since 2000, compared with 14 percent of annual growth in the 1990s, despite the coffee price collapse, buoyed mainly by fish & fish products, the "other" exports category, tourism, oil re- exports and to a lesser extent flowers, tobacco and maize. 1.25 The investment pattern shows rising private construction, falling machinery and equipment, and falling public construction (Table 1-4). Bigsten et al. (1999) noted that in 1998 investment rates in Ugandan firms were in line with Africa's average, but profits at half the average14. They attributed this to the extra confidence Ugandan firms had in their economy. The trends in public vs. private investment and consumption growth may in part be due to the privatization program ­ which would have transferred public investments into private hands and should have created efficiency gains which could have reduced private consumption. In that case, the productivity element of growth may be increasing. Nevertheless, when investment growth is deflated to 1997/98 prices, it appears lower than in either the more rapid growth period (1992/93 ­ 1996/97), or in the rehabilitation period (1986/87 ­ 1991/92). 1.26 Total investment remains low by international standards, and its composition may no longer be as productive as it was in the recovery period. The share of private investment in GDP over the period 1997/98-2004/05 is the same as it was over the period 1992/93-1996/97, but a fall in the machinery and equipment component was offset by an exactly equal rise in private construction. The decline in public construction investment therefore accounts for the overall fall compared to the rapid growth period.15 The fall in public structures investment is analyzed further with sectoral data in chapter 7. 14 Bigsten A, Collier, Dercon, Gauthier, Gunning, Isaksson, Oduro, Oostendorp, Patillo, Soderbom, Sylvain, Teal, Zeufack, (1999). 15Obwona M., J. Nannyonjo and S. Bahemuka (2005) 12 Table 1-4: Uganda's Investment Growth (1997/98=100) 1982/83 to 1986/87 to 1992/93 to 1997/98 to 1985/86 1991/92 1996/97 2004/05 Fixed Capital Formation 12.2 16.1 17.2 15.9 of which - Private 7.4 8.5 11.9 11.9 - Public 4.8 7.6 5.3 4.1 Construction 7.8 9.9 11.3 11.9 of which - Private 4.9 5 7.3 9.1 - Public 2.8 4.8 4 2.8 Machinery & Equipment 4.4 6.3 5.9 4 of which - Private 2.4 3.5 4.6 2.8 - Public 2 2.8 1.3 1.3 Source: Uganda Bureau of Statistics 1.27 Exports are diversifying, with most of the net total increase in export earnings coming from tourism & fish. Albeit from a low base, the recent across-the-board increase in non-traditional exports in the face of falling prices for Uganda's traditional exports (coffee, cotton and tea), could be an encouraging sign of economic transformation. By far the main driver of the increase in export revenues has been fish and fish products. However, neither the 200 percent increase in fish product manufacturing in the Industrial Production Index, nor the 295 percent increase in US dollar-valued fish exports is reflected in a proportionate growth in fishing in the GDP accounts: the real value of fishing in monetary GDP is estimated to have grown by only 20 percent since 1997/98. Tourism also accounts for a large share of the increase in exports from 1998/916. The technological component of manufacturing in the fisheries industry is likely higher than that of the coffee sector which it has replaced in export volume; hence the technological composition of exports should have increased. An important indicator of technology adoption, adaptation and innovation will be whether technology in the fish & fish product export industries continues to develop, e.g. with the introduction of new product lines. Another is whether technology is spreading to improve the sustainability of fishing practices. It is not clear that either is happening yet. 16World Bank (2006), Diagnostic Trade Integration Study (DTIS). Tourism statistics in Uganda's Balance of Payments need further scrutiny, because neither the nationality nor the numbers of arrivals in Uganda's game parks and other tourism attractions match well with statistics on visitor arrivals that are used to derive tourism exports. 13 Figure 1-8: Structure of Exports, 1994/5-2004/5 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1993/94 1995/96 1997/98 1999/00 2001/2 2003/4 Cofee Cotton Tea Tobacco Fish Hides & skins Simsim Maize Beans Flowers Others Source: BoU Quarterly reports, various issues. Aggregate Income: And What to Do About the Slowdown? 1.28 Uganda's Gross Domestic Income (GDY) has reflected fluctuations in both the terms of trade and net transfers from abroad. In the period from 1990 to 1998/99, GDP in US dollars grew faster than GDY in US dollars. Since then however, GDY has grown faster, driven by the increase in worker remittances, official transfers (including HIPC assistance), NGO transfers and a reduction in private outflows. Since 1998/99, worker remittances and NGO transfers have accounted for a similar share of the change in net current inflows as aid. Net transfers (official and private) saw two surges. The first, in 1993/94, came from private transfers, as flight capital returned. Net transfers then declined as a share of GDP during the terms of trade boom (Figure 1-7). Since 1998/99 aid flows, debt reduction, increased FDI, and a sharp rise in remittances have been boosted GDY. This offset the income effects of the negative terms of trade that the country experienced over the period. Overall, GDY growth decerated from an average of 7.0 percent in the 1990s, to 5.5 percent over the 5 years to 2004/05. 1.29 These external shocks to Uganda's national income will have affected the distribution of income, and the economy's supply response. First, the transport and distribution sectors have been expanding rapidly with growth, as should be expected in a rural economy with a sparsely distributed population. With rising oil prices, Uganda has been importing a higher volume of petroleum products at higher prices. Unless competition was strong, traders would tend to pass their increased transportation costs on to farmers and consumers; either in lower farm gate prices and higher market prices, or by not showing up at all in more remote areas. Either way, rural communities would likely suffer worsened terms of trade. Second, lower coffee prices would have a direct negative impact, which has been compounded by widespread coffee wilt disease, which reduced coffee yields. Third, whereas the aggregate economy received compensatory FDI and remittances, it is unlikely these would have compensated those who lost through the terms of trade. Lastly, the impact of scaled-up aid for 14 poverty reduction depends upon the incidence of public spending. The recently completed Poverty Assessment suggests that poor groups in rural areas did benefit from social programs under the PEAP such as UPE and water programs. 1.30 The actual impact on national income, national expenditure, and output growth are more complex than simply looking at world commodity prices and balance of payments aggregates. A fuller understanding of the first and second round effects on incomes and demand of different groups of producers and households would require the use of a CGE model, and is beyond the scope of this report17. Four effects are worth considering, along with the path of the exchange rate. The first two factors would have reduced the incomes of farming communities since 1999, and in turn would have taken away what had been a strong source of demand in the rural economy in the preceding 5 years. The third factor would have benefited urban workers. The impact on inequality of the fourth, depends upon who benefited from aid-financed public spending. Since most of these factors works against rural communities, we should expect - and in fact we do observe - a widening of inequality in Uganda since 199918. Figure 1-9: Growth in GDP, GDE and GDY and the Terms of Trade over 1990/91-2004/05 16.0% 160.00 Transfers as share of GDP(leftscale) 14.0% 140.00 12.0% 120.00 10.0% 100.00 ) TOT(1999/00=100) %( (right scale) ge 8.0% 80.00 hnac 6.0% 60.00 4.0% 40.00 2.0% 20.00 0.0% - 1991/92 1992/93 1993/94 1994/95 1995/96 1996/97 1997/98 1998/99 1999/00 2000/01 2001/02 2002/03 2003/04 2004/05 GDP at Market Prices GDE GDY TOT index Net Transfers/GDP 1.31 How should Government react to the growth slowdown? That growth is slowing down, and inequality is rising, is obviously a problem, whatever its cause. The cause appears to be exogenous, and its impact has been disproportionately felt by rural groups. Since the urban economy around Kampala is doing relatively well, the correct policy response will differ from if the economy was stagnant, or if the slowdown was caused by domestic factors, or was policy- induced. Some academics and policy makers have used the slowdown to question whether Uganda's policy framework is working for growth and poverty reduction. The evidence presented here (and subsequently in this report) suggests four key conclusions for policy makers; (i) The reform program is working, in difficult external circumstances. (ii) Trend growth has not significantly slowed down. 17 A CGE model was developed under this study and will be running within the next year to address research issues such as these. 18World Bank (2005) Uganda Poverty Assessment. 15 (iii) It would be wrong to roll back or reverse policies that have sustained two decades of growth in Uganda. (iv) A special effort is needed to resuscitate rural growth to address inequality caused by the fall in terms of trade. 1.32 Prudent exchange rate management, coupled with investments to raise rural incomes and exports should feature in the policy response. The appropriate response to a lasting decline in the terms of trade is to adjust ­ by allowing the currency to depreciate19; a policy which in the early reform days, combined with price liberalization, seems to have helped farmers, especially coffee farmers. In this regard, officials in the Bank of Uganda have rightly been concerned about managing the recent hike in net transfers from aid and worker remittances without allowing the exchange rate to appreciate too strongly. Appropriate exchange rate management is taken up in Chapter 8. Chapter 2 looks in more detail at policies to stimulate agricultural productivity, whilst Chapters 6 and 7 provide recommendations for improving rural finance and infrastructure. The rest of this chapter adopts a growth accounting approach to take the analysis of recent growth deeper; into factor accumulation and factor productivity. It ends the macro stock take by setting out where Uganda is now, and what the implications of current endowments are for growth strategy. D. ACCOUNTING FOR GROWTH ­ MACRO ANALYSIS 1.33 Long run growth is about factor accumulation and technical change, and about the incentives entrepreneurs have to invest in them. The right choice of public interventions for growth depends on whether future growth is more feasible through accumulation and more intense employment of factors of production, or from increased efficiency in the use of available factors. In East Asia there is evidence of more intense factor use and efficiency-enhancing technologies. In Latin America the turnaround in growth in the 1990s was driven by one-off policy-induced reversal of negative growth in total factor productivity in the 1980s. Long-run growth is therefore not just about macro policies. 1.34 Stabilization through demand-management and monetary restraint are necessary but insufficient for long run growth. This report has so far stayed quiet on Uganda's exemplary stable macroeconomic management. The omission was deliberate. Along with the market and trade friendly structural reforms mentioned in section 1.0, macro economic stability in Uganda has doubtless underpinned the solid growth performance. It is a critical foundation for price signals to work, for investor confidence, and for FDI. We know from global experience dating back to the 1970s that not having macro stability is enough to wreck a country's growth prospects; but having stability is not enough for growth20. Economic textbooks tell us that demand management policies for macroeconomic stability affect the cyclical fluctuations of an economy around its potential or full-employment level of output. Long run growth in contrast, depends upon the supply of factors of production in an economy, and upon technological change. 1.35 Nor is long run growth about the endless pursuit of progressively more comprehensive structural reforms. The tradition of macro economists in assigning policies to 19See Paul Cashin and Catherine Pattillo, 2000, "Terms of Trade Shocks in Africa: Are They Short-Lived or Long-Lived?" IMF Working Paper 00/72 20De Long and Summers20 (1992) note that whereas poor macro policies can cause dismal performance in an economy, good macro policies are insufficient for an outstanding growth performance. 16 long run growth probably dates back to Easterly et al. 1991.21 Now 14 years on, Easterly (2005) is amongst those questioning the extent to which gradual changes in policies really do matter for growth. Capabilities definitely matter, and they matter a lot. Countries which maintain openness to trade and investment, maintain sustainable public finance and sound money, and who protect contracts and property rights do tend to earn a long term growth dividend.22 Rodrik23 however, goes further on the limits of standard policy reform prescriptions. He suggests that neither casual empiricism nor statistical analysis suggests we should expect a determinate, uniform, and non- context specific relationships between specific policy reforms and economic growth. He suggests that most developing economies exhibit considerable "slack" and can therefore respond very vigorously to relatively small-scale changes in the business environment if binding constraints can be relaxed. Growth Accounting Using Solow's Decomposition: It's the TFP Residual Again 1.36 If factor accumulation, factor efficiency, and factor mobility into new products and through new technologies are what matter most for long-term growth, what's the story for Uganda? We start with a review of various growth accounting exercises for Uganda using variants on Solow's decomposition, before delving into physical capital accumulation and total factor productivity (TFP) in more detail. 1.37 TFP growth accounts for most of Uganda's real GDP per capita growth during the 1990s 24(Table 1-5). Kasekende et al. (2004)25 conclude that 90 percent of Uganda's growth in output per worker in the 1990s can be `explained' by the TFP residual. They arrive at the same result using a regression-based decomposition of real GDP growth over the same period using the Hoeffler (1999) augmented Solow decomposition26, i.e. the large growth in per capita real GDP over the 1990s of 3.3 percent resulted from productivity growth. They calculate that `base' factors (initial income, investment rates, education attainment, and replacement investment) would have given negative growth in per capita income. Berthelemy and Soderling (2002) use an error correction model formulated from the long run co-integration between GDP per unit of labor with capital per unit of labor and proxy variables for factor productivity. They got similar results; as did Tahari et al. (IMF 2004)27. While growth in the preceding two decades relied on capital accumulation, 80 percent of growth in GDP during the reform period came from TFP. Fortunately, since it uses broadly the same methodology as Kasekende et al. the World Bank's own calculations28 also suggests that while TFP growth has over the long term (1965-2004) 21 William Easterly, Robert King, Ross Levine and Sergio Rebelo "Do National Policies Affect Long-Run Growth? A Research Agenda (World Bank, 1991). 22Quote - Larry Summers (2003), from Rodrik (2004). 23Rodrik, D. (2004), "Growth Strategies" 24This conclusion is broadly shared by Dunn (2002), Bethelemy and Soderling (2001), Keefer (2000) and most recently by Tahari et al., IMF Working Paper 04/176 (2004), and Kasekende, Atingi-Ego and Sebudde (2004). 25From Bosworth and Collins (2003). The basis for these growth accounting exercises is a benchmark Cobb-Douglas aggregate production function in physical capital K and effective labour, h*L, : Y = AK (h*L(1- ) where =0.35 r in per worker terms y = Ak (h(1- ) ) ) 26Hoeffler augmented Solow decomposition is a regression of the form dlny = lnyt-1 + x't + z'+ et 27Tahari, A. Ghura, D. Akitoby, B. and Aka, E. (2004) Sources of Growth in Sub-Saharan Africa, IMF Working Paper WP04/176 28"Measuring growth in total factor productivity."; PREM Note 42 17 contributed negatively to growth, the growth phase of the 1990s is mainly attributable to productivity growth (Figure 1-11)29. 1.38 Unsurprisingly, TFP growth also seems to have slowed since the late 1990s, signaling a possible lowering in social returns to investment in Uganda. Dunn (IMF 2004) suggests that while factor productivity rose rapidly in the 1990s, it has since slowed down. Obwona, Nannyonjo and Bahemuka endorse this view in their recent study. They conclude that TFP growth was just 0.4 percent in the period from 1997-99 to 2004.30 They add that this is reflected in an increase in the ICOR from an unusually impressive 1.8 in the early reform period (1992-1996) to a still impressive 2.2 since 1996. Investment can increase growth through an increase in the share of output that is invested, and through an increase in the social return on the investment; d (g) = r *d(I/Y). For Uganda the private investment share of real GDP remained constant in the period after 1997/98 when the ICOR deteriorated. Hence one can infer that either public capital was falling, or social returns to investment were falling, or both. Public investment's average share of real GDP did in fact fall ­ from 5.3 percent in the period 1992/93 to 1996/97 to 4.1 percent in 1997/98 to 2004/05. The ICOR deteriorated by more than the fall in the real share of investment in GDP, implying a fall in aggregate returns to investment. Given that public investment is generally a complement to private investment, this could suggest that falling public investment is behind a decline in real social returns to investment. Further decomposition of investment is needed to uncover more. Table 1-5: Growth Accounting Decomposition of Uganda's Growth31: Percentage Contribution Dependent Dependent Physical Human TFP Source/Study Period variable variable capital capital growth Kasekende et al. 1960-97 Output per -0.23 134.7 69.5 - worker 304.3 Kasekende et al. 1990-97 Output per 2.5 2.4 6.4 91.6 worker Berthelemy and 1986-96 GDP per unit 3.8 19.4 .... 80.6 Soderling (2002) labour Dunn (2004) Tahari et al. 1960-2002 GDP growth 5.1 13.7 29.4 54.9 (2004) 29Most growth accounting studies conclude that the TFP contribution to growth is larger when growth is higher, see for instance Loayza, Fajnzybler and Calderon (2005) for Latin America, and global estimates by Bosworth and Collins (1996). 30Obwona, M. Nannyonjo, J and Bahemuka S. (2005). The trend and financing of investment at the macro level in Uganda: The implications for sustainable growth, EPRC. 31See background paper: Macro stocktake 18 Figure 1-10: Comparison of Growth Performance and TFP Trend in Uganda (1960-2000) 10.00 10.00 5.00 0.00 5.00 -5.00 -10.00 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 0.00 % -15.00 -20.00 -5.00 -25.00 -30.00 -10.00 -35.00 -40.00 -45.00 -15.00 TFP growth GDP growth(5-yr miving average) Source: Kasekende et al. (2004) Trends in Physical Capital Accumulation 1.39 Decomposition of gross fixed capital formation reveals a significant increase in real commercial buildings investment, and falls in equipment investment as a share of GDP. This is further evidence of lower capital equipment per worker. Structures investment dominates the investment picture in terms of shares and recent growth. The share of machinery and equipment investment has been falling since 1996/7, with all of the fall in the share of GDP being due to a fall in private equipment investment. This could be regarded as a property boom, most likely around Kampala and Entebbe. The finding justifies further investigation. It may reflect the emergent structural transformation of the economy towards services after the manufacturing sector rebounded, but it could be potentially concerning. 1.40 The fall in equipment investment could signal a growth concern. Much of the change in economic performance across countries comes from productivity growth, and some theorists attribute the uptake of new productivity-enhancing technologies to investment in equipment rather than structures. De Long and Summers (1992) show that machinery and equipment is closely associated with output per worker in economies. They note that equipment embodies technology transfer, and learning-by-doing. They suggest that the direction of causality seems to run from equipment to growth, and that equipment's effects on growth are structural ­ ie they are not the by-product of other contributors such as policy improvement. They also note that relative prices between equipment and other investments matter for the ensuing pattern of capital formation. Ortiguera (2003) also finds strong links between; a) investments in equipment and income growth, and b) the rate of decline of equipment prices and income growth. He finds that the production of equipment goods leads to faster technological change than the production of final goods, because improvements in the production of capital goods lower their relative price. New research from Hausman et al. (2005) suggests that it matters what a country exports, and to who32. They note that countries that have grown rapidly from exporting have tended to sell 32Hausmann, Hwang and Rodrik (2005) "What You Export Matters", NBER Working Paper. 19 successively higher technology products to countries richer than themselves. Such products are generally manufactured, and so their production employs equipment. 1.41 In Uganda, nearly all capital goods are imported ­which makes the exchange rate and taxes important in determining investment prices. The country's landlockedness makes equipment expensive. This fact creates a peculiarity ­ in that the exchange rate has an effect of the relative price of equipment and structures investment, as do border taxes33. For Uganda, taxes seem heavily skewed towards equipment rather than property or rental income, and the exchange rate depreciation caused by declining terms of trade could also be marking up capital prices relative to buildings investments. 1.42 The slow rate of accumulation of equipment, coupled with rapid population growth, suggests capital shallowing in the economy. A fast rate of growth of labor implies that a greater share of national product must be devoted to investment in physical capital and education just to keep the average level of skills and amount of physical capital used per worker constant. Berthelemy and Soderling conclude that early TFP growth in Uganda characterizes the one-off effects from successful adjustment policy, and that sustaining growth in Uganda requires capital accumulation and accelerated structural change. Kasekende et al., and Dunn also conclude that future growth will require stepping up the country's investment efforts to promote capital deepening. Human Capital Trends 1.43 What about human capital accumulation? There are two main factors; the size of the working population, and the average education of workers. In addition, the number of dependents per worker (dependency ratio) is an important consideration for growth and national welfare. Placing a high economic responsibility upon workers tends to reduce the propensity to save and invest. 1.44 Uganda's labor force has been growing at an average annual rate of 3.2 percent per year during the reform period. (Figure 1-12). This is a slightly higher rate of growth than the average since the 1960s (3.15 percent). There are two discernible turning points in the early 1970s and in the early 1990s which may relate to the impacts of the expulsion of Uganda's Asian population and the impact of high mortality from HIV/AIDS respectively. 33The construction sector also has import content. 20 Figure 1-11: Annual Growth in the Labor Force (1960-2006) Uganda: Estimates of Annual Labor Force Growth 1960-2006 3.9% 3.7% 3.5% 3.3% 3.1% 2.9% 2.7% 2.5% 60 63 66 69 72 75 78 81 84 87 90 93 96 9 2 5 19 19 19 19 19 19 19 19 19 19 19 19 19 199 200 200 Source: Staff Estimates based upon UN Population Data 1.45 The average skills of the workforce have also been increasing. Using data for 2000, Figure 1-12 shows that for both the 15-19 years cohort and for the 30-39 year cohort, average years' schooling increased between 1995-2000. Since the younger cohort enjoyed a higher educational attainment, and since enrollments in primary education have increased dramatically since the start of the UPE program in 1998, the average years' schooling of the workforce in Uganda should continue to rise healthily in the coming decade. Secondary school enrolment remains low relative to the fast growing economies in Africa (averaging 14.6 percent, compared to the sub-Saharan average of 24.0 percent ­ Table 1-6). The challenge for Uganda will be to maintain education standards and find jobs for a growing and more skilled future workforce. 21 Figure 1-12: Educational Attainment of Workers (1995-2000) by Cohort 15-19 years 30-39 years 100% 90% 80% 80% 70% 60% 60% 50% 40% 40% 30% 20% 20% 10% 0% 0% 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 grade grade Uga 1995 Uga 2000 Uga 1995 Uga 2000 Source: World Bank Research Project on Education Attainment and Enrolment around the World (www.worldbank/research/projects/edattain.htm) 1.46 But Uganda has the highest dependency ratio in the world. Half of the population is children (see Figure 1-14). The main reason for this is the high fertility rate. At 6.8 births per mother on average in 2002, Uganda's fertility rate is the third highest in the world. The situation in urban areas is markedly better; fertility and dependency are lower in Uganda's cities than in rural areas. The dependency ratio currently stands at 1.12 (ie there are 1.12 dependents per worker. This compares with 0.84 for Kenya and 0.85 for Tanzania, and the average for sub- Saharan Africa (which is already high by world standards) of 87. The dependency ratio increased during the reform period (Figure 1-15), and it increased by considerably more in rural than in urban areas. World Bank staff estimates suggest that the dependency ratio will continue to rise ­ and will not peak until 2013 (World Bank (2006) forthcoming). A major policy challenge will be to hasten the demographic transition and generate a per capita growth premium form this (see Klasen (2005)). 22 Figure 1-13: Population Pyramid of Uganda in 2002 Source: UBOS. The 2002 Uganda Population and Housing Census. Figure 1-14: Labor Force and Dependency Ratio (1980-2006) 1.2 51% 50% 1.1 49% 1.0 48% 0.9 47% 46% 0.8 Dependency Ratio (Left Axis) 45% 0.7 Labor Force % (Right Axis) 44% Under 15s % (Right Axis) 0.6 43% 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 Source: UN Population data 1.47 Human capital accumulation has exceeded physical capital accumulation through three decades (Figure 1-16). Human capital was boosted in the late 1990s by the return of residents following the end of the war. Positive growth rates have been maintained through public policy on education, and private demand for education returning to traditional levels. 23 Figure 1-15: Comparison Human and Physical Capital Growth since 1965 9.00 8.00 7.00 6.00 5.00 4.00 % 3.00 2.00 1.00 0.00 1960 1965 1970 1975 1980 1985 1990 1995 2000 -1.00 -2.00 Physical Cap Stock Growth Human Capital Growth Source: Kasekende et al. (2004) Table 1-6: Selected Human Capital Development/Accumulation Indicators (average2000-2003) Country/region/group Education effort Education Outcomes Public Primary Secondary Tertiary Labour force Expenditure enrolment enrolment enrolment with primary on Education education in GDP(%) a. East African region Kenya 6.6 68.4 21.3 2.2 ... Uganda 2.5 52.7 14.6 2.0 62.0 Tanzania 2.2 47.3 5.4 0.5 43.0 b. African countries among the fastest growing countries in the world Mozambique 2.4 47.3 8.2 0.6 ... Mauritius 4.1 96.9 64.1 6.0 65.6 Botswana 2.2 80.9 45.5 3.2 63.2 Tunisia 7.6 94.2 68.6 15.1 42.9 c. Other regions/groupings Sub-saharan Africa 3.4 53.5 24.0 2.7 45.6 East Asia & Pacific 2.2 95.6 56.3 7.5 47.9 South Asia 2.8 84.0 44.6 7.8 43.8 Low Income 2.8 76.2 29.1 6.9 44.4 countries Source: World Development Indicators (2003) 24 Understanding the TFP Residual ­ Towards and explanation of the drivers of growth in Uganda 1.48 The TFP residual in standard growth accounting is "a measure of our ignorance"34. It captures all factors which contributed to growth other than the accumulation of human and physical capital. Given Uganda's recent economic history, the TFP residual most likely captures; 1. "catch-up" - an increase in utilization of the economy's resources (including farm land) during post-conflict rebound; 2. "policy-induced rebound"; improved efficiency in resource allocation from increased trading and openness to international trade); 3. terms of trade fluctuations; 4. "reconstruction-induced rebound"; higher investment returns as economic infrastructure was rehabilitated; 5. transformational growth, including: · shifts in labor between sectors and lines of production · efficient firms expanding at the expense of inefficient firms as outward-oriented policies expose them to competition · new investments bringing a higher quality of capital35. · productivity-enhancing technical change; either within farms and firms through technology adoption, or from the entry of new farms and firms, and new product innovation. Short-term Factors versus Longer-term Transformational Aspects 1.49 Quantitative assessments identify short-term factors including terms of trade and rebound through policy reforms. Some authors have tried to estimate the determinants of TFP, although such attempts are fraught with difficulty because growth accounting is at best imprecise. Data gaps in measuring physical and human capital are passed on to the TFP residual, and get picked up in coefficients in TFP regressions. Berthelemy and Soderling, and two of the background papers to this study have attempted a quantitative explanation of the drivers of TFP36. Of the factors listed above, each finds that structural transformation and openness have been important long-term factors, whereas terms of trade and policy factors have been significant short-term influences. The story that emerges is that Uganda has bounced back, thanks to good policies and good luck, and has experienced a slowdown despite good policies, through a recent run of bad luck. Structural transformation has started, but Uganda has a long way to go to generate employment in jobs which yield higher productivity. 1.50 Apart from policy and prices, other macro factors which may have driven TFP growth in Uganda include the quality of growth, rising hi-tech products in both exports and imports, increasing number of firms that exports and rural urban migration,and diversification within and out of agriculture. 34Easterly and Levin (2001), "It's not Factor Accumulation: Stylized Facts and Growth Models". 35 The World Bank's Investment Climate Assessment (2004) notes that Uganda's capital stock in manufacturing is young ­ consistent with re-habilitation of businesses by returning Asian entrepreneurs. 36Obwona et al. (2006) and Sebudde (2006). 25 1.51 First, the quality of growth in Uganda seems to be improving. Many of the features of a dynamic and transforming economy seem visible in Uganda's recent growth experience. Changes in Uganda's product mix, labor movements across sectors, a changing composition of imports, and efficiency indicators for formal sector firms all point to a more positive growth picture. 1.52 Second, new exports have been discovered, with higher technology content than Uganda's traditional exports. Exports remain weakly diversified, and still remain too low in volume terms to drive economic growth (see trade chapter). A common measure of the extent of concentration or diversification, is the Herfindahl Index, ranging from 1 to zero ­ where 1 represents extreme concentration (ie one export). Nevertheless, the number of products exported by Uganda has risen substantially from 11 in 1986, to 21 in 1990 and averaging over 50 since 2004. Although still natural resource-based, several non-traditional exports have involved product processing, nudging the export basket towards higher technology exports37 38 . These emergent successes could be worth encouraging, because exporting firms experience significant efficiency gains, and the number of firms exporting does not seem to be increasing due to high start-up costs (Ha (2006) finds results consistent with Gauthier (2001) on this point). 1.53 Third, the share of manufacturing firms exporting seems to have decreased between 1998 and 2002/03, although the value of their total exports has improved. Ha (2006) concludes that firm productivity decreased marginally between 1998 and 2002/03. He finds that imports were a significant contributor to productivity. Firm size and foreign ownership determine exports. He also concluded that a firm's productivity does not seem to affect its export market share. Exports were only weakly found to cause productivity. 1.54 Fourth, the share of high technology products in imports increased from 11 percent of imports in 1996, to 16 percent in 2004. This occurred despite the rise in oil prices which raised the value of resource-based imports. It largely reflects the boom in telephony, ICT and financial services following liberalization of Uganda's telecommunications sector in 1998, the banking reforms of 2001, and the re-habilitation of the electricity companies following reform in the energy sector. Growth and investment in manufacturing of metals and plastics has also contributed to machinery imports, but it is the services boom that has driven an increase in hi-tech imports. 1.55 Fifth, there has been gradual net migration to the cities, but by no means a mass- migration from rural areas. Extrapolations from the 2002 Census show that the urban population in Uganda in 2005 is still only around 15 percent of the total. UBOS estimates that the population of Kampala has been increasing by around 40-50,000 people per year since the 2002 Census, a rate of growth of around 4 percent, compared with urban population growth for the whole country, of about 5.3 percent and national population growth of 3.4 percent. There are estimated to be over 2 million more people living in urban areas now than in 1991 with about a quarter of them living in Kampala. 37 Chandra, V. and Boccardo J. (2006), "Export Diversification and Competitiveness In Uganda", forthcoming. 38New empirical work by Rodrik and Hausman suggests that it matters what a country exports, and to whom. Hwang (unpublished, forthcoming) finds unconditional convergence at the level of individual products, suggesting that incomes in countries that produce products typically produced by richer countries, start to converge with the richer countries. This strengthens the case for supporting successful higher technology exports. 26 1.56 In the 2002/03 UNHS survey, half of Uganda's households had migrated out of their location of birth, and 44 percent of households living in rural areas had migrated at least once. However, most of these migrations had taken place long in the past. Only 10 percent of households had migrated in the previous five years. For all migrant households, those in the richest quintile were more likely than those in the poorest quintile to report a recent migration. In urban areas 10 percent of the poorest quintile are migrants. 1.57 Within agriculture, farmers have been diversifying into higher return crops, increasing yields, and selling more of their output in markets. A much larger share of crop production (about 70 percent) is now being marketed. There has been a diversification in exports of agricultural commodities. Productivity has been increasing since 1996, although the pace of technical change remains far too slow. 1.58 Household labor in rural areas is moving out of agriculture and into higher return informal household-based services. Households have shifted their labor out of crop agriculture and into higher return activities in livestock and fishing and trade and distribution services (Tables 1-7 and 1-8 show the shifts in labor and the higher real returns which have encouraged it). This has increased the productivity of the labor force which remained in agriculture, and in most cases, has increased the productivity of the labor which exited agriculture.39 1.59 There is, however, little evidence to suggest that shifts in formal employment have added much the TFP growth. New registered businesses are being created, but they have created limited job opportunities in the formal sector. In fact, employment by formal firms40 actually fell in the 1990s in all sectors except trade, hotels and restaurants, and manufacturing. Firms in manufacturing and trade created most employment, gross output and value added, with most value added being created by larger firms. Since in contrast, most employment in recent years has been created in small firms, it is unlikely that labor movement within the formal sector has contributed very much to TFP growth. 1.60 The next section of the growth stock-take switches from macro to micro analysis. It considers productivity and labor movements on farms, the productivity of informal household enterprises (informal firms), and productivity and labor movements in formal sector firms. A variety of data sets are used for the analysis of farms, which draws on deeper analysis under chapter 2. For informal household enterprises, the work uses the Informal Sector Survey of the 2002/03 UNHS (3,191 enterprises in the sample). For the formal sector analysis, we use the Business Register for employment data (166,000 firms) and the 2001/02 Uganda Business Inquiry (over 4,400 firms) for information on productivity, capital and costs. First, the section starts by re-capping why policy makers need to concern themselves with productivity and labor movements. 39See World Bank Poverty Assessment, and also Boumeester, K and Burger, K (2006). 40Those with >5 employees. 27 E. ACCOUNTING FOR GROWTH: EVIDENCE FROM FARMS AND FIRMS Why is Productivity so Important for Policy Making and Why Is Labor Productivity Especially Important? 1.61 Policy makers should be concerned with productivity improvements because they can dwarf capital accumulation in their growth impact. As set out in De Long and Summers41 a very simple relationship is useful in thinking about the relationship between investment and growth: Dg = r D(I/Y). The immediate increase in an economy's growth rate (g) from an increase in its investment share in national income, is the product of two things: first, the increase in the share of output that is invested (I/Y), and; second, the social rate of return on the investment (r). If an economy increases (I/Y) by 3 percent of GDP and the investment yields a 10 percent rate of return, its immediate output growth rate will rise by 0.30 percentage points. But it's long run growth rate will not increase by that much, because; a) there may be decreasing returns to scale in investment, and b) the capital stock depreciates, so the investment does not permanently increase capital. This is a big investment increase for a small impact on output. In contrast, the impact of an increase in productivity on growth is much more profound, which is why we see TFP dominating growth accounting exercises in the Solow decompositions. A nation that sees productivity grow three percent per year, sees income per worker rise eight fold over a lifetime; income only doubles if annual growth is one percent"42. 1.62 To maximize growth through encouraging additional investment, it is very important relative to the quantity of investment, that policy makers raise the quality of investment. 43 Public policy should seek to encourage investments with high social returns. There are good reasons to believe that equipment investment might have a strong association with growth through its impact on raising productivity (equipment is associated with learning-by- doing externalities). New and improved technologies require new and improved types of capital equipment. Technological change is capital using, and TFP cannot increase without an increase in capital intensity as well. 1.63 To improve the quality of growth, public policy should therefore set the right price and policy incentives for private investments which yield the highest social return. In Uganda's case there may be a need to look in more detail at incentives to invest in equipment rather than buildings. Policy makers should also select public investments as efficiently as possible. In order to identify investments with high enough social returns to have a substantial impact on growth, it is necessary to find investments with substantial external benefits ­ i.e benefits not captured by the entity undertaking the investment. Identifying and promoting such strategic investments is a critical way in which public policy can promote growth. "Leaving it to the private sector" is appropriate if the policy and price signals they pick up are right for investment. Getting the environment right involves structural and microeconomic policy, hence it is important that policy-makers gain evidence on productivity and returns to capital from microeconomic data (firm surveys in particular). 1.64 The pattern of economic growth matters for poverty reduction. Labor intensive growth benefits people most because labor is the most important, and sometimes the only asset of people, especially poor people. In Uganda many people, including poor people, also own and use land, or have some form of land rights. This section therefore considers trends in returns to farm 41De Long, J.B and Summers, L.H, (1992) "Macroeconomic Policy and Long-Run Growth 42De Long and Summers (1992). 43De Long, J.B and Summers, L.H, (1992) 28 land and labor, and in the movements of labor in the economy, and it considers the future prospects for each. 1.65 Growth in labor productivity matters most for growth in welfare. In a poor country where labor is the abundant factor of production, labor productivity more than anything will `drive' TFP growth. This is true at the level of the individual, the household, the sector or the economy as a whole. It holds even truer if growth in labor productivity results in higher real returns to labor; i.e. if increased labor productivity increases the demand for labor, and thus drives up real wages of workers. Workers with better health, education and equipment, are generally more productive. 1.66 Fast population growth can reduce welfare if human and physical capital per worker falls, leading to lower labor productivity. In Uganda's case, rapid population growth is driving fast growth in the work force, as Uganda has entered the demographic transition. More workers can mean more production, i.e. overall growth. But having more workers in itself may not be welfare-enhancing. It is very important to draw distinctions between national economic growth, per capita income growth, and growth in labor productivity. High population growth will usually lead to high overall GDP growth; because production will rise when there are more people to produce it. So long as the real price of the product being produced does not fall through increased supply, this can happen so long as the new people joining the workforce produce the same amount of the same thing as those who came before them; i.e. it can happen if labor productivity (output per worker) stays the same. 1.67 New workforce entrants can only produce as much value as those who came before, if they either have; a) similar physical and human capital endowments with which to produce the same product, or b) new efficient ways of producing the same product, or; c) if they produce a product with more value. A healthier and better educated workforce tooled with more equipment, more land, more roads and more bridges, will produce more output per person on average than a sick, under-nourished and impoverished workforce operating with fewer tools and machines, on depreciating roads and bridges. Again, there are two concepts in play here: a) factor accumulation - the new working age people need to be at least as well educated and well equipped as their predecessors for their productivity not to fall with rapid population growth; b) factor employment ­ the workers need to be employed in activity that is at least as valuable as that of existing workers. If additional workers don't at least maintain the same level of productivity, national productivity falls, income per capita and welfare falls too ­ people consume less per head, demand falls, production falls and so incomes per head fall some more. 1.68 Labor mobility and relative factor proportions (capital and land to labor ratios) are important for productivity-led growth. For a rural economy like Uganda, the extent of labor mobility into more productive activities will be determined by the demand and supply conditions in agriculture, and the returns to labor in other activities. The land to labor ratio in agriculture and the productivity of agricultural practices will determine labor demand. The attractiveness of alternatives and the rural population growth rate will determine the supply of unskilled agricultural labor. Alternatives to agricultural employment for unskilled rural workers are; unskilled labor in the formal farming sector, government employment, and employment in the informal sector. Productivity and Labor on Farms: What's the evidence? 1.69 For those who remained in agriculture, productivity has been increasing since 1996, although the pace of technical change remains too slow. Area expansion was the main driver 29 of output growth in the rebound phase of growth up to 1996: yields declined between 1989 and 1996. Since 1996 area expansion has accounted for more than half the increase in agricultural output; but aggregate yields have contributed significantly too, reversing the downward trend the first half of the 1990s. The use of modern inputs has grown quite quickly since 1995, but it remains among the lowest in the World. The intensity of fertilizer use in Uganda is less than 10 percent of even the average intensity for Africa, which itself is low. Other inputs such as seeds, capital inputs, and other chemicals - although higher than fertilizer use ­ remain very low. (Chapter 2 offers explanations for the low use of fertilizers, and recommendations for how to address this). Nevertheless, land productivity in agriculture is rising faster than in most developing countries, and is rising twice as fast as Uganda's agricultural labor productivity. Population pressure on the land makes productivity improvements in agriculture imperative for future growth in the sector (see chapter 2). 1.70 A structural transformation in agriculture is gradually underway. A much larger share of crop production (about 70 percent) is now being marketed. Growth since 1998 in land area under different crops shows land is increasingly being allocated to higher value (higher return) crops. Furthermore there has been a diversification in exports of agricultural commodities. As coffee prices fell, Uganda's exports of fish, tea, tobacco, cotton, flowers, cocoa beans and vanilla have increased. In addition, new informal exports of maize and beans and bananas were found in a recent survey to be highly significant. Chapter 2 explores these factors in detail. 1.71 Farmers have been diversifying into higher return crops, increasing yields, and selling more of their output in markets. Uganda's farmers still have scope to improve crop yields - so long as price incentives remain positive. In producing more for markets however, farmers have found themselves more susceptible to real crop price volatility. Data on informal household enterprises points toward new household enterprises having been set up as a response to low returns in agriculture44. A combination of; a) lower incentives to farmers from real crop price volatility, and b) rapid population growth, could easily put a brake on growth in labor productivity in farming, despite the opportunities Ugandan farmers have to increase crop yields. 1.72 High population growth is reducing average farm sizes, which may be lowering productivity in farming. Small farms in Uganda yield high productivity, as one would expect. They also market less of their produce. There is some evidence that beyond a certain dependency ratio on small farms, population pressure is leading to increased intercropping. Intercropping in turn is resulting in a lower input use, and so a lower overall level of crop production, and possibly soil depletion. Population density and high dependency is also associated with higher diversification, largely into food crops, presumably as a risk-mitigation strategy (see chapter 2). With limited new land available for extensification in agriculture, and with the population set to explode, these negative trends in productivity with population growth present a very important challenge for future growth in natural-resource dependent Uganda. It seems likely with the projected population exceeding 100 million people in the next 40 years (from 27 million today), that future generations of people will need to leave the land to avoid a downward cycle of soil infertility, declining productivity and increasing subsistence. Towns will need to increasingly provide jobs for them, and a major challenge in the towns will be supply of adequate water and sanitation, transport and trade infrastructure, and renewable energy supplies (urbanization will lead to increased fuelwood and charcoal demand unless there are cheap substitutes). 44Boumeester and Burger (2006), "The Informal Sector Survey In Uganda", CEM background paper. 30 Table 1-7: Agricultural Value Added Growth Rates (1990-2004) Growth in Population Per Capital Value Added Growth (%) Growth Uganda 3.8 3.3 0.5% Tanzania 3.5 2.4 1.1% Kenya 1.2 2.2 -1.0% Ethiopia 1.8 2.5 -0.7% South Asia 2.7 1.8 0.9% Sub-Saharan Africa 3.4 2.4 1.0% East Asia & Pacific 3.0 1.5 1.5% Latin America & Caribbean 2.2 1.6 0.6% 1.73 Large scale commercial or plantation farming is still the exception in Uganda, despite good productivity levels and healthy productivity growth. It employed a tiny fraction of the agricultural labor force and farm land in 2001/02. Aside from Jinja, Mukono, Masindi and Kasese, Kabarole and Tororo Districts, large scale farming represents less than 2 percent of farmland in use. It is much less for most Districts. Looking only at large farms, just 55 registered commercial farms employed almost 13,000 workers, around 90 percent of total formal employment in commercial farming. Workers on commercial farms enjoy less volatile earnings, and proximity to a commercial farm is associated with higher marketed share of production for smallholders in the vicinity (see Porto 2005)45. Formal employment in commercial agriculture represented 10.5 percent of total formal employment in 2001/02, compared with 17.3 percent in 1989. This change was driven by both a rise in formal employment in services and industry, and a fall in the absolute number of people formally employed in agriculture and fishing according to the surveys46. Yet returns to capital in commercial agriculture calculated from the UBI data, seem to be amongst the highest of all the sectors in the economy, in part because wages are relatively low and are rising more slowly than input and output values. Output per worker in 2001/02 in current prices, although just 43 percent of labor productivity in manufacturing, exceeded that of the formal fishing industry and was rising faster than for the economy as a whole, and faster than the high employment sectors of manufacturing, trade, and hotels & restaurants (see Table 1-17). 1.74 The average education level of farmers also improved. So the story on farms seems to be ­ more workers in absolute terms, but a smaller proportion of the workforce. Farmers have more skills, and likely more equipment than they did at the start of the reform period, but they are facing limits to land fragmentation, soil depletion, and use a low level of inputs. Productivity and Labor in Household Firms 1.75 Employment in informal and household based enterprises has increased very rapidly. Informal enterprises are abundant among Ugandan households. No less than 35 percent of the households in rural areas and 19 percent of the households in urban report having such an enterprise. To illustrate the rapidity of growth in informal enterprises, of the 1.7 million enterprises showing up in the 2002/03 UNHS informal sector survey 56 percent of existing household enterprises were set up in the 5 years to 02/03, with 39 percent opening between 2000- 02/03 (see Table 1-9). In the year 2000, at least 187 thousand small enterprises were started in 45Porto (2005) 46This surprising result may be due simply to sampling given the low number of farms in the 2001 Business Register. 31 Uganda which still existed in 2002/03. In the same year of 2000, however, 158 thousand enterprises were closed. Table 1-8: Growth rate of GDP and Change in Shares of Production and Labor Force by Sector: Share 1992/93 1999/00 2002/03 GDP (factor cost) 100.0 100.0 100.0 Agriculture 48 41 39 Industry 12 18 18 Services 40 41 43 Labor force 100.0 100.0 100.0 Agriculture 82 80 69 Industry 5 4 7 Services 14 16 23 NB ­ construction is here classified as industry, consistent with the IIP (UBOS), classifying it as a service, would have the share of industry unchanged at 10 percent between 1999-2003. Table 1-9: Change in Average per Capita Consumption by Sector of Household Head (in constant, 1997/98 prices) Average per AE Change in average Percent of hh-heads consumption Consumption* Sector 1992 1999 2002/03 92-99 99- 92/93- 1992 1999 2002/03 02/03 02/03 Crop agriculture 20,682 28,759 25,659 4.8 -3.7 2.2 71.6 68.9 52.5 Noncrop agriculture 24,319 32,538 39,288 4.2 6.5 4.9 2.8 3.0 5.1 Mining/ construction 37,832 38,895 42,303 0.4 2.8 1.1 1.7 2.2 2.3 Manufacturing 28,517 44,774 38,929 6.7 -4.6 3.2 4.7 4.0 8.1 Trade, hotels, 42,939 58,439 48,929 4.5 -5.7 1.3 8.2 10.0 22.3 restaurants Transport/ 35,300 55,376 197,739 6.6 52.8 18.8 1.6 2.5 0.6 communications Government services 37,576 56,671 51,024 6.0 -3.4 3.1 7.6 5.8 2.8 Other services 45,842 67,463 66,684 5.7 -0.4 3.8 1.8 3.6 6.4 All 25,154 35,725 36,461 5.1 0.7 3.8 100.0 100.0 100.0 Formal 33,994 54,817 52,623 7.1 -1.4 4.5 19.6 16.6 17.5 Informal 42,297 52,512 47,844 3.1 -3.1 1.2 10.8 14.0 28.4 Agriculture 20,591 28,687 27,049 4.9 -1.9 2.8 69.6 69.5 54.1 "* Average annual growth in per AE consumption. **The formal sector includes wage earners. The informal sector includes self-employed and unpaid family workers in non-agriculture, unpaid family workers in urban areas who do not report a sector, and domestic workers in urban areas. Agriculture includes selfemployed and unpaid family workers in agriculture, domestic workers in rural areas, unpaid family workers in rural areas who do not report a sector, and those in rural areas who do not report either employment status or sector. " 32 Table 1-10: Household-Based Enterprises by Age Year Started: Number % Before 1980 108,003 6% 1980-1989 212,951 12% 1990-1995 292,505 17% 1996-1997 152,405 9% 1998 120,801 7% 1999 159,710 9% 2000 186,659 11% 2001 194,281 11% 2002/03 284,745 17% Total 1,712,060 100% Source: Informal Sector Survey 1.76 Household enterprises now account for about 21 percent of total employment; expansion in start-ups seems to have been sensitive to coffee prices and real interest rates. Based upon calculations for this study from the informal sector survey of the UNHS, informal enterprises comprise about 21 percent of total employment. The average enterprise is young; at 8 years, younger than the average formal sector firm (9.7 years). Years with low levels of the real interest rate are also those with more starters. Each percentage point change of the real interest rate goes along with almost a percentage increase in starts. Macro analysis shows a role for agriculture as the source of finance for new starters. Higher prices for coffee appear to have induced more people to start new enterprises47. 1.77 For rural households, informal and household based enterprises are mostly used to supplement income from agriculture, which remains an important income source and a source of food (the main expenditure item). The average value added of an enterprise is just one third of the average consumption expenditure of the household. The average correlation of value added with household expenditure is just 11 percent in urban areas, and 6 percent in rural areas. For 35 percent of the rural households with a home-based firm, agriculture still is the most important source of income, while 54 percent of rural households indicate the informal enterprise to play this role. In urban areas this latter percentage rises to 74 percent. Table 1-11: Ratios of Value Added (VA) to Capital and Worker Forestry Restaurant Livestock Manufacturing Trade Household-based Mean VA/capital 0.30 0.45 0.16 0.34 0.21 Median VA/capital 0.63 0.36 0.00 0.12 0.18 Mean VA/worker 99,272 250,187 219,007 259,820 319,697 Median VA/worker 72,000 150,000 14,000 43,200 129,600 Non household-based Mean VA/capital 0.90 1.90 0.01 1.66 -0.47 Median VA/capital 0.18 2.46 -0.01 2.56 -0.92 Mean VA/worker 586,543 330,640 14,219 560,444 -165,954 Median VA/worker 90,000 144,000 -12,000 243,600 -132,000 47Consistent with findings following the Kenya coffee boom in the mid 80s in Bevan, D., P. Collier and J.W. Gunning (1989) Peasants and Governments. 33 1.78 Although it exceeds returns in own-account agriculture, labor in the informal sector is far less productive than labor in formal firms. The "1989 Manpower and Employment in Uganda" survey showed returns to informal sector in urban areas were only 13 percent below earnings in the formal sector48. This is definitely no longer the case. Separate establishments for manufacturing in the informal sector (i.e. non-household based in the ISS) have a value added to worker ratio that is about a quarter of the standard for Ugandan manufacturing industry (a standard which itself was considered in the World Bank ICA to be very low by international comparisons). It amounts to half the per worker per year that the RPED survey showed for micro firms in 2003 ($578). Smaller manufacturing firms should have lower labor productivity, if only because they have less capital per worker in general. The RPED survey showed that for medium sized firms (50-99 employees), small (10-49) and micro (<10) firms, value added per worker was $1379, $897 and $578. The outcome of $311 for the non-household-based micro enterprises in the ISS establishments ­ which have an average of only 2.1 workers, seems to fit this series. The pattern may also hold for the household-based enterprises with only 1.5 worker on average, where value added per worker was estimated from the ISS at just $144. Neither the household-based enterprises nor the separate (non-household-based) seem to show any signs of increasing economies of scale in labor ­ value added per worker in the 1 person firm is higher than the two person firm, and 4 employees have similar value added per worker to a one-person firm. In contrast to the earnings function in RPED data, for firms in the ISS, younger household heads show considerably higher levels of value added per worker. Returns to capital are high, especially in the household-based sector; the most obvious reason for this is that capital is under- recorded, but it may also be an indicator of capital scarcity. 48World Bank (1995), "Uganda: The Challenge of Growth and Poverty Reduction" 34 Table 1-12: Average Consumption by Sector of Household Head Agriculture Self-employed Wage earners Non-agriculture (formal sector) (informal sector) Average monthly cons. per AE District Kampala 1992 52,271 78,041 63,809 District Kampala 1999 163,452 84,263 100,074 District Kampala 2002/03 252,442 89,076 114,050 Average annual consumption growth District Kampala 1992-1999 17.7 1.1 6.6 District Kampala 1999-2002/03 15.6 1.9 4.5 District Kampala 1992- 2002/03 17.1 1.3 6.0 Average monthly cons. per AE District Mukono 1992 24,872 37,765 25,111 District Mukono 1999 32,347 45,893 46,920 District Mukono 2002/03 24,310 40,160 50,091 Average annual consumption growth District Mukono 1992-1999 3.8 2.8 9.3 District Mukono 1999-2002/03 -9.1 -4.4 2.2 District Mukono 1992- 2002/03 -0.2 0.6 7.1 Average monthly cons. per AE District Wakiso 2002/03 62,519 87,812 88,499 Average monthly cons. per AE Other urban 1992 24,881 40,494 38,588 Other urban 1999 45,679 63,438 69,608 Other urban 2002/03 36,057 55,868 66,969 Average annual consumption growth Other urban 1992-1999 9.1 6.6 8.8 Other urban 1999-2002/03 -7.6 -4.1 -1.3 Other urban 1992- 2002/03 3.8 3.3 5.7 Note: Consumption is measured in constant, 1997/98 prices. Districts not surveyed in 1999 are excluded from 1991, 2002 samples. 1.79 Successful informal enterprises have helped their owners escape poverty, but they don't generate enough employment to draw other poor Ugandan workers out of poverty. Owning a successful household business seems to have helped many households escape poverty49. But setting up a business is risky given the turnover rate, and requires initial capital, which is generally not available through the banking system (see chapter 6). The wages of paid workers in the informal sector (separate establishments) are between USh 20,000 a month for casual workers and USh 30,000 per month for regular workers. This is close to the value of shillings 23,800 per month that Ellis and Bahiigwa (2003) present as the poverty line in rural areas for 2001, implying that workers taking on such wages are either poor and have no other option, or perhaps have a farm of their own, but need cash. Wages paid in the informal sector are lower than can be explained by labour productivity, and the levels paid do not differ between 49World Bank 2005, Poverty Assessment. 35 male and female workers. The differential between wages and productivity is high by international standards. Paid workers form only a very small proportion of all workers in the informal sector, suggesting that much of the labor in informal enterprises is unpaid family labor. The livestock sector scores highest, with 19 percent of all workers regularly paid, but in the other sectors the percentage is around 7. Since wages are low and employment opportunities are thin, employment in the informal sector micro enterprises is unlikely to draw poor Ugandans out of agriculture and out of poverty. Uganda needs to look to the formal sector for this. Table 1-13: Monthly Wages of Paid Workers In ISS 2002/03 Regular Casual Forestry 21,900 24,261 Restaurants 24,269 18,006 Livestock 31,245 24,410 Manufacturing 27,774 23,006 Trade 32,782 21,837 All 29,948 21,613 Productivity and Labor in Formal Sector Firms 1.80 An expansion in formal employment might also show up as TFP growth. TFP would have increased considerably if; (i) labor was moving from low productivity informal sectors to the higher productivity formal sectors, (ii) labor was moving from low productivity firms within a particular sector, to higher productivity firms; or (iii) labor productivity in the formal sector as a whole was rising. This section tries to find out which of these may be true. We first analyze employment and start-up data from the Uganda Business Register (2001)50 for over 166,000 registered firms in all sectors and across all Districts of Uganda. We compare aggregates of this firm-level information with aggregate sector-level data collected under the 1989 Census of Business Establishments51. Finally, we analyze sector-level productivity indicators for the 4,400 firms surveyed in the Uganda Business Inquiry.52 We conclude that; · aggregate formal employment growth in the 1990s was disappointing, and needs to increase much more rapidly to absorb a much higher flow of semi-skilled workers; · some of this could be due to closures of unproductive firms that could not survive competition; · labor productivity is rising in the formal sector. 1.81 New businesses are being created, but they have created limited jobs opportunities in the formal sector. Out the formal sector firms that existed in 2001/02, the 83 percent established after 1989 provided 71 percent of all formal sector employment. The sixty two percent of all firms that were created after 1995 provided 44 percent of all jobs, whilst the 46 percent of firms started after 1998 created just 36 percent of all formal sector jobs. Younger firms are smaller, and most firms are relatively young. 50UBOS (2003) 51UBOS (1992) 52UBOS (2004) 36 1.82 New large firms are not emerging. Entry of new firms employing 10 people or more is minimal (Figure 1-14). Most small businesses (those employing from 1 to 9 people) that existed in 2001/02 were created in the previous 3 years. The opposite is true for the large firms: less of them were created recently. The medium size firms demonstrate a flat trend: approximately equal number of those was established at different points of time. Figure 1-16: Formal Sector Firms: Start-Ups by Age Number of Businesses by Year of Establishment and Firm Size. Uganda Business Register, 2001/02 (logarithmic scale) ) ,s e 100,000 1-9 sines alcs 10,000 bu ci 10-49 fo mht 1,000 50-199 100 ogari 10 200 and L above Number 1 before 1989 1989-95 1996-98 1999-2002 Year of establishment 1.83 Most employment, and most of the additional jobs created in recent years, seem to be in small firms53. Firms that opened or newly registered between 1989 and 2001/02 comprise 56 percent of estimated total formal employment. Within the firms which started since 1989, some sixty percent of total employment is in micro firms (i.e. firms with less than 5 employees). This means that in total 48 percent of employment in registered firms in 2001/2 was in micro firms. Micro firms account for 84 percent, 81 percent, 75 percent and 64 percent of firms starting up in 2002, 2001, 2000 and 1999 respectively. There is nothing unusual about firms starting small. The same phenomenon occurs in the US and UK economies. The more important questions for pro-poor growth are which sectors these businesses are starting up in, whereabouts in the country they are located, and whether they succeed and grow as they get older. Figure 1-17: Size Distribution of Firms Started Since 1989 Distribution of Employment In Firms Started Since 1989 (Business Registry 2001) 70% 60% 50% 40% 30% 20% 10% 0% <5 5-9 10-19 20-49 50-99 100- 200- 500- 1,000 199 499 999 and above 53This is inferred. We do not know for sure by how much older firms have grown, and it is feasible, although implausible that new firms could have started then shrunk. Further work is under way to assess the extent to which old firms are more productive, invest more, and grow. 37 1.84 Firms in Uganda do seem to grow over time. Available data in Uganda prevents an accurate analysis of whether firms in particular sectors have grown over time. We know only how many workers a firm employed in 2001/02, and when it was started. We do not know how many people it employed upon starting54. However, in line with other studies, we can impute that if average employment per firm increases with age, firms on average do grow. Younger firms appear to be smaller, and the majority of firms are relatively young. Out of all formal sector firms that existed in 2001/02, those created after 1989 constituted 83 percent and provided 71 percent of formal sector employment. Corresponding numbers for firms created after 1995 are 62 percent and 44 percent, for firms started after 1998 - 46 percent and 36 percent. Based upon this, there is evidence that firms which started since the reform period have indeed grown on average. 1.85 Firm growth seems to be more pronounced in Kampala and Wakiso, and is less consistent elsewhere. Figure 1-16 shows that the average size of manufacturing firms during the reform period increased quite steeply with age in Kampala and Wakiso (Entebbe corridor) and from 1990 in Mukono too, but the traditional industrial belt of Jinja and Mbale shows the opposite for the 1990s, whilst Mbarara, Masaka, Lira and Gulu show little sign of firm growth. These Districts account for 70 percent of all formal employment in 2001/02. Figure 1-18: Age and Average Firm Size: Whole Economy Mean Number of Employed in Formal Sector by Year of Firm's Establishment (annually, 1975-2002). Uganda Business Register, 2001/02 70 edy 60 ol 50 p 40 em fo 30 erb 20 m un 10 - 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 20 20 20 year of establishment 54Work by Hesse (2006) for this study shows using limited data available from the 2002/3 RPED dataset that firms which grow are generally more productive. 38 Figure 1-19: Manufacturing Firms Age and Average Firm Size: Selected Districts Manufacturing Firms Average Employment By Age (SelectedDistricts) 14.0 Kampala 12.0 Mbarara 10.0 Jinja 8.0 6.0 4.0 2.0 - 1985-1989 1990-1994 1995-1999 2000-2002 Manufacturing Firms Average Employment By Age (Selected Districts) 10.0 9.0 Wakiso 8.0 Mukono 7.0 Mbale 6.0 5.0 4.0 3.0 2.0 1.0 - 1985-1989 1990-1994 1995-1999 2000-2002 Manufacturing Firms Average Employment By Age 9 8 Masaka 7 Gulu 6 Lira 5 4 3 2 1 - 1985-1989 1990-1994 1995-1999 2000-2002 39 1.86 Overall employment in formal sector firms is low, and the rate of growth in formal jobs, at 5.4 percent from 1989 to 2001, was slower than real GDP growth. At this rate of growth in job creation, it would take Uganda almost 40 years to create 2 million jobs. More ominously, it would take some 13 years just to create the ½ million jobs needed to employ the annual flow of new entrants to the labor force over the next 5 years. 1.87 Formal employment did not grow very quickly during the rapid growth period of the 1990s. In 1989, with 135,45155 people working in the formal sector in 2,787 registered firms, formal employment constituted 1.6 percent of the labor force.56 By 2001 a total of 254,28257 people were employed in, or 2.2 percent of labor force, worked in the formal sector and additional 236,725 people or 2.0 percent of labor force - in urban informal sector.58 Most firms outside of self-employed agriculture are in urban informal sector. In 2001, out of all registered formal and urban informal firms, 92.2 percent were urban informal (with less than five people employed), 6.7 percent were small formal firms (employing 5-19 people employed) and just 1.1 percent were medium and large firms (with 20 and above employees, see Table 1-13). Table 1-14: Percentage of Firms, Number of Employed and Percentage of Employed in Labor Force by Firm Size, 2001/02 Percentage Number of Share of total of all firms workers labor force Urban informal sector firms (number of employed<5) 92.2% 236,725 2.0% Small formal firms (number of employed 5-19) 6.7% 90,241 0.8% Medium size and large formal firms (number of employed 20 and above) 1.1% 164,041 1.4% All formal sector firms 7.8% 254,282 2.2% All firms 100% 491,007 4.2% Source: Business Register 2001. 1.88 In fact, employment by formal firms59 actually fell in the 1990s in all sectors except trade, hotels and restaurants, and manufacturing. Some of this fall may have been due to restructuring following trade liberalization and privatization. It is not possible to say for certain because firm-level data in Uganda are insufficient to look at trends firm-by firm (there is no panel data). Making the most of Uganda's unbalanced RPED surveys, Ha (2006) suggests that even new entrants in the manufacturing sector between 1998 and 2002 have found the going tough, and have experienced some shrinkage in employment. He also notes that on average large firms grew, and small firms shrank somewhat. What we can say for certain is that there has been no major increase in TFP from net labor movements to formal sector firms. 55This figure includes the highly diverse "other services" category which was excluded from Table 1-14. 56Data for urban informal sector are not available for 1989. 57This figure includes the highly diverse "other services" category which was excluded from Table 1-14. 58One can see from these data that formal sector employment increased from 1989 to 2001/02 by 88%. During the same period, the labor force increased by 37% and GDP ­ rose by 41%. 59Those with >5 employees. 40 Table 1-15: Comparable Numbers for Shares of Formal Employment and Numbers Employed * Share of Employment In Firms > 5 Total Numbers Employed By Sector In Firms > 5 Employees 1989 2001 Sector 1989 2001 By Sector : Agriculture 17 11 Agriculture 15,279 11,415 Industry 53 41 Fishing & Forestry 8,171 1,318 Services 30 48 Mining and Quarrying 1,597 45 Total 100 100 Manufacturing 48,798 142,385 By Region Utilities (electricity, Gas 6,092 2,517 & Water) Kampala 49 39 Building and Construction 15,005 3,339 Central (ex-Kampala) 26 24 Trade 11,000 38,450 Eastern 12 13 Hotels and Restaurant 6,149 23,711 Western 9 19 Transport & 15,130 4,402 Communications Northern 3 4 Total 100 100 Total 127,221 227,581 NB ­ excludes "other services" category. The data are subject to gaps in Census recording between the periods ­ for instance formal employment in mining in 2001 looks very odd. 1.89 The Central Region (mostly Kampala) has the highest employment and number of firms, but the Western region has done well recently, and has several large firms (Figure 1-17). The regional distribution of employment became more balanced between 1989 and 2000/01, mostly because of employment creation in the West. In 1989, 75 percent of formal employment and 60 percent of formal sector firms were located in Central region, and within Central region 65 percent of regional formal employment and 69 percent of regional formal firms were in Kampala. The Eastern region contributed 12 percent of total formal employment in the country, the Western region 9 percent, and the Northern region only 3 percent. The Central region remains dominant and the North still has the lowest formal employment (4 percent of national formal employment) and number of firms (6 percent of registered firms. The share of formal employment in the Western Region rose significantly, from 9 percent in 1989 to 19 percent in 2001. Little has changed in the relative shares of the Eastern and Northern regions. Of the Districts, Kampala and Wakiso account for 50 percent of total employment. Most employment (76.5 percent) in large and medium size firms (those with 50 or more employees) is in just 12 Districts with municipalities: Iganga, Jinja, Kabale, Kabarole, Kampala, Masaka, Mbale, Mbarara, Mubende, Mukono, Tororo, Wakiso. Table 1-16: Formal Employment and Number of Firms in Formal Sector by Region and in the Capital, 1989 and 2000/01 Regions Kampala Central Eastern Northern Western country 1989 percentage of total employment 49% 75% 12% 3% 9% 100% percentage of all firms 42% 60% 15% 8% 16% 100% 2000/01 percentage of total employment NA 64% 13% 4% 19% 100% percentage of all firms NA 64% 15% 6% 16% 100% 41 Figure 1-20: Regional Pattern of Employment by Firm Size Regional Breakdown Of Formal Employment By Firm Size 2001/02 100% 80% Central 60% Eastern 40% Northern 20% Western 0% <5 5-9 10-19 20-49 50-99 100- 200- 500- 1,000 199 499 999 and above 1.90 Most employment, gross output and value added got created by firms in manufacturing and trade, and most value added was created by larger firms. Trading firms dominate total employment, but most of them are micro firms. Trade, manufacturing, services and hotels & restaurants together comprised 71 percent of total employment and 68 percent of formal employment. Manufacturing and trade together contributed 60 percent to total output and 61 percent to formal sector output. 1.91 While most firms are small, most of output and value added is created by larger firms. In 2000/01, informal sector firms (below 5 people employed), while constituting 92 percent of all firms in formal and urban informal sectors, produced only 28 percent of output and 30 percent of value added. Firms with 5-19 people employed constituted 7 percent of all firms and produced 15 percent of output and also 15 percent of value added. Firms with 20 and above people employed added up to 1 percent of firms and contributed 57 percent of output and 55 percent of value added). Figures 1-18 and 1-19 amply illustrate the share of value added which is accounted for by the big shift in labor into micro enterprises in service sectors subtracting the employment and value added of micro firms gives a very different picture, with manufacturing easily the most important sub-sector in terms of value added. 42 Figure 1-21: Sector Pattern of Employment in 2001/02 Employment by Sector as Percentage of Total Employment. 2001/02 Business Register. 40% 30% 20% all firms formal sector 10% 0% riculture FishingQuarryingac s ing Hotels ans t s tur Utilities truction ade por ance Tr nicationInsur rvice Se Services Ag uf & ns Tr Mining Man Co Commuce and onal rs Finan Pe Figure 1-22: Sector Distribution of Value Added in 2001/02 Value Added by Sector as Percentage of Total Value Added. 2000/01 U B I. 50% 40% 30% 20% 10% 0% lture ing tion e e ad Tr rvices s ricu Fish Quarrying Utilities Hotels ansport ications uranc rvice un Ins Se Ag & nufacturing nstruc Tr ss Se ning Co Ma Comm and onal ce ine rs Mi Bus Pe all firm s form al s ector Finan and 1.92 Formal employment in firms which were established since 1989 has been mostly in retail, restaurants, and education sectors. These three sub sectors together account for 60 percent of total formal employment in "new" firms. Ninety percent of employment in firms which opened since 1989 has come in just 15 sectors. Two thirds has come in just four; retail services, hotels, bars and restaurants, education services and repairs, personal, household, and other services. Table 1-16 provides the contrast with firms established before 1989; tea processing and horticulture are amongst the largest employers in older firms 43 Table 1-17: Sector Shares of Formal Employment­ For Registered Firms Established Before and Since 1989 Before 1989 Since 1989 Cum Cum % % Tea Processing 14% 14% Retails Services 38% 38% Growing Of Fruit, Nuts, Beverage 12% 25% Hotels, Bars And Restaurants 13% 51% And Spices Primary Education 6% 31% Education Services 8% 59% Retail Sale: Food, Beverages And 5% 37% Repairs, Personal Household 5% 64% Tobacco Services Hospital Activities 5% 41% Coffee And Tea Processing 4% 68% Restaurants, Bars And Canteens 4% 45% Public Service Activities 3% 71% Building Of Complete Constructions 3% 48% Wood & Wood Products 3% 74% General Secondary Education 3% 52% Agriculture (Fruits, Nuts, 3% 77% Beverage) Other Retail Sale In Non-Specialized 3% 55% Health Services 2% 79% Stores Wholesale Of Agricultural Raw 3% 58% Manufacture Of Food And Animal 2% 82% Materials Feed Maintenance And Repair Of Motor 3% 60% Wholesale Services 2% 84% Vehicles Hotels; Camping Sites etc 3% 63% Motor Repair And Spare Parts 2% 86% Farming Of Cattle, Sheep, Goats, 3% 65% Manufacture Of Metal Products 2% 87% Horses, Manufacture Of Furniture 2% 67% Agriculture (Vegetables, 1% 89% Horticulture) Medical And Dental Practice 2% 69% Manufacture Of Clothing Except 1% 90% Activities Fur 1.93 There is little evidence of TFP gains from labor shifts within the formal sector. Manufacturing and utilities increased their share of formal sector value added between Overall, the 1990s saw a big shift of labor from agriculture into informal sector services, and a modest shift into formal sector manufacturing, trade, and hotels and restaurants 1989 and 2001/02. The share of transport & communications, buildings & construction and agriculture fell. The driver of increased share for manufacturing was increased manufacturing employment. For utilities, a huge increase in value added per worker (34 percent per year) seems to have been driven by increased capital ­ most likely through construction of the Kiira Dam, and a substitution of labor for capital. The workforce in utilities more than halved, but their staff costs (especially Directors' costs) increased by some 44 percent per year in real terms, whilst profits increased 40 percent per year. Formal employment increased most in manufacturing, trade and hotels, and fell in relatively higher productivity sectors of utilities, construction, and transport and telecommunications. Since 2001 there has been an increase in skilled labor in telecommunications and banking, which is not captured in these results. The main increase in real capital stock in the 1990s seems to have flowed into utilities and transport and telecommunications. 44 Table 1-18: Formal Sector - Real Output, Labor, Capital, Profits and Inputs per Worker 1989 vs 2001 by Sector. Utilities Fishing Mining (electricity, Transport & and Gas & Building and Hotels and & Agriculture Forestry Quarrying Manufacturing Water) Construction Trade Restaurant Communications Total 1989 1,197 250 2,239 8,831 1,714 4,034 10,817 2,620 6,457 5,963 2000/01 5,148 3,443 19,909 12,046 59,771 41,613 12,020 5,986 26,399 11,296 annual Q growth 13% 24% 20% 3% 34% 21% 1% 7% 12% 5% 1989 293 102 285 673 223 307 660 666 905 547 2000/01 685 934 4,355 1,309 17,620 5,941 1,679 866 3,488 1,466 annual L growth 7% 20% 26% 6% 44% 28% 8% 2% 12% 9% 1989 643 154 1,040 1,951 718 1,083 4,362 1,625 3,605 1,895 2000/01 2,659 2,061 11,038 4,000 44,671 11,867 6,975 2,320 12,536 5,050 annual VA growth 13% 24% 22% 6% 41% 22% 4% 3% 11% 9% 1989 349 53 755 1,278 495 776 3,702 959 2,700 1,348 2000/01 1,974 1,127 6,682 2,691 27,051 5,926 5,296 1,453 9,048 3,583 annual Profit growth 16% 29% 20% 6% 40% 18% 3% 4% 11% 8% 1989 555 96 1,199 6,880 996 2,950 6,456 996 2,852 4,068 2000/01 2,489 1,382 8,872 8,046 15,100 29,746 5,045 3,666 13,862 6,247 annual Inputs growth 13% 25% 18% 1% 25% 21% -2% 11% 14% 4% 1989 68 9 1,813 1,112 1 1,019 1,311 2,249 405 849 2000/01 1,738 - 765 23,713 343,767 21,335 13,469 9,773 32,656 15,543 annual K growth 31% -100% -7% 29% 187% 29% 21% 13% 44% 27% Share 1989 1% 0% 3% 50% 0% 14% 13% 13% 6% 100% Of Capital 2000/01 1% 0% 0% 27% 41% 1% 9% 6% 15% 100% Share 1989 12% 6% 1% 38% 5% 12% 9% 5% 12% 100% Of Labor 2000/01 5% 1% 0% 63% 1% 1% 17% 10% 2% 100% 1.94 Unit labor costs grew faster than per worker output for the economy as a whole. Unit labor costs grew faster than productivity in the formal sector in Mining and Quarrying, Manufacturing, Utilities (electricity, Gas & Water), Building and Construction and Trade (Table 1-17). However, since operating costs per worker grew more slowly than output for per worker, both profits per worker actually grew faster than Conclusions from Ugandan farms and firms: · Much of the growth in productivity in Uganda in the early 1990s seems to have come initially from capacity utilization and investment in response to policy reform and rehabilitation, including the rebound of manufacturing. · Later, structural transformation in the form of a movement of labor into non-farm enterprises ­ especially in trade ­ and revolutionary reforms to telecoms and banking sectors have probably accounted for much of the observed improvements in TFP, along with a gradual transformation in agriculture in response to increased demand for higher value foods. · Much of this transformation also has its roots in policy reforms and rural infrastructure investments, which together provided better incentives for trading. Policy reforms also 45 seem to have laid the foundation for more efficiency in manufacturing firms, which now seems to be showing up in lower input costs and higher profits per labor unit in the manufacturing sector compared to1989. · There are few macro and structural reforms left to be implemented! The reform agenda needs to get deeper into specific sector issues which protect competition and private enterprise, and encourage productive investment (some of these are discussed in the background papers on service sectors). · There is widening inequality, little evidence of growth in the average employment level of firms outside of Kampala, and some evidence of an overall fall in the returns to capital in formal sector firms in some industries. · We conclude that Uganda now needs to enter an investment phase. Carefully selected public infrastructure investments are a key priority, especially in electricity and road transportation. These should focus on rural towns with growth potential, Greater Kampala, and trading routes, to reduce the costs of external trade. They should seek to reduce the high costs of infrastructure in Uganda. · The twin aim of public investments should be to stimulate formal sector jobs which attract workers from agriculture, and to encourage commercial ventures which reduce the volatility in demand and prices for rural produce. A special attempt should be made to attract viable exporting companies who could bring new technologies and create out- grower schemes. · Privatization and competition from imports may have slimmed down the formal sector, but it should have emerged more competitive as a result. Indeed there are early signs of the emergence of a range of new processed export products, which would be worth nurturing, without targeting specific firms. · A positive growth agenda such as this will create opportunities for rent seeking and cronyism. Such individual interests should not distract policy makers from protecting the national interests of Uganda, which lie in ensuring a level playing field for competition. · A change in behavior is needed: with more open consultation with and amongst business clusters on identifying coordination gaps, opportunities for technological innovation, and market failures, rather than firm-specific investment incentives. · Uganda's economy has come a long way on `policies and prices': the priority now is to invest in high return projects to increase the set of economic opportunities in the economy; ie to create new sources of growth, with the accent on export potential. Uganda needs investment and behavior change for growth. 46 2. AGRICULTURE SECTOR PERFORMANCE A. INTRODUCTION 2.1 This chapter diagnoses growth in the agriculture sector, with a particular emphasis on its contribution to poverty reduction. Consistent with chapter 1, it seeks to provide an understanding of Uganda's recent growth experience, dating back over the past two decades. Then, consistent with chapter 3, and within the limits of available data and time constraints, this chapter also seeks to identify key constraints which could be restricting growth in the agriculture sector. The chapter then tries to draw implications for policy. B. AGRICULTURE AND PRO-POOR GROWTH 2.2 Agriculture has a critical role to play in growth and poverty reduction. The international empirical evidence on the importance of agriculture for poverty reduction and growth is well documented (World Bank 2005a and Figure 2-1). Agriculture contributes most directly to the reduction of income poverty. The other (non-income) dimensions of poverty are equally important but are beyond the scope of this analysis. A full analysis of Uganda's performance on the various dimensions of poverty is discussed in detail in a recent poverty assessment report (Poverty Assessment, World Bank, 2005b). Figure 2-1: Agricultural Growth Remain the Engine of Growth for Poverty Reduction in Low Income Countries Agricultural Growth remains the engine of growth for poverty reduction in low income countries 0 -1Ethiopia -0.5 0 0.5 1 1.5 2 2.5 3 -0.5 ytr -1 Zambia Senegal oveplarur S l -1.5 Burkina F India -2 of )ry/ ge (% -2.5 Bangladesh anhc -3 of Uganda eta -3.5 Ghana R Vietnam -4 -4.5 Growth AgGDP/per cap, 1990s Source: World Bank 2005: Operationaling Pro-Poor Growth 2.3 Between 1992/93 and 1999/2000, Uganda's growth strategies were impressively pro- poor (Okidi, et al. 2005). The recent albeit modest reversal is also tied to the performance of the agricultural sector. According to the 2002/03 household survey, about 86 percent of the 47 population (83 percent of households) were rural, and the headcount poverty in rural areas was 42 percent, compared to the national poverty estimate of 38 percent (World Bank PA, 2005). Rural and aggregate poverty declined from 60 percent and 56 percent, respectively, in 1992 to 37 percent and 34 percent in 1999, before rising slightly to the levels of 2002/03. Slightly over 52 percent of the total population and 61 percent of the rural population lives in crop farming households. Among all sectors, poverty is the highest among those dependent on crop agriculture, and the largest number of absolute poor are dependent on crop agriculture. This implies that what happens in crop agriculture would have a direct impact on poverty. 2.4 Agriculture contributes directly to poverty reduction by reducing real prices of food and increasing the real incomes of the population directly dependent on agriculture as their main source of income. A decline in real prices of food increases the real incomes of all, but particularly the poorest who spend a large proportion of the income on food. Lower real food prices also underpin expansion in employment throughout the economy (through reduced inflationary pressure on wages). Increasing incomes of agriculturists has a direct impact on the majority of the population that is still employed in agriculture. Agriculture also has a higher multiplier effect as higher incomes spur demand for non-agricultural goods. Higher agricultural incomes are effected through increased productivity, for both food and non-food crops. 2.5 For higher farm incomes, the rise in productivity must offset the decline in prices of agricultural commodities. Higher productivity in cash crops is critical to help reduce the raw materials for domestic agro-based industry and to increase the competitiveness of exports. 2.6 Higher productivity in both the food and non-food sectors is critical for poverty reduction and growth. To achieve productivity growth, attention needs to be directed on both the supply and demand. On the supply side, raising farm (land and labor) productivity highlights the role of agricultural technology (generation and dissemination), but also market efficiency (to improve incentives for producers through higher real producer prices) both of which are equally important. 2.7 Ensuring adequate producer incentives, by keeping producer prices from falling faster than productivity gains, requires outlets for agricultural products and hence a focus on the demand side. Demand side comprises domestic demand (for food and raw materials) and export (for industrial crops). Improved marketing efficiency will help also lower final consumer and raw materials prices as well as make Ugandan exports more competitive. And while improved markets will help stimulate the demand side, higher real non-farm incomes, particularly among the urban and rural poor, will be important to increase the demand for agricultural commodities. Actions on all three fronts are needed to increase farmer incomes through higher productivity, even as the final consumer and producer prices decline. C. AGRICULTURE SECTOR 2.8 The share of agriculture in the economy has declined overtime but still remains substantial. Over the past 25 years, the structure of the Ugandan economy has changed significantly. Agriculture contributed the dominant share (53 percent) to the national economy in the early 1980s, but it currently contributes 36 percent of GDP. This is an expected correlate of growth and development, but the structural transition from a rural base has been relatively limited. Agriculture continues to employ almost 70 percent of the labor force, accounts for 90 percent of the exports and provides a large proportion of the raw materials for its industrial sector. Until 1999/2000, the share of agriculture in employment was 80 percent, virtually the same as in 48 1992/93. The decline by 2002/03 marks a structural shift, the drivers of which need to be understood better. 2.9 Agriculture continues to exert considerable influence on overall GDP Growth: Given the large proportion of people whose livelihood is dependent on agriculture, its performance matters greatly for general welfare, and still impacts on overall economic growth. Table 2-1 shows the growth rates of the three main sectors of the economy and their contributions to the overall growth. The share of agriculture in overall growth has declined as industry has grown, particularly through the 1990s. It still directly contributes 21 percent of the observed growth between 2001-2005 and also probably accounts for a significant proportion of the growth indirectly, through linkages with the service and industrial sectors. Evidence of the importance of agriculture for overall growth is the high correlation between total GDP and agricultural GDP growth rates of 77 percent between 1991/92 and 2004/05. And within agriculture, food crop growth continues to be important, with a correlation of 96 percent with agricultural GDP and 66 percent with overall GDP. Agricultural and trade services show a correlation of 38 percent. Table 2-1: GDP Growth and Sector Contributions 1987-2005 1983-1986 1987-1990 1991-2000 2001-2005 GDP at Factor Cost 6.0 1.3 5.7 6.4 5.4 Sector Growth Rates Agriculture 3.81 1.44 4.77 3.89 2.89 Industry 9.13 -0.23 8.43 10.24 7.48 Services 7.17 1.71 6.02 6.38 5.45 Sector shares in Aggregate GDP Growth Agriculture 0.29 0.60 0.44 0.28 0.21 Industry 0.29 -0.03 0.23 0.30 0.32 Services 0.43 0.43 0.34 0.43 0.48 2.10 Food crops continue to dominate agricultural value added. Over 2000-2005, on average over two-thirds of agricultural GDP was from food crops and only 9 percent from industrial crops. The economic dominance of the sub-sector combined with the concentration of the poor in agriculture and particularly in crop agriculture highlights the importance of growth in the crops and in particular food crops for poverty reduction. Figure 2-2: Composition of Agricultural GDP Composition of Agricultural GDP Industrial crops 11% Fishing 6% Forestry 5% Food crops Livestock 65% 13% (Note: Shares are averages from 2000-2005) 49 2.11 Over the past almost two decades (1987- Table 2-2: Agricultural Value-Added 2005), agriculture has performed well, growing Growth Rates, 1990-2004 at 3.8 percent. It has grown faster than population Uganda 3.8 (averaging about 3.3 percent), and has thus been a Tanzania 3.5 major contributor to the success of Uganda in its Kenya 1.2 poverty reduction efforts in the 1990s. Even Ethiopia 1.8 relative to other countries (in the region and South Asia 2.7 average for other regions worldwide), Uganda's Sub-Saharan Africa 3.4 long term agricultural growth trend has been East Asia & Pacific 3.0 impressive (Table 2-2). [But per capita growth is Latin America & Caribbean 2.2 far less impressive]. 2.12 Disaggregating the long-term trend into shorter periods shows that growth has been fairly robust since 1987. The sub-period growth rates shown in Table 2-1 indicate a significant and sustained improvement after 1986. There was a notable change from the low 1.4 percent annual growth before 1986 to 4.8 percent as peace and stability were established. The strong performance continued through the 1990s, averaging 3.9 percent per year. After 2000, growth was slower than the population growth rate of 3.4 percent. 2.13 The policy reforms initiated in the early 1990s and sustained since have been highly effective. This long and sustained period of growth has earned Uganda the distinction of being one of the most successful sub-Saharan African countries in reducing poverty (Figure 2-1). This growth process demonstrates the success of the policy framework adopted and maintained by Uganda. A conducive policy environment, led by exceptional progress in stabilization and market liberalization, has been a key factor in Uganda's success. Without any obvious change in the policy or economic environment, the slowdown after 2000 thus calls for deeper analysis. 2.14 Food crops are still a dominant contributor to growth. The sub-sectoral contributions to overall agricultural growth are given in Table 2-2. Food crops have traditionally been the dominant sub-sector. In the 1990s, driven by the rise in the world price of coffee (which peaked in 1997), the contribution of industrial crops increased significantly but food crops were still the main contributor. In the post-coffee boom period, i.e., over the past five years, food crops remain the main contributor, accounting for 58 percent of the sector growth rate, but the contribution of industrial crops has declined significantly, while livestock has become more important. 2.15 The decline contribution of industrial crops is a source of concern. This is especially disconcerting given reversal in coffee prices in recent years and the impressive growth in new industrial crops. A clear policy priority is to address the coffee wilt disease (CWD) problem (see text box on coffee). 50 Table 2-3: Sub-Sectoral Contributions to Agricultural Growth 1987-2005 1983-1986 1987-1990 1991-2000 2001-2005 Agriculture 3.8 1.4 4.8 3.9 2.9 Growth Contributions Industrial crops 0.53 0.05 0.27 0.77 0.26 Food crops 2.41 1.32 3.62 2.30 1.69 Livestock 0.50 -0.22 0.50 0.48 0.53 Forestry 0.19 0.07 0.19 0.19 0.21 Fishing 0.22 0.28 0.21 0.22 0.21 Sub -sectoral shares in Growth Industrial crops 0.14 0.04 0.06 0.20 0.09 Food crops 0.63 0.92 0.76 0.59 0.58 Livestock 0.13 -0.15 0.11 0.12 0.18 Forestry 0.05 0.05 0.04 0.05 0.07 Fishing 0.06 0.19 0.04 0.06 0.07 Box 2-1: Some Facts about Coffee and the Ugandan Economy · In the mid-1990s, liberalization of the coffee sector, combined with high world prices, and easier availability of higher yielding varieties, had a significant positive impact on poverty in Uganda; · Between 1997-2000, growth in crop agriculture more broadly had a large impact on poverty, contributing about 80 percent of the reduction in poverty incidence (Deininger and Okidi, 2003; Okidi, et. al. 2005). · The coffee price boom in the mid-1990s benefited the coffee growers and had broader positive economic impacts. · The subsequent decline in coffee prices was in part compensated by the rise in the share of farmgate in export prices (from 30 percent in the 1980s to over 70 percent in 2004), but production suffered due to coffee wilt disease (CWD), resulting in a decline in the volume of coffee exports. · Uganda has a comparative advantage and a niche in the world coffee market, but current supply is much below what it should be (Baffes, 2006). 2.16 Agriculture is particularly vulnerable to weather variability. While the longer term trends are impressive, a closer look at the annual rates of growth shows that the ride over the past two decades has not been particularly smooth. Agriculture closely tracks the growth rate of food crop production and they, on average have grown at the same rate (Figure 2-3). 51 Figure 2-3: Agricultural Growth Volatility and Poverty Estimates 14.00 70.00 12.00 60.00 10.00 50.00 8.00 6.00 40.00 4.00 30.00 2.00 20.00 0.00 10.00 -2.00 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 0032 2004 -4.00 0.00 Agriculture Food crops Period Average Poverty Headcount 2.17 Overall volatility in growth rates has declined but agriculture remains relatively more susceptible to shocks than other sectors. The degree of variability, measured as the coefficient of variation (CV) of sectoral growth rates has been compared for the period 1987-97 and 1998-2005 (Figure 2-4). In the early part of Uganda's recovery (1987-97), industrial performance was more volatile, perhaps reflecting growing pains. After 1998, however, all sectors were relatively more stable, except industrial crops, whose volatility increased. This reflects the fact that in addition to weather shocks, industrial crops are also subject to external market fluctuations, which directly affect export crops such as coffee. Food crops and overall agriculture remained relatively more unstable compared to other sectors. Figure 2-4: Change in Growth Volatility 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 ure rops gn yr esc c cult GDP Crops Livestock Forestry Fishi Indust Servi Agri Ind Food 1987-97 1998-2005 Total 2.18 The correlation between agricultural performance and poverty provides some insights to the estimated levels of poverty. There are some insights for the ongoing analysis of poverty trends in the comparison of the annual growth rates in agriculture and poverty levels (Figure 2-3). The three point estimates of poverty are based on household surveys for the years 52 the 1992, 1999 and 2002/03. Based on these surveys, national poverty is estimated to have fallen from about 60 percent in 1992 to 34 percent in 1999, and then risen again to 38 percent in 2002. 2.19 There is an inverse relationship between the poverty rate and agricultural growth. The survey year 1992/93 was a particularly bad year for production and corresponds to the highest measured poverty rate. The year 1999/00, which saw a large decline in the poverty rate, was the second in a row of three very good years of production (with almost identical growth rates of 6 percent per year for food crops). The year 2002/03 was also a positive but below average growth year (especially for food crops), and this corresponds to the small rise in the poverty rate in that survey year. 2.20 The critical point is that the overall trend in poverty is strongly downwards, suggesting that the basic economic strategy over the past decade or so has been essentially on the right track. Given the importance of food production and agricultural income to the welfare of Ugandan households, and the fluctuation in the performance of these sectors depending on exogenous shocks, the three point estimates need to be interpreted with caution. The fluctuations also reinforce that agricultural performance is a good surrogate measure of poverty. The annual fluctuations in agriculture thus imply that around the long term poverty trend, there appears to be a lot transient poverty. This is consistent with the "churning" effect identified in the Poverty Assessment (World Bank, 2005b). 2.21 There is a need to better understand the factors driving volatility in agricultural performance, and what they mean for household welfare, and then to identify policy options or actions that might appropriately address the underlying causes. Among these, the wide fluctuations in agricultural production suggest greater attention to devising policy options and actions to help reduce the vulnerability of households to the volatility in incomes. In considering alternatives, it is important to understand how volatility is changing over time. D. DRIVERS OF PAST AGRICULTURAL GROWTH 2.22 The recent slowdown in agricultural performance is understandably raising concerns at the policy level. To help formulate appropriate policy actions, it is critically important to understand the underlying factors and identify the most appropriate response. 2.23 The analysis here suggests it may be due to a decline in prices. As productivity rises, a decline in real prices is to be expected, and as noted earlier, this is desirable, as long as that productivity increases more than offset the fall in real prices. To understand how the aggregate value has evolved, consider the graphical presentation of the trends in indices of output, price, value of production and output per capita (using 2000 as the base year) for the major food and industrial crops. As the aggregate output is dominated by food crops, the comparison of output per capita provides a good approximation of food production per capita. 53 Figure 2-5: Agricultural Performance: Trends 120 110 100 90 80 70 60 50 40 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Output per capita Price Output Value 2.24 Output has grown steadily since 1996, but prices have fluctuated significantly. Aggregate real value of agricultural production, the normally used barometer of performance, has grown at about 5.8 percent a year since 1989/90. Value of production, however, closely follows the price trend, indicating a dominant influence of prices on aggregate performance. At the same time, output (a value weighted aggregation of all crop outputs) has grown at a more modest rate of 3.1 percent. There are two distinct periods that can be discerned from the output trend: that of stagnation prior to 1996 with growth in output at 0.1 percent per annum, and that of positive growth after 1996 when out grew at 4.7 percent. 2.25 Even the last three years, which have been a cause for concern for policy makers, output has maintained the upward trend, except for a slight dip in 2003. This dip and the slightly below trend in 2004, reflects the twin impacts of a prolonged drought over the past 2 years, and a likely negative reaction to the decline in producer incentives (because of the sharp decline in prices) in the preceding years. 2.26 Volatility in aggregate production value thus appears to be driven by price variability more so than production variability. For the period 1990-2004, the variability in the value of output, measured as the coefficient of variation (CV) around the trend, is estimated to have been 24 percent. This was driven mainly by volatility in prices, with a CV of 20 percent, and to a much lesser extent by output variability, with a CV of only 8 percent.60 The covariance between prices and output is also negative, as expected. Further almost 80 percent of the variability in output is attributable to yield variability, and much less to changes in total area cultivated. 2.27 The trends in prices are consistent with the trends in output. In the early1990s, per capita output fell, assuming a steady or growing level of demand (with population growth), and as a result prices trended sharply upwards. After about 1996, output per capita rose, and supply side 60For food crops, variability in value of production is 25%, which is again dominated by price variability of 20% and much lower production variability of 9.5%. 54 pressure resulted in a downward trend in prices (the trend between 1997 and 2004). Initially, between 1997 and 2000, there was a modest downward trend in prices, but because of output growth, value of production (and by association, incomes) continued to rise. 2.28 In 2001 and 2002, there was an unusually large and sustained decline in prices, which drove down aggregate value of production. It is critical to understand the reason for this decline in prices. From the trends in food and industrial crop yields and prices depicted in Figure 2-6, the yields during these years do not appear to be abnormally high and they are in line with the trend from 1997 to 2002. However, production indices from FAO data indicate that these two were good years in neighboring countries (particularly Kenya, Rwanda and Burundi ­ confirm). This could have created a regional surplus but with knock-on effects on Uganda (with essentially open borders with these countries). The most important impact was probably through reduced demand for Ugandan products in Kenya. Another important effect on Ugandan markets was the change in purchases by the World Food Program (WFP), which has a large local procurement program and dominates the Ugandan maize and beans markets. In 2002, WFP purchases were significantly below the rising trend (about 8,000 tons compared to over 30,000 in 2001). Figure 2-6: Yields and Prices of Food Crops and Industrial Crops in Uganda(1990-2004) Uganda Food Crop Output & Real Value Uganda Industrial Crop Output & Real Value Indices Indices 120 160 100 140 120 80 100 60 80 40 60 20 40 Output Value Baseline 2.29 Understanding the price dynamics in agricultural markets remains an important area for study. The events described above do not fully explain the large and sustained price fluctuation. Volatility in market prices can have serious negative consequences for farmers to undertake productive investments and adopt technology for productivity enhancement. It is important to understand the nature and causes of price volatility to be able to develop appropriate policy responses, and thus remains as a high priority area for further analytical work. Agricultural Growth Decomposition 2.30 To identify the drivers of past and recent growth in agriculture, it is important to understand its composition of the aggregate trends. The growth experienced between 1989/90 and 2003/04 is further decomposed by type of crop. The period is split into two ­from 1989/90 to 1996/97 and from 1997/98 to 2003/04. This provides about equal observations in the two periods and also coincides with the peak of coffee prices in 1997/98. The split could be done in 1999/2000 to coincide with the debate on poverty trends, but that would leave fewer observations in the second period. 2.31 Ugandan agriculture is highly diversified, but the aggregate composition of output has not changed over the years. From the aggregate structure of agriculture in Uganda in 55 Figure 2-7, one can infer that traditional staple foods (matoke, cassava and sweet potato) dominate with the share in aggregate value changing little over the years. The major changes that took place between 1989-97 and 1997-2004 are a reduction in the share of matoke from 31 percent to 23 percent, and a significant increase in the share of cassava from 9 percent to 25 percent. Almost all other crops have declined in share to some extent, yielding acreage to cassava. Figure 2-7: Structure of Major Crops by Value (% of total gross crop value) Maize Irish Potato Sorghum 5% Sweet Potato 4% 2% Rice 12% 2% Millet 5% Cassava 15% Vegetables 8% Beans & Peas 6% Oilseeds 7% Matoke 27% Industrial Crops 7% Note: shares are averages from 1990-2004 2.32 The food sector dominates overall agricultural performance. The results of the sectoral growth decomposition for each crop by area, yield, output, price and value are summarized in Tables 2-4 and 2-5, while the contributions of components of output and value of individual crops to total growth are given in Tables 2-6 and 2-7, respectively. The last two columns of Tables 2-4 and 2-5 show that not only do food crops dominate in terms of their share of area but also in terms of their share in the value of total production. 2.33 The first period (1990-97) is characterized by very large gains, with area expansion as the main driver of output growth. Both industrial crops and food crops experienced significant growth. Because of the small share of industrial crops in aggregate value added, overall sector growth was much closer to the food crops' growth rate. For industrial crops, the growth was due to increase in both production (output) and prices, with output growth contributing relatively more. For food crops, output stagnated until 1997/98, and growth in value was driven by price increases. Area expansion has been the main driver of output growth, with yields showing a significant declining trend. This result is again driven by food crops, with all except matoke experiencing declining or stagnant yields. The decline in land productivity outweighed area expansion for food crops. In contrast, industrial crops did remarkably well, with area expanding at the same rate as food crops, but yields increasing at a very high growth rate. It is noteworthy that the output growth of almost all industrial crops was faster than that of coffee. 56 2.34 Between 1998-2004, overall performance was very modest, but growth in food output has been a major source of growth. Prices fell for both food and industrial crops, but much more so for industrial crops. Food crop output did well growing at (3.8 percent), but there was a marked slow down in industrial crop output (2 percent compared to the earlier rate of 11 percent). It is notable that the growth in food output was a result of gains in a broad range of crops (with big gains recorded in cassava, maize, irish potato, rice, beans and peas and oilseeds). The performance of food crop yields showed a significant reversal (from -2.7 percent to 1.3 percent) but yield growth remains low. Cassava was a major contributor to the increase in food output, with disease resistant varieties spreading quickly and widely, reversing the earlier trend of negative growth in area and yields to positive growth in area and a significant improvement in yields. 2.35 A significant change in the recent years is the decline in area under industrial crops, with a significant decline in coffee acreage. Initially, this reflected a combined impact of the coffee wilt disease and reduced incentives as world coffee prices dropped. More recently, the coffee prices have picked up but production continues to decline. Yield growth for industrial crops made up for the lost acreage. 2.36 Area expansion continued to account for more than half the increase in output in the second period, but aggregate yields also contributed significantly to output growth. Tables 2-6 and 2-7 show the difference in the contributions of the different drivers of agricultural growth rate. The roles of output and price have similarly changed between the periods, with prices falling in the second period and most of the growth coming from output growth. Table 2-4: Principal Crop Growth Estimates from 1989/90-1996/97 value area Crop area yield output price value share share Food Crops 2.5% -2.7% -0.3% 5.7% 5.5% 0.93 0.90 Matoke 1.4% 1.3% 2.7% 8.6% 11.3% 0.28 0.26 Cassava -2.9% -4.0% -6.9% 16.8% 9.9% 0.15 0.06 Sweet Potato 3.6% -2.9% 0.7% -2.5% -1.8% 0.12 0.09 Irish Potato 8.3% -1.4% 6.9% -1.8% 5.1% 0.04 0.01 Maize 6.4% -1.7% 4.7% 6.9% 11.6% 0.05 0.10 Sorghum 2.6% -5.3% -2.6% 7.9% 5.3% 0.02 0.04 Rice 5.5% 0.0% 5.5% -1.1% 4.5% 0.02 0.01 Millet 0.7% -3.3% -2.5% -0.9% -3.4% 0.05 0.07 Vegetables 2.8% -0.5% 2.3% -1.1% 1.2% 0.08 0.01 Beans & Peas 3.4% -8.4% -5.0% 6.2% 1.2% 0.06 0.14 Oilseeds 4.0% -2.2% 1.8% 1.6% 3.4% 0.07 0.11 Industrial Crops 2.4% 8.6% 11.0% 7.6% 18.6% 0.07 0.10 Cotton -5.2% 19.3% 14.1% -10.0% 4.1% 0.00 0.03 Sugar 13.1% -1.2% 11.9% -6.5% 5.4% 0.00 0.02 Cocoa 3.2% 11.4% 14.6% -12.7% 1.9% 0.00 0.00 Coffee 3.5% 6.0% 9.4% 11.5% 21.0% 0.05 0.05 Tea -0.3% 15.1% 14.8% -0.4% 14.3% 0.00 0.00 Tobacco 4.7% 2.7% 7.4% -5.6% 1.8% 0.00 0.00 Total 2.5% -1.9% 0.5% 5.9% 6.4% 1.00 1.00 57 Table 2-5: Principal Crop Growth Estimates from 1997/98-2003/04 value area Crop area yield output price value share share Food Crops 2.5% 1.3% 3.8% -1.9% 1.9% 0.93 0.90 Matoke 1.3% 0.1% 1.3% -2.7% -1.3% 0.28 0.26 Cassava 2.0% 4.7% 6.6% -0.8% 5.8% 0.15 0.06 Sweet Potato 2.2% 0.8% 3.0% 3.6% 6.5% 0.12 0.09 Irish Potato 5.6% 0.8% 6.4% -7.4% -1.0% 0.04 0.01 Maize 3.5% 1.9% 5.4% -4.3% 1.1% 0.05 0.10 Sorghum 2.0% -0.9% 1.1% 1.9% 3.0% 0.02 0.04 Rice 5.9% 0.2% 6.1% -2.8% 3.2% 0.02 0.01 Millet 0.8% 0.8% 1.7% -1.3% 0.4% 0.05 0.07 Vegetables 1.9% -0.5% 1.4% -6.5% -5.1% 0.08 0.01 Beans & Peas 3.5% 1.5% 5.0% -1.6% 3.4% 0.06 0.14 Oilseeds 4.2% 1.6% 5.7% -1.5% 4.3% 0.07 0.11 Industrial Crops -3.8% 5.8% 2.0% -7.7% -5.7% 0.07 0.10 Cotton -2.3% 21.6% 19.3% 2.0% 21.3% 0.00 0.03 Sugar 0.9% -0.4% 0.5% 0.2% 0.7% 0.00 0.02 Cocoa 2.7% 5.5% 8.2% -1.4% 6.8% 0.00 0.00 Coffee -8.4% 6.5% -1.9% -10.5% -12.4% 0.05 0.05 Tea 0.2% 7.6% 7.8% 4.6% 12.4% 0.00 0.00 Tobacco 10.7% 5.8% 16.4% 0.2% 16.6% 0.00 0.00 Total 2.0% 1.7% 3.7% -2.3% 1.4% 1.00 1.00 Table 2-6: Output Growth Decomposition for Individual Crops 1989/90-1996/97 1997/98-2003/04 crop area yield output area yield output Matoke 34.0% 33.3% 67.3% 9.5% 0.4% 9.9% Cassava -39.3% -54.0% -93.4% 7.9% 18.9% 26.8% Sweet Potato 38.7% -30.9% 7.9% 7.0% 2.6% 9.6% Irish Potato 26.4% -4.5% 21.9% 5.4% 0.7% 6.1% Maize 25.5% -6.7% 18.8% 4.2% 2.3% 6.5% Sorghum 5.8% -11.5% -5.7% 1.3% -0.6% 0.7% Rice 8.4% 0.1% 8.5% 2.7% 0.1% 2.8% Millet 3.3% -14.6% -11.3% 1.1% 1.1% 2.3% Vegetables 20.1% -3.5% 16.5% 4.1% -1.1% 3.0% Beans & Peas 16.5% -41.0% -24.5% 5.1% 2.3% 7.4% Oilseeds 24.2% -13.6% 10.6% 7.6% 2.8% 10.4% Cotton 0.3% 1.0% 1.3% -0.1% 0.2% 0.1% Sugar -1.9% 7.2% 5.2% -0.3% 2.4% 2.2% Cocoa 0.5% 0.0% 0.4% 0.0% 0.0% 0.0% Coffee 15.1% 53.0% 68.1% 3.8% 7.8% 11.5% Tea 1.1% 1.8% 2.9% -0.8% 0.6% -0.2% Tobacco -0.1% 5.6% 5.5% 0.0% 0.8% 0.9% Total 178.5% -78.5% 100.0% 58.6% 41.4% 100.0% Note: Estimates calculated using production weights. 58 Table2-7: Value Growth Decomposition for Individual Crops 1989/90-1996/97 1997/98-2003/04 crop output price value output price value Matoke 12.9% 41.0% 53.9% 18.7% -37.6% -18.9% Cassava -18.0% 43.5% 25.5% 50.7% -6.1% 44.7% Sweet Potato 1.5% -5.1% -3.6% 18.1% 21.7% 39.8% Irish Potato 4.2% -1.1% 3.1% 11.6% -13.3% -1.7% Maize 3.6% 5.3% 8.9% 12.3% -9.8% 2.5% Sorghum -1.1% 3.3% 2.2% 1.4% 2.4% 3.8% Rice 1.6% -0.3% 1.3% 5.3% -2.5% 2.8% Millet -2.2% -0.8% -2.9% 4.3% -3.2% 1.0% Vegetables 3.2% -1.5% 1.7% 5.7% -26.4% -20.7% Beans & Peas -4.7% 5.9% 1.2% 14.0% -4.6% 9.4% Oilseeds 2.0% 1.9% 3.9% 19.7% -5.0% 14.7% Cotton 0.3% 0.2% 0.4% 0.1% -0.5% -0.4% Sugar 1.0% -0.7% 0.3% 4.1% 0.4% 4.5% Cocoa 0.1% 0.0% 0.0% 0.0% 0.0% 0.0% Coffee 13.1% -11.4% 1.7% 21.8% -3.8% 18.0% Tea 0.6% 0.7% 1.2% -0.3% -1.8% -2.1% Tobacco 1.1% 0.0% 1.0% 1.6% 1.0% 2.6% Total 19.2% 80.8% 100.0% 189.1% -89.1% 100.0% Note: Estimates calculated using value weights. 2.37 Summary: Aggregate trends and decomposition of growth highlights important reversals in key underlying trends in the performance of the agricultural sector. · Crop productivity as defined by output per unit of land (yields) reversed in the mid-1990s from a negative to a positive trend. · With area expansion slowing down marginally in the second period, the major change has come as a result of yield increases for virtually all food crops, while maintaining yield growth for industrial crops. · Despite the good performance in output, growth in gross value has been slowed down by price declines. · The aggregate decline in food crop prices is desirable, as their value has still been growing. · The substantial decline in export crop prices has been a drag on overall growth. · Most importantly, the trends in recent years suggest that the worrisome slowdown after 2000 has been due to a significant and unexplained negative price fluctuation, while output has continued to grow at a steady pace. · These trends endorse the current policy framework that has helped promote yield and productivity growth. · To address the slowdown, the appropriate policy response would be to target the factors underlying the prolonged price decline in 2001-2002. · There is a need to undertake this analytical work to avoid similar declines in future, as Uganda continues to grow. 59 Sources of Growth 2.38 Growth in agriculture would come from increased factor inputs (area), improved productivity of existing crops (higher yields) or shifts to higher valued crops. A good measure of the contribution of factor productivity as opposed to expansion of factor inputs is total factor productivity (TFP). However, a lack of good quality aggregate time series data prevent reliably estimating Total Factor Productivity in agriculture and how it has evolved overtime. Trends in area expansion and partial factor productivity are discussed in this section, and shifts among crops are discussed in the next section. Area Expansion 2.39 A major part of the past growth came from area expansion but this cannot continue for long. There are limits to environmentally sustainable area expansion, especially with a population growth rate as high as Uganda's. The decomposition of growth shows that area has expanded at over about 2.2 percent a year consistently since 1987. And while there was a significant improvement in yield trends after the mid-nineties, area expansion continues to account for over half of the current growth. Based on data from the 1995 and 1999 Uganda National Household Survey (UNHS) crop module data, average area cropped per farm (for the first season) has fallen from 1.4 hectares to less than 1 hectare, reflecting rising population pressure on land. Consistent with this trend, the number of parcels of land owned by households has declined on average from 1.84 to 1.54. 2.40 As the younger population enters the work force in the coming years - effectively doubling it - labor will have to leave agriculture. Aggregate data shows that arable land per agricultural worker is now under 0.54 hectares, while arable land per rural person is less than half or 0.26 hectares. This reflects the age structure of the Ugandan population, with over half the population under 15 years of age who will come into the work force over the next 10 years (approximately). Since 1987, agricultural work force has increased by about 45 percent, but availability of land per worker has fallen by about 25 percent. 60 Figure 2-8: Land Use 90000 Small-scale farmland 40 80000 34.8 35 70000 30 60000 Grassland 25 50000 Woodland 21.2 20 Open water 40000 Tropical High 16.7 15.3 15 30000 Forests - fully stocked 10 20000 Papyrus Reeds Tropical High Bush Large-scale Swamps 5.9 Farmland 10000 Forests - degraded 5 Plantations 2.5 Built up areas 1.2 2 0 0 0.1 0.3 0.2 2.41 Continued reliance on extensification of agriculture as a source of growth is likely to be environmentally disastrous. It could lead to enormous conflicts with diminishing grasslands and other areas for cattle grazing for the pastoralists. There is already limited scope for area expansion. About a third of the tropical high forests are already degraded, 17 percent of the land remains as woodland and 2 percent as wetlands (see Figure 2-8). If it can be assumed that the remaining woodland, grassland and land under bush (currently accounting for about 44 percent of Uganda's geographical area) could be brought under the plow, just the current population (without any additional growth) would require adding at least another 35 percent of the total area to smallholder farmland over the next 10-15 years, leaving at most 10 percent of the total area of Uganda as woodland, grassland or bush. Even these crude calculations and unrealistic assumptions show that area expansion is not a viable source of future growth. 2.42 Future growth will thus have to rely on a combination of more intensive agriculture and movement of labor out of agriculture to urban and rural non-farm activities. More intensive agriculture to raise land productivity is clear. And this is discussed further in the next section. At the same time, continued population expansion (and even the momentum of the current population structure) will put immense pressure on land as more and more seek to make a living out of smaller and smaller farms. This is simply not a viable scenario for sustainable and improving livelihoods. There is a clear need to address the issue of rapid population growth, but also to help people off the land and out of farming. Providing education is a key part of this strategy, and is already being done. The strategy also needs to address the issues concerning the environment for the creation of off-farm employment. Growth of the non-farm economy, and preferably following a pattern that is not centered on Kampala as the primary urban area, remains a high priority area. Changes in (Partial) Factor Productivity 2.43 Data on yields needs to be interpreted with caution, as they likely underestimate yield levels. Published yield data are based on household survey data (1995 and 1999 UNHS) and are likely estimated with error because of inaccurate estimates of area under each crop. The data do not allow identifying production from sole cropped plots as distinct from intercropped 61 plots. Further, area under individual crops on intercropped plots is not estimated. Without an accurate estimate of the area allocated to specific crops, yields cannot be estimated reliably. Thus it is likely that the published data underestimate crop yields. For example, maize yields are estimated at 1.0-1.2 tons (depending on the season) per hectare using the 1999/00 UNHS data. Using an independent household survey, which allows distinction between sole cropped plot output and intercropped plot output, maize yields are estimated at just over1.6 tons/ha for solecropped maize plots. At the same time, the same survey gives an average across all plots of 1.2 tons/Ha, which is similar to the published estimates from UNHS data. 2.44 Uganda needs to raise land and labor productivity. The two primary factors are land and labor. Raising land productivity is clearly high priority based on the discussion above. But the ultimate goal is to raise living standards, which is more appropriately measured as labor productivity. This section discusses the trends in land and labor productivity. Land productivity can be measured as yields and a commonly used measure is cereal yields. An alternative measure, which is broader and also relevant from a broader economic growth perspective is agricultural value added per hectare. Both these measures are examined for Uganda. 2.45 Cereal yields: Growth in recent years is much improved over previous years, and aggregate yields compare well with the rest of SSA, but there is substantial scope for improvement. Uganda is endowed with relatively favorable weather (a bimodal and relatively reliable rainfall pattern) and agro-ecological conditions, especially soils. Thus, despite having among the lowest rates of fertilizer use in the world (less than 10 percent of even the Sub-Saharan Africa average), crop yields are better than neighboring countries and Sub-Saharan Africa as a whole. However, they remain low in comparison with other regions of the world. The growth in cereal yields has also been low in absolute terms as well as relative to other countries. Yield growth has averaged 0.1 percent since 1990, but as noted in the decomposition discussion above, the performance in the second half of the period was significantly higher. Figure 2-9: Cereal Yields (Average 1999-2004) 5000 3 4500 2.5 4000 3500 2 3000 1.5 2500 2000 1 1500 0.5 1000 0 500 0 -0.5 Asia SSA EAP LAC Kenya Uganda S. Tanzania Ethiopia Cereal Yields (Kg/Ha) Growth rate 2.46 Changing cropping patterns make estimating yield growth rates difficult. Yield growth is a key indicator used to judge the performance of agriculture and the effectiveness of agricultural services (specifically agricultural research and advisory services). As noted earlier, farm sizes and the number of parcels owned per household are falling, reflecting rising population pressure on land. This decline in the availability of land has led to a change in the patterns of farming. The 1995 and 1999 UNHS data show a shift from sole cropped plots to more 62 intercropped plots (Figure 2-11). Also the crop mix is changing, with a marked reduction in sole food crops and cash crops, and an increase in the proportion of area under food-food based intercrops (Figure 2-10). Thus, with these non-insignificant changes taking place and the lack of an accurate estimate of area under each crop, not only is it impossible to get an accurate measure of technical efficiency based on yields at a point in time, but also it is difficult to estimate yield growth rates. Thus an assessment of performance based on crop yields is not appropriate. Figure 2-10: Composition of Crops Grown per Plot 70.00% 60.00% 50.00% 40.00% 1995 30.00% 1999 20.00% 10.00% 0.00% Food Cash otherFood- Food- Cash- Food- Cash- Food cash cash other Other 2.47 Nevertheless, available evidence suggests that average yields obtained by Ugandan farmers are well below their attainable potential (as obtained from research trials). There remains ample room for growth through yield increases even for traditional crops as demonstrated by the comparison of the estimated (pure stand) yields from REPEAT survey and yields obtained in research trials(see Table 2-8). Realistically obtainable yields will be below those obtained in research trials, but there is potential to increase output between 100-300 percent (or more) depending on specific crops. This potential validates the government's focus and commitment to agricultural advisory services and research, two key pillars of the PMA. There is a need to strongly maintain this focus and redouble the efforts to expand the NAADS program that has suffered from resources constraints, as well as to fully support the establishment of the new NARS system that was recently launched after the passing of an innovative and progressive NARS bill by the Parliament. 63 Figure 2-11: Sole Crop Verus Intercrop 350.0% Other Industrial 300.0% 250.0% 200.0% Potatoes Matooke 150.0% Solecrop Maize 100.0% Intercrop Pulses Other cereals Coffee Other 50.0% Cassava 0.0% Oilseeds -50.0% -100.0% Table 2-8: Comparison of Existing Yields to Research Trials Current yields on farmers' Experimentation Station yields Crop fields (REPEAT survey Kgs/Ha (verify) 2002/03) Kgs/Ha Maize 1609 5000-8000 Beans 641 2000-4000 Groundnuts 1222 2700-3500 Banana 1200 4500 Coffee 1019 3500 Figure 2-12: Land and Labor Productivity 240 1800 y = 1406.3e 0.0091x 1600 220 1400 200 1200 180 1000 y = 180.71e 0.0179x 160 800 y = 110.62e 0.0355x 600 140 400 120 200 100 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Agriculture value added per worker (constant 2000 US$) Ag Value Added per Hectare of Agricultural land (constant 2000 US$) Cereal yield (kg per hectare) 64 Land and labor productivity: Aggregate land productivity (value added per hectare) has risen almost twice as fast as labor productivity (at 3.6 percent and 1.8 percent, respectively) and significantly faster than growth in estimated crop yields. Factor productivities more broadly defined as valued added per unit of factor employed are a better measure of the overall performance of the sector. Trends in aggregate crop yields, land productivity defined as agricultural value added per hectare, and labor productivity defined as agricultural valued added per agricultural worker are summarized in 2.48 Figure 2-12. The divergence between yield growth and the broader measure of land productivity indicates a structural shift to higher value crops. 2.49 Uganda has done better in land productivity than labour productivity. However, the growth rates in land and labor productivity for Uganda are comparable to a number of other developing countries (Figure 2-13). Figure 2-13: Uganda's Land Productivity as Compared to other Countries 4.5 3.5 ry/t 2.5 ercen P 1.5 0.5 -0.5 so ana l da ia il r ga sh sia az Fa mbia India etnam Br na Gh Sene Ugan Za done Boliv Vi Salvado Burki nglade In Ba El AgGDP/Worker AgGDP/Hectare 2.50 The growth in labor productivity has been constrained by the population growth rate, which is among the highest in the world. As in the case of yields, however, there remains much scope for improvement. Figure 2-14 below shows the levels of agricultural GDP per worker in Uganda compared with neighboring countries and averages for other regions for the world (averaged over a1999-2003, years for which data are available). The level of labor productivity in Uganda is higher than some of its neighbors, but it is lower than most developing countries. The growth rate, however, has been higher than most, with the only exception being the Latin America and the Caribbean region countries (taken as an average). 2.51 So far labor productivity has been higher than the estimated agricultural value added per capita. This is because agricultural labor force has grown at a slower pace than total population. Some of this is because of outmigration but also because at present over half of Uganda's population is below working age, and hence is not yet in the labor force. Looking to the future, sustaining the trend in labor productivity growth will require a combination of slower population growth and much more rapid out-migration. Uganda needs to create jobs outside of agriculture. 65 Figure 2-14: Agricultural Value Added per Worker 3000 2762 2.5 2.3 2 2500 1.8 1.5 1.3 1.4 2000 1 0.5 1500 -0.01 0.01 0 1000 -1.4 -0.5 404 321 -1 500 225 275 150 122 -1.5 0 a a -2 ia As SSA Carib Kenya h Uganda Ethiopi Tanzani Sout Level in 2000 US$ Growth rate Lat.Am.& Structural transformation: commercialization of Ugandan agriculture 2.52 Agriculture is increasingly shifting from subsistence to commercial orientation suggesting that PMA is working. The divergence between growth rates of agricultural value added per hectare and crop output per hectare (yields even though inaccurately measured) indicates that a structural transformation is taking place in Ugandan agriculture. This is significant as it reflects the growing success of the government strategy of promoting commercial agriculture ­ the explicit objective of its multi-sectoral policy framework of the Plan for Modernization Agriculture (PMA). There are three dimensions of this transformation ­ one is the increased monetization of agriculture, with a larger share of production now being marketed than before; the second is a shift to the production of higher valued crops; and the third is the diversification of export of agricultural commodities. Monetization: 2.53 The overall degree of commercialization, or extent of market participation has been rising steadily as observed from the trend in the monetization of agriculture. The monetization of traditional food crops has also risen overtime. At present, about 58 percent of agricultural output and 46 percent of food is marketed (monetized). Livestock and fisheries have traditionally been largely in the monetary economy (two-thirds for the former and almost 90 percent for the latter). 66 Figure 2-15: Agricultural and Food Crop Monetization 60% Crop 55% 50% GDP/Food 45% Ag. Production total 40% of 35% Share 30% 1990 1991 1992 1993 1994 1995 1996 1997 1998 199 2000 2001 2002 2003 2004 2005 9 Monetary Ag GDP Food Crop Monetization Shift to Higher Value crops 2.54 "Value" lies in the eye of the beholder. From the farmers' point of view, it is the profitability of crops (adjusted for risk) that drives their decisions on production and productivity. Thus to assess a move towards "higher value" crops, it is important to note that the concept of higher value crops does not necessarily rule out traditional food crops. Based on a combination of changes in productivity and prices some traditional "low value crops" such as cassava and sweet potatoes have been important contributors to sector growth after 1997/98, as discussed in the growth decomposition analysis earlier. Table 2-9: Average Production 2.55 And the contribution of oilseeds was almost as much as that Value per Crop Ha of coffee, while vegetables declined in importance. Table 2-9 Vegetable 2884 indeed shows that average (from 2000-2004) value of production Cassava 2791 per hectare for some traditional crops can be quite high. The negative growth in vegetable gross value of production probably Tobacco 2764 reflects lack of appropriate marketing infrastructure for perishable Irish Potato 1934 commodities such as vegetables. Tea 1143 Rice 957 Sweet 2.56 At the aggregate level, land is increasingly being Potato 854 allocated to higher value crops as indicated by the growth in Matoke 698 area under different crops over the past 6 years (since 1998). Coffee 613 There is a negative correlation between area allocation and value Sorghum 550 added per hectare across crops (Figure 2-16). Of course, land Millet 504 allocation across different crops is influenced by many factors that Oilseeds 373 need to be understood using micro level analysis, as is doen in the Maize 334 next section. Beans & Peas 259 Sugar 136 Cocoa 127 Cotton 65 67 Figure 2-16: Shift in Area Allocated to Higher Value Crops Shift in Area Allocated to Higher Value Crops 15.00% 3500 3000 10.00% 2500 5.00% 2000 0.00% 1500 a toa a blesa Te Pot Rice toesa & 1000 Matoke Coffee Millet Peas Sugar Cocoa Cotton -5.00% Cassav Tobacco Sweet Pot Sorghum Oilseeds Beans Irish Maize Veget 500 -10.00% 0 Growth rate 98-04 Value per ha 2.57 The policy framework adopted by Uganda to create an enabling environment for producer incentives to drive agricultural growth has been appropriate and should be strengthened. The rising trends in monetization are encouraging as they indicate that the mindset of smallholders is changing from subsistence to a more commercial orientation. And they appear to be responding to incentives. Going forward, the sources of growth will not be the same depending on market conditions and productivity changes. Thus, given consumer preferences, it is likely that some of the past sources of growth will no longer be viable (e.g., cassava as a food staple as incomes rise because of its negative income elasticity of consumption). Yet, with technological change, additional demands for the same products will likely come on line, such as cassava processing for starch to be used in beer production. Thus it is difficult to predict what the next "winner" will be. Nor should it be the objective of policy to do so. The primary objective of policy needs to remain on strengthening the enabling environment for producers to respond to incentives to drive and sustain future growth in the sector. Export Diversification 2.58 The increasing diversification of exports also provides evidence of the changing structure of production and farmers' response to changing incentives. There has been a significant decline in the importance of coffee over time, and a rise in the exports of other commodities (Figure 2-17). After a sharp decline from 1996 to 2000 (due to decline in coffee earnings as prices declined), the value of agricultural exports have risen at an annual rate of almost 10 percent. Non-coffee exports show a steady increase from 1995 at an annual rate of about 10 percent a year. 68 Figure 2-17: Change in Exports 60% 50% 40% 30% 1995-1998 2001-2004 20% 10% 0% o k Tea Fish Coffee Cotton Tobacc Flowers Oilseeds Livestoc Horticulture Cereals/beans 2.59 Formal trade statistics, however, present only a part of the export story. A significant volume of informal cross-border trade is not captured in the formal statistics. A recent survey by UBOS shows that over a period of 6 months, about $27 million worth of goods were informally exported, while about $8 million worth were imported. Of these, 71 percent of the exports and 36 percent of the imports were agricultural goods. Among exports, the main commodities were maize, beans, fish, bananas, and other cereals. Based on these estimates, annualized agricultural exports would be about $38 million, in addition to the $98 million recorded formal trade of non- industrial agricultural crops (i.e., leaving out crops such as coffee, tea, cotton, tobacco, flowers, cocoa beans and vanilla). This is important as it reflects a significantly larger share of agricultural food commodities which are often not considered as having much potential for trade. 2.60 A high priority is to identify the constraints to increased regional trade and pursue regional agreements for increasing agricultural trade. Regional trade provides significant potential for helping commercialize even farmers who currently grow "subsistence" crops. As the only surplus producer in the region, surrounded by countries facing a structural deficit in staple crops (maize, beans and possibly bananas and other crops), Uganda is in a unique position to exploit its comparative advantage in growing these crops. It is also in a unique situation to exploit the high transportation cost barrier facing most countries in the region, and to use its advantage to compete against imports from outside the region. The biggest constraint of being a land-locked country is the high transport costs faced for imports and exports. However, proximity to countries that provide a market for its commodities gives Uganda the competitive advantage over imports from outside the region. E. UNDERSTANDING GROWTH CONSTRAINTS Supply Side Analysis: Stimulating Agricultural Production 2.61 Growth in agricultural sector depends on the choices made by households, and specifically on the factors driving those choices. In this section, two sets of analyses are carried out to enhance the understanding of agricultural production decisions. The first set of analyses focuses on farm household decisions making concerning crop choice and marketing, and examines: (i) the determinants of overall diversification on-farm; (ii) the share of land allocated 69 specifically to commercial crops; and (iii) the share of output marketed. An understanding of these issues is critical for formulation of appropriate policy measures to promote commercialization. All three dimensions are important as they reveal different aspects and levels of the household decision making process, although in a dynamic sense they are inter-related. The scope for this analysis, however, is limited to looking at each aspect individually, leaving the study of the dynamic aspects of decision making to a later stage. All three analyses are investigated using reduced form regressions to identify key variables that may influence household decisions. The second set of analyses is focused on farm productivity, and examines: (i) the basic determinants of agricultural productivity; and (ii) the factors influencing farmers' decisions regarding input use and technology adoption. Policy recommendations based on the results are offered. Crop choice and market participation 2.62 Recent household survey data show a high degree of degree of diversification.61 The estimate of the Simpson index, a measure of diversification, is quite high at 0.70.62 The most widely grown crops are beans (grown by 76 percent of farmers), maize (75 percent), matoke (67 percent), sweet potato (52 percent), and cassava (48 percent). Coffee is grown by 31 percent of the sample and other industrial crops by 8.4 percent. In terms of area, cereals are grown on average on 22 percent of the farm area, bananas on 25 percent, legumes on 19 percent and roots on 18 percent. On average, 10 percent of farms grow cash crops. Vegetables and fruits are grown by few households among those surveyed (15 percent for both) and command a small share of the farm area (3 percent). 2.63 The structure of production at the micro-level shows a high concentration in staple crops, as may be expected in a poor rural economy. Clearly there are still signs of households' focus on food security. However, as noted above, not all of these are necessarily "low value" crops. While subsistence requirements continue to play an important role in household decisions to allocate land among crops, a number of factors including risk aversion, market opportunities and profitability probably explain why a majority of households grow at least one and usually more basic staples (e.g., cereals, matoke, cassava, and roots). This could also reflect lack of adequate integration and outreach of markets, which raises food security concerns and risks facing households, forcing them to diversify into multiple staple crops. 2.64 At the same time, however, a very large percentage of farmers participate in the market. Overall, about 77 percent of the households report selling at least a part of their produce, with the percentage particularly high for vegetables (96 percent), cash crops (91 percent) and fruits (85 percent). The high percentage of even staples (close to 80 percent) suggests that households may be reacting to a strong demand for staples, perhaps from local and urban centers, that provides the incentive for farmers to concentrate on these crops. 2.65 The proportion of output marketed varies by type of crop. As may be expected, the proportion of cash crop output marketed is highest, with almost 100 percent of coffee and over 90 percent of vanilla marketed. Very high proportions of some other crops, including vegetables (such as onions and tomatoes) and rice, are also marketed (ranging from 70 ­ 90 percent). The 61 The data used are from the 2003 REPEAT survey, which covered three regions of Uganda (Eastern, Central and Western). 62The Simpson Index is defined as 1-(Aij/Aij)2, where Aij is the area devoted to crop i on plot j. The index takes the value of 0 for an undiversified crop portfolio (only one crop grown), and rises asymptotically to 1 as the portfolio is increasingly diversified. 70 proportions of staples such as matoke and cassava are much lower at about 15 percent; the proportion of maize marketed is about 26 percent. 2.66 The patterns of farm size, productivity and diversification are complex. The table below gives the averages for some key variables, notably farm area, diversification index, and proportion of area under different crops, by six farmer groups. These groups are defined according to farmer productivity, using percentile ranges for the value of production per hectare of farm size. Table 2-10: Averages by Value of Production per Acre Groupings Area Div. % % % % % Percentiles (acres) Index cash cereals legumes roots banana 0-10 4.41 0.59 0.08 0.36 0.21 0.21 0.09 11-25 6.02 0.70 0.09 0.29 0.21 0.16 0.18 26-50 5.54 0.73 0.11 0.21 0.21 0.17 0.25 51-75 5.04 0.73 0.11 0.20 0.18 0.18 0.29 76-90 4.29 0.72 0.09 0.16 0.18 0.18 0.34 91-100 3.14 0.67 0.09 0.17 0.15 0.22 0.32 2.67 The relationship between farm size and productivity is non-linear, but generally shows an inverse relationship. The smaller farms are the more productive, with productivity falling as farm size grows. However, at the bottom end of the productivity scale farm size falls, indicating that not all small are necessarily more productive. 2.68 The least productive farmers are also least diversified. The relationship between productivity and diversification is also non-linear. Diversification increases to about the median productivity farmer group. The most productive farms again start to specialize, or show lower levels of diversification. These patterns probably reflect changes in the average share of farm allocated to each crop in the different productivity groups. Area under cash crops initially rises with productivity and then falls. The proportion of area under cereals is the highest for the least productive group and is consistently inversely related with rising productivity. Area under legumes shows a similar trend, but increasing area under bananas is associated with rising productivity. The proportion of area under root crops initially falls with rising productivity and then rises. Determinants of on-farm Diversification 2.69 Larger farms, those in more densely populated areas and those living in more remote areas are likely to be more diversified. Farmers in endemic risks areas are likely be less diversified, likely because of a focus on specific less risky crops. Annex 2.1 gives the regression results of the determinants of the degree on-farm diversification. It should be noted that the results are from a reduced form specification; a more structural model would be helpful in understanding better the path of influence of different factors, but that is a much more sophisticated and time consuming modeling exercise that needs to be pursued as part of the ongoing analytical work. The dependent variable is the farm level of crop diversification as measured by the area-based Simpson index. The regression equation includes a number control variables, most notably the district fixed effects that capture a wide range of economic and geographical non-observable factors, including price variability. In addition, within each district the agro-climatic factors may vary and hence agro-climate zone indicators are included. As 71 defined here, crop diversification reflects both the physical dispersion of the different parcels of land owned by the household and the choice of crops. Thus, the positive impact of the number of parcels cultivated helps control for this source of diversification. · Larger farms are more diversified as are farms located in more densely populated areas. The latter may reflect that higher population pressure leads household to diversify their production patterns, in part as a risk aversion measure but also to meet family consumption requirements. · The distance from district towns is positively correlated, indicating that households in more remote areas are more likely to diversify. · Those who live farther from agricultural markets are more likely to diversify. · Along with the fact that those who sell on farm are likely to be more diversified, this indicates that more remote households, and more specifically those whose access to markets is limited, are more likely to diversify. · None of the demographic variables are significant determinants of diversification; the only variable weakly associated with less diversification is the indicator for male-headed households. · Nor are any of the connectivity variables significant (e.g., ownership of radio, mobile phones or bicycles). · The impact of exogenous shocks (i.e., factors that damage production) shows significant effects. The damages are reported by households for their crops in the survey year. Since the diversification decision is made at the start of the growing period (i.e., prior to the incidence of weather or pest or other damage is incurred on individual crops), this set of variables is used to capture potential endemic risks in each location. · With that interpretation, it appears that areas that suffer from droughts, diseases, and pests are likely to be less diversified, probably because they concentrate on fewer but more drought resistant crops. Determinants of share of land allocated to commercial crops 2.70 Crop choice is determined by market access, endowments, and prevailing risks. Area allocation to individual crops provides some explanation of the results from the generically defined level of diversification in the previous section. Annex 2.2 gives the results of models explaining the proportion of area under the five main crop groups ­ cash crops, cereals, legumes, roots and bananas. These results are from fixed effects regressions to identify factors influencing the share of farm area under different categories of crops to provide additional insight to the degree of commercialization. Prices are an important determinant of crop allocation decisions; however, for lack of a complete set of prices, these effects are controlled for through district specific effects. 2.71 Proportionally greater area is devoted to cash crops and legumes as farm size increases and population density decreases. Cereals and bananas are allocated proportionally less (coefficient less than 1), while area allocated to root crops falls rapidly. These show that as farm size decreases, farmers tend to retreat into subsistence. High population density also places a greater premium on cereals and legumes, and to a lesser extent on cash crops, but reduces the area to root crops. The large influence of land pressure on cereals reflects food security concerns. The share of banana (matoke) appears to be uninfluenced by population pressure. 72 2.72 High market access promotes the production of cash crops and banana, while reducing the dependence on root crops. Both cash crops and banana have high commercial value and hence this confirms the importance of market access for commercialization. Holding market access constant, distance to markets is positively correlated with the production of cash crops, which probably reflects the degree of specialization in cash crops that is likely to occur in areas farther from populated areas. This is consistent with the negative impact of distance to markets for cereals and legumes, both relatively low value and bulky crops. The positive impact on root crops reflects subsistence focus away from markets, with the same holding for bananas, although bananas are also easily transportable and marketable. The indicator for selling on farm, as opposed to in markets, is also consistent with the distance to market, and has similar patterns across crops. This likely reflects the fact that cash crops, being commercial, also have a larger interest among traders and hence a greater propensity to be marketed at the farm gate as opposed to in the market center. At the same time, the low value-high bulk cereals are probably less attractive to traders and hence are more often sold at the market place than at the farm gate. 2.73 Farmers' age and household size matter. Among the demographic variables, farmer's age (a proxy for farming experience) is an important determinant of allocation to cash crops and bananas as opposed to the other crops. Household size, a proxy for food security concerns, shows up as important only for root crops (largely cassava). 2.74 Exogenous shocks such as weather and pests/diseases also have predictable impacts, reflecting important risk aversion behavior. Droughts reduce allocation to cash crops, root crops and bananas, but increase the focus on cereals. This likely reflects subsistence requirements take precedence in the presence of drought risk. At the same time, prevalence of diseases affects cereals and legumes negatively but increases the reliance on cash crops. Pest damage reduces allocation to all crops with revealed preference towards bananas. Determinants of the share of output marketed 2.75 Distance to district towns is an important determinant, with households closer to towns marketing less, while likely relying more on non-farm income for their cash requirements. This puts the focus on transport and market infrastructure to reduce marketing costs for households in more remote areas to benefit more from commercialization. 2.76 Households that have higher production per unit of land market a higher proportion of their product. The value of production per hectare, or plot level productivity, was included to capture factors that may affect outcomes beyond the factors explicitly included in the regression. Farm level productivity also has its own influence on a household's decision to market its output, as a higher level of productivity is likely to lead to a larger quantity marketed. The results show that this is a statistically significant factor. 2.77 A large number of farmers, other than those producing cash crops, are participating in markets. Overall, the proportion of output that is marketed is also a good indicator of the degree of commercialization or market participation. Annex 3 gives the regression results describing the market participation analysis. The estimation uses plot level data, as farmers were asked how much of the output from each plot was sold. Since the quantity sold was zero for a significant number of plots (about 38 percent of the 9,137 observations available for estimation), a Tobit regression was estimated. The regression controlled for district fixed effects (not reported) in addition to the agro-ecological zones and crop groups indicated in the results table. 73 2.78 Root and fruit crops are marketed less than cash crops, vegetables, and cereals, and those living farther from a district town market more of their output. Households farther away from the district town (the main urban economic center in the district) tend to market a greater proportion of their output. This likely reflects the fact that households closer to the district town rely more on non-farm income for their cash needs and probably produce for own consumption needs. This is an important result since the regression controls for the different types of crops grown. Thus the fact that commercial crops are grown more in areas farther from town centers (because of specialization as noted in the previous section), and these are likely to be marketed more than other crops, is controlled for in this regression. A similar effect holds for distance to markets. 2.79 More remote households receive a lower price for their products, and hence may need to sell more of their output to generate sufficient income to meet non-food consumption needs. That the households with high market access get a better price for their products, and those selling on farm get a lower price, is confirmed using preliminary regressions of prices on these variables (using the REPEAT survey data ­ results not reported). These results were also found in Porto et al. (2006) from 1999 UNHS data. 2.80 Current estimates indicate that for cereals and beans, farm gate prices are 17-20 percent lower than at the market place, indicating a high transport cost element. At the same time, high market access raises prices received by farmers by about 20-40 percent (estimated to be significant for some but not all crops). Further analysis is needed to better understand the factors driving household marketing decision, using better data and statistical methods. 2.81 Food security and risk aversion are important concerns. Households headed by older people and larger households tend to market less of their produce, both reflecting food security concerns and risk aversion. As may be expected, larger farms participate more in markets. Households that have borrowed money also participate more, reflecting the need to generate cash to repay their loans. Finally, all damage variables have negative coefficients, with five of the six variables statistically significant. This is also intuitive, as lower outputs are less likely to be marketed as households try to meet their own requirements. But also, as before, if weather or pest/disease damages are endemic, this could also reflect a retreat to self-subsistence. 2.82 Summary: The above findings on farm level diversification, allocation of area to different crops, and the degree of market participation (proportion of output sold) indicate some common issues that are important. · Market Access and Infrastructure: They highlight the importance of access to markets, indicating the need for improved infrastructure, to help in the commercialization process. · Population Pressure: The second key point relates to population pressure, as it forces households to either retreat into subsistence production and use of outputs, with their focus more on assuring household food security. · Risk and volatility: A consistent theme emerging in influencing household choices is weather-, pest-, and disease-related risks that affect not only the degree of diversification and crop choice, but also the degree to which farm output is marketed. Importantly, damage from excess water has a larger quantitative impact than droughts, indicating need for better water and natural resources management. 74 2.83 These effects clearly swamp the effects of household characteristics such as education (an important determinant of other sources of income), physical assets (other than land), and access to information. The main implications emerging are the need to increase investments in public goods targeting pest and disease control, while also focusing on natural resource management issues and household risk management. Determinants of agricultural productivity 2.84 As noted earlier, there remains considerable potential to raise farm productivity through increased efficiency of production and yields. This section is focused on understanding factors influencing farm productivity, especially important in light of the above results positively linking higher yields to the degree of market participation by households. Proper use of inputs, and hence agricultural advisory services, and natural resource management practices are important for higher productivity. Weather impacts, high population density and declining farm sizes appear to be main constraining factors. Using plot level data on crop production from the REPEAT household survey, factors contributing to higher production are investigated. The value of plot production is explained, and the regression is estimated across all plots. The results from the regression holding LC1 level fixed effects constant are given in Annex 4. Dummy variables for the type of crop grown on each plot are included to control for crop-specific effects. Agro-ecological indicators are also included to hold these factors constant. 2.85 The regression results show that the value of production is lower on intercropped plots (i.e., plots with multiple crops as opposed to a single crop grown). Thus the declining farm sizes and increased intercropping evident from the macro data are constraining growth in overall production. As expected, plot size has a positive impact. For key productivity enhancing variable inputs (i.e., improved seeds and fertilizers), dummy variables are used to capture the average effect of their use. The marginal effect using quantities provides qualitatively similar results. Using quantities would reflect the combined effect of the usefulness of the input as well as the efficiency with which they are used. The latter is planned to be investigated at a later time. 2.86 Improved seeds and natural fertilizers have a large and quantitatively important impact on production, but chemical fertilizers appear not to be used properly. As currently used, natural fertilizer use on average is associated with about 40 percent higher production, all else held constant. Similarly, improved seeds increase production on average by about 21 percent. Chemical fertilizers have a statistically insignificant estimate, but a large effect quantitatively. This suggests that currently the chemical fertilizers are not being used properly, resulting in a highly variable impact. Irrigation does have an impact, but it should also be noted that at the moment a very small percentage of area is irrigated. Investment in farm capital is also very low, with hand implements as the main form of capital for most farmers. Thus, the insignificance effect is not a surprise. 2.87 Mulching and zero-tillage, as well as slash and burn practices have significant and positive impacts on farm productivity. This suggests that there remains room for improving soil fertility through better management practices, but it also indicates that farmers who are not practicing these methods, and who are also not using fertilizers, are likely depleting soil fertility quite rapidly. Overall, 63 percent of the farmers are using at least one NRM practice, with 25, 15, 2 and 43 percent of the plots using mulching, slash and burn, zero tillage and compost, respectively. Zero tillage is the most beneficial of these, but this is practiced on only 2 percent of the plots. Compost appears to have the least benefit but is used on 43 percent of the plots. This remains an area for further research and investigation, as it is possible that the proper composting practices are not being used. 75 2.88 The variables that might affect farming efficiency, namely market access, distance to plots and population density, have statistically significant effects. The distance to plot, in terms of time from homestead, has a negative impact (as it likely reduces the incentive to give distant plots more attention); high market access areas have a substantial positive impact on the value of production; and high population density has a negative impact on value of production. These reflect the indirect influence through crop choices (discussed earlier), but also affect economic efficiency as they influence farmer incentives. They could also reflect other unobservable effects, such as general and economic services associated with easier access to markets (economic dynamism). The Boserup hypothesis would dictate that population pressure, through smaller farm sizes, increases the incentive to become more efficient, but it is also possible that population pressure is leading to soil depletion and reduced productivity. Which of these hypotheses reflects the true underlying factors needs further investigation, but also requires panel data that is currently unavailable. 2.89 Weather shocks and education have significant impacts on production. Education has a strong and large positive impact. Dummy variables indicate that drought and other weather impacts have a significant impact on production, confirming the importance of these variables in the household decision making discussed earlier. 2.90 More effective and demand-driven technical advice (as opposed to simple delivery of extension services) has a strong positive impact. Credit is not an important factor. Two key policy variables that usually attract a lot of attention are credit and agricultural technology extension or dissemination services. The data show that credit does not have any impact on production. Including access to extension services does not have any impact (not reported), but when a variable indicating whether households applied the extension advice (the true test of effective extension services), the impact is significantly positive. 2.91 The main finding here is that delivery of extension services is not sufficient, and that the relevance of the advice provided, as revealed through actual adoption by farmers, is more important. This result is also consistent with emerging evidence on the impact of National Agricultural Advisory Services projects (NAADS), which shows that NAADS is delivering more effective demand driven and farmer controlled advisory services and that the program is beginning to have significant impacts on farmer income (via moves to higher value production enterprises and rising productivity). 2.92 Fertilizers are perhaps not being used efficiently. Further and better analysis is needed to guide policy actions, especially on the determinants and economics of fertilizer use. The above analysis is based on cross-sectional data. While it is useful and can provide a number of insights, some key policy issues can only be isolated using a panel data set that allows controlling for a number of unobservable variables. One such key variable is the impact (average or marginal) of chemical fertilizer use. As noted, the estimated coefficient is quantitatively important but statistically insignificant (both the average and marginal impacts. Whether this means fertilizer is being used inefficiently requires further study with more robust statistical methods. One key area for further investigation is whether fertilizers are profitable ­ under current practices as well as under more efficient use of fertilizer.63 63The REPEAT survey provides an opportunity to do this with a recent resurvey of the households used for this analysis. It is also possible to link back to an earlier IFPRI survey, which was more limited in its coverage, but does provide a subset of households used in this survey. These investigations are planned for a follow up study to this analysis. 76 Determinants of input use 2.93 The use of modern inputs is growing in Uganda, but its intensity remains among the lowest in the world. The use of fertilizer is growing at a rapid 18 percent rate of growth since 1982 (and at much faster if only the late 1990s are considered). Yet fertilizer intensity is less that 10 percent of even the average intensity for Africa. Other inputs such as seeds, capital and other chemicals remain very low. Figure 2-18: Fertilizer Use (100 gms per Ha) 20 15 10 y = 0.3529e 0.1808x 5 0 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2.94 Whether or not the cost of fertilizer is a binding constraint remains a subject of debate, with insufficient evidence to support the case. As the table below shows, only 4 percent of the farmers in 1999/2000 reported using inorganic fertilizer. Improved seeds were used by a much larger but still low proportion of 18 percent, and pesticides used by about 9 percent. A lot of policy attention is focused on the "cost' of fertilizer and attempts to subsidize fertilizer to promote its use. But it is not evident that cost is the binding constraint in the use of fertilizers. The table below also shows that more farmers are incurring as high a cost (on a per acre basis) using other inputs that cost as much or more than fertilizers. 2.95 It is not at all clear under these circumstances that credit for fertilizer will increase its use. These results, combined with the production function findings above which show that the impact of fertilizer used is statistically insignificant, suggest that there may be other reasons for farmers not using fertilizer. At the same time, improved seeds and natural fertilizers were found to have a significant impact and the table above shows that more farmers are indeed using these inputs. 77 Table 2-11: Household Use of Improved Agricultural Inputs in 1999/00 Household Adoption Number of Rate (adopting hh Households USh per Acre as a % of total hh) Single input adoption 666,652 9,112 23% Two input adoption 221,337 26,969 8% Three input adoption 68,460 50,785 2% Full input adoption 15,422 51,138 1% Fertilizers 261,346 15,137 9% Manure 169,573 14,936 6% Inorganic 75,117 13,293 3% Manure+Inorganic 16,655 25,303 1% Pesticides 249,183 12,277 9% Improved Seeds 504,952 7,018 18% Land Preparation (tractors, oxen, etc.) 360,910 16,211 13% Source: 1999/00 UNHS 2.96 Relevant technical advice has a significant positive impact on input use. Ongoing pilot programs (e.g., Agricultural Productivity Enhancement Project (APEP) program on maize production in Eastern Uganda) also find that chemical fertilizers can be profitable if used properly. Additional evidence of the importance of appropriate uses of inputs is the finding by International Food Policy Research Institute (IFPRI) from a recent survey conducted as part of the NAADS mid-term evaluation (Nkonya et al. 2005). The survey found that among households who were aware of modern inputs or technologies and had received advice on their use, adoption rates were significantly higher than households who were also aware of these inputs or technologies but had not received advice on their proper usage. This is a significant finding in that appropriate advisory services, not only on enterprise selection but also on input use, can have a significant productivity impact. 2.97 Infrastructure has a strong and consistent impact on adoption of modern technology, as does effective delivery of advisory services; but credit consistently fails to show up as a determining factor. Given the critical importance of modern inputs to productivity growth, this section looks at the determinants of adoption of these inputs. Two data sets are used for this analysis. One is the National Service delivery Survey (NSDS) data, which are more limited in information on input use, but nevertheless have certain types of information that allow tests of some important hypotheses. The other is the REPEAT household survey, which also has certain unique types of information for testing additional hypotheses. Analysis using NSDS 2002/03 Survey 2.98 The NSDS data do not have any information on fertilizer use, but do have information on the use of improved seeds and on other inputs (chemicals, pesticides, etc.). Using Probit equations, the results on factors affecting the adoption of improve seeds and all inputs are given in Annexes 5 and 6, respectively. 2.99 Identifying the impact of credit is difficult. The NSDS contains data on a series of questions asking households if they needed credit; if so, whether they applied for credit; and if so, whether they got the full amount applied for, a partial amount or nothing. Normally, including credit in a regression establishes correlation but determining causality is very difficult. That is, it is not clear if higher credit leads to increased input use or whether households actually seek out 78 credit once they have decided to use inputs (e.g., once they are empowered with knowledge on their effective use). This endogeneity is difficult to overcome as the appropriate instruments for the credit received are often not available. Another problem is that credit access is often determined by a number of factors that also relate to input use (farm size for collateral purposes, household wealth, education, etc.). Thus what is captured as a "credit" effect may actually represent the impact of some other unobservable characteristic. This is confirmed by an auxiliary relationship (the results of which are not reported here) between expressed need for credit as well as of the credit amount applied and a number of household characteristics and infrastructure/access variables. The results show a strong correlation between dependent variables and better roads, farm size, participation in livestock activities, and the level of per capita total consumption expenditures. Since these also affect adoption decisions, it is difficult to disentangle the pure credit effect. 2.100 It is not evident that credit constraints are the most binding on household decisions to adopt inputs. The data at hand allow testing for the hypothesis that credit rationed households end up using less inputs. Using full information available in the survey, the impact of the various interrelationships is minimized by explicitly including key variables to tease out possible causal effects. With regards to credit, in addition to whether the household applied for credit, another variable is included to capture those households who expressed a need for credit but did not apply. Thus, while the application of credit is likely to be associated with increased adoption, those who expressed a need for credit but did not apply are expected to be negatively affected, i.e., they should have a significant and negative coefficient. Additionally, two variables are included to indicate those households who applied but got only partial or no credit. The hypothesis here is that these households, being rationed, should have a negative coefficient for adoption (relative to those households who applied and got the full amount applied for). The results confirm a positive and significant correlation between the variable indicating that a household applied for credit the adoption of improved seeds. However, none of the other three credit variables are significant, raising doubts about the causal effect from credit availability and adoption of improved seeds. On the contrary, in the regression for all inputs combined (Annex 2-6), households who did not apply for credit also have a positive and significant coefficient, which is inconsistent with the credit constraint hypothesis. 2.101 Infrastructure and agricultural advisory services have consistent and strong impacts on adoption. The quality of roads matters, as progressively poorer quality roads (murram, feeder and community roads, respectively) have progressively negative impacts on adoption in both equations. Distance to roads is also a disincentive for seed adoption but not for all inputs taken together. Extension services, including the presence of NAADS, over and above extension services, have a positive and significant impact, confirming the importance of appropriate advice on input use. 2.102 Crop indicators, farm size, number of adults, gender and age affect input use. A number of crop indicators are significant, indicating that not all inputs are applicable or adopted for all crops (reflecting in part the availability or applicability of such inputs for certain crops). Farm size has a positive impact in both equations, but the number of adults matters only for seed adoption, while males are more likely to adopt inputs more broadly defined. Age has a negative impact in both regressions, also statistically significant only for all inputs taken together. 2.103 The NSDS results for input use confirm results for productivity and diversification; infrastructure and advisory services matter, but it is not clear that credit does. Of course, access to credit will help, but the driving force seems to be profitability through access to markets and relevant information. Given the hard choices to be made on public expenditures, the 79 rationale for a large scale investment in providing credit services to farmers is not likely to have a large payoff, while it will probably have significant opportunity costs in terms of other key investments that could be made with scarce resources. Analysis using REPEAT Survey 2.104 The data from the REPEAT survey provide an alternative source of information on the determinants of adoption. This survey collected information on fertilizers (both natural and chemical), as well as on the use of improved seeds. Again probit equations are used (the results are in Annexes 2-7, 2-8 and 2-9) to model the probability of adoption of each input relative to a wide set of explanatory factors. The results are intuitive given the cost and familiarity of different types of households with chemical and natural fertilizers. But the main conclusions on key policy issues remain the same as those derived from the NSDS data. These findings confirm the same for the case of fertilizer use ­ which information was missing in the NSDS data. 2.105 The results for improved seed adoption are similar to the NSDS results but provide additional insights: intercropping, large families and weather and pest shocks reduce the likelihood of adoption. Mixed plots (intercropped) are less likely to adopt improved seeds; hence, as farm sizes get smaller and more farmers resort to intercropping, the likelihood of using modern technology to raise productivity is diminished. Availability of credit does not have any impact on adoption of improved seeds. Plot size, total wealth, farm capital and irrigation use are significant determinants of improved seed adoption. Large households, and families with older household heads, are less likely to adopt. Areas suffering from weather shocks and pests are less likely to adopt. Interestingly, households who practice mulching or zerotillage are less likely to adopt modern technology, while those practicing slash and burn are more likely to adopt improved seeds. 2.106 The further the distance to district towns, the less the likelihood of adoption, but higher population density is positively associated with adoption of improved varieties. Radio and mobile phone access is negatively associated with improved seed use, which is a surprising result. The access to markets variable is insignificant, perhaps because the distance to district town is capturing this effect. 2.107 The factors explaining the adoption of chemical fertilizers are very similar. Credit again has no impact, while market access (as opposed to distance to market towns) is now significant. NAADS has a positive impact (though significant only at 10 percent level). Natural resource management practices have the opposite effect of those on the adoption of improved seeds, but have intuitive signs. Mulching does not show any influence on chemical fertilizer adoption; slash and burn practice reduces the demand for fertilizer; and zerotillage increases the demand for fertilizers. Other farm and demographic variables have a similar effect as on the adoption of improved seeds. 2.108 The determinants of adoption of natural fertilizers also are the same as for chemical fertilizers. Again, credit is not significant. The main differences are that for natural fertilizers, market access variables (either the distance to district towns or a market access indicator) are not significant. Wealthier households demand less natural fertilizer, and farm size and household size both reduce natural fertilizer demand. Mixed crops, on the other hand, use more natural than chemical fertilizer. Education increased the likelihood of adoption of natural fertilizers but was insignificant for chemical fertilizers. Mulching reduces the likelihood of adoption, while slash and burn increases it. Farmers in high populated areas are more likely to adopt natural fertilizer, as are those who own more cattle. 80 Demand Side Analysis: Incentives for Productivity Growth 2.109 To stimulate production and productivity growth, the incentives facing the farmers need to be attractive. The correlation of trends in area allocation among crops with net returns per hectare demonstrates this. Prices and risks are key elements of the market environment that are important for farmers. Increased supplies with inefficient or segmented markets or markets that fail to absorb the increased supplies can create severely adverse incentives. Thus, to fully understand the potential for growth in the agricultural sector, the demand side of the equation must also be taken into account. 2.110 High price variability identified earlier suggests there may be demand side problems, specifically a lack of effective demand. Price fluctuations could be suggestive of markets not clearing adequately, either because the markets are functioning imperfectly or because there isn't enough purchasing power. The two sources of demand for agricultural products, including food products, are domestic demand and exports. By far the main source of demand is domestic, which is in turn related to domestic incomes. Exports of industrial crop exports are primarily to the world market, while food crop exports (mainly maize but also other foods) are primarily to regional markets. Domestic consumption patterns 2.111 An important determinant of the price for domestic food commodities is the structure of domestic consumption (i.e., household per capita expenditures). Patterns of consumption and their impacts on domestic prices helps assess the prospects for various commodities and also helps guide public policy decisions that may be targeted as supply side or demand side constraints. This section describes the structure of consumption and provides estimates of Engle elasticities for main food crops (to explain how consumption levels change as expenditures or incomes rise. Figure 2-19: Household Per Capita Expenditures Social Own Produce, Activities, 7% 26% Non-Food, Staples, 9% 33% Fresh Food, Non-Fresh 14% Food, 11% 2.112 Consumption patterns: A large share of household expenditures goes on food. The 2003 REPEAT household survey data show that households on average spend about 58 percent of their total per capita expenditures on food. Of the total, 26 percent is accounted for consumption 81 of own produced products (Yamano et. al. 2004). Of the cash expenditures, another 34 percent is spent on purchase of food. UNHS data for 2003 show that on average about 51 percent of household expenditure is spent on food and only 4 percent on non-consumption. The share of rural expenditures is higher (55 percent) than in urban areas (47 percent). Of the food expenditure, about 29 percent is from own production and 67 percent is purchased food (about 4 percent is "free food"). Purchased food is 85 percent of the food bill for urban households, and 54 percent for rural households, who consume a larger proportion of own produce. Ownfood is a relatively larger part of the food consumption expenditure for the lowest quartile (based on per capital total expenditures), at 43 percent. This declines with rises in income (36 percent, 25 percent and 11 percent for successively higher income quartiles). But surprisingly, the share of food in total expenditure of all four quartiles is about the same (51 percent, 54 percent, 52 percent and 51 percent from the bottom to the top quartile, respectively). 2.113 A very large proportion of the income at the margin is spent on food reflecting a continuing need to raise food productivity to lower real prices for broad welfare impacts. The total expenditure elasticity of demand for food, controlling for household size, is estimated to be very high at 0.74. (The estimated elasticity of food consumption is 0.76 using the REPEAT survey data.) For non-food consumption and for non-consumption expenditures the elasticity is about the same at 1.17. In urban areas, households spend 70 percent of their marginal shilling on food; in rural areas, this rises to 81 percent. Per capita expenditures are still very low: in 2002/03 monthly per capita expenditures are estimated at USh 23,474 in rural areas and USh 70,167 in urban areas. This indicates the low income levels, with high marginal propensity to spend on food. Demand for agricultural products 2.114 Meat and fish have the highest share of food expenditures, followed by matoke, cassava, and pulses. Expenditures on other commodities are lower, ranging from 1 percent to 7 percent, as given in the table below. The results from the regression analysis are also summarized in the table below. Engle curves were estimated by regressing the share of each commodity against the log of total food expenditures on log of household size (see results in Table 2-12). Prices are not available but the results obtained are consistent with those for fixed effects models, with district fixed effects used to control for price variation across districts. Because of a large number of zero quantities reported (e.g., different staples are consumed in different regions based on tradition and agro-ecological suitability), Tobit regressions were used. Foods were classified into 14 groups, some as individual commodities, others as categories representative of a group of commodities. The sign of the coefficient indicates the commodity expenditure elasticity as greater than one/less than one when the estimated coefficient sign is positive/negative. 2.115 Matoke is a preferred staple, whereas cassava is an inferior staple. Maize consumption also falls as total food expenditures (and hence incomes) rise. Consumption of other cereals (rice, breads, etc.) increases as expenditures rise. The results for livestock products, oils, fruits, beverages and other foods are as expected ­ with elasticities greater than one indicating rising shares of these commodities as incomes rise. Roots, pulses and sugar/salt/spices appear to be inelastic in demand, indicating that at lower levels of income, they are substituted for the higher value animal proteins and other starches. What is not usually observed is the fall in the share of vegetables. This suggests that perhaps households at all levels are producing and consuming vegetables but as incomes rise, they prefer to switch to other foods. This is also consistent with a diversified agriculture with a continuing subsistence orientation. Some of these consumption patterns may also be consistent with isolated markets (lack of adequate transportation facility or high transfer costs) that make local areas diversify into a number of 82 commodities and promote local consumption. The fact that weather and soil conditions are reasonably good across the country allows farmers to grow multiple commodities, either autarkic reasons or for economic reasons (e.g., high transaction/transport costs provide incentives for production for own consumption). Table 2-12: Results of Engle Curve Estimation Commodity Share in Food Log of food expenditures Log of household size Expenditure Matoke 0.10 0.064*** -0.014*** Maize 0.06 -0.005* 0.034*** Cassava 0.09 -0.092*** 0.102*** Other cereals 0.06 0.027*** 0.005* Roots (potatoes) 0.07 -0.026*** 0.057*** Meat and fish 0.14 0.077*** -0.023*** Milk and eggs 0.05 0.093*** -0.026*** Oils and fats 0.03 0.013*** -0.003*** Fruits 0.02 0.030*** -0.004** Vegetables 0.07 -0.035*** 0.018*** Pulses 0.12 -0.058*** 0.058*** Sugar, salt and spices 0.07 -0.003*** 0.002** Beverages 0.06 0.057*** -0.050*** Tobacco 0.01 -0.007 -0.011** Other foods 0.06 00.280*** -0.34*** ***, **, *: significance at 99%, 95% and 90%, respectively. 2.116 Summary: food expenditures dominate household consumption expenditures, reiterating the need for the following. · Continued focus on increasing productivity and production of basic food commodities to reduce real prices for welfare gains. · The elastic demand for certain commodities identifies potential areas where additional production could be absorbed domestically (assuming marketing, transaction and transport costs are not prohibitive). These include matoke, livestock products, fruits and other cereals. · The importance of matoke highlights the need to focus on Banana Bacterial Wilt and other diseases that have reduced production in recent years. · The elasticity of other cereals indicates that the recent focus on rice (with growing urban demand) is justified. · Livestock products have the highest elasticity and offer the best market prospects. But the livestock sector remains underdeveloped and livestock marketing remains inefficient. The livestock sector could be significant potential source of growth for agriculture. · For other crops with inelastic demand, cost-effective and competitive supply to regional and global markets is the best option. · Maize has a ready regional market and a substantial volume is already being traded, both formally and informally. Prospects for maize marketing could be significantly improved through trade negotiations and agreement with Kenya for maize trade. · The prospects for maize as animal feed also remains severely underexploited but its prospects are tied to the development of the poultry and animal feed industry. 83 · The market potential for other staples, such as cassava, is pessimistic but it has performed well over the last few years indicating that to some extent it remains an important food security crop. · There are also prospects, currently under review by private investors, for cassava processing as an industrial starch. its continued relevance remains to be tested (as for processing for starch, and other products). · Similarly, the regional demand for vegetable exports needs to be determined. The performance of markets and market efficiency 2.117 This section looks at how well agricultural markets perform in Uganda, and hence how well they transmit market signals. Lowering real food prices for consumers is a key development objective. One of the primary contributions of agriculture towards poverty reduction is the potential to lower real food prices via increases in production. But for this productivity based growth to be sustainable, it is important that farmers have adequate incentives to pursue productivity enhancing practices. 2.118 The efficiency with which markets operate determines farmer incentives. The level of retail prices and farm gate prices, the variability in producer prices and high input costs are critical elements of farmers' incentive structures, and they are, in turn, a function of how well markets operate. Productivity increases are a result of a combination of higher yields through improved technology and shifts in farm and crop enterprises to higher valued activities. Yield increases can be brought about through improved technical efficiency of production (adoption of improved crops and practices) and economic efficiency (through more efficient use of inputs in response to market signals). Adoption of technology is critically dependent on profitability of the input use, and this is a function of market incentives, primarily the result of how well markets function. 2.119 Lowering high marketing margins serves the dual purpose of lowering retail prices and raising farm gate prices. Transport costs are important determinants of marketing margins, and it is estimated that these costs are a large proportion of the total marketing costs in Uganda. How well markets clear, in terms of geographical or inter-temporal arbitrage, determines the level of price variability, an important factor in the decision making process of risk-averse poor farmers. 2.120 At the macro level, food prices relative to non-food prices show a long term decline. This is encouraging as it suggests rising food production is putting downward pressure on food prices. The figure below shows that after rising in the early part of the 1990s, food price inflation leveled off around 1998, and has started to rise in the past two years. Non-food prices have maintained a steady upward trend. The relative prices show a slight declining trend in food prices, with a notable deviation from the trend due to the fall in food prices during 2001-2002. The sharp decline in food prices between 2001 and 2002 is a reflection of market imperfections, the exact reasons for which are not fully known and need to be investigated further. As noted earlier, in part this was due to the shift in the demand for bulk food by WFP and good agricultural production in neighboring countries. 2.121 The detail with which the performance of agricultural markets in Uganda can be studied is limited by the availability of a long enough time series for relevant prices. From the available data, this analysis first looks at the ratio of producer prices (national average) to export prices for selected commodities; the trends in whole price differentials between markets; 84 and the incentives for arbitrage. It finally tests for market integration and assesses how well the wholesale markets are functioning. Figure 2-20: Trends in Price Indices 140 3 120 2.5 100 2 80 1.5 60 1 40 20 0.5 0 0 Jan-91 Jan-92 Jan-93 Jan-94 Jan-95 Jan-96 Jan-97 Jan-98 Jan-99 Jan-00 Jan-01 Jan-02 Jan-03 Jan-04 cpi_nfood cpi_food food/non-food Expon. (food/non-food) Export margins 2.122 For the main export crops, both formal and informal, the share of the export price received by farmers is a useful way to look at the likely performance of markets. The trends in the ratio of nominal producer prices (national average) to export prices are given in Annex 2-7. Among basic staples, the share of bananas is low but more or less constant since 1990; the share of maize is relatively high, but also fairly constant. The different levels could reflect transport costs, since bananas are more prevalent in the western part of the country, while maize production is concentrated in the eastern region, closer to Kenya, the main (albeit informal) export market for maize. 2.123 The share of producer prices for beans and coffee show a remarkable improvement over time, going from about 50-60 percent to almost 90 percent for beans and 80 percent for coffee. The improvements in coffee marketing, with the liberalization of the sector since the early 1990s, have been a major factor for coffee. The beans trend probably reflects the proximity to Mbale and the main markets bordering Kenya. Thus, with increasing demand in the east for beans and improved market performance, the share of the export price reaching farmers has increased over time. 2.124 Cocoa, cotton, and tobacco show a declining share of producer price in export price, only tea has a flat trend. The share of cocoa has fallen sharply from a high of 75 percent to a low of 15 percent. Cotton, tobacco and tea have a substantial proportion of processing costs and hence the share reaching the farmers is lower than other crops. But the declining shares for cotton and tobacco cannot be easily explained. These could reflect increased processing costs but this needs to be investigated further. 85 2.125 The trends across commodities do not reveal a clear cut pattern, nor allow speculation of what might be the driving factor. For example, how much and what the role transport costs is versus improvements in market functioning and efficiency cannot be established from this mix of trends.64 Even among the relatively low value, high bulk commodities like bananas, maize and beans, the trends are very different. Similarly, the different trends among the cash crops make it difficult to draw any conclusions about potential transport cost constraints or the impact of fuel prices. Different commodities are likely to be affected by sub-sector specific issues, suggesting that the performance of each commodity needs to be looked at separately and in depth. Spatial margins 2.126 Beyond the clearing of markets across space, it is important that markets function efficiently. This will reduce price volatility facing farmers, and lower margins will benefit both farmers and consumers. To assess how market performance has changed over time, the trends in marketing margins for four major commodities ­ beans, maize, groundnuts and matoke ­ are examined (see Annex 2-8). The data were collected from several markets and on selected commodities from around the country from 1991-2005 (sources ­ foodnet, etc.). In the early 1990s, the data were collected on a monthly basis, and from the mid-1990s onwards, on a weekly basis. However, there are a number of gaps in the data. Therefore, monthly averages are used from January 1995 to December 1998, and from January 2001 to December 2005. Wholesale marketing margins are calculated as the ratio of Kampala wholesale price to other markets' wholesale prices. For presentational purposes, Annex 2-8a presents the trends from 2001-2005, using the average for four markets in each of the three directions from Kampala ­ Southwest (including markets in Mbarara, Rakai, Masaka and Kabale); North (Gulu, Arua, Lira and Masindi); and East (Soroti, Mbale, Iganga and Tororo). The trends for the longer time series, 1995 to 2005, are given in Annex 2-8b (data for matoke are not available for 1995-1998). 2.127 There has been a general improvement in the marketing margins especially over the longer term, indicating improvements in market functioning. The shorter time series also shows a general improvement, with some exceptions. For beans, the markets to the North and Southwest show an improvement, with margins reducing to quite low by 2005. The only divergent trend is in the East (ratio is less than one and falling). This could reflect an increasing share of the output going to the Kenyan border from Eastern markets, raising the price in the East relative to Kampala, while Kampala is increasingly better integrated with markets to the North and West. Maize prices show an improvement in the margins between East and Southwest, but for the North, the trend shows an improvement till mid-2004, after which the price differential starts to rise again. The Kampala price is higher than in the North, and the rise in price differential indicates the difficulty producers in the North have to market their maize in Kampala. Markets for groundnuts are improving all around (the margins to the North were low to begin with, indicating well performing markets). Matoke price margins are the most volatile and also show the biggest movements. Matoke markets have improved dramatically in the Southwest. Markets to the North are improving as well, but widening in the East. 2.128 One general conclusion from this analysis is that wholesale markets appear to be functioning well and improving (for the most part). Decreasing marketing margins could be the result of lower transport costs, or improved marketing efficiency and hence lower transaction 64One also needs to account for the increased efficiency in the transport sector and the increased costs due to higher fuel prices. 86 costs, which would indicate increased competitiveness. As we shall see in chapter 6, there is a lot of scope to reduce transport costs. 2.129 A second conclusion is that marketing margins appear to have been reduced to quite low levels by the end of 2005 for the North and Southwest, and are higher in the East. Taking the average price differential for the period (2001-2005) as a whole, the figure below shows that with the exception of matoke, other price differentials are not very high, and fairly stable across markets, some of which are widely dispersed. Figure 2-21: Average Monthly Wholesale Price Differential to Kampala (Jan 2001 ­ Dec. 2005) 1.60 1.40 1.20 Bean Gnut 1.00 Maize 0.80 Matoke Millet 0.60 0.40 2.130 These results are somewhat at odds with the general perception that because of high transport costs in Uganda the marketing margins would also be quite high. It should be noted that the markets included in this analysis are major wholesale markets in regional/district towns. They are likely well connected, although roads to some towns may not be tarmac or as good as one might like. Given this, it appears as if the price differentials that transport costs are not prohibitive. But it should also be noted that there are peculiarities in the transport market that influence transport costs and these vary by town and time of year. Thus for equidistant towns, depending on whether they are on a major trucking route (e.g., to Rwanda or DRC), the backhaul freight may not be much, which reduces transport charges (in one direction, usually coming from the North or SW to Kampala) and probably explains the low marketing margins to these areas. At the same time, transport from the East to Kampala is usually very costly as most trucks carry full loads to Kampala (from Kenya) (Kiddedde 2006). 2.131 Marketing margins are also quite stable, with the exception of matoke and maize. The figure above shows the average monthly price differential over a five year period. It may be possible that prices and price differentials are highly volatile and are perhaps being "averaged" out. However, the coefficient of variation of price differentials over the same period for each market, shown below, does not indicate very high volatility for most crops. The prices for matoke and maize are the most volatile, reflecting their particular market peculiarities. For other crops, the variability is within reasonable bounds considering it is taking account of both the inter-seasonal and inter-year price variability. 87 Figure 2-22: CV of Monthly Wholesale Price Differential to Kampala (Jan. 2001 ­ Dec. 2005) 1.60 1.40 1.20 Bean Gnut 1.00 Maize 0.80 Matoke Millet 0.60 0.40 2.132 Despite the rising price of petrol, marketing margins have fallen suggesting that marketing (and perhaps transport sector efficiency) has dominated in compensating for higher fuel costs. This is investigated in the figure below plots the average of the wholesale price differential for beans between Kampala to three markets (Mbarara, Mbale and Jinja) plotted against the real price of petrol (Kampala). Figure 2-23: Average Wholesale Marketing Margins for Beans & Real Petroleum Pr 1.50 1,650 1.40 1,600 Bean margin 1.30 1,550 trend (left scale) (Kampala/Mbale-Jinja-Mbarara) 1.20 1,500 (UShs/Ltr) 1.10 1,450 1.00 1,400 0.90 1,350 Petroleum price trend (right 0.80 1,300 scale) 0.70 1,250 -02y Jan-01 Jul-01 Mar-01 Sep-01 Jul-02 Jul-03 Jul-05 May-01 Nov-01 Jan-02 Mar-02 Sep-02 Ma Nov-02 Jan-03 Mar-03 Sep-03 May-03 Nov-03 Jan-04 Mar-04 Jul-04 May-04 Sep-04 Nov-04 Jan-05 Mar-05 May-05 88 2.133 The above analysis suggests a 15-20 percent marketing margin between major market centers and Kampala, which is mostly due to transport costs. This compares well with the estimated marketing costs for maize in a transactions study conducted in 2002 (NRI/IITA). In that analysis, the marketing costs between secondary markets and tertiary markets (Mbale, Kampala, Kenya) are estimated to be between USh 10-30 per kg depending on the secondary and tertiary markets). The study associates all these costs to transport costs at that level of the marketing chain. 2.134 A large part of the transport cost is incurred before reaching the secondary markets, i.e., between farm gate and primary markets, and between primary and secondary markets. The above analysis between secondary and final consumer markets does not capture the major share of overall transport costs. And while even at the higher level the costs could be brought down, it is important to look at the relative prices between farm gate to primary markets and primary markets to secondary markets. Unfortunately there are no comparable time series data to conduct such an analysis. But the transaction costs study provides some insights to this. The study estimates that transport costs from the farm gate to the primary market are about USh 10 per kg for maize, which is 15-20 percent of the farm gate price. The total marketing costs, including trader margins of between 4-9 percent, range from 30-43 percent of the farm gate price. Of this, 50-75 percent is the share of transport costs alone. Between primary markets and secondary markets, the marketing costs are USh 20-25 per kg, of which 25-40 percent is transport costs. Trader margins were again estimated to be reasonable at USh 3.5-5 per kg or under 5 percent of the final secondary market price. 2.135 Getting from the farm gate to the primary market, and then from primary to secondary markets, costs about the same as getting from secondary to tertiary markets, which are at a much greater distance than the first two links in the supply chain. Overall, maize producers get about 45-50 percent of the final tertiary market price (not the retail price as retail margins have not been included). 2.136 Thus, the bigger challenge is to tackle the rural transport problem, which adds much more in proportional terms to the post-farm gate costs than the transport between urban centers/markets. Clearly a reduction of transport and transaction costs at all levels of marketing chain will help improve farmer incentives and market efficiency, but as a priority, targeting rural access to markets or reducing the costs of rural transport should be targeted. How much leeway there is or what impact the reduction in different segments of the transport market will have needs a more thorough study of the components of transport costs and potential for cost reduction (fuel prices, vs transport vehicles costs, vs vehicle operating costs). Chapter 6 takes this issue up. 2.137 Another aspect of marketing is the retail-wholesale price differentials. In general these are larger and more volatile than the wholesale price differentials across market centers. The figure below shows the price spreads for beans, comparing the retail to wholesale prices in Kampala and then the Kampala wholesale price to other markets' wholesale price (averaged across markets). The retail-wholesale margins are generally higher in Kampala than in other market centers. The trends show that the retail spreads increased until about 1998; after 2001, the margins have had a flat trend but have been fluctuating. At the same time, wholesale margins declined after 2001. Retail margins are also more volatile, indicating a fair amount of risk in retail business. 89 Figure 2-24: Price Margins for Beans 2.25 2.00 1.75 1.50 1.25 1.00 0.75 0.50 kamp_r/kamp_w kamp_w/oth_w Inter-temporal price variability and arbitrage 2.138 The third major dimension of market efficiency is the degree to which there is inter- temporal arbitrage ­ to reduce the gap between harvest season low prices and pre-harvest season high prices. If the storage markets are functioning properly, or are improving, then it is expected that the intra-year price rises would become progressively less with time. Using monthly prices (averaged across markets) for different commodities, the price rise between the lowest price month and the highest price month, in percentage terms, is plotted for different years below. Figure 2-25: Returns to Intra-Year Storage: Average across Markets (Percent Difference between Min and Max Prices) 300% 250% 200% Beans Gnuts 150% Maize 100% Matoke Millet 50% 0% 1991 1992 1993 1994 1995 1996 1997 1998 2000 2001 2002 2003 2004 2005 90 2.139 The intra-year price rises have become less pronounced over time (as shown by the declining trends) but the absolute magnitudes of inter-temporal margins are quite high. Maize prices still rise as much as 50 percent between seasonal lows and highs. Considering that these price rises are often for as short a period as three months, the annualized returns to storage are quite high. A large part of the country also enjoys two growing seasons, implying even greater returns to storage activity. 2.140 Despite the decline in the inter-temporal margins, the continued high returns call for a further investigation into the working of the commodity storage markets. One major constraint noted by market observers is that access to finance for storage of commodities is very difficult to obtain. Thus despite significant storage capacity in Uganda, there is little storage and stocking of food grains, which severely limits large formal trading activity. Besides trade finance, finance by small and medium size traders to invest in storage is severely limited ­ either because the commercial banks are not willing to lend to small traders or the cost of borrowing is prohibitively high. This remains an area of high priority for analytical work, and is taken up in chapter 7. Degree of Market integration 2.141 The above analysis shows that market performance has improved and markets appear to be functioning well. Statistical analysis verifies this. The graphical and trend analysis shows that marketing margins on average are improving except in a few specific circumstances. Also the degree to which price margins decline appears to be market and commodity specific. But it does not provide answers to whether the markets are integrated or efficient in terms of the transmission of price signals. This requires a statistical assessment of the degree of market integration. To answer these questions, a statistical model belonging to the family of error correction models (ECM) is estimated for 10 markets (relative to Kampala) for 4 major crops. The same time series are used as were used for the trend analysis of marketing margins above. Most prices were test to be stationary, but some were not. The ECM model is robust to potential non-stationarity. It also provides several insights that most other models do not: it tests for market integration (i.e., whether or not prices in different markets co-move in the long run); it allows an estimate of the degree of instantaneous price transmission between two markets; and it allows testing for the number of periods (months used here) it will take for the prices to equalize (to a pre-set level of differential).65 Alternatively, the last measure of the speed of transmission can be judged by how much of the price change in one market is transmitted to the other market. The model also allows inference as to the direction of price change, or which market appears to be the price setter. 2.142 The results show that all markets are integrated with Kampala, and for all crops. The detailed results are given in Annex 2-9. The feedback (ECM) coefficient for all market pairs is highly significant. Also for most market pairs, the instantaneous adjustment coefficient is also significant (it is insignificant for 8 of 40 pairs). While the ECM coefficient shows that most markets are in a long term relationship (are integrated), it does the degree of integration varies considerably. The degree to which prices are transmitted instantaneously, that is within period 1, and the speed of adjustment (the degree to which price is transmitted at the end of period 3) are show graphically in Figure 2-26. 65Most market integration models stop with the short-run or instantaneous relationship, but this can be misleading as this may simply represent a response to a common exogenous shock but not be able to test whether there is a longer term relationship between markets. 91 Figure 2-26: Market Integration Market Integration: Instantaneous Price Transmission 100% 80% 60% Maize 40% Beans 20% Gnut Matoke 0% Arua Gulu Lira -20% Masindi KabaleMbarara Masaka JinjaIganga Mbale -40% Market Integration: Price Transmission by Period 3 100% 90% 80% 70% Maize 60% Beans 50% Gnut 40% Matoke 30% 20% 10% 0% Arua Gulu LiraMasindi KabaleMbararaMasaka JinjaIganga Mbale 2.143 The degree of price transmission varies by crop and by market. Masindi achieves the highest degree of instantaneous price transmission for maize, beans and groundnuts, but not for matoke. Jinja, Iganga and Mbale also have good integrate for maize and to some extent Mbale. Mbale is well connected for beans and groundnuts as well. This pattern shows a high degree of integration with markets in the East, a major thoroughfare and marketing corridor to Kenya. Matoke markets are well connected to Masaka and Iganga, and to a lesser degree Jinja. These markets are in close proximity to Kampala and hence the high degree of instantaneous adjustment. 2.144 The main conclusion is that markets are quite well integrated overall. Most perform do well for maize and beans, with the exception of Arua, which is the most remote among the markets studied. Gulu and Lira are well connected to Kampala for maize, and more so than Mbarara is. Matoke markets are more fickle, though except for the north, most others are 92 reasonably well connected to Kampala. Overall Masindi is the best integrated, along with Mbale. As noted earlier, Masindi lies on the trunk road to DRC and has a lot of truck traffic, most of which is returning without much cargo. This provides a good opportunity for low cost transport and probably explains the degree of integration with Kampala. In general, markets in the East are most connected. Long distances to the north and West (Kabale) are less but still reasonably well connected. F. SUMMARY AND THE WAY FORWARD 2.145 Agriculture will continue to play a critical role to play in growth and poverty reduction in Uganda. Past performance has been impressive on both growth and poverty reduction counts. The recent slowdown (post-2000) is a cause for concern, and needs to be properly analyzed to understand the key factors behind the decline in growth rates. 2.146 The slowdown after 2000 was due to a significant and unexplained negative price fluctuation, while output has continued to grow at a healthy and steady pace. It is important to fully understand the factors underlying the prolonged price decline in 2001-2002 and take appropriate action to address this issue. 2.147 Whereas the main driver of past growth was area expansion, in recent years the contribution of yield and productivity growth has increased significantly. Ugandan agriculture is becoming increasingly more commercial and shifting to higher valued crops. These developments, along with continued strong growth in output, indicate that the current policy framework and strategy of the PMA is yielding results. This needs to be strengthened to promote future productivity growth. 2.148 Looking to the future, the debate on whether the development strategy should focus on food crops or exports is not a productive one. It is essential to focus on broad based productivity growth given the current structure of the economy and the strategic importance of both the food crops and industrial or export crops. What is needed is a balanced development strategy. A critical role of agriculture in poverty reduction is to maintain downward pressure on real food prices. But to ensure adequate producer incentives and enhance producer welfare, productivity gains have to outweigh declining producer prices. This requires outlets for agricultural products and hence a focus on the demand side. Demand side comprises domestic demand (for food and raw materials) and export (for industrial crops). Improved marketing efficiency will help also lower final consumer and raw materials prices as well as make Ugandan exports more competitive. And while improved markets will help stimulate some demand, higher real non-farm incomes, particularly among the urban and rural poor, will be important to increase the demand for agricultural commodities. Thus, for a robust strategy for pro-poor rural growth, actions on all three fronts needed to increase farmer incomes through higher productivity, even as the final consumer prices (and producer prices) decline. 2.149 One of the biggest challenges to sustaining agricultural growth in Uganda is the rising population. In the past, with available land, growth was sustained through area expansion. In recent years yields have contributed to output growth, but area expansion still remains the major source of growth. But the limits to environmentally sustainable area expansion have probably already been reached. Without significant possibilities for further area expansion, future growth will have to rely on a combination of more intensive agriculture and movement of labor out of agriculture to urban and rural non-farm activities. The latter will be important, as population pressure is already leading to uneconomical farm sizes, subsistence orientation of 93 households and reduced adoption of productivity enhancing technologies because of higher risk aversion. 2.150 The second major challenge is the volatility in agricultural growth. Agriculture is naturally more susceptible to weather variability. In the case of Uganda, it appears to be also significantly subject to market (price) shocks. Large exogenous shocks can have devastating and long-lasting impacts. It is critical to better understand the factors driving the volatility and their impact on household welfare. The analysis would then help identify appropriate policy actions to address the underlying causes. The various analyses undertaken in this paper suggests that that both price and weather shocks are important in household decision making and outcomes. 2.151 There are many options that can help cushion the impacts of exogenous weather related shocks. These include development of irrigation to reduced the impact of droughts, but also to help enhance productivity through a stable water supply. Another option may be to put in place a safety net program in the form of a public works or employment scheme that the affected people can turn to in times of extreme and widespread weather outcomes (such as a major drought). Such safety nets provides income earning opportunity for households in distress and can be used to help build labor intensive public infrastructure for future productivity growth. An alternative, to deal with more localized shocks may be weather indexed insurance, which pays out in times of an objectively defined weather outcome based trigger. Finally, helping promote the development of the non-farm economy would help provide income diversification and help cushion the impact of a negative shock to agricultural incomes. 2.152 The third main challenge is to improve rural infrastructure and market access. This is an important and consistent theme emerging from almost all analyses (productivity enhancement, diversification, adoption of inputs and market performance). Overtime, the performance of agricultural major markets has improved significantly, and evidence shows that most markets are performing well. However, the transport costs from the farm gate to primary and secondary markets remain high. Thus, the main challenge is to tackle the rural transport problem, which adds much more in proportional terms to the post-farm gate costs than the transport between urban centers/markets. High fuel costs may be an important element in this, as motorization rates are very low (see chapter 6). 2.153 Fourth, an important segment of agricultural marketing is commodity storage market, which remains problematic. The intra-year price rises have become less pronounced over time as shown by the declining trends but the absolute magnitudes of inter-temporal margins are quite high. Despite the decline in the inter-temporal margins, the continued high returns call for a further investigation into the working of the commodity storage markets. 2.154 A Fifth theme emerging in influencing household choices is weather-, pest-, and disease-related risks that affect all aspects of farmers' production decisions. Importantly, in addition to addressing the drought risks mentioned above, damage from excess water has a larger quantitative impact than droughts, indicating need for better water and natural resources management. Main implications emerging are the need to increase investments in public goods targeting pest and disease control, while also focusing on natural resource management issues and household risk management. Two examples of where pests and diseases are costing the economy significantly are the Coffee Wilt Disease (CWD) and the Banana Bacterial Wilt (BBW). The decline in coffee production is a direct impact of CWD, costing Uganda about half its potential coffee earnings. Similarly, being a preferred staple, the decline in matoke production is also significantly hurting producers through foregone income. 94 2.155 Sixth, the analysis in this paper suggests that the fertilizers are likely not being used efficiently and hence the lack of more widespread adoption. In the productivity debate in Uganda, considerable policy attention is devoted to the promotion of fertilizers. However, there is insufficient evidence that the cost of fertilizer is a binding constraint in its use. Further and better analysis is needed to guide policy actions, especially on the determinants and economics of fertilizer use. 2.156 Seventh, it is not evident that credit constraints are the most binding on household decisions to adopt inputs, including fertilizer. On the other hand, infrastructure and relevant and effective agricultural advisory services show consistent and strong impacts on technology adoption. The evidence on the strong impact of effective and demand-driven technical advice (as opposed to simple delivery of extension services) calls for a more rapid roll out and strengthening of the NAADS program which has been constrained by resources.. 2.157 Eighth and finally, agricultural exports have performed well, with increasing diversification in formal exports, despite the decline in coffee. There is also a volume of informal exports of traditional food staples. Looking to the future, this suggests that Uganda stands to gain not only by focusing efforts on exporting "high value" formal exports, but also through regional trade which offers significant potential for helping commercialize even farmers who currently grow "subsistence" crops. 95 3. GROWTH DIAGNOSTIC A. GROWTH DIAGNOSTIC: BACKGROUND ­ THEORETICAL FRAMEWORK 66 3.1 Recent empirical research suggests that growth accelerations may be unleashed by relaxing a limited number of country-specific key constraints67. This empirical observation occurs in developing economies in particular, because they typically exhibit a high degree of "slack". For example, firms in developing countries typically operate below their productive capacity and hold extra stocks, and there is typically an ample supply of unskilled or underemployed labor in the economy. 3.2 It is certainly true in Uganda that manufacturing firms have spare capacity. In the 1998 Regional Program for Enterprises Development (RPED) firm survey, average capacity utilization was estimated by responding firms to be 56 percent for the period 1995-97. Respondents to the 2003 RPED survey revealed average capacity utilization at 57 percent for the period 2000-2002. A survey by UNIDO based on 158 firms for the period 2001-2003 indicated capacity utilization at 74 percent. Despite being rather rough measures, these results do suggest spare capacity for Uganda. 3.3 On average globally, most growth accelerations are not sustained beyond about 8 years. This suggests that after the current binding constraint is relaxed, the economy starts to run up against other situation specific constraints, some of which in turn become binding on the country's growth. These then need to be removed to avoid bringing the growth spurt to an end. 3.4 This line of research suggests that country-specific growth strategies need to focus on identifying and removing the binding constraints. In a country with stable macro economic policies like Uganda, this will mean addressing a shortlist of structural reforms and `public good' type investments that bind growth the most at any given time, and which can be dealt with without provoking negative effects which compromise future growth. The "binding constraint" will change over time; hence the growth assessment needs to be regularly updated. In some cases the constraints may vary by sector, or by area of economic activity, and possibly even by region. 3.5 The search for a set of potentially binding constraints is particularly challenging in a growing low income country like Uganda. For a start, the country is growing already: the quest for constraints to growth is perhaps relatively easier when investment and growth are showing less dynamism. Nevertheless, given global evidence that growth spurts don't last, the quest to find and remove the obstacles to growth is important. Second, Uganda is a very poor country in which a lot of things need fixing ­ policy makers and the foreign consultants who advise them - can more or less take their pick from thousands of possible solutions to a myriad of development problems. 66See World Bank "Development Lessons of the 1990s", 2005, and Rodrik, Dani "Growth Strategies", Harvard. 67Hausmann, Pritchett and Rodrik (2004) analyze 83 global instances of growth episodes between 1957 and 1992 where growth in an economy accelerated at least 2 percentage points to a level of at least 3.5 percent per capita for at least 8 years. They found that it takes little to trigger growth, but that most episodes slow down. 96 3.6 This makes it all the more important to narrow down the list of priorities for growth. Chapter 4 of the Poverty Eradication Action Plan (PEAP) on "Enhancing production, competitiveness and incomes" is 41 pages long, and contains a long and yet still not exhaustive list of desirable reforms. In addition the Government is close to completing a re-draft of the Medium Term Competitiveness Strategy, and already has several sector strategies in infrastructure, agriculture, etc (the Plan for Modernization of Agriculture already runs into several chapters). Most of the actions in these various strategies have a compelling rationale, and in most cases Uganda would fall short of satisfactory indicators. But just because everything needs to improve, does not mean that there is the capacity to implement everything simultaneously. Nor does it mean that if something in the list is fixed, it will have the same impact on growth as another alternative action. As a case in point, privatization of telecommunications in 1997/98 seems to have set in train a growth spurt ­ it would appear to have been the right thing to have done at the right time, to unleash service sector growth. 3.7 To identify likely binding constraints to economic growth in Uganda, this chapter adopts the conceptual framework by Hausmann, Rodrik and Velasco (2005). Their framework for growth diagnostics (which we call HRV) begins with the simple proposition that growth continues so long as the expected returns to asset accumulation (i.e. investment) exceed the costs of financing the investment. The HRV framework is demonstrated in Figure 3-1. It starts with the simple proposition that growth in income depends on the interface between each of three determinants of returns to investment: Private income = appropriability (1- ) * productivity k * accumulated factors (A) i.e: Y = (1-) * A * k 3.8 Consistent with neoclassical growth theory, in the HRV framework there are three ways in which a country can grow68. It can accelerate factor accumulation, it can improve appropriability of the returns to accumulation, or it can improve productivity. To increase accumulation, policy makers can seek to; (i) increase domestic savings (increase public savings, improve the financial system, social security reform), (ii) facilitate access to foreign savings (through official international lending, foreign direct investment, open capital account), and (iii) promote education (by increasing expenditure allocations, and improving the effectiveness of education services). To improve appropriation of returns, policy makers can ; (i) ensure low and predictable taxes, (ii) reduce the probability of expropriation through macro volatility, (iii) improve contract enforcement, protect property rights and justice, (iv) reduce corruption and crime, and (vi) assure political stability and good governance. To improve productivity, policy makers can; (i) invest in skills development, (ii) open up to foreign trade and investment, (iii) provide complementary public investment, (iv) uphold intellectual property rights, (v) promote innovation, (vi) resolve coordination failures, and (vii) promote R&D and foreign direct investment. 68Draws heavily from presentation by Ricardo Hausman, May 2005 on "Growth Diagnostics" in the PREM 2005 Conference. 97 Figure 3-1: Growth Diagnostics Problem: Low levels of Private Investment and Entrepreneurship High cost of finance Low return to economic activity Low social returns Low appropriability bad international bad local finance finance government market failures failures information poor bad coordination externalities: externalities infrastructure geography "self discovery" micro risks: macro risks: low poor low property rights, financial, human domestic Intermediation corruption, monetary, fiscal capital saving taxes instability 3.9 Key public interventions in the HRV framework are those that remove the binding constraints or largest distortions in the economy. In the ideal world all distortions would be removed, and all investment-enhancing infrastructure and institutions would be put in place simultaneously. However the HRV framework recognizes that not all growth-enhancing interventions can be implemented simultaneously, and with perfect foresight. Furthermore, it recognizes that different policy changes can have very different growth effects. This is because the growth impact of an intervention depends on the level of the policy or investment that is being changed, as well as the other fundamentals (technologies, preferences) and "distortions" (government failures and market failures) in the economy which shape appropriability, productivity and accumulation. The welfare returns to policy interventions which address non- binding constraints can be low, or even negative. But because the factors that determine growth (appropriability, accumulation, productivity) are complements, the marginal return on the binding constraint will be high, and this will drive down marginal returns to other factors. 3.10 To keep growth from slowing, the challenge is to identify what could be most binding at any given time and correct for it. To help do this for Uganda, the following diagnostic exercise investigates a number of questions ­ emerging from Figure 3-1 - to identify why investment is not higher. · Is it inadequate access to finance or inadequate returns to investment? · If it is access to finance, is this because of a lack of access to savings or because of poor intermediation? 98 · If it is low returns, is this because social returns to investment are low, or because private investors cannot appropriate the returns? · If it is low social returns, is this due to a scarcity and high price/cost of factors of production (human capital, technical know-how, or infrastructure)? · If it is poor appropriability, is the problem high taxes, instability, learning and coordination externalities, high costs of enforcing property rights, poor contract enforcement, etc? 3.11 The growth diagnostic involves looking iteratively for direct and indirect evidence of symptoms of binding growth constraints. As mentioned above, a low income country like Uganda, is likely to reveal evidence of constraints in many areas. It is quite likely as growth continues, as the PEAP is implemented, and as the economy evolves, that the nature of constraints to growth will change. The diagnostic should therefore be done carefully and iteratively, and it should be updated regularly by Uganda's authorities, perhaps in updating the PEAP. Three other things are worth bearing in mind in undertaking the diagnosis: 1. The growth that does occur in the presence of a binding constraint should be in activities that are least intensive in that constraint. 2. Thriving firms should more often exhibit behavior designed to get around the binding constraint than slow growing firms. 3. Also, as Hausmann and Velasco (2005) point out, since the factors determining investment returns are complements, a constraint arising from one of them will cause a high marginal return on that factor, and a low return to the other factors. Some likely symptoms to look for include: · If low education is a serious problem, the return to skills should be high and unemployment of skilled people should be low, assuming no other restrictions or distortions in the labor market. · If investment is savings constrained, interest rates on deposits and lending should be high, and short-term changes in available savings (e.g. inflows of foreign resources) should stimulate growth spurts. · If poor transport is a serious constraint, we should observe bottlenecks in busy locations, high private costs for transport, and perhaps low transport flows in more remote markets. · If coordination failures are serious, we might observe unfunded projects despite potentially high returns. · If high taxes are constraining appropriation, we should observe high informality and evasion. · If poor legal institutions were to blame, we should observe high demand for informal mechanisms of conflict resolution and contract enforcement. · If poor financial intermediation is the constraint, we should see internalization of finance through business groups, etc. · If access to capital goods is the problem, we should see high returns to capital stock and lower returns to labor. 99 · If poor public capital is the constraint, we should observe firms self-provisioning infrastructure (eg own electricity generators, water supply, transportation), reducing the returns to their investments. B. INITIAL RESULTS ­ TOWARDS IDENTIFYING BINDING CONSTRAINTS TO PRIVATE INVESTMENT AND GROWTH IN UGANDA Is it High Costs of Financing? 3.12 Symptoms of Uganda's financial market are: · high interest rates on private lending, · low interest rates on savings deposits, · very short maturity structure of bank credit (64 percent < 6 months duration), · low private credit to GDP ratio (just 7 percent in 2004/05), · low lending to deposit ratio, · Ugandan firms rely heavily on internal finance for investment and working capital credit market participation is very low, · Large firms do not appear to be credit constrained, but small firms are. The evidence 3.13 Real interest rates on commercial bank lending are high, and have been so for many years. However, real interest rates on bank deposits were ­ until recently69 - very low, with real returns on deposits negative when bank charges are taken into account. In 2003/04, the average bank lending rate was 21.7 percent, compared to an average deposit rate of 2.2 percent, with average annual inflation at 5.1 percent. The decomposition of these very high spreads is discussed in chapter 6. 3.14 Ugandan banks have high net interest margins and high overhead costs. Net interest margins and overhead costs are higher in Uganda - both on average, and for the same commercial banks ­ than in Kenya and Tanzania. They are also higher than the average for low-income and Sub-Saharan African countries (see chapter 6). Table 3-1: Selected Indicators of Financial Intermediation in 2004 across Countries Financial Private Liquid Bank Loan- Overhead Interest intermediation Credit/GDP Liabilities/GDP deposits/GDP deposit costs margin across countries, ratio 2004 Uganda 5.9% 19.0% 14.8% 39.9% 9.5% 13.8% Kenya 24.7% 38.7% 32.0% 73.2% 5.7% 6.6% Tanzania 7.8% 22.1% 16.7% 46.7% 6.7% 8.0% Sub-Saharan Africa 19.7% 31.1% 24.4% 61.1% 6.8% 9.5% Low income 14.9% 30.8% 22.0% 113.5% 5.9% 8.1% Source: World Bank's Financial Sector Assessment Program (FSAP) Update 2004 69In 2005, the Government moved all donor accounts to the Bank of Uganda. This removed a substantial foreign and domestic currency deposit base from the commercial banks. Many banks then began to compete for deposits, and for some this drove deposit rates up to positive real levels. Nevertheless the fees and charges imposed on transactions in some of these accounts remained high, and remained a significant source of income for the banks. 100 3.15 Manufacturing firms in 2002/03 cited high costs of finance and collateral as constraints to business. Over 60 percent of manufacturing firms in the RPED survey regard cost of finance as a major or severe constraint, with over 45 percent citing access to finance (collateral requirement). This was particularly the case for Ugandan owned firms and non-exporters. Out of all possible constraints to operation, cost of finance ranks as the most severe in the 2003 RPED survey. Furthermore, by far the most common complaint by the hundreds of thousands of households who recently set up household-based enterprises, is that access to credit and working capital is the most serious constraint on opening and running a business70. Manufacturing firms also reported facing high costs of borrowing from banks. The average reported rate in the previous three years of the survey was 20 percent. Ugandan firms do seem to face very high costs of finance, and undertake low credit market participation. But this is not the same as concluding that firms are credit constrained ­ i.e. some firms do not participate in credit because they don't need it. 3.16 Borrowing from banks is an insignificant source of credit for rural households. Farmers and rural households require short-term credit for working capital and longer term credit for investments in processing, tree crops, machinery etc. Studies have shown that rural Ugandans do not seek credit from formal financial institutions71. The most common sources of finance and the purpose of borrowing in Uganda vary significantly by region; for example households in the North are heavily dependent upon cooperatives and Government agencies for credit for agricultural inputs), whereas households in the Central region relied on family and friends for enterprise loans. Results for the country as a whole in 1999 (last available data) are shown below. Like the results for manufacturing, this seems to suggest that formal bank lending in Uganda is of limited importance for the rural community. Table 3-2: Sources and Uses of Credit ­ Rural Households Source of Credit in 1999 Share Purpose of Loan Share Source Of Enterprise Share Rural Households (1999) Start-Up (Informal Sector Survey 2002/03) - Relatives and friends 46% Expansion of Enterprise 29% Own Saving 92% - NGO / Cooperative society 21% Education & Health 27% Loan From Family / 4% Friends - Community / group 13% Agricultural Inputs 24% Loan From Money 1% Lender - Government agency 12% Consumption 17% Loan From Bank Or 1% Financial Institution - Firm/employer/money lender 6% Housing 3% Other 1% - Banks 2% Source: Mpuga (2004) and Boumeester & Burger (2006) 3.17 Credit market participation by Ugandan manufacturing firms is very low, especially for small firms (Table 3-3). This is consistent with the findings of Collier and Gunning (1999). Very few surveyed micro or small firms were granted a loan in 2002, their access to overdraft facilities and trade credit is minimal, and their collateral value to loan size ratio is on average higher than that of the combined level for medium and large firms. Since the access to bank loans is very limited for micro and small firms, they have to rely more extensively on internal 70Boumeester and Burger (2006). 71See Mpuga (2004), `Demand for Credit in Rural Uganda: Who Cares for the Peasants?' 101 funds and retained earnings to finance both their working capital (inventories, accounts receivable and cash) and new investments (not shown). Firm size correlates highly with new investments undertaken in 2002; ie larger firms invested more. Overall, only 23 percent of firms have access to an overdraft facility, internal funds constitute 80 percent of working capital and debt/ overdraft to banks only 7 percent. Table 3-3: Credit Market Participation and Firm Size (by no. of Employees Micro Small Medium Large (1-5) (6-25) (26-100) (100+) All Received loan in last year (2002) 0.10 0.07 0.09 0.28 0.11 % of firms with overdraft facility 0.05 0.15 0.40 0.69 0.23 Debt/ overdraft to banks as % of working capital 0.02 0.05 0.07 0.28 0.07 % of firms with informal sector debts 0.01 0.01 0.02 0.00 0.01 Informal debt as % of working capital 0.01 0.00 0.00 0.00 0.00 % of firms currently receiving trade credit 0.10 0.10 0.18 0.41 0.14 Trade credit outstanding as % of working capital 0.04 0.04 0.07 0.13 0.05 % of firms providing collateral (formal loans) 0.67 1.00 0.92 1.00 0.93 Internal funds or retained earnings as % of working capital 0.85 0.84 0.79 0.53 0.80 Collateral value to loan size ratio 2.02 1.12 0.83 1.22 1.21 % of firms invested in last year 0.36 0.39 0.58 0.71 0.46 Observations 83 126 57 32 298 Source: World Bank, RPED survey, Uganda, 2002/2003 3.18 Large firms are much more likely to apply for and receive credit (Table 3-0).72 This is natural since typically the returns to their investments are higher, and lending to large firms is less risky and therefore more `bankable'. Also, the fact that firms do not apply for loans does not necessarily mean that they are credit constrained. We therefore investigate credit constraints using the methodology from Bigsten et al. (2003)73. In Table 3-5 we consider as credit constrained, those firms in the categories who did not apply for credit because they had `Inadequate collateral,' `Process too difficult,' and `Didn't think I'd get one'. Firms who either do not need current or additional debt, or who think they are already heavily indebted, or who regard the interest rate as too high are treated as unconstrained firms without credit demand. Using this classification, the severity of credit constraints in Uganda diminishes with firm size (Table 3-6). 72Bigsten et al. (2003) find for their sample that micro firms require a return on fixed capital of more than 200% in order to have the same likelihood of obtaining a loan as large firms. 73 They have analyzed credit constraints in African manufacturing companies in the six countries Cameroon, Ghana, Kenya, Zimbabwe, Burundi and Cote d'Ivoire. 102 Table 3-4: Formal Credit Market Participation by Firm Size (Percentages of Firms) Micro Small Medium Large All Did not apply 0.64 0.67 0.54 0.31 0.60 Applied and did not receive 0.05 0.01 0.02 0.00 0.02 Applied and received 0.31 0.32 0.44 0.69 0.38 Source: World Bank, RPED survey, Uganda, 2002/2003 Table 3-5: Why did Firms not apply for Loans? By Firm Size (Percentage of Firms) Micro Small Medium Large All Inadequate collateral 0.13 0.11 0.07 0.00 0.10 Don't want to incur debt 0.15 0.23 0.27 0.11 0.21 Process too difficult 0.23 0.16 0.03 0.00 0.15 Didn't need one 0.11 0.14 0.37 0.44 0.19 Didn't think I'd get one 0.06 0.07 0.07 0.00 0.06 Interest rate too high 0.15 0.20 0.13 0.11 0.17 Already heavily indebted 0.02 0.00 0.00 0.00 0.01 Other 0.15 0.09 0.06 0.34 0.11 Source: World Bank, RPED survey, Uganda, 2002/2003 3.19 Credit constraints are highest amongst micro and small firms, but overall credit constraints do not appear to be binding for Ugandan manufacturing firms (Table 3-4). Only 7 percent of medium and no large firms are considered credit constrained, and over 70 percent of large firms receive a loan following application.74 It also appears that the degree of credit constraints for micro and small firms is not very high, since almost a third of them receive a loan, and a high proportion does not face any credit demand. Table 3-6: Credit Constraints by Firm Size Micro Small Medium Large All No credit demand 0.36 0.48 0.48 0.30 0.43 Demand, but rejected* 0.33 0.20 0.07 0.00 0.19 Received loan 0.31 0.32 0.45 0.70 0.38 *Includes firms that suggested that a loan application would be rejected by banks Source: Own calculations 3.20 By regional standards, credit constraints in Uganda are relatively low, access to overdrafts is low, and the share of working capital from internal funds is high. A cross- country comparison of RPED surveys from African countries suggests that credit constraints in Uganda are relatively low by regional standards, but that credit market participation or access to 74The sample for the large firms is small with only 32 observations. 103 finance is a potential bottleneck for Uganda's manufacturing firms. Except Kenya with an average of 9 percent, Uganda's average percentage of credit constraints ­ at 19 percent - is below all African countries in Table 3-7. In contrast the percentage of Ugandan firms with access to an overdraft is well below that of most African countries. Finally, an above-average percentage of working capital is financed by internal funds in Uganda. Table 3-7: Cross Country Comparison (Independent of Firm Size) Received Internal Collateral to Credit Non- Bank loan in Overdraft Funds Loan Size Country Constraints Application last year Access in WC Ratio Kenya (2003) 0.09 0.45 0.18 0.74 0.48 1.80 Madagascar (2005) 0.33 0.80 0.10 0.34 0.76 1.37 Mali (2003) 0.31 0.61 0.04 0.44 0.83 1.17 Mozambique (2002) 0.38 0.60 Na 0.12 0.90 1.31 Senegal (2003) 0.30 0.64 0.12 0.60 0.74 1.08 Tanzania (2003) 0.31 0.58 0.10 0.32 0.74 1.15 Uganda (2003) 0.19 0.60 0.11 0.23 0.80 1.21 Zambia (2002) Na na 0.16 0.51 0.61 2.29 Source: World Bank, RPED surveys, and own calculations 3.21 Credit is important to Ugandan firms; evidence suggests growing firms are demanding it. Using RPED firm data from 2002, Habyarimana (2004) investigates the effect on the performance of Ugandan firms which lost a banking relationship after the problems at the then Uganda Commercial Bank (UCB). The growth rate of employment for such firms fell by 10-15 percent relative to unaffected firms over the three years after the incidence. These findings suggest that banking relationships are beneficial for Ugandan firms. Given that many micro and small firms do not have close relationships with financial institutions, their growth potential has not been optimized. Diagnostic analysis It's not low Savings: good international finance offsets low domestic savings: 3.22 Macro level analysis suggests that Uganda's economy is not constrained by low savings. Again, this does not mean savings are high; they are not. It means that an additional inflow of savings is not necessarily going to lead to a growth spurt. At about 9 percent of GDP, domestic savings are low, largely because Uganda's revenue performance has lagged in recent years, and because of scaled-up spending on Universal Primary Education. However, official transfers more than compensate for the public deficit, so the public sector's demand for external savings is not competing with private sector. Government has no access to international capital markets, nor does the Government use domestic debt to finance expenditures. The rapid build up in domestic debt in recent years has been driven by Bank of Uganda interventions in the money market to mop up excess liquidity arising from foreign inflows. 3.23 Foreign savings and private transfers seem more than adequate to meet domestic investment requirements. Cumulatively, foreign reserves have been built up to the tune of $US425 million from 2001-05 and Uganda has made cumulative net repayments to the IMF of $150 million in 4 years. Furthermore, commercial bank deposits in Uganda were 19.3 percent of GDP in 2003/04, compared to private credit of just 5.8 percent of GDP. The loan/deposit ratio, at 104 39 percent was just over half the average for Sub-Saharan Africa. This does not suggest a macro savings constraint. On the contrary, it suggests an excess of inflows over intermediation. It Could Be Poor Intermediation: 3.24 Uganda shows evidence of quite low credit demand, and even lower credit supply, at high lending rates, and with high interest rate spreads. Low credit demand could be symptomatic of low investment returns. However, Uganda has relatively high levels of investment ­ mostly financed out of internal finance ­ which suggests intermediation problems. High interest rates on lending might suggest low savings, but not when interest rates on deposits are simultaneously very low as they are in Uganda. Very low lending to deposit ratios also suggest that Uganda could be experiencing intermediation constraints. We therefore take this issue up in detail in chapter 6. Is It Low Appropriability? 3.25 Symptoms observable in Uganda include: · high informality; · relatively low costs of doing business; · high degree of confidence in the judiciary; · macro stability with low inflation, and some recent financial volatility; · business associations more focused on lobbying than on product standards and technologies; · product discovery is increasing; · the spread of new technologies in manufacturing is slow; · apparently sound private infrastructure investments are going unfunded; · relatively low unofficial payments are made to officials, but medium size firms and exporters complain the most about officials; It's not Macro risks: 3.26 Macro risks in Uganda are definitely not the binding constraint, but there is no room for complacency on exchange rate and interest rate volatility, nor inflation. The Bank of Uganda's 2003 private investment survey revealed healthy confidence. The vast majority of investors indicated an intention to expand or maintain their investments. The survey showed that the level and equity share of FDI was increasing in 2001. FDI has since increased on average by 18 percent per year, with the annual flow in 2004/05 almost doubling compared to 2000/01. Investors saw the strongest negative effects on investment as coming from exchange rates, corruption, interest rates and inflation, and the strongest positive effects coming from political stability and economic policy. 105 Table 3-8: Investor Perceptions of Likely Direction of Investment in 2003 Likely Direction of: Expand Maintain Contract Diversity by region 49.1 47.9 3.0 Diversity by Sector 41.9 54.8 3.3 Staff Training 62.2 32.2 5.7 Research & Design 49.9 39.4 10.7 Investment in technology 67.6 26.0 6.5 Recruitment 56.5 37.0 6.5 Imports 44.3 45.5 10.2 Exports 61.7 32.2 6.1 Turnover 89.5 8.2 2.4 Profit 86.9 10.2 2.9 Source: Private Sector Investment Survey by Bank of Uganda, Uganda Bureau of Statistics and Uganda Investment Authority It's not high taxes: 3.27 Taxes in general are not high, but the sector composition of the tax burden needs to be reviewed from the standpoint of growth. Marginal rates of tax in Uganda are not high. Uganda's tax mobilization effort has been slow, indicating low compliance, extensive concessions, or a combination. The tax structure and tax rates are simple and efficient. The corporate tax rate is 30 percent, VAT is 18 percent, the top rate of income tax is 30 percent, and there are just three bands. Tariffs on imports are low by international comparisons, and are largely uniform. Taxes on equipment and machinery almost certainly exceed taxes on property, and this should be reviewed. Taxes are frequently mentioned by entrepreneurs as burdensome, but analysis of the characteristics of complaining firms and non-complaining firms is inconclusive, suggesting this is a not a specific issue. Analysis of RPED data shows that on average, total tax payable from profits is 15 percent lower in Uganda than the Sub-Saharan average. 3.28 Taxes on transportation are disproportionately high. Transportation is heavily taxed in Uganda; in such a rural economy, this amounts a tax burden on farmers. Table 3-9 which is drawn from the 2001 supply and use tables prepared by UBOS shows that 4 of the top 10 taxed manufacturing activities in Uganda are in the transport sector. As a rural and landlocked country which depends disproportionately on transportation, this is inappropriate for growth. The finding is confirmed by recent detailed transport sector studies conducted by the Bank; Johanson (2005) and Hussein and Kopicki (2006) under this project, and Raman (2005) under the DTIS (results are shown in chapter 7). They find that vehicle operating costs (at 50 percent) are a higher share of transport costs in Uganda than in most countries (the international average is 30 percent). They put this down to high fuel costs (in part due to high fuel taxes) and inefficient fuel use. In turn the inefficiency of the vehicle fleet is due to its age; transporters tend to import older and less fuel efficient vehicles because the cumulative burden of import duties and withheld VAT makes importation of newer vehicles uneconomic (see World Bank DTIS 2006 for details). 3.29 High tax rates on mobile telephony may now be limiting its spread. With taxes now accounting for 30 percent of the cost of a call, after Turkey, Uganda has the second highest tax burden for mobile telephony in the world.75 In rural areas, only 2 percent of the population owns a mobile phone compared to 16 percent in urban areas. Farmers, whose livelihoods could benefit 75 World Telecommunications Development Report 2006: Measuring ICT for Social and Economic Development, (Geneva, Switzerland: International Telecommunication Union, 2006). 106 greatly from the use of mobile telephony, are likely being taxed out of the market. In 2005, the number of mobile subscribers increased by 29 percent, far below the 50 percent and 97 percent growth rates accrued in 2004 and 2003 respectively. The 2005 mobile subscription growth rate has been the lowest growth rate experienced in the industry since 1997. Table 3-9: Top 10 Taxed Manufactures (VAT Plus Import Duty as a share of supply) 1 Petroleum Refining, Manufacture Of Products Of Coal 42.0% 2 Other Manufacturing N..E.C. 18.6% 3 Manufacture Of Wines 17.0% 4 Manufacture Of Motor Vehicles, Bodies And Parts 14.1% 5 Manufacture Of Electrical Equipment, Apparatus And Supplies 12.4% 6 Reproduction Of Recorded Media 12.3% 7 Manufacture Of Tyres, Tubes And Other Rubber Products 12.2% 8 Manufacture Of Leather, Products And Footwear 11.4% 9 Manufacture Of Glass And Other Non-Metalic Mineral Products 9.8% 10 Manufacture Of Ships And Boats 9.2% Source: Uganda Bureau of Statistics It's not corruption: 3.30 There is high informality in the economy, and employment in the informal sector is booming. Typically, informality signals firms wishing to avoid taxes or regulations. Neither of these seems unduly burdensome in Uganda (ICA, FIAS report on costs of doing business). The costs of business registration are abnormally high in Uganda, which may be a factor. The number of procedures to start a business is high; 17 procedures take on average 36 days. The average is 11 procedures taking 63 days. Informality can also be a means of avoiding corrupt and predatory officials. More work is needed to understand what stops informal firms from growing and `formalizing'. This should be a topic for the forthcoming investment climate update. 3.31 Proxies for corruption from RPED data suggest it is not the binding constraint for Ugandan manufacturing firms. Table 3-10 provides a cross-country comparison of micro risks. On average Ugandan firms spend 6 days in meetings with tax officials, higher than the Sub- Saharan average of 5.4 days, and also higher than other regions. However, perhaps surprisingly given the general perception of corruption in Uganda, and given scores in the World Economic Forum's Growth Competitiveness Index76 Manufacturing firms in Uganda report making lower unofficial payments as percentages of sales to get things done than most African countries, and countries in the Middle East & North Africa and Latin America & the Caribbean. Furthermore, only 7 percent of the Ugandan firms give gifts to tax inspectors, which is far lower than the Sub- 76 www.weforum.org ­ the competitiveness index is a survey-based instrument. WEF rates Uganda 79th out of 104 countries, and places Uganda in the bottom half of 25 African countries (14th). Uganda just makes the top half for macroeconomic environment (12th) and technology (10th). Uganda's overall score is dragged down by a low rating on the Public Institutions Index. (18th). Within public institutions, out of 21 countries, Uganda scores badly on sub-indices for; (i) favoritism in decisions of government officials (16th), (ii) organized crime & extortion (18th), (iii) irregular payments in import and export (19th), (iv) irregular payments in utilities (15th), and (v) irregular payments in tax collection (16th). 107 Saharan average of 17 percent and the other regions.77 Security and protection costs and the time to resolve a legal dispute are in the average with most African countries, whereas bureaucracy as measured by the time senior management spends with regulation requirements appears to be lower. 3.32 Corruption is, however, disproportionately felt by medium-sized firms, foreign firms, and exporters, and it is a serious problem in the public service delivery, including in infrastructure. Uganda's manufacturing industry is largely comprised of micro firms, who do not seem to be as severely affected by corruption. Large firms also seem relatively less troubled, and hence the overall score may be low for structural reasons. However, to the extent that exporters and medium-sized firms will likely be the drivers of employment growth in Uganda, these findings are disturbing. In addition, to the extent that public investment in skills and infrastructure are important for growth (see below), corruption is an important constraint which must be dealt with swiftly if scaled up spending on public investments are to generate commensurate growth. Uganda Revenue Authority seems a particularly important source of the corruption experienced by firms. Table 3-10: Cross- Country Comparison of Selective Micro Risks Time Unofficial Firms Security Dispute Senior spent in payments expected to and resolution management Economy or Region meetings for firms give gifts in protection time time spent with tax to get meetings costs (% (weeks) dealing with officials things with tax of sales) requirements (days) done (% inspectors of regulations of sales) (%) (%) Uganda (2003) 6.00 2.54 6.93 2.35 8.42 4.99 Kenya (2003) 4.17 3.85 37.26 2.80 11.38 13.48 Madagascar (2005) 2.65 7.69 12.69 0.92 4.39 22.18 Mali (2003) 5.33 3.45 30.23 2.05 4.41 8.84 Tanzania (2003) 12.40 1.48 22.41 2.85 8.34 15.74 Sub- Saharan Africa 5.41 2.61 17.19 1.75 7.98 10.85 Middle-East & North Africa 4.66 6.26 27.74 na 23.36 12.99 South Asia 3.37 1.57 44.27 0.83 9.20 8.15 Latin America & Caribbean 3.92 7.07 12.02 8.60 4.80 12.46 Source: World Bank, RPED Surveys 77 Even though 55% of foreign and exporting firms perceive corruption as a major or severe constraint in Uganda, there is no statistical difference between all and foreign firms in the amount of unofficial payments to get things done and also the percentage of firms that give gifts to tax inspectors. In addition, exporting firms pay 4% of sales for unofficial payments and 16% of them give gifts to tax inspectors. These figures are higher than the Ugandan average but partly lower than most African countries. For instance, 39% and 47% of exporting firms in Kenya and Mali, respectively, have to provide gifts to tax inspectors. Overall, the findings provide evidence that corruption especially hits foreign and exporting firms in Uganda which might indicate the presence of more bureaucratic barriers than for domestic or non-exporting firms. 108 Table 3-11: Firms' Evaluation of Some Selective Constraints (% of respondents evaluating constraint as major or very severe) Tax Corruption Tax Crime, Business licensing Rates administration theft, and and operating disorder permits Full-sample 48.3 38.2 36.1 26.9 10.1 Foreign firms 43.3 55.0 42.2 37.3 13.4 Domestic Firms 49.6 33.3 34.5 23.5 9.2 Exporters 48.9 56.4 42.9 36.4 8.9 Non-Exporters 48.4 35.0 35.1 25.3 10.4 Micro (<10) 53.3 24.8 52.3 22.9 11.1 Small (10-49) 47.6 46.3 30.3 26.8 8.7 Medium (50-100) 33.3 56.5 48 38.5 15.4 Large (100+) 47.2 38.7 37.1 30.6 8.3 The firm size is measured in terms of employees Source: World Bank, Investment Climate Assessment, Uganda, 2004 It's not property rights or contract enforcement: Significantly improving the costs of doing business will help in alleviating possible growth constraints in the Ugandan manufacturing sector but it will not provide the intended push-effect since other growth constraints such as cost of finance and inadequate infrastructure are currently more binding. 3.33 Contract enforcement indicators are good. Whereas it takes on average 209 days to enforce a contract in Uganda, it takes 439 days on average for Sub-Saharan countries and 226 for OECD countries (Table 3-12). The figures for the cost of contract enforcement support this finding as they are 50 percent less in Uganda than the Sub-Saharan average. Confidence in the judiciary is relatively good in Uganda ­ 70 percent of firms in the ICA reported confidence, compared with 55 for Tanzania and just 44 percent in Kenya. Collier and Gunning (1999), suggest that manufacturing firms in Africa often lack social capital for contract enforcement, and hence companies resort to social networks which improve the asymmetries of information between business partners. Uganda may well benefit from such strong social networks. Almost 70 percent of Ugandan firms are entrepreneur owned, and even though 75 percent of them are run by indigenous Africans, the entrepreneurs of Asian ethnicity that control the remaining 25 percent own most of the large firms. Asian entrepreneurs in general have a higher level of education, more experience, and are often able to finance the start-up company with external loans ­ all of which leads to a higher growth rate than for firms of indigenous Africans.78 One implication of this might be the presence of a close Asian social network that fosters business links and enables Asian companies to lower the risk and cost of conducting business. 3.34 Property rights do not seem to be a disproportionate problem for Uganda. The time and cost of registering property are favorable in Uganda with the cost in percentage of the 78These findings are robust across East African countries. (Ugandan ICA, 2004) 109 property value almost reaching OECD level. Land administration and titling are inefficient, with the time taken to procure land relatively high at 12 months. 3.35 The costs of starting and closing a business in Uganda are high, at around 118 percent of GNI per capita, compared with 215 percent for Sub-Saharan Africa, and only 64 percent in the Middle East & North Africa, and barely 7 percent in OECD countries. Even though low by regional standards in Africa, the costs of starting a business are high in Uganda, especially compared with countries such as China (14 percent) and India (62 percent). Furthermore, the costs of closing a business are high in Uganda with 30 percent of the estate value but if we compare the recovery rate in contrast, a business closure in Uganda still recoups 40 percent of the investment, higher than Sub-Saharan Africa with 16 percent and all other regional averages except the OECD countries.79 This finding is consistent with observed high legal fees paid by businesses in the Uganda Business Inquiry, and may warrant further analysis of the costs of legal services in Uganda. Table 3-12: Cross- Country Comparison of Cost of Doing Business Surveys Enforcing Registering Paying Taxes Starting a Closing a Business Contracts Property Business Time Cost Time Cost (% Total tax Cost (% Cost Recovery (days) (% (days) of payable (% of GNI (% of Rate (% of property of total profit per estate) of dollar) debt) value) capita) Economy or Region Uganda 209 22.3 48 5.1 42.9 117.8 30.0 39.8 Kenya 360 41.3 73 4.1 68.2 48.2 22.0 15.0 Madagascar 280 22.8 134 11.0 58.9 54.3 na 0.0 Mali 340 34.6 44 20.0 44.0 190.7 18.0 6.4 Tanzania 242 35.3 61 12.2 51.3 161.3 22.0 22.4 Sub- Saharan Africa 439 41.6 118 12.6 58.1 215.3 19.5 16.1 Middle-East & North 432 17.7 52 6.8 35.1 64.2 13.4 28.8 Africa South Asia 386 36.7 124 6.3 35.3 40.5 7.3 19.7 Latin America & 461 23.3 77 4.8 52.8 56.2 17.0 28.2 Caribbean OECD: High Income 226 10.6 32 4.8 45.4 6.8 7.4 73.8 Source: World Bank, Cost of Doing Business Surveys It's not "self discovery"80 3.36 Uganda does not seem to have a binding problem discovering new export opportunities. The 97 non-traditional products that were discovered between 1976 and 2004 accounted for 40 percent of Uganda's overall export basket during 2001-04. During the initial recovery period, these `discovered products' averaged only 2.2 percent of Uganda's exports. This level of discovery is a huge achievement in the face of declining prices for Uganda's main traditional exports of coffee and cotton. It was greatly assisted by grant aid support to budding 79 One reason for the high closing costs as percentage of the estate value could be that Ugandan estate prices especially in the Kampala region have increased over proportionally in comparison to some African countries in recent years. 80Refer to Hausman and Rodrik (2002) "Economic Development as Self-Discovery", KSG Working Paper No. RWP02-023 110 exporters through for example USAID's "IDEAS" project, which allowed private producers to overcome the costs of information externalities. Even though the bulk of the products are tied to its primary and natural resource-base - reflecting Uganda's comparative advantage ­ there has been a steady emergence in the number and value of new products signals diversification within these broad categories. This trend will help to dampen the volatility of export growth associated with traditional exports such as coffee and minerals. Government should encourage a transparent and contestable aid-financed experimentation facility along the lines of that started under the IDEAS project. 3.37 The discovery of low, medium and high tech products is an interesting development. While negligible, the gradual increase in their share could be a signal of nascent manufactured exports, an encouraging development for a low income, primary product-dependent exporter. Among the notable export discoveries were maize and related products, high value fruits and vegetables (although each had a small share, collectively they accounted for a significant share), flowers, tobacco, fish products, cement, sheep and lamb skin leather, a variety of cotton products, iron and simple iron products (iron sheets, machine parts), chemical products, and paper products. Table 3-13: Share of Discovered (non-traditional) Products in Total Exports (%) Tech SITC products 76-85 86-90 91-95 96-00 01-04 2001 2002 2003 2004 pp. Primary+high value 0.2 2.0 2.8 5.3 13.45 11.4 13.0 14.0 15.4 Rb1 & 2 Natural resource based and processed 0.0 0.1 5.4 10.7 24.8 25.6 25.7 20.7 27.2 Lt1 &2 Low-tech 0.03 0.12 0.09 0.28 1.3 0.71 0.66 1.7 1.98 Mt1&2, Medium and Ht1,2&3 High-tech 0.0 0.0 0.0 0.3 0.8 0.6 0.6 1.2 0.6 Total share of nontraditional products 0.2 2.2 8.3 16.6 40.3 38.3 39.9 37.6 45.1 Source: Chandra and Boccardo (forthcoming) 3.38 In contrast, there is dispersion in technical efficiency amongst firms in the same sector. Productivity of the best firms determines the distance a firm is from the industry frontier. For Uganda the dispersion is very wide (World Bank, ICA). On average, firms are only 50 percent as efficient as the best practice firm, and this number is consistent across firm sizes and whether or not a firm exports. This suggests that information flows amongst business associations are not so focused on technology adaptation for mutual productivity gains. In fact, preliminary analysis of firm survey data suggests that more business association members value their membership to get information on Government regulations (73 percent) and to lobby Government (69 percent) than for accrediting standards or product quality (59 percent). This behavior needs to change for the mutual gain of all Ugandan industry. It may be to do with fierce competition for the limited market for manufactured goods, which are supplied mostly to the domestic and neighboring markets. One way to overcome it may be to work with export associations. 111 Are there coordination failures? 3.39 Coordination failures81 do seem to be a problem in Uganda with respect to infrastructure gaps. There are examples of some coordination successes in Uganda, where firms in an industry have overcome coordination problems, but a better approach is needed, particularly to overcome Uganda's serious infrastructure gaps. In the late 1990s, the fisheries industry worked together with Government to overcome the threat of a salmonella outbreak, which led to a ban on East African fish exports to Europe. Fisheries and floriculture firms have worked together to develop cold storage at the international airport to handle the extra bulk of these expanding export sectors. The Uganda Grain Traders Ltd was formed as a consolidated company of individual traders to capitalize on WFP bulk contract for maize, and to exploit opportunities in the Kenyan grain market. 3.40 But there are numerous examples of coordination problems too, particularly in overcoming Uganda's infrastructure deficit with private solutions. For instance, internal transport logistics would benefit greatly from up-country storage warehouses linked to transport facilities, but none have opened up. Better connection between water, road and rail modes would improve external transportation, but has not emerged. Road transport service companies have been opening up at a rapid rate. Nevertheless, their failure to offer services which enhance the competitiveness of transport dependent businesses is reflected in efforts on the part of specific groups of exporters to "in-source" their essential transport services rather than to rely on third parties to provide them. There are also obvious coordination gaps in the power sector, which have not been overcome despite the availability of guarantees and subsidized capital in the Rural Electrification Fund; supposedly viable proposals from private generation companies for mini- hydro schemes have gone unimplemented, whilst rural industry complains about the lack of power. 3.41 The problem for Government is what to do about these coordination gaps. This is a classic industrial policy dilemma. It is beyond the terms of reference of this study to provide detailed recommendations in specific sectors, but it is worth proposing some ground rules, to which we will return in chapter 4. The solution is not necessarily public interventions with public capital ­ even if the Government could afford these, they would not have the detailed sectoral knowledge to identify and design appropriate interventions. To intervene in the right way; e.g. with complementary infrastructure, or with guarantees, Government needs to be wired in to the private sector's dialogue on needs. But officials need to be sufficiently distanced from the private sector so that they are not "captured". As Hausmann and Rodrik put it, Government needs to "find an intermediate position between full autonomy (to the public sector) and full embeddedness (by the public sector with the private sector). Too much autonomy for the bureaucrats, and you have a system that minimizes corruption, but fails to provide the incentives that the private sector really needs. Too much embeddedness for the bureaucats, and they end up in bed with (and in the pockets of) business interests. Moreover, we would like the process to be democratically accountable and to carry public legitimacy". 81Well explained in Rodrik (2005), "Industrial Policy for the 21st Century". Profitable new industries can fail to develop unless upstream and downstream investments are coaxed simultaneously; when this depends upon private agents acting cooperatively, they may either not see the potential returns, or be unable to absorb all of the risks. 112 It's Not Low Education... But Skills Gaps Are Emerging 3.42 Symptoms observable in Uganda include: · strong human capital demand - as reflected in primary enrollments under UPE; · increase in the skills premium for highly qualified labor; · but returns to primary, secondary education have not been increasing since 1999; · high underemployment amongst secondary school graduates, low underemployment amongst technical skills graduates; · low returns for unskilled labor and falling returns to "some primary" education ­ perhaps in response to increased supply; · net outward migration, with significant inward migration of consultants and skilled workers, especially in telecommunications, engineering, and banking; · relatively high real wages, rising faster than productivity in manufacturing and trade (where input costs have come down), but rising more slowly than productivity in other sectors; · growing firms employ more secondary graduates. Evidence from Firm-level data 3.43 The labor market in Uganda is not restricted. Only 6 percent of the total workforce is unionized in the Ugandan manufacturing sector, and lost days from production strikes are minimal in comparison to most other African countries and the regions. 3.44 Levels of training by Ugandan firms are low. Fewer Ugandan firms offer formal training than the Sub-Saharan average, and in general the percentage of skilled workers in African companies receiving training is below that of other regions. The fact that the optimal employment level for Ugandan firms is 50 percent higher compared to the current level might imply some problems with a skills gap.82 In other words, Ugandan firms can not find the workers with the right skill set. 3.45 Faster growing firms have a higher percentage of secondary educated employees (49 percent) compared to below-average growing firms (37 percent).83 Slower growing firms operate at lower capacity (57 percent) than faster growing firms (64 percent) and over 36 percent of the slow growing firms state in the RPED survey that they are short of their desired staff, compared with 14 percent for faster growing firms. In other words, fast growing firms do not seem to face a significant shortage of employees. There is, however, no evidence that fast growing firms have higher wage rates for their skilled production workers84. 82The very high Ugandan number for the optimal employment level is partly driven by a few outliers. For example, one company has three employees but regards the optimal employment level as 200 employees. 83We use employment growth as a proxy for firm growth since we have more observations for employment than sales, and the period for employment growth (2002-1998) is longer than sales growth (2002-2000) so this provides a better perspective on firm growth than with short-term sales growth figures. Also, faster growing firms have a higher sales growth so our usage of employment growth as a proxy for firm growth is justified. 84The UBI dataset would give more reliable data on the link between wages and firm growth. 113 3.46 Forty seven percent of Ugandan companies report skills as either a moderate, major, or very severe constrain to business. In Uganda, sixty percent of top managers have no university training, compared to 6.8 percent in Zambia and 30 percent in Kenya and Tanzania. It seems likely that improved educational levels for top management, given good quality university training, should increase industrial productivity and competitiveness. Diagnosis 3.47 The Universal Primary Education program in Uganda (UPE) has yielded strong results in terms of enrollment (Figure 3-3). Graduates of UPE are now entering the workforce, slowly increasing the average number of years' schooling, although this remains very low in Uganda. Education seems to have been closely associated with poverty reduction in the rebound phase. Appleton (2001) 85found quite high returns to primary education in household data from 1999. World Bank (2006) confirmed this trend for 2003 in both rural and urban areas. Figure 3-2: Enrollments in Primary Education Have Increased 150 ) 1991 ts(% 2002 en 100 ml ro en 50 ss Gro 0 Primary Secondary Tertiary Source: WDI 3.48 Recently however, average real returns to education economy-wide seem to have stopped rising. Figure 3-4 uses consumption data from the UNHS as a proxy for income, and displays real income for different schooling categories. Since 1999, for all but those with post- secondary schooling, returns to education do not seem to have risen86. As the dependent variable is household consumption, these returns reflect not only the impact of education on earnings from labor activities, but also the impact of education on the entire household production process ­ how the household deploys labor, investments in labor saving technology in the household, the effect of labor in lowering fertility, etc. 3.49 If low education was a serious problem for economic growth, the returns to education in the workforce should be high, and unemployment rates for those with schooling should be low. This only seems to be the case for the more highly skilled and those with vocational training; in fact the Ugandan Labor Force Survey suggests that a significantly higher share of secondary graduates are underemployed than those with more vocational 85 Simon Appleton (2001) What can we expect from Universal Primary Education? In Collier and Reinikka. 86 This result is confirmed in Mugume and Canagarajah (2005), who find convexity in the earnings function. 114 training87. Economic efficiency would suggest the need for investments in tertiary education, where returns are higher, but before we can conclude this it is important to understand why returns to education have stopped rising in a growing economy. 3.50 Falling returns to education could be due to a number of factors. First, in any economy where the supply of human capital rises rapidly relative to the supply of physical capital, returns can be expected to fall. This is clearly the case in Uganda. Second, this result can occur through selectivity: those graduating primary education before it became universal would have had unobservable characteristics which exert a positive effect on their earnings (family support, determination, or natural ability if for instance they were selected to attend school by family heads who were rationed in the education they could afford to provide their children). Third, the fall may be due to adverse selection; poorer children are now going to school and on average leave with lower attainment. Fourth, if low returns to primary education are the result of falling standards, investment in improving standards may be needed. On the supply side, according to Mugume and Canagarajah (2005), enrolment in secondary schools has doubled from 1996-2003, whereas the number of university graduates has increased by 66 percent during the period 2000-2003. The number of people with primary education increased two-and-a-half fold between 1992/3 and 2002/03, as did the number with some secondary (Figure 3-5). Combined with the number of workforce age people with some primary education ­ which increased by 46 percent ­ the increase in `semi-skilled' people amounted to 30 percent of the total population of working age88. Figure 3-3: The Higher Returns to Higher Education Depict a Skills Premium 100,000 ent val 80,000 Ushs) qeuit 98 60,000 adul 40,000 per on(1997/ti 20,000 verage consum A 0 1992 1999 2002/03 Postsecondary Completed secondary Some secondary Completed primary Some primary No formal 87Secondary education is still something of a luxury on Uganda ­ it could be that secondary graduates coming from higher income groups are simply better able to hold out for a longer job search. 88NB ­ the age group 10-64 is used in the poverty assessment because children from the age of 10 are typically employed by their parents on farms or in family businesses. In no way does use of this number imply by the authors of this report their acceptance of 10 year olds as being of working age; quite the opposite is true. 115 Figure 3-4: Supply of Workers with Primary and some Secondary Education has Increased Markedly 8,000 7,000 '000s)s 6,000 old Some primary 5,000 No formal years Completed primary 64 4,000 10-( Some secondary Completed secondary erskro 3,000 Post secondary w 2,000 of.o N1,000 - 1992/93 1999/00 2002/03 3.51 Low demand for educated labor relative to its supply is another possible cause. Decreasing returns to investment from some other cause could be reducing the returns to skills. We already know that in the period from 1999 real prices for farmers fell, and many rural workers diversified into household enterprises. Figure 3-5 looks at a consumption proxy for earnings from non-farm employment89. What stands out is that between 1999 and 2002/03, non-farm earnings have increased in real terms on average only in Kampala. Within Kampala, earnings increased most for wage earners. Wage earners in other urban areas outside of Kampala actually saw their earnings drop in real terms between 1999-2002/03. This suggests there is something unique about returns in Kampala relative to the other municipalities in districts in Uganda. We suggest this is a problem of inadequate infrastructure outside of the city, which has become the binding constraint for Uganda as growth in the economy has continued in the 1990s. 3.52 Government should take a phased and targeted approach to expanding secondary education. Basic education does not appear to be Uganda's binding constraint for growth, hence Government should phase in its expansion, or risk increasing youth underemployment. A first priority is to seek to get more efficiency and effectiveness from primary education in order to expand post-primary. Second, Government should consider the transition to work from post- primary education; an assessment should be made of the appropriate supply of vocational secondary education. Third, Government should look into technical skills gaps and higher education to anticipate what the labor market demand might be in future. In the short-term growing firms in Uganda can, and do, import skilled labor. 89Income data are not available for Uganda in the 2002/03 UNHS. 116 Figure 3-5: The Higher Returns to Higher Education: Evidence of a Skills Premium? Earnings from Employment: Kampala vs other Urban (Consumption per adult Equivalent 1997/8 prices) 120,000 Informal Sector Kampala 100,000 Informal Secotr 80,000 Other Urban Wage Earners 60,000 Kampala Wage Earners 40,000 Other Urban Formal 20,000 Informal 0 1992 1999 2002/03 Is It Poor Infrastructure? 3.53 Symptoms observable in Uganda include: · internal provision of infrastructure services is common; companies run their own transport fleets, and firms generate their own electricity; · load shedding is now routine; · industry has a low reliance on electricity; · traffic bottlenecks are part of daily life in Kampala; · transport costs are high; · Uganda has very low infrastructure indicators, especially for road transport and electricity; · fuel represents a high share of vehicle operating costs; · volatile prices for agricultural produce ­ in part due to a lack of storage; · rapid growth in telephony, but with expensive external communications; · relatively healthy returns to capital, which exceed the costs of borrowing; Evidence of High Indirect Costs: 3.54 Ugandan firms face some considerable cost disadvantage; from Uganda being landlocked, and from high energy and indirect costs90. Using the RPED surveys, Eifert et al. (2005) calculates both gross value- added (sales less raw materials) and net value-added (gross value-added less energy and indirect costs). Indirect costs include land rent, transportation, telecom, water and further operating expenses. In high growth countries such as China (85 percent) and India (71 percent), net value-added is a high share of gross value-added, indicating 90This will be taken forward in a CEM follow-up study for the Uganda using UBI data. 117 that indirect costs are a low burden. Table 3-14 shows the relative importance of infrastructure charges to total costs for a sample of registered firms. Table 3-14: Share of Infrastructure Charges in Total Cost, by Sector (2000/01 List) Water & Post & Electricity Sewerage Transport Telecom Agriculture 0.2% 0.9% 10.3% 0.1% Fishing 0.0% 0.0% 4.3% 0.1% Mining & Quarrying 5.6% 0.0% 7.0% 0.6% Manufacturing 7.4% 0.5% 9.7% 1.9% Utilities 6.6% 0.5% 14.5% 1.9% Construction 1.6% 0.4% 22.4% 2.6% Trade 2.6% 0.3% 6.2% 12.3% Hotels 2.8% 6.3% 12.8% 2.3% Transport 1.0% 0.4% 16.3% 6.1% Communications 0.7% 0.1% 11.6% 7.4% Finance and Insurance 0.6% 0.1% 6.2% 2.8% and Business Services 2.3% 0.8% 8.3% 4.2% Personal Services 2.1% 0.3% 4.0% 2.1% Total 3.0% 1.7% 9.1% 5.9% Source: UBI 3.55 Despite getting more efficient in input use, Uganda's firms face relatively high raw material costs, which may be due in part to high transport and storage costs. We saw in Chapter 1 (section 1.92) that the real value of input use in Uganda's manufacturing firms increased at one third of the value of output during the 1990s, suggesting substantial efficiencies in production. Nevertheless, Eiffert et al.(2005) find that Uganda lags behind somewhat in gross TFP, indicating relatively high materials costs. These perhaps result ­ for the import content - from Uganda's landlocked status, Uganda's dependence upon the inefficient port at Mombassa, and the notoriously inefficient rail service from Mombassa to Kampala. The domestic component of raw materials will also embody a high transport cost element, driven in part by high fuel costs and inefficiencies in logistics (the coordination failures noted above). 3.56 Ugandan manufacturing firms also face disproportionately high indirect costs relative to other countries, largely due to energy costs. The ratio of gross TFP to net TFP in Uganda is 63 percent. This is better than Kenya (42 percent) and Nigeria (51 percent), but lower than Tanzania and the more successful exporting countries. Eifert et al. (2005) then estimate a production function to account for these differences, and assess the TFP measures against the benchmark of China. Uganda's productivity measured by gross TFP is about 47 percent of China's, and is far lower than the best African performers Kenya or Senegal (Figure 3-6). However, including energy and indirect costs in production function, Uganda's productivity performance falls to 30 percent of China's. Lastly, the authors decompose the cost structure and profits for their sample countries and argue that a reduction of indirect costs will have a larger effect on profit margins than sharply reducing labor costs for most African countries, including Uganda. 118 Table 3-15: Net Versus Gross TFP, Adjusted Prices Source: Eifert et al. (2005) Evidence of Poor Electricity Provision 3.57 Electricity use by households in Uganda is stunningly low, but even worse in rural areas (Table 3-16). Only 8.6 percent of households nationally, reported having any access to electricity for lighting in the 2002 Population Census. Just over 3.2 percent of the total population reported access to modern cooking fuels, which is a startling number for a natural- resource dependent poor country with the world's third highest population growth rate. 3.58 Electricity self-generation is the norm for large firms, and is costly, especially at current world oil prices. Reinikka and Svensson (2002) showed using a Ugandan RPED survey from 1998 that manufacturing firms subject to poor public capital- specifically firms that suffer from power supply problems- are not able to make productive investments. Firms with their own generators did not invest significantly in further productive capacity, and in firms without generators, the investment rate was negatively influenced by the number of lost days due to power outages. Finally, a firms' probability of owning a generator was significantly positively correlated with the number of days of power outages. We could not replicate this picture with data from the 2002/03 firm survey, since there was no clear link between outages and generator ownership, and those who owned and used a generator did not invest any less than those without. However, since 1998, generator ownership has increased - over 60 percent of medium, and 93 percent of large firms now have a generator compared with 45 and 92 percent, respectively, in 1998. 119 Table 3-16: Household Fuel Use Fuel For Cooking Total Rural Urban Fuel For Lighting Total Rural Urban Electricity 1% 0% 4% Electricity 8% 3% 39% Gas 0% 0% 1% Gas 0% 0% 0% Paraffin 1% 1% 4% Paraffin (lantern) 11% 9% 25% Charcoal 15% 7% 67% Paraffin (tadooba) 75% 82% 33% Firewood 82% 91% 22% Candle 1% 0% 2% Cow Dung / Grass (reeds) 0% 0% 0% Firewood 5% 6% 0% Bio gas 0% 0% 0% Cow Dung / Grass (reeds) 1% 1% 0% Others 0% 0% 2% Other 0% 0% 0% Source: Population Census, 2002 3.59 Firms that self-generate face high costs of generator fuel, and invest less as a result. There is some evidence91 that larger firms and exporters face higher generator fuel costs. Furthermore, there is a positively significant relationship between investment scaled by capital and generator costs. The inclusion of sector and regional dummy variables yields unsurprising results.92 Energy intensive industries such as plastics face higher costs of running generators, whereas low energy intensive industries such as paper & publishing and wood, have lower costs. Finally, firms that operate in the central region with a more adequate infrastructure and reliable electricity have lower costs of generator fuel than firms in other regions. Overall, the estimations are able to explain 60 percent of the cross-sectional variation of the costs of running a generator. 3.60 Running a generator is between 2 and 6 times more expensive than obtaining electricity from the public grid. For the full sample of firms that owns generators and for which data is available, running a generator is three times more expensive than obtaining electricity from the public grid. For smaller firms that employ up to 25 people self-generated energy is five times more expensive. Not surprisingly, for larger firms the price difference becomes smaller even though it is still fairly substantial. 3.61 The costs savings for Uganda's firms from being able to substitute self-generated electricity for public grid electricity are huge. This is true even though tariffs for grid- purchased electricity in Uganda are not cheap. Table 3-17 presents some preliminary results for the savings which firms could have achieved in 2002/03 if they could have substituted electricity generated with electricity from the public grid. Smaller firms would have saved USh 7 million on average a year for 10 percent less self-generated electricity. Large firms with more than 100 employees would have saved on average USh 40 million per year. 91Hesse (2006) for this project estimates a basic OLS equation with robust standard errors to avoid heteroskedasticity. The dependent variable is the log cost of generator fuel regressed against various determinants. 92In the RPED survey firms operate in the following sectors: Agro industry, chemicals, construction, furniture, metals, paper & publishing, plastic, textile and wood. 120 Table 3-17: Costs of Running a Generator Versus Obtaining Electricity from the Public Grid Small Medium Large Full sample % of self generated electricity 21.82 15.56 22.76 20.00 Average costs of 1% firm electricity from the public grid (USh) 169,505 613,203 2,959,072 1,061,348 Average fuel costs for generating 1% of firm electricity (USh) 900,165 3,359,242 6,951,427 3,267,312 How much more expensive is self- generated electricity? 5.31 5.48 2.35 3.08 Cost Savings if 10% less self-generated electricity (USh) 7,306,594 27,460,394 39,923,550 22,059,640 Source: World Bank, RPED Surveys; Own Calculations 3.62 Extrapolating to 2005, the potential savings from medium and large firms become still more significant. Updating with 2005 data on Ugandan diesel and public grid energy prices indicates that the cost savings for large firms from switching 10 percent of their generated energy to the public grid would be around USh 64 million. This exercise is crude and simple since it excludes employment costs and other factors. But the findings strongly suggest that the high costs of running a generator in response to poor electricity availability might put Ugandan firms at a severe competitive disadvantage. Table 3-18: Costs of Running a Generator versus obtaining Electricity from the Public Grid in 2005 Small Medium Large Full sample Average costs of 1% firm electricity from the public grid (USh) 215,272 735,843 2,337,667 1,099,237 Average fuel costs for self- generating 1% of firm electricity (USh) 1,143,209 4,266,237 8,828,312 4,149,486 How much more expensive is self- generated electricity? 5.31 5.80 3.78 3.77 Cost Savings if 10% less self-generated electricity (USh) 9,279,374 35,303,942 64,906,454 30,502,495 Source: World Bank, RPED Surveys; Own calculations of extrapolating the 2002 figures to 2005 by including changes in diesel prices and energy tariffs. Evidence from the petroleum sub-sector 3.63 High petroleum prices also drive up indirect costs in Uganda (Table 3-19). Further work is needed to determine the drivers of high petroleum product prices in Uganda, and how best to reduce them through public action. It is unclear for instance how much of the cost of fuel in Uganda is driven by high margins, high taxes, diseconomies of scale in importing, and the costs of having neither a pipeline from Eldoret to Kampala, nor a railway capable of competing with it93. What is clear is that there are high social costs in terms of maintenance and road congestion from the current practice of trucking petroleum products into Uganda. 93The best source of information is included in the PDC Consultants Report ­ which concludes that high retail margins, and high taxes are responsible for high pump prices in Uganda. 121 Table 3-19: Trend of Petroleum Prices for a Slecetion of Countries and Regions Country/region Super gasoline (US$ per liter) Country/region Diesel fuel (US$ per liter) 1998 2000 2002 2004 1998 2000 2002 2004 Low income 0.47 0.61 0.55 0.79 Low income 0.31 0.44 0.41 0.66 Lower middle 0.46 0.49 0.50 0.65 Lower middle 0.27 0.35 0.34 0.44 income income Middle income 0.46 0.55 0.54 0.68 Middle income 0.29 0.41 0.39 0.55 Sub-Saharan 0.53 0.65 0.64 0.85 Sub-Saharan Africa 0.41 0.47 0.51 0.77 Africa World 0.50 0.61 0.58 0.78 World 0.34 0.45 0.44 0.64 Burkina Faso 0.68 0.68 0.83 1.18 Burkina Faso 0.50 0.46 0.62 0.94 Zambia 0.53 1.00 0.72 1.10 Zambia 0.49 1.00 0.60 0.98 Burundi 0.72 1.01 0.58 1.04 Burundi 0.66 0.71 0.54 1.08 Uganda 0.86 0.86 0.83 1.02 Uganda 0.68 0.75 0.70 0.88 Rwanda 0.72 0.89 0.84 0.98 Rwanda 0.72 0.84 0.84 0.99 Malawi 0.51 0.69 0.66 0.95 Malawi 0.45 0.68 0.62 0.88 Tanzania 0.63 0.75 0.67 0.93 Tanzania 0.57 0.73 0.61 0.87 Congo, Dem. 0.50 1.00 0.70 0.92 Congo, Dem. Rep. 0.50 0.93 0.69 0.81 Rep. Kenya 0.70 0.71 0.70 0.92 Kenya 0.54 0.60 0.56 0.76 Ethiopia 0.36 0.46 0.52 0.60 Ethiopia 0.25 0.27 0.32 0.42 Source: World Bank's World Development Indicators, 2005 3.64 The price of alternatives to modern fuel is rising much less slowly than the price of modern fuel. One would expect some substitution in demand from traditional to modern fuels as incomes rise. But set against this is the price effect caused by the exogenous shock to world prices of paraffin and LPG, and the recent increases in electricity tariffs. The net effect would probably be to increase demand for traditional fuels (ie wood and charcoal), as would increasing household size, especially amongst poor households. However, monthly data from the CPI suggest generally stable prices for fuel wood. Charcoal in the likely face of increased demand. Prices to the West of Kampala actually seem to be trending down in nominal terms. Either fuel wood is being burnt more efficiently, or supply has been increasing. As a natural resource based economy with low fertilizer use, depending upon the source of wood, a more rapid rate of wood burning could potentially reduce farm productivity in Uganda. 122 Figure 3-6: Nominal Prices for Traditional Fuels 1990-2005 350 KLA Charcoal 300 Mas Charcoal Mbarara Charcoal 250 Mbarara Fire wood Mbale Charcoal 200 Jinja Charcoal Series9 150 Poly. (Mbarara Charcoal) Poly. (Mas Charcoal) 100 Poly. (KLA Charcoal) Poly. (Mbarara Fire wood) 50 Poly. (Jinja Charcoal) Poly. (Mbale Charcoal) 0 91 1990 1990 29-t 5 8 9 93 94 96 97 00 01 01 02 04 05 ov- ec- ov- ug- N Oc p-eS -9l r-9 30-t ug- Ju A un-J r-9a MonthAp May- M b-eF n-aJ D N Oc p-eS A Source: Consumer Pice Index Data from UBOS Evidence from the transport sub-sector 3.65 Being both a landlocked, hilly, and agrarian country, Uganda's transport sector needs to be highly efficient to offset the competitive disadvantages of geography and topography. Instead, transport costs in Uganda seem to be high. Rail services are poor and unreliable, forcing many exporters to use road links to Mombassa. In some cases transport costs can be 50 percent of the value of goods depending on bulkiness and weight94. Ugandan firms on average lost 1.8 percent of domestic sales and 1.1 percent of exports because of delays in transportation services. 23 percent of non-exporting firms cited transport as a major or severe obstacle to business, compared to 35 percent of exporters. 3.66 Uganda's international freight transport rate95 is high even among land-locked countries. From a group of countries including SSA (Figure 3-8), only Ethiopia had a higher freight transport rate than Uganda during 1999-2003. All the other landlocked countries shown-- Malawi, Mongolia and Kazakhstan--had lower freight transport rates. The contrast is even starker when compared with exporting countries in Asia. 94 Siggel and Ssemogerere (2004), Uganda's Policy Reforms, Industry Competitiveness and Regional Integration: a Comparison with Kenya, Journal of Trade and Economic Development, 13:3. 95This is an extract from the World Bank DTIS report ­ volume 1 chapter 6. The freight transport rate is given by (freight credit + freight debit + other transportation services credit + other transportation services debit + insurance credit + insurance debit) / (merchandise exports + merchandise imports), using IMF Balance of Payments Statistics. 123 Figure 3-7: Freight Transport Rates of Selected Countries (in percent) 25.0 1999 2000 2001 2002 2003 20.0 15.0 10.0 5.0 0.0 Kenya Uganda Tanzania Ethiopia Ghana Malawi South China Thailand BangladeshKazakhstanMongolia IndonesiaPhilippines Africa Source: DTIS(2006), based on IMF Balance of Payments data 3.67 Rail services are a major cause of high external transport costs. Although there is enough track and rail ferry capacity to carry all of Uganda's export trade by rail, only 27 per cent of Uganda's exports actually go by rail. Operations are inefficient, and the availability and reliability of rolling stock and infrastructure are poor. Estimates of average wagon cycle time for Port-Kampala-Port on the two rail corridors in the period 2001-2005 are about 34 days for the Northern (Kenya) Corridor, double or three times what is easily achievable, and 43 days on the Central (Tanzania) Corridor. In contrast, imports to Kampala by road take only 7 days from Mombassa or Dar-es-Salaam, and exports on the Northern Corridor take about 3 days to Mombasa.96 3.68 Poor rail services have created a costly imbalance between import and export trade. New high-value non-traditional exports go out by air, yet the aircraft arrive half empty, meaning the cost of air freight are borne disproportionately by exporters. By contrast, Ugandan trucking firms lose out to their Kenyan competitors from the deficit of exports over imports in road transportation via Mombassa. Transporters based at Mombasa have the advantage that they would normally be at their home base waiting for import traffic, while trucks from Uganda with exports to Mombasa would have to wait for return cargoes without a confirmed booking in Kenya. 3.69 Since around 90 percent of freight is carried by roads, rising road transport costs could put a brake on economic growth. Poor access and quality of the domestic road network is one driver of high costs. In the 2002-household survey, two thirds of households reported having access to a usable road year round, but only 17 percent report their closest feeder road in good condition, and a mere 3 percent report their closest community road to be in good condition. Whereas the density of the total road network is good, the paved network density is among the 96Most imports move in organised convoys unlike exports, and this is the main reason why land transit times for imports are higher than for exports on the Northern Corridor. 124 lowest in the Africa region. At 7 percent, the percentage of paved roads in Uganda is one tenth that of Korea in 1999. Furthermore, the percentage of paved roads in Kampala is just 12 percent. Within the road sector, national roads make up 15 percent of the network but carry about 80 percent of the total traffic. 3.70 Vehicle prices and vehicle operating costs are too high. In fact vehicle operating costs are 50 percent of total transport costs in Uganda, which is far above the global benchmark of 30 percent. Increasing traffic congestion around Kampala, the transport hub for most of Uganda's trade, is one of the drivers of high vehicle running costs. Others include the high price of petroleum products, high CET tariff on freight vehicles, and delays in VAT refunds to transporters. Road transport costs are further raised by high operating costs resulting from old age and poor quality of vehicles, some of which are arguably imported because of high tariffs on newer trucks, and the long wait times from storage and coordination problems referred to earlier. Diagnosis 3.71 We do not see low average returns to capital from the 2001/02 Uganda Business Inquiry. Low infrastructure would normally reflect in low investment returns for firms and industries which are infrastructure-intensive. On the contrary, weighted returns to capital97 averaged around 21 percent for formal sector firms in 2001/02 prices. Figure 3-8: Returns to Capital in Manufacturing Sector versus Infrastruture availability by District 100% 90% 90% 80% 80% 70% Returns To Capital In 70% 60% Manufacturing 60% 50% Electricity for lighting 50% 40% 40% 30% 30% Access to Tap / Piped Water 20% 20% 10% 10% 0% 0% A E A RA L R JINJA PAL LI A KISO R KONO MBA U WA KAM M MBA 97Measured as profit as a share of market value of capital stock ­ as measured in UBI (2004). 125 3.72 Yet, there is very wide variance in returns to capital across regions within a particular sector. There is wide variance across sectors as one would expect - but the variance within sectors by region is also striking. As demonstrated in Figure 3-8, returns to capital in manufacturing were noticeably higher (and indeed very high) in Kampala in 2000/01, averaging 93 percent, and curiously low in Jinja, the old industrial center, where they averaged 25 percent. Furthermore the correlation coefficient between the returns to capital in manufacturing in the main Districts where manufacturers are located, and the proportion of households using electricity for lighting is 0.69. 3.73 One likely explanation for high returns despite poor public capital is that thriving firms in Uganda capitalize on the country's comparative advantage, and adapt to cope with binding constraints. Hausmann uses the metaphor of camels and hippopotami: just as one only observes camels in climates that are scarce in rainfall, he suggests that the firms that survive in a country are those that do not rely on the binding constraint. This seems to be borne out by Ugandan firm data. As Table 3-14 shows, electricity use is a relatively low share of total costs for Uganda's firms. Transportation is a much more important cost driver ­ a constraint that firms cannot get around. Firms seem to have adapted to do without electricity, and firms and industries that require it, simply don't seem to open up in Uganda. 3.74 High returns to capital and low returns to semi and low skilled labor are consistent with the syndrome of "under-investing State". As seen above, the social returns to connecting businesses to the grid are most likely very high, especially if the power produce by the grid was to be cheaper hydro power. 3.75 Congestion in traffic around Kampala, and self-provision of electricity nation-wide are also symptoms consistent with an under-investing State. Private returns to capital are higher where public infrastructure is better, and private returns on average exceed on the high costs of capital from the banking sector. 3.76 This growth diagnostic for Uganda concludes the following: · Under-investment in infrastructure is the binding constraint to growth in Uganda; · Electricity is the number one priority ­ with major investments needed in towns outside of Kampala to expand job creation; · Trunk roads and main roads around Kampala need to be better maintained and expanded at key bottlenecks; · The costs of power and fuel are too high; · Financial intermediation is a future constraint that could quickly bind if infrastructure constraints are removed; · Coordination gaps are leading to inefficiencies in infrastructure, and seemingly skills training. 3.77 Chapters 6 and 7 analyze infrastructure and financial sectors in more detail, and provide policy recommendations to remove constraints to growth. 126 List of Priority Specific Interventions [to be completed and sequenced after discussions with client: 1. Scale up investments in power generation, overcoming coordination gaps associated with ERT, and low investment record to date; prioritize and sequence investments according to which are on the critical path ­ for instance, improving internal infrastructure is a priority, but it might not lead to higher processed exports if Mombassa port and the transport links to it remain poor. 2. Scale up electricity connections to rural growth centers once the base load for existing customers is adequate; Government might also consider franchised Infrastructure Enhanced Zones in the short term, until problems with generation and transmission from the grid are fixed. These could make available a full complement of infrastructure services, including a stand-alone reliable power supply (any spare capacity could be sold to the grid), dedicated telecommunications, a container depot, a freight handling facility; 3. Develop a strategy to reduce price of modern fuels, prioritizing a low cost solution to the Eldoret-Kampala oil pipeline. 4. Improve the efficiency of traditional fuel use by households, and the efficiency of charcoal production. 5. Accelerate modernization of the railway and ensure rail concessionaire is capable of petroleum importation to compete with pipeline; 6. Scale up road investments in Kampala and on trunk roads, first by addressing backlog of maintenance then by increasing paving. 7. Reduce burden of corruption on exporters and medium-sized firms, targeting URA for anti-corruption effort. 8. Formalize a transparent and contestable aid-financed fund for reducing discovery costs for exporters, building on USAID's IDEAS project. 9. Work more formally, collaboratively, and transparently with export associations to identify any coordination gaps, and to share latest technologies. Given the different specificity of production methods, it will be important to approach each separately rather than all together to avoid recommendations that are common to all, but binding to no-one in particular. 10. Cautiously roll out the expansion of post-primary education, considering the demand for vocational skills and the transition from post-primary education to the workforce. Identify likely labor market demands for skills, and target strategic skills and technology capabilities which will be required for Uganda to become internationally competitive in its areas of comparative advantage. These are likely to include skills in renewable power generation, storage and transport logistics, telecommunications and ICT, road maintenance, accounting and banking. 11. Implement reforms to increase competition within the banking sector (see chapter 7). 127 4. PATTERNS OF GROWTH AND PUBLIC SPENDING IN UGANDA: ALTERNATIVE SCENARIOS FOR 2003-202098 A. SUMMARY AND CONCLUSIONS 4.1 In this chapter, we explore the economywide effects of alternative scenarios for growth patterns, domestic policies and foreign aid during the period 2003-2020. The analysis is based on the Ugandan version of an economy-wide simulation model ­ MAMS (Maquette for MDG Simulations). 4.2 The BASE scenario, which is designed as a business-as-usual scenario, is calibrated to an annual GDP growth rate at just above 6 percent. As a general rule, government services and related spending also grow at a rate of 6 percent. Foreign aid (government grants and borrowing) grows roughly at the same rate as GDP, leaving the foreign debt ­ GDP ratio unchanged except for a slight downward adjustment in response to foreign debt relief. Under these circumstances, Uganda is able to achieve MDG 1 (reducing the poverty rate by half or more) and make considerable progress on other MDGs (in health, education, and water). The fact that the country still falls short of achieving the other MDGs largely reflects that real growth in MDG-related services is lower than what is required according to Uganda's needs assessment (UNMP 2004). 4.3 The scenarios are divided into two sets. In the first set, we address the consequences of alternative growth patterns. In the second set, we explore issues related to alternative allocations and levels of government spending, improvements in government efficiency, and different levels of foreign aid. 4.4 More specifically, the first set of simulations address the question: what is the impact of more rapid total factor productivity (TFP) growth targeted on a subset of the economy ­ agriculture, industry or transportation services ­ or affecting the entire private sector? The results indicate that the impact of stronger TFP growth tends to be benevolent ­ across the different simulations, which differ in terms of sector targeting, all household categories and most factors gain in welfare and wages or rents. Nevertheless, the results indicate that the factors that are concentrated in the sectors for which TFP growth expands see their wages improve if the targeted sector is "dynamic," with growth supported by strong demand in domestic or foreign markets. This finding is particularly pertinent to agriculture. This is because at least part of domestic demand (that of the household sector) has relatively low income elasticities for agriculture and would, over time, switch their consumption basket to other products as incomes improve. Somewhat paradoxically, unless agricultural exports easily expand, the returns to factors that are intensively employed in agriculture grow more rapidly if TFP grows more rapidly in non- agriculture or across all sectors.. Given that the most of the poor live in rural areas and depend on agriculture, a multi-faceted approach to poverty reduction may therefore encourage growth in (i) agricultural subsectors with good export prospects; (ii) industries that use processed agricultural products; and (iii) the parts of the transportation and other services sectors that facilitate low-cost access to agricultural markets abroad and at home and to inputs used in agricultural production. In 98 This chapter draws on a background paper for the Country Economic Memorandum prepared by Lofgren, Hans and Diaz-Bonilla, Carolina "Patterns of Public Spending: Alternative Scenarios for 2003- 2020. 128 contrast to this, poverty reduction would not be served by higher productivity in the production of basic foodstuffs destined for domestic markets. For operational purposes, these general conclusions need to be complemented by more detailed analyses at a more disaggregated level. 4.5 In the second set of scenarios, we address questions regarding the consequences of (i) alternative allocations of government spending across different functions, including human development (HD) and infrastructure; (ii) increased government efficiency; and (iii) scaled-up programs with and without additional foreign aid. 4.6 The results indicate that, other things being equal, a reallocation of government spending from infrastructure (represented by roads and agriculture) to human development (represented by primary education, health, and water), lowers growth in GDP, private final demand (consumption and investment). Poverty increases whereas indicators of human development MDGs (education, health, and water access) improve. The scope for realizing gains in one area by shifting spending is ultimately empirical, depending on relative marginal gains from spending in different areas. The potential gains may, however, be quite limited due to interdependencies and financing constraints. For example, growth losses from less spending on infrastructure reduce government revenues, limiting the resources for increasing spending on human development sectors. The gains from this expenditure switching are only gross, since they also have to make up for reduced, complimentary household out-of-pocket spending on human development sectors. Similarly, increasing expenditure on infrastructure at the expense of human development sectors may run into diminishing marginal returns in infrastructure in the short term, and in the long run, reduce economic growth due to slow human capital accumulation arising from slower increase in the average level of education of the labor force. In our simulations, we varied growth in government recurrent expenditure on "infrastructure" (defined to include roads and government services to agriculture) relative to the BASE scenario by ±90 percent and adjusted expenditure on human development (defined to include primary education, health, and water) within the available fiscal space. To exemplify the impact, when infrastructure spending was at its lowest level, the 2020 poverty rate was 22.8 percent whereas the net primary completion rate reached 78.8 percent. In contrast, when infrastructure spending was at its highest level, the final-year poverty rate was 20.4 percent and the net primary completion rate 58.7 percent. The change in 2020 primary enrollment is less drastic, around 4.2 percent, reflecting the fact that the deterioration in part is reflected in higher repetition rates, as opposed to dropout rates, and slower progress through the system. 4.7 On the other hand, rather than reallocating spending between these high-priority and productive, it may be possible for the government to cut down on its unproductive activities and/or raise its over-all efficiency. This way, it can spend more within all the high priority areas and make more rapid progress across the full range of MDGs, although still without being on track to fully achieve them by 2015. There would still be trade-offs but it may be easier to address these in a setting where there is room for a general expansion of productive government services. 4.8 The government can become more productive, in terms of its objectives, in two related ways. First, it may be able to reallocate resources from wasteful activities, i.e., activities that do not contribute to the achievement of its objectives ­ an example would be government officials who administer programs that impose regulations with little or a negative impact on economic performance. Secondly, the government may be able to produce larger output quantities with unchanged input quantities or the same output quantities using smaller input quantities. We simulated the first case by reducing growth in "other government" activities (which include general government administration and defense) and the second case by reducing the 129 requirements of labor, capital and other inputs per unit of out produced throughout the government. In both cases, the resources that are released permit expansion in all high priority areas, adding to economic growth and private consumption and leading to more rapid progress across the full range of MDGs, also in human development areas. According to our simulations, a 50 percent reduction in the growth of other government services to 3.1 percent per year or, alternatively, a percentage increase in government efficiency, registers considerable progress along several fronts: the poverty rate declines by 2-3 percent-age points (an 8-11 percent decrease in the number of poor); the net primary school completion rate increases by 15 percent-age points (a 20 percent increase in the number of within-cohort primary school graduates); gross enrollment rates increase at higher levels of education; the under-five mortality rate decreases by 1.4-1.6 percent-age points (15-18 percent decline in the rate); and access to an improved water source increases by 11 percent-age points (17-18 percent increase in the number of people with access). 4.9 The degree to which a government economy relies on foreign aid has numerous implications, including the scope for expansion of government programs that the donors are willing to finance, the extent to which domestic taxes have to be raised and the dynamics of trade flows and foreign indebtedness. First, we simulated the impact of lower growth in foreign aid (both grants and borrowing), reaching US $1.1 billion in 2020 (as opposed to US $1.8 billion for BASE). This cut was combined with reduced spending on infrastructure and higher direct taxes. The impact was negative on all indicators except for foreign indebtedness (which by 2020 had fallen from 70 percent to 64 percent of GDP) and exports ­ a stronger export orientation may have long-run benefits. GDP growth declined, primarily due to lower growth in the private capital stock but also because of less TFP growth and less employment of labor at all educational levels (for the least educated due to more unemployment and for the more educated due to slightly less rapid supply growth). Household consumption per capita decerated by 0.5 percent-age points per year, leading to an increase in poverty. The other MDG indicators all deteriorated, but to a lesser degree. 4.10 These results suggest that additional foreign aid can contribute positively to economic objectives, especially if the new resources are targeted to services in productive areas and if they can be managed without any negative impact on over-all government efficiency. Under these assumptions, we implement a scenario with foreign aid increases in 2004-2020 that are twice as large as under the BASE scenario (reaching a total aid figure of US$2.9 billion in 2020 as compared to US$1.8 billion for the BASE scenario). The main qualitative difference between this scenario and the preceding scenarios with increased productivity is that more foreign aid permits a larger trade deficit, brought about by some combination of slower export growth and more rapid import growth, induced by appreciation of the real exchange rate. Although private consumption and incomes play a more prominent role here, the magnitute of improvements in MDG indicators is similar to the two scenarios with improved government productivity. 4.11 Finally, we test the impact of an identical expansion in government MDG services and infrastructural investments financed by higher direct taxes in the absence of more foreign aid. The main consequence is that this increase in direct taxes significantly reduces growth in household consumption and savings, reducing growth in private investment, private capital stocks and GDP. More specifically, compared to the scenario with an expansion in foreign aid, the real growth rate for household consumption declines from 5.6 percent to 5.4 percent with a slightly larger decline for private investment. At a more disaggregated level, the welfare loss is faced by the households who have to pay the taxes (in this case the richest half in rural and urban areas). On the other hand, this scenario still realizes improvements in the indicators for primary completion, enrollment at higher levels of education, under-five mortality, and water access, albeit weaker 130 than under foreign-grant financing. From a macro perspective, these outcomes point to the fact that, if government consumption and/or investment expand in the absence of an increase in the trade deficit or an increase in GDP growth, then private consumption and/or investment have to decrease. This observation confirms that, in the absence of more foreign aid or productivity improvements, a government that pursues multiple objectives faces difficult trade-offs. B. INTRODUCTION 4.12 The Uganda Government faces tough decisions about which sectors to promote and how to allocate scarce public resources across different types of spending. In this chapter paper, we used an economy wide model, MAMS99, for Uganda to analyze the long-run consequences of these choices by simulating alternative scenarios over the period 2003-2020. The scenarios were designed to shed light on the impact of alternative patterns for sectoral productivity growth and the allocation and over-all level of government spending, with alternative levels of foreign aid. In the analysis, we monitored the impact on national account aggregates, growth and trade at the level of major production sectors, factor wages and employment, and household welfare. 4.13 The main contribution of the MAMS model is that it links the provision of human development (HD) services by different providers, inside and outside government, to education and health outcomes (including a subset of MDG indicators), while considering the role of other determinants, such as growth in the incomes and consumption of households. 4.14 On a cautionary note, although we consider our qualitative conclusions as robust, one should view the specific quantitative results with caution given parameter uncertainty and the fact that the model provides a highly simplified representation of the Ugandan economy. In other words, the analysis should be considered as indicative, highlighting different issues that policymakers should consider when designing policies. It is also important to bear in mind that, given that many phenomena that are not considered also will change during the simulation period (for example world prices), the model generates counterfactual projections, not forecasts. 4.15 This chapter is divided into three parts. The first presents the BASE scenario. The second analyzes a set of scenarios based on alternative patterns of total factor productivity (TFP) growth in the private sector. The third addresses alternative government scenarios in terms of spending patterns and levels, efficiency, and availability of foreign aid. Appendix A describes the model and the database in some more detail whereas Appendix B provides additional simulation results. C. BASE SCENARIO 4.16 The main role of the BASE scenario is to provide a benchmark for comparison. Under this scenario, growth in TFP was adjusted on the margin (reaching 1.5 percent per year) to generate an annual growth rate of just above 6 percent for real GDP, i.e. a rate that is close to the actual growth rrealized during the period 1990-2003. 4.17 Government is assumed to expand its spending according to a set of simple rules. Government consumption, outside of education, are all set to grow at around 6.2 percent per 99The model on which this analysis is based is called MAMS (Maquette for MDG Simulations). MAMS is a dynamic CGE (Computable General Equilibrium) model, developed at the World Bank and designed for economywide analysis of the impact of poverty reduction and MDG strategies on the macro economy, poverty and human development (HD). 131 year.100 For primary education, Government increases its demand sufficiently to permit a gradual 80 percent increase in total (government and non-government) resources per student, an increase that is deemed necessary to make substantial progress in reaching (geographically or socially) marginal groups and to reduce repetition and dropout rates. For secondary and tertiary education, government services will increase sufficiently to maintain unchanged resources per student. 4.18 Government finances its activities from domestic and foreign sources in a manner that is designed to be compatible with macroeconomic stability. Tax collection is increased gradually, from 11 percent of GDP in 2003 to 18 percent in 2020 via increases in the effective rates of domestic direct and indirect taxes. For imports tariffs, the effective rates and the GDP share are roughly unchanged.101 Government domestic borrowing, part of which is interest-bearing, is close to 1 percent of GDP and grows at a rate that is close to the rate of GDP growth.102 The trajectory of foreign government borrowing is such that the base-year foreign debt ­ GDP ratio (71 percent) is targeted until foreign debt relief sets in (a total of US $109 million during the period 2006- 2008), at which point the targeted ratio is gradually adjusted downward (to 69.6 percent) in order to avoid a return to the initial debt burden. Foreign grant aid is assumed to grow at the same pace as GDP. As a result, total foreign aid (defined as the sum of government's external grants and borrowing) increases by 6 percent per year from US $700 million in 2003 to US $1.8 billion in 2020. For the full simulation period, most of the aid (63 percent) is in grant. 4.19 In the BASE scenario, summarized in Table 4-1, most national account aggregates and parts of government consumption or demand grow at rates of 5-7 percent. Growth in private investment is determined by the availability of financing, net of government financing needs. The required annual growth rate for recurrent government demand for primary education is quite moderate (5.1 percent per year) due to the influence of growth in non-government spending and a gradual decline to zero for the number of out-of-cohort entrants to first grade. As a result of the increase in resources per student, the net primary completion rate improves drastically, from 15.5 percent in 2003 to 76 percent in 2020.103 Given very rapid enrollment growth for secondary and tertiary education (at 8-9 percent per year), slower growth in non-government demands, and the policy of maintaining unchanged resources per student, government demand for secondary and tertiary education grows rapidly, at 14-15 percent per year. 100For the sectors where capital spending is more important than recurrent spending ­ water, agriculture and roads ­ service expansion refers to growth in the government capital stock; for the other government functions, it refers to growth in recurrent real government demand (consumption). Across all government- related functions, growth in output is proportional to growth in the government capital stock, which is determined by government investment. 101Changes in direct tax rates are free to adjust to balance government receipts and outlays. 102MAMS is a "real" model ­ nominal (as opposed to relative) prices do not matter and the monetary sector is not covered. In order to capture a key aspect of government financing, the ability of the government to obtain a certain amount of resources for free, it divides domestic government borrowing into two categories, interest-bearing , referred to as bond sales, and non-interest-bearing, referred to as Central Bank borrowing (providing resources to the government via money creation withhout any debt-servicing obligation). Bond sales are set to permit the stock of bonds to grow at the same rate as GDP. Bond sales are matched by an equal amount of Central Bank borrowing. Overall, this results in domestic borrowing growth at a rate that is close to GDP growth. Such a scenario for domestic financing should be compatible with continued macroeconomic stability. 103Given that the primary cycle in Uganda has 7 years, the 2003 net primary completion rate is defined as the product of the within-cohort entry rate in 1997 times the product of the primary passing rates in each grade during the period 1997-2003. 132 4.20 In the labor market, real wages grow for all three labor types at rates of 1-2 percent per year, most strongly for workers with tertiary education, next followed by those at the lowest education level. In conjunction with these wage developments, the rate of unemployment (which also covers underemployment) declines for all three labor segments, signaling an over-all tightening of the labor market.104 Among non-labor factors, rent (or wage) growth is particularly strong for land (given a rate of supply growth that, at 1 percent, is lower than those of other factors), while capital rents grow at the more moderate rate of 1.2 percent per year.105 Aggregate real household consumption per capita grows at 2 percent per year, with a moderate pro-poor and pro-rural bias, reflecting the composition of endowments for the different household types:106 (i) the higher the quartile to which a household belongs, the larger the income shares for capital and relatively educated labor, and the smaller the importance of the least educated labor segment; and (ii) only rural households own land, with larger per-capita endowments for the upper two quartiles. 4.21 Among the MDGs, this performance is sufficient to generate substantial improvements, not only for MDG 2 (the net primary school completion rate), but also for MDGs 1 (poverty), 4 (the under-five mortality rate), and 7a (access to safe water).107 With the exception of MDG 1 (with the poverty rate defined as a function of per-capita consumption with an elasticity of -1), however, the modeled MDGs are not achieved by 2020 (or by 2015), reflecting the fact that the growth rates for the MDG outcome determinants (including relevant services, per-capita consumption and the availability of public infrastructure) are insufficient. 4.22 At a more disaggregated level, real GDP growth is also quite uniform, in the range of 5-7 percent per year, with the above-mentioned exception for higher education. Growth in agriculture is below average due to the tendency of households to reduce the share of agricultural commodities in their consumption basket as incomes increase. GDP growth in transportation services and agriculture is influenced positively by the endogenous TFP impact of growth in government capital stocks in road and services to agriculture ­ annual TFP growth in agriculture is 2.0 percent and as high as 7 percent for transportation services, as compared to 1.7 percent for other private sectors. For all private sectors, there is also a positive relationship between TFP and increased openness (measured by the share of the sum of exports and imports in GDP); however, under the BASE scenario, openness does not change much over time. In government sectors, TFP growth is exogenous and set at zero for the BASE scenario. The distributional consequences of 104 The model allows for the existence of labor unemployment. As long as unemployment is above the minimum level, the labor supply cure is upward-sloping: employers can hire the desired quantity at a reservation rate that is inversely related to the unemployment rate. As a result, the impact of changes in labor demand is spread across wages and employment (as opposed to formulations with a fixed unemployment rate, under which wages would fully absorb demand- (and supply- )side shocks. Reliable data for un- (or under-) employment are not available. The model database assumes considerable unemployment (30%) for the least educated and largest segment of the labor force (those with less than completed secondary) and more moderate rates for those with more education (10%). The minimum rate is 5% for all segments. See Annex for Chapter 4 for more details. 105 The total (agricultural) land supply is set to grow at an exogenous rate of 1% per year, a rate that is lower than the recorded growth rate of 2.2% for the period 1990-2004 ­ this reflects our assessment that land supply growth will be considerably lower in the future. Private capital growth is driven by private investment which is determined by the availability of financing (the sum of private savings and FDI), net of government borrowing. 106 The chapter uses real household consumption per capita as the indicator of consumption-related household welfare. Equivalent variation (EV) results are computed but not reported since they generate the same results in terms of rankings of welfare outcomes across simulations and household types. 107Due to data limitations, the model does not consider the MDGs for maternal mortality and sanitation (MDGs 5 and [part of] 7, respectively). 133 alternative growth patterns are largely explained by the pattern of factor use across activities. In terms of value-added shares, land and the least educated labor segment are intensively used in agriculture, capital in industry and transportation services, and more educated labor in government services. Other private services are moderately intensive in capital and relatively educated labor. 4.23 Annual growth rates for real exports and imports, both aggregate and disaggregated, are close to 7 percent and 6 percent, respectively. Non-trade sources of foreign exchange earnings and outlays are either exogenous or policy-driven (including the policy of maintaining a certain foreign debt ­ GDP ratio), in effect defining the required trade balance (in foreign currency). The real exchange rate (which depreciates at an annual rate of 0.3 percent) influence export and import volumes, assuring that this trade balance is realized. 4.24 During the simulation period, the composition of the labor force and employment is changing toward a higher average level of education. Employment for the least educated grows at 4.5 percent per year as opposed to 4.8-4.9 percent for the two more educated labor types. In terms of the labor force (as opposed to employment), the growth contrast is sharper due to a relatively strong reduction in unemployment (or underemployment) for the least educated: annual labor force growth for the latter is at 3.5 percent as opposed to 4.5-4.6 percent for labor with higher education. In the private sectors, employment of labor and private capital grows at close to 4.5 percent per year. For government services, the growth rate is a couple of percentage points higher ­ in the absence of TFP growth, more rapid factor employment growth is needed to maintain real growth rates that are similar to those of the private sector. It should also be noted that private capital is used also in what is labeled as "government services," reflecting that, inter alia, part of health and education is owned by the private sector. 4.25 Table 4-1 summarizes the changes in economic structure between 2003 (according to the model database) and 2020 (according to the BASE simulation). In 2003, private services represented the largest part of value-added (38 percent), followed by agriculture and industry (both at 23-24 percent), with the remaining part of GDP (15 percent) represented by government services. The fact that, for industry, as opposed to the other sectors, the production share is larger than the value-added share indicates that it has stronger backward linkages to other sectors (value-added is a smaller share of its output value). Agriculture remains dominant in terms of employment whereas trade (both exports and imports) is dominated by industry. Under the BASE simulation, the structure in 2020 is quite similar. The main change is a growing importance for government services in terms of shares in value-added, production, and employment. This change, however, is driven by model assumptions, some of which are changed in model simulations. 134 Table 4-1: Structure of Uganda's Economy in 2003 and 2020 (base simulation) Nominal share (%) of value- prod- employ- exports in imports in added uction ment exports output imports demand 2003 Agriculture 24.0 18.1 69.3 24.7 8.2 3.4 3.3 Industry 23.3 34.8 7.3 48.1 9.2 52.6 23.7 Petroleum 8.5 100.0 Transportation 4.0 3.5 1.9 13.2 37.7 Other private services 34.0 29.6 15.7 26.1 6.6 6.9 3.8 Primary education 2.3 1.9 0.8 Secondary education 1.9 1.6 0.5 Tertiary education 2.0 1.6 0.5 Health 2.2 2.5 0.9 Government water services 0.0 0.0 0.0 Government agriculture services 0.1 0.1 0.0 Government road services 0.3 0.3 0.1 Other government services 6.0 5.9 2.8 1.1 1.4 0.5 1.4 Non-competitive imports 14.9 100.0 Total 100.0 100.0 100.0 100.0 6.7 100.0 16.5 2020 Agriculture 22.5 16.9 67.9 22.5 10.2 3.3 3.3 Industry 21.8 34.5 7.2 49.9 12.0 53.8 23.8 Petroleum 9.4 100.0 Transportation 1.9 2.4 0.9 10.3 39.0 Other private services 32.0 28.1 15.5 26.8 8.7 6.7 3.8 Primary education 2.4 1.9 0.9 Secondary education 4.0 3.1 1.1 Tertiary education 3.9 3.0 1.1 Health 2.9 2.8 1.2 Government water services 0.0 0.0 0.0 Government agriculture services 0.1 0.1 0.1 Government road services 0.4 0.3 0.2 Other government services 8.1 6.9 3.9 0.7 0.9 0.6 1.4 Non-competitive imports 16.0 100.0 Total 100.0 100.0 100.0 100.0 8.4 100.0 16.2 Difference (2020-2003) Agriculture -1.5 -1.3 -1.5 -2.1 2.0 -0.2 0.0 Industry -1.5 -0.4 -0.1 1.8 2.8 1.2 0.1 Petroleum 0.8 0.0 Transportation -2.1 -1.0 -0.9 -2.9 1.3 Other private services -2.0 -1.5 -0.2 0.7 2.1 -0.2 0.0 Primary education 0.2 0.0 0.1 Secondary education 2.1 1.5 0.6 Tertiary education 2.0 1.4 0.6 Health 0.7 0.3 0.3 Government water services 0.0 0.0 0.0 Government agriculture services 0.0 0.0 0.0 Government road services 0.1 0.1 0.1 Other government services 2.1 1.0 1.1 -0.4 -0.5 0.1 0.0 Non-competitive imports 1.1 0.0 Total 0.00 0.00 0.00 0.00 1.63 0.00 -0.26 135 D. ALTERNATIVE PATTERNS FOR SECTORAL PRODUCTIVITY GROWTH 4.26 The first set of experiments, defined in Table 4-2, addresses the impact of alternative patterns of accelerated productivity growth for the private sector, which, as noted above, is disaggregated into agriculture, industry, transportation services, and other private services. The scenarios do not address the costs of achieving these productivity gains. To the extent that government resources are needed, it is implicitly assumed that these can be mobilized by reducing spending in non-productive areas. The results are summarized in Tables 4-3 - 4-6. Table 4-2: Definitions of Simulations with Accelerated TFP Growth Name Description tfp-agr increased TFP growth for agriculture tfp-agr-pt increased TFP growth for agriculture + perfect transformability of agricultural output (between exports and domestic sales) tfp-ind increased TFP growth for industry tfp-pser increased TFP growth for private services (transportation and other private services) tfp-pall increased TFP growth for all private sectors (agriculture, industry and private services) + perfect transformability of agricultural output (between exports and domestic sales) 4.27 In the first simulation of increasing TFP grpwth for agriculture (TFP-AGR), the exogenous component of TFP growth in agriculture is increased by 1.8 percent per year. This is a doubling of this component relative to the BASE scenario, under which total agricultural TFP growth was slightly higher (at 2.0 percent) due to some positive, endogenous developments (primarily expansion of roads and government services to agriculture). 4.28 As expected, more rapid productivity growth in agriculture has a positive impact on growth in total GDP and absorption (i.e., total domestic final use or demand; the sum of private and government consumption and investment), both of which record annual growth increases of around 0.5 percent (Table 4-3). Given minimal changes in government demand, the absorption increase benefits private consumption and investment. As a result of higher private consumption and incomes, the poverty rate declines significantly (by 5 percent) while minor improvements are realized for other MDGs. At the sector level, agricultural growth increases by more than 1 percent per year while other private sectors see their growth increase by 0.4-0.6 percent (Table 4-4). With higher private incomes and output levels, there is also some expansion in non-government demand for "government" services (some of which are supplied by the non-government actors), complementing government spending on these sectors.108 The initial supply increase for agriculture (due to more rapid productivity growth) puts downward pressure on agricultural output prices and value-added, reducing agricultural production incentives and factor demands. Conseqyently, wages decline for factors that are intensively used in agriculture, especially if they are immobile. This explains why land rents decline for this scenario and why the factor with the second lowest rate of wage growth is the least educated segment of the labor force. Accordingly, employment growth in agriculture is lowest for this scenario. Migration of agricultural workers to other sectors mitigates their wage losses ­ the outcome would be even less favorable for the least 108On the other hand, in education, the government may also have to increase its spending to maintain targeted resource levels per student to the extent that enrollment increases. This explains the minor increase in government spending on tertiary education under the scenario TFP-AGR. 136 educated labor if such migration were not possible to the extent assumed. Given these wage developments, welfare improvements are stronger for urban and better-off households. Table 4-3: Summary at Macro/Aggregate Level ­ BASE Versus TFP Simulations Simulations* Indicator 2003 base tfp-agr tfp-agr-pt tfp-ind tfp-pser tfp-pall bn annual growth Real Absorption 13,484 6.0 6.4 6.6 6.5 6.5 6.6 macro Private consumption 9,232 5.6 6.2 6.4 6.3 6.2 6.3 Government consumption 1,799 7.3 7.3 7.3 7.3 7.2 7.2 Private investment 1,903 6.2 6.9 7.2 7.2 6.8 7.1 Government investment 553 6.0 6.0 6.0 6.0 6.0 6.0 Stock change -3 Exports 1,463 7.2 8.1 8.7 8.1 7.9 8.0 Imports 3,088 5.7 6.3 6.9 6.4 6.3 6.5 GDP at factor cost 10,871 6.2 6.7 6.8 6.7 6.7 6.7 Real exchange rate index 100 0.3 0.3 -0.6 0.2 0.3 0.0 bn annual growth Real gov- Primary education 260.2 5.1 4.9 4.9 4.9 4.9 4.9 ernment Secondary education 75.2 14.8 14.7 14.6 14.6 14.5 14.5 services Tertiary education 85.7 13.7 13.8 13.5 13.5 13.3 13.4 Health 192.3 6.2 6.2 6.2 6.2 6.2 6.2 Water 2.9 6.2 6.2 6.2 6.2 6.2 6.2 Agriculture** 15.6 6.2 6.2 6.2 6.2 6.2 6.2 Roads** 57.6 6.2 6.2 6.2 6.2 6.2 6.2 Other government 1,109.6 6.2 6.2 6.2 6.2 6.2 6.2 % of GDP GDP in final year Government Taxes 11.3 17.4 16.8 16.3 16.1 15.7 15.5 revenue Foreign aid 11.3 10.8 10.3 9.9 10.3 10.5 10.7 '000 annual growth Factor Labor -- chi2 = 0.00 Ln(Div) Coef. Std. Err. z P>z nparcel 0.04 0.01 2.63 0.008 lnarea 0.34 0.02 16.00 0.000 highpopden 0.14 0.05 2.76 0.006 lndistkm 0.05 0.02 2.16 0.030 highmrktacc 0.06 0.05 1.05 0.295 sellonfarm 0.20 0.08 2.63 0.009 lnmktdist 0.01 0.01 2.21 0.027 mhead -0.10 0.06 -1.75 0.080 lnage 0.06 0.06 0.97 0.330 lnedu 0.04 0.05 0.93 0.354 lnhhsize -0.01 0.04 -0.22 0.825 lnwealth -0.02 0.02 -0.73 0.463 radio -0.01 0.05 -0.23 0.817 bicycle 0.00 0.04 -0.02 0.988 mobile -0.05 0.10 -0.49 0.624 droughtdam -0.24 0.07 -3.53 0.000 diseasedam -0.19 0.10 -1.93 0.054 waterdam 0.08 0.16 0.48 0.630 pestdam -0.29 0.07 -4.08 0.000 othweathdam -0.01 0.18 -0.06 0.948 otherdam -0.51 0.23 -2.21 0.027 acl1 -1.07 0.31 -3.41 0.001 acl2 -1.19 0.31 -3.88 0.000 acl3 -1.13 0.31 -3.61 0.000 acl4 -0.92 0.31 -2.92 0.004 acl5 -1.24 0.32 -3.89 0.000 acl6 -1.05 0.31 -3.37 0.001 _cons (dropped) sigma_u 0.1439 sigma_e 0.7196 rho 0.0384 (fraction of variance due to u_i) 245 Annex 2. 2: Determinants of Proportion of farm area under crop groups Variable Pcash Pcereals Plegumes Proots Pbanana nparcel 0.111 0.259 -0.036 0.279 0.017 z 1.510 2.720 -0.330 3.060 0.190 prob>|z| 0.131 0.007 0.745 0.002 0.848 lnarea 1.237 0.840 1.396 0.469 0.913 10.580 5.660 8.110 3.190 6.720 0.000 0.000 0.000 0.001 0.000 highpopden 0.736 1.205 1.067 -1.236 0.481 2.670 3.230 2.550 -3.790 1.420 0.008 0.001 0.011 0.000 0.156 lndistkm 0.037 0.016 0.100 -0.188 0.156 0.310 0.100 0.540 -1.360 1.030 0.760 0.924 0.587 0.173 0.302 highmrktacc 0.934 0.225 0.672 -0.775 1.097 3.270 0.570 1.530 -2.320 3.060 0.001 0.569 0.125 0.020 0.002 sellonfarm 2.869 -1.788 -0.153 1.929 2.456 6.870 -3.180 -0.240 3.930 4.800 0.000 0.002 0.809 0.000 0.000 lnmktdist 0.188 -0.189 -0.075 0.215 0.095 6.850 -5.030 -1.770 6.740 2.770 0.000 0.000 0.076 0.000 0.006 mhead 0.815 -0.603 -1.121 -1.306 -0.001 2.480 -1.470 -2.330 -3.110 0.000 0.013 0.142 0.020 0.002 0.998 lnage 0.956 -1.349 -1.070 -0.119 1.436 2.980 -3.360 -2.280 -0.290 3.900 0.003 0.001 0.023 0.770 0.000 lnedu 0.231 0.740 1.001 -0.095 0.210 0.910 2.320 2.690 -0.300 0.720 0.363 0.020 0.007 0.767 0.473 lnhhsize -0.326 0.337 0.020 0.843 -0.465 -1.460 1.200 0.060 2.980 -1.810 0.144 0.229 0.952 0.003 0.070 lnwealth -0.287 -0.246 -0.509 -0.261 0.143 -2.280 -1.550 -2.750 -1.640 0.980 0.023 0.121 0.006 0.101 0.327 radio -0.153 0.176 0.709 0.213 -0.103 -0.580 0.540 1.850 0.640 -0.340 0.559 0.589 0.064 0.524 0.731 bicycle 0.396 0.161 -0.019 -0.438 -0.090 1.600 0.520 -0.050 -1.410 -0.320 0.109 0.605 0.959 0.160 0.752 mobile -0.841 -0.880 -0.837 1.451 -0.258 -1.530 -1.270 -1.040 2.090 -0.410 0.126 0.203 0.299 0.037 0.684 droughtdam -1.256 1.083 0.724 -2.236 -1.268 -3.300 2.280 1.300 -4.630 -2.900 0.001 0.023 0.193 0.000 0.004 246 Variable Pcash Pcereals Plegumes Proots Pbanana diseasedam 2.431 -3.216 -2.470 0.284 -0.745 4.470 -4.720 -3.100 0.410 -1.190 0.000 0.000 0.002 0.680 0.233 waterdam -1.938 0.249 3.607 -0.846 -3.605 -2.110 0.220 2.690 -0.720 -3.430 0.035 0.828 0.007 0.471 0.001 pestdam -1.276 -0.831 -1.556 -3.285 0.733 -3.200 -1.660 -2.670 -6.510 1.600 0.001 0.096 0.008 0.000 0.110 othweathdam 0.008 1.138 1.231 -0.209 0.735 0.010 0.920 0.850 -0.160 0.650 0.993 0.357 0.395 0.869 0.517 otherdam -0.905 2.392 -3.585 -0.990 -0.518 -0.690 1.470 -1.880 -0.590 -0.350 0.488 0.140 0.060 0.553 0.728 acl1 -3.783 5.031 1.130 0.000 -14.745 -5.940 4.670 0.440 . -7.280 0.000 0.000 0.659 . 0.000 acl2 -3.909 2.945 -1.910 3.207 -16.017 -6.720 2.930 -0.760 6.940 -8.030 0.000 0.003 0.447 0.000 0.000 acl3 -1.823 2.329 -0.953 2.913 -15.140 -3.190 2.290 -0.370 6.540 -7.490 0.001 0.022 0.709 0.000 0.000 acl4 -2.517 3.615 1.639 -3.077 -14.688 -4.300 4.470 0.640 -5.020 -7.160 0.000 0.000 0.523 0.000 0.000 acl5 0.000 0.000 0.113 -3.610 -14.898 . . 0.040 -5.550 -7.100 . . 0.965 0.000 0.000 acl6 -3.821 2.796 0.464 -0.700 -13.745 -6.430 2.590 0.180 -1.340 -6.820 0.000 0.010 0.855 0.181 0.000 _cons -8.465 -4.499 0.000 -1.356 0.000 -4.870 -1.950 . -0.620 . 0.000 0.052 . 0.533 . 247 Annex 2. 3: Determinants of Proportion of Output Marketed Tobit estimates Number of obs 9137 LR chi2(63) 2055.73 Prob > chi2 0 Pseudo R2 0.0532 Log likelihood = -18306.658 Ln(Crop Share Marketed) Coef. Std. Err. t P>t cereals 5.04 2.28 2.21 0.027 legumes 1.01 2.27 0.45 0.655 roots -7.50 2.30 -3.27 0.001 vegetables 22.26 2.92 7.62 0.000 fruits -5.97 2.82 -2.12 0.034 banana -0.17 2.23 -0.08 0.938 cashcrop 24.77 2.33 10.62 0.000 acl1 -3.34 3.72 -0.90 0.371 acl2 -6.86 3.50 -1.96 0.050 acl3 -6.77 3.75 -1.80 0.071 acl4 -8.29 1.74 -4.77 0.000 acl6 1.33 3.90 0.34 0.733 radio -0.55 0.66 -0.82 0.410 bicycle 0.88 0.62 1.42 0.157 mobile -1.78 1.32 -1.35 0.179 highmrktacc 0.20 0.86 0.23 0.815 highpopden -0.19 0.81 -0.24 0.813 sellonfarm 1.30 1.27 1.02 0.306 lnmrktdist 0.23 0.10 2.29 0.022 lndistkm 0.67 0.36 1.86 0.063 mhead -1.20 0.80 -1.49 0.136 lnage -5.58 0.84 -6.66 0.000 lnedu 0.06 0.65 0.08 0.932 lnhhsize -2.94 0.57 -5.16 0.000 lnarea 3.71 0.28 13.25 0.000 lnwealth 0.50 0.32 1.57 0.116 credit 1.80 0.51 3.54 0.000 lnyieldv 1.40 0.15 9.26 0.000 season -1.47 0.47 -3.12 0.002 droughtdam -2.72 0.71 -3.81 0.000 diseasedam -1.84 0.91 -2.01 0.044 waterdam -4.17 1.41 -2.96 0.003 pestdam -2.35 0.70 -3.38 0.001 othweathdam -0.74 1.64 -0.45 0.653 otherdam -4.73 2.43 -1.94 0.052 _cons 2.21 6.82 0.32 0.746 _se 18.789 0.264 Ancillary parameter Obs. summary: 5628 left-censored observations 3509 uncensored observations 248 Annex 2. 4: Determinants of Farm production value Random-effects GLS regression Number of obs = 9936 Group variable (i): lc1code Number of groups= 94 R-sq: within 0.3303 Obs per group: min= 20 Between 0.6559 Avg= 105.7 Overall 0.3642 Max= 197 Random effects u_i ~ Gaussian Wald chi2(44)= 5048.36 corr(u_i, X) = 0 (assumed) Prob > chi2= 0 Ln (Farmprodv) Coef. Std. Err. z P>z season 0.04 0.03 1.11 0.265 mix -0.18 0.04 -4.33 0.000 lnahhsize 0.03 0.04 0.72 0.469 lnarea 0.31 0.02 15.39 0.000 iseed 0.19 0.05 3.67 0.000 nat 0.33 0.07 4.45 0.000 chem 0.25 0.16 1.54 0.122 irrigate -0.07 0.19 -0.36 0.719 lnfcapital 0.02 0.01 1.46 0.143 lnage 0.06 0.06 0.97 0.330 lnedu 0.20 0.05 4.19 0.000 mhead 0.01 0.06 0.17 0.863 droughtdam -0.26 0.05 -4.86 0.000 diseasedam -0.09 0.07 -1.30 0.193 waterdam -0.10 0.10 -0.95 0.340 pestdam 0.03 0.05 0.59 0.558 othweathdam -0.42 0.12 -3.43 0.001 otherdam -0.57 0.16 -3.52 0.000 mulching 0.12 0.05 2.24 0.025 slashburn 0.11 0.04 2.44 0.015 zerotill 0.25 0.12 2.15 0.032 compost -0.02 0.04 -0.51 0.608 beans -0.55 0.06 -8.64 0.000 gnut -0.32 0.10 -3.30 0.001 maize -0.50 0.06 -7.85 0.000 millet -1.99 0.12 -16.83 0.000 sorghum -2.15 0.12 -17.88 0.000 ipotato 1.88 0.14 13.71 0.000 spotato 1.62 0.07 21.66 0.000 matoke 1.97 0.06 31.70 0.000 matokeb 0.39 0.10 3.99 0.000 matokes 0.01 0.12 0.10 0.921 coffeer -0.68 0.10 -7.18 0.000 coffeea -1.13 0.11 -10.12 0.000 lnpartime -0.01 0.00 -4.93 0.000 highmrktacc 0.41 0.09 4.44 0.000 lnpopden -0.18 0.04 -4.47 0.000 credit -0.02 0.04 -0.45 0.653 extapplied 0.17 0.04 3.80 0.000 acl1 0.69 0.19 3.66 0.000 249 Random-effects GLS regression Number of obs = 9936 Group variable (i): lc1code Number of groups= 94 R-sq: within 0.3303 Obs per group: min= 20 Between 0.6559 Avg= 105.7 Overall 0.3642 Max= 197 Random effects u_i ~ Gaussian Wald chi2(44)= 5048.36 corr(u_i, X) = 0 (assumed) Prob > chi2= 0 Ln (Farmprodv) Coef. Std. Err. z P>z acl2 0.24 0.17 1.38 0.168 acl3 0.62 0.16 3.89 0.000 acl5 0.18 0.21 0.86 0.387 acl6 0.20 0.18 1.14 0.254 _cons 8.30 0.40 20.77 0.000 sigma_u 0.355 sigma_e 1.691 rho 0.042 (fraction of variance due to u_i) 250 Annex 2. 5: Adoption of Improved Seeds (NSDS) Number Probit estimates of obs = 8935 LR chi2(39) = 590.18 Prob > chi2 = 0 Log likelihood - Pseudo = 4732.944 R2 = 0.0587 iseed Coef. Std. Err. z P>z apcred 0.25 0.06 4.46 0.000 parsati 0.15 0.12 1.31 0.190 nosati -0.06 0.09 -0.60 0.551 crnotapp -0.02 0.04 -0.46 0.642 roadmur -0.18 0.08 -2.20 0.028 roadfeed -0.18 0.07 -2.37 0.018 roadcom -0.26 0.07 -3.52 0.000 lnroadkm1 0.00 0.00 -1.83 0.068 extvisit12m 0.60 0.04 14.65 0.000 naads1 0.05 0.01 3.26 0.001 animal 0.03 0.04 0.71 0.478 fish 0.17 0.14 1.21 0.226 lncland 0.09 0.02 4.77 0.000 lnadults 0.08 0.04 2.29 0.022 male 0.06 0.04 1.40 0.162 lnage -0.07 0.05 -1.62 0.104 nmatooke 0.09 0.03 2.58 0.010 nmaize 0.08 0.04 2.13 0.033 nccer 0.01 0.03 0.40 0.686 ngnut 0.03 0.04 0.86 0.390 nbean -0.01 0.04 -0.30 0.762 npot -0.04 0.04 -1.03 0.305 ncassava -0.08 0.04 -2.23 0.026 nrice 0.04 0.07 0.56 0.572 nsimsim -0.12 0.06 -2.07 0.038 norange -0.03 0.06 -0.55 0.582 npineapple 0.05 0.06 0.83 0.407 nmango 0.05 0.04 1.28 0.201 ncabbage 0.14 0.06 2.38 0.018 ntomato 0.14 0.05 2.84 0.004 ncotton 0.15 0.06 2.55 0.011 ncoffee 0.11 0.04 2.59 0.009 ntea -0.14 0.18 -0.80 0.424 ntobacco 0.03 0.07 0.47 0.638 ncattle 0.03 0.05 0.61 0.543 ngoat 0.04 0.04 0.99 0.320 npig -0.09 0.04 -2.11 0.035 npoultry 0.07 0.04 1.91 0.056 nmilk 0.05 0.05 0.86 0.391 _cons -0.66 0.19 -3.54 0.000 251 Annex 2. 6: Adoption of Modern Inputs (NSDS) Number Probit estimates of obs = 8680 LR chi2(39) = 1892.23 Prob > chi2 = 0 Log likelihood Pseudo = -5070.375 R2 = 0.1573 input Coef. Std. Err. z P>z apcred 0.34 0.06 5.87 0.000 parsati -0.09 0.13 -0.70 0.481 nosati -0.04 0.10 -0.38 0.707 crnotapp 0.16 0.04 4.01 0.000 roadmur -0.24 0.09 -2.82 0.005 roadfeed -0.24 0.08 -3.05 0.002 roadcom -0.28 0.08 -3.63 0.000 lnroadkm1 0.00 0.00 0.17 0.863 extvisit12m 0.80 0.05 17.28 0.000 naads1 0.07 0.01 5.17 0.000 animal 0.36 0.04 9.89 0.000 fish 0.01 0.14 0.07 0.944 lncland 0.09 0.02 5.00 0.000 lnadults 0.04 0.04 1.22 0.223 male 0.12 0.04 3.04 0.002 lnage -0.17 0.04 -3.83 0.000 nmatooke 0.15 0.03 4.40 0.000 nmaize 0.10 0.04 2.85 0.004 nccer 0.04 0.03 1.29 0.198 ngnut 0.07 0.03 2.10 0.036 nbean -0.02 0.04 -0.50 0.618 npot -0.07 0.04 -2.00 0.046 ncassava -0.19 0.03 -5.36 0.000 nrice 0.14 0.07 2.04 0.042 nsimsim -0.31 0.05 -5.70 0.000 norange 0.01 0.06 0.15 0.878 npineapple -0.04 0.06 -0.73 0.468 nmango 0.07 0.04 1.54 0.123 ncabbage 0.22 0.06 3.46 0.001 ntomato 0.21 0.05 4.21 0.000 ncotton 0.40 0.06 6.86 0.000 ncoffee 0.16 0.04 3.91 0.000 ntea 0.16 0.18 0.91 0.360 ntobacco -0.15 0.07 -2.14 0.033 ncattle 0.45 0.05 9.22 0.000 ngoat -0.06 0.04 -1.67 0.095 npig -0.06 0.04 -1.48 0.139 npoultry 0.15 0.03 4.24 0.000 nmilk 0.21 0.06 3.68 0.000 _cons 0.06 0.18 0.33 0.745 252 Annex 2. 7: Share of Produce Prices in Export Prices Producer Margins for Bananas Producer Margins for Maize 0.60 1.10 y = 0.2356e-0.0094x y = 0.6807e-0.0112x 0.50 o)itar o)itar 0.90 e eci 0.40 ic 0.70 prt 0.30 prt porx 0.50 0.20 porxe- m-eraf( 0.30 0.10 mraf( 0.00 0.10 1990 1992 1994 1996 1998 2000 2002 2004 1990 1992 1994 1996 1998 2000 2002 2004 Producer Margins for Beans Producer Margins for Coffee 1.00 1.00 )oi y = 0.518e0.0253x 0.90 )oi 0.90 rat y = 0.5713e0.0256x rat 0.80 ceirptro 0.80 0.70 0.70 ceirptro 0.60 exp 0.60 exp 0.50 arm-f( 0.50 arm-f( 0.40 0.40 0.30 1990 1992 1994 1996 1998 2000 2002 2004 1990 1992 1994 1996 1998 2000 2002 2004 Producer Margins for Cocoa Producer Margins for Cotton 0.90 0.40 )oi y = 0.8186e-0.0854x y = 0.3029e-0.0312x 0.75 0.35 rat o)itar ceirptro 0.60 e 0.30 ic 0.45 prt 0.25 exp 0.30 porxe- 0.20 arm-f( 0.15 mraf( 0.15 0.00 0.10 1990 1992 1994 1996 1998 2000 2002 2004 1990 1992 1994 1996 1998 2000 2002 2004 Producer Margins for Tea Producer Margins for Tobacco 0.14 0.90 y = 0.5194e-0.0291x )oi y = 0.0828e-0.0038x 0.80 0.12 )oi rat rat 0.70 ceirptro 0.10 0.60 0.08 ceirptro 0.50 exp 0.06 exp 0.40 arm-f( 0.04 arm-f( 0.30 0.02 0.20 1990 1992 1994 1996 1998 2000 2002 2004 1990 1992 1994 1996 1998 2000 2002 2004 253 ht ts ht ts SW Nor Ea SW Nor Ea 50-ceD a 50-tcO 50-peS alp 50-luJ laap 50-nuJ ma 50-rpA ma K K 50-raM ot 50-naJ ot 40-ceD evitaler e 40-tcO ivt 40-peS 40-luJ laer 40-nuJ ot 40-rpA ot 40-raM sd 40-naJ 30-ceD enrT ezi 30-tcO sdnerT ekot 30-peS nigra 30-luJ Ma 30-nuJ Ma 30-rpA ingra 30-raM M M 2005 gni 30-naJ g 20-ceD 20-tcO etkra 20-peS 20-luJ intekra 20-nuJ M M 20-rpA 20-raM 2001-Dec e al 20-naJ 10-ceD Jan eslo 10-tcO leaseloh 10-peS Wh 10-luJ W 10-nuJ 10-rpA 10-raM 10-naJ 00-ceD Kampala to 00 80 60 40 20 00 80 60 40 00 80 60 40 20 00 80 60 40 2. 1. 1. 1. 1. 1. 0. 0. 0. 2. 1. 1. 1. 1. 1. 0. 0. 0. 254 ht ts ht ts relative SW Nor Ea SW Nor Ea average 50-ceD 50-tcO laap a 50-peS 50-luJ alp 50-nuJ ma 50-rpA ma 50-raM regional K K ot 50-naJ ot 40-ceD e ivt 40-tcO 40-peS laer 40-luJ evitaler 40-nuJ ot 40-rpA ot 40-raM 40-naJ sd s 30-ceD Differentials: sdnerT sna 30-tcO enrT 30-peS 30-luJ 30-nuJ Price ingra Be oundnutr 30-rpA nigra G 30-raM M M g 30-naJ 20-ceD 20-tcO gni 20-peS 20-luJ 20-nuJ Wholesale intekra etkra M M 20-rpA in 20-raM leaseloh e 20-naJ al 10-ceD 10-tcO eslo 10-peS Trends W 10-luJ Wh 10-nuJ 10-rpA 10-raM 8a: 10-naJ 00-ceD 2. 00 80 60 40 20 00 80 60 40 00 80 60 40 20 00 80 60 40 2. 1. 1. 1. 1. 1. 0. 0. 0. 2. 1. 1. 1. 1. 1. 0. 0. 0. Annex ht ts SW Nor Ea ala 5002/10/10 mpa K ot 4002/10/10 evit laer 3002/10/10 ot sdnerT s 2002/10/10 1002/10/10 ingra oundnutr G M 0002/10/10 2005 g intekra ht 9991/10/10 ts SW Nor Ea M 8991/10/10 1995-Dec Jan elaseloh 7991/10/10 ala W 5002/10/10 6991/10/10 mpa K ot 4002/10/10 Kampala 5991/10/10 to evit 2.00 1.80 1.60 1.40 1.20 1.00 0.80 0.60 0.40 laer 3002/10/10 ot ht 255 ts relative SW Nor Ea sdnerT 2002/10/10 ezi 1002/10/10 average ingra Ma M 0002/10/10 ala g 5002/10/10 mpa regional K intekra 9991/10/10 ot 4002/10/10 M evit laer 3002/10/10 elaseloh 8991/10/10 7991/10/10 ot W Differentials: sdnerT 2002/10/10 6991/10/10 sna 1002/10/10 5991/10/10 Price ingra Be 0002/10/10 2.00 1.80 1.60 1.40 1.20 1.00 0.80 0.60 0.40 M g Wholesale intekra 9991/10/10 M in elaseloh 8991/10/10 7991/10/10 Trends W 6991/10/10 8b: 2. 5991/10/10 00 80 60 40 20 00 80 60 40 2. 1. 1. 1. 1. 1. 0. 0. 0. Annex pd3 0.47 0.57 0.83 0.8 0.71 0.84 0.67 0.96 0.93 0.89 0.42 0.49 0.95 0.87 0.82 0.9