Berlin Workshop Series 2008 45398 Agriculture and Development Edited by Gudrun Kochendörfer-Lucius and Boris Pleskovic Themes for the 10TH ANNUAL BERLIN WORKSHOP SERIES Berlin, Germany "SPATIAL DISPARITIES AND DEVELOPMENT POLICY" September 30­October 2, 2007 Macro-trends: Spatial Patterns of Economic Activity, Income, and Poverty New Economic Geography and the Dynamics of Technological Change-- Implications for Least Developed Countries Perspectives: Rural-Urban Transformation--Leading, Lagging, and Interlinking Places Spatial Disparity and Labor Mobility Africa Rethinking Growth and Regional Integration Learning from Europe's Efforts at Integration and Convergence Spatial Policy for Growth and Equity Agriculture and Development Berlin Workshop Series 2008 Agriculture and Development Edited by Gudrun Kochendörfer-Lucius and Boris Pleskovic THE WORLD BANK Washington, D.C. © 2008 The International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org E-mail: feedback@worldbank.org All rights reserved 1 2 3 4 11 10 09 08 This volume is a product of the staff of the International Bank for Reconstruction and Development / The World Bank. The findings, interpretations, and conclusions expressed in this volume do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgement on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development / The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; telephone: 978-750-8400; fax: 978-750-4470; Internet: www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2422; e-mail: pubrights@worldbank.org. ISBN: 978-0-8213-7127-5 eISBN: 978-0-8213-7128-2 DOI: 10.1596/978-0-8213-7127-5 ISSN: 1813-9442 Contents ABOUT THIS BOOK vii INTRODUCTION 1 Gudrun Kochendörfer-Lucius and Boris Pleskovic WELCOMING ADDRESS 11 Gudrun Kochendörfer-Lucius KEYNOTE ADDRESS Current Thinking on the Scope and Approach of WDR 2008 17 François Bourguignon OPENING ADDRESS 25 Astrid Kühl INTRODUCTORY REMARKS 27 Michael Hofmann Part I: Agricultural Policy, Subsidies, and Trade Agriculture and Development: The Perspective of a Think Tank on the WDR 2008 Outline 33 Michael Brüntrup Part II: Markets and Institutions Market Access for Small Farmers: The New Standards Challenge 41 Benoît Daviron and Isabelle Vagneron Globalization, Privatization, and Vertical Coordination in Food Value Chains 49 Johan F. M. Swinnen and Miet Maertens Part III: Agriculture, Natural Resources Management, and the Environment Designing Improved NRM Interventions in Agriculture for Poverty Reduction and Environmental Sustainability in Developing Countries 57 H. Ade Freeman V VI | CONTENTS Aspects of Land Degradation in Lagging Regions: Extent, Driving Forces, Responses, and Further Research with Special Reference to Ethiopia 71 Gunnar Köhlin Agriculture and the Rio Conventions 85 Laurence Tubiana and Sheila Wertz-Kanounnikoff Part IV: Agriculture-Rural Nonfarm Linkages and Rural Poverty Agricultural Growth, Employment, and Wage Rates in Developing Countries 97 Raghav Gaiha and Katsushi Imai Are Farm-Nonfarm Linkages Still Strong Enough to Tackle Rural Poverty? 131 Daniel Bradley Agriculture: Rural Nonfarm Linkages and Poverty 139 Alexander Schejtman Part V: Decentralization, Local Governance, and Rural Development Decentralization, Local Governance, and Rural Development 147 Jean-Louis Arcand Decentralization, Agriculture, and Poverty Reduction: Harsh Realities? 157 John Farrington Decentralization and Development: Emerging Issues from Uganda's Experience 161 John A. Okidi and Madina Guloba Decentralized Rural Governance in a Post-Socialist Economy: A Case of Community-Based Agricultural Marketing in Poland 175 Martin Petrick and Andreas Gramzow Part VI: Emerging Policy Issues Opportunities for European Bioethanol in a Global Environment (2006) 187 Oliver Henniges Nonlinear Effects of Weather on U.S. Crop Yields, Implications for Climate Change, and Why These Effects Matter for Developing Countries 193 Wolfram Schlenker and Michael Roberts Emerging Policy Issues within the Realm of Science Policy for Africa 203 Monty Jones CLOSING REMARKS 215 Alain de Janvry APPENDIXES Appendix 1: Program 221 Appendix 2: Participants 229 About This Book The World Bank and InWEnt (Capacity Building International, Germany) hold a Development Policy Forum each September in Berlin. This meeting, known as the "Berlin Workshop," provides a forum for the European research community to contribute its perspectives to early discussions in preparation of the World Bank's annual World Development Report. The Workshop offers new ideas and distinctive perspectives from outside the World Bank. Participants in the Workshop come from a range of academic, governmental, think-tank, and policy-making institutions in Europe, the United States, and the Russian Federation, as well as from the World Bank and the German development institutions. Conference papers are written by the participants and are reviewed by the editors. Participants' affiliations identified in this volume are as of the time of the conference, September 4­6, 2006. The planning and organization for the Workshop involved a joint effort. We extend our special thanks to Derek Byerlee and Alain de Janvry, codirectors of the World Bank's World Development Report 2008. We wish to thank Aehyung Kim and Marisela Monoliu Munoz for their advice and suggestions. We also would like to thank the conference coordinators, Klaus Krüger, Joachim Müller, Katja Wehlte de Hernandez, and Theresa Bampoe, whose excellent organizational skills kept the workshop on track. Finally, we would like to thank the editorial staff, especially Stuart Tucker and Cindy Fisher from the Office of the Publisher, Martha Gottron, and Grit Schmalisch for all of their work on this volume. VII Introduction GUDRUN KOCHENDÖRFER-LUCIUS AND BORIS PLESKOVIC The Berlin Workshop Series 2008 presents a selection of papers from meetings held on September 4­6, 2006, at the ninth annual Berlin Workshop, jointly organized by InWent--Capacity Building International, Germany, and the World Bank in prepa- ration for the World Bank's World Development Report (WDR) 2008. The work- shop brings diverse perspectives from outside the World Bank, providing a forum in which to exchange ideas and debate in the course of developing the WDR. Participants at the 2006 Berlin Workshop gathered to discuss challenges and successes pertaining to agriculture and development. Agriculture is the major sector contributing to economic development in many poor countries. Three out of every four poor people in developing countries live in rural areas. As globalization acceler- ates, development policies should tackle future challenges in agriculture arising from the scarcity of natural resources and globalization. In her welcoming address, Gudrun Kochendörfer-Lucius highlights the paramount importance of redefining the framework for agriculture, providing us with food for thought and putting forward suggestions that need greater reflection and more detailed discussion. Her contribution focuses on three main topics. First, she presents some considerations on global agricultural development and trade. She describes the different approaches to agricultural development, stressing that we should discuss the outcomes and effects of these approaches and evaluate which nations or which pop- ulation groups are benefiting, as this could help to develop target group­oriented strategies in poverty alleviation and agriculture. Second, she takes a critical look at how agriculture and the rural sector can be an effective engine for growth. Another issue on the agenda is to determine what agriculture needs in the way of technology, infrastructure, and financial support to become a growth engine? These new insights should contribute to an appropriate formulation and implementation of tailored Gudrun Kochendörfer-Lucius is Managing Director of InWEnt--Capacity Building International, Germany. Boris Pleskovic is Research Manager, Development Economics, at the World Bank. Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 1 2 | GUDRUN KOCHENDÖRFER-LUCIUS AND BORIS PLESKOVIC agriculture-for-development programs. Finally, she looks at development in connec- tion with systematic capacity building and training, pointing out the need to define sound capacity-building measures in terms of agriculture as well as to determine how these could be used more effectively. In his keynote address, François Bourguignon discusses how agriculture encom- passes everything that matters in development. With 70 percent of the world's poor people living in rural areas and depending directly or indirectly on agriculture, agri- culture is the center of development and poverty reduction. Bourguignon asks why and how we should invest public resources in agriculture or implement agriculture- oriented development policies to reduce poverty. In addressing these questions, he examines five issues. First, he argues that agriculture is an engine of growth and that productivity gains in agriculture have a positive impact on the whole economy, low- ering wages and facilitating development in the nonagricultural sector based on what he calls the "old" development paradigm and the spillover effects of agriculture. Second, a comparative advantage in some crops permits the local economy to become more efficient and to generate a surplus that can be invested elsewhere in the econ- omy under the "new" development paradigm associated with international trade. Third, agriculture is the sector that encompasses the largest number of the poor. Finally, he highlights the need to be clear about what issues are to be addressed by policies. In her opening address, Astrid Kühl notes that agriculture, with its achievements and problems, has taken a back seat in the European Union (EU) over the years. Fur- ther, she states that, because of agricultural subsidies and protected markets in the EU, competition does not provide a development impulse to agriculture in develop- ing countries. Kühl concludes that rural areas and the farming population in devel- oping countries offer a much suppressed and neglected potential. In his introductory remarks, Michael Hofmann notes that the majority of the poor live in rural areas and that agricultural development has received declining attention in recent years. Hofmann emphasizes the importance of making significant progress in terms of agricultural markets and subsidies in trade negotiations in order to bene- fit developing countries. He argues that there is a need for the analysis of who would benefit from more investment in agriculture, a need to examine how urban and rural areas are mutually dependent, and a need to examine migration flows related to the exodus from rural areas. Hofmann concludes that we need to look for nonconven- tional solutions and be candid in our approach and analysis. Derek Byerlee and Alain de Janvry also gave an introductory address in which they observed that the success of many countries in diversifying their economies started with major productivity gains in agriculture. Agriculture provided the basis for liveli- hoods to a majority of humanity, offered important business opportunities to private investors, and provided the cornerstone for food security. Providing an engine of growth for countries with a large agricultural sector relative to other sectors remains as vital and unique today as it has been historically. Despite all this, the world often has turned its back on agriculture too early, as though other sectors could assume its development functions more effectively. The resulting costs of underinvestment, INTRODUCTION | 3 policy neglect, and policy biases against agriculture have been staggering in forgone economic growth, unrealized business opportunities, environmental degradation, and the recurrence of hunger. The implication for growth and the development com- munity is that agriculture must play a redefined and far more significant role in national and international development agendas in order to capture the various ben- efits it offers. The content of Byerlee and de Janvry's address can be found in the overview of the World Development Report 2008. Session I, on agricultural policy, subsidies, and trade, presents emerging issues in the global trade of agricultural products. Michael Brüntrup enlists four areas for further research: (a) the role of agriculture and agricultural policy in and for development, including lessons learned from pre- vious decades and new developments in international development policies; (b) more explicit discussion and policy on the role of the public sector versus markets in agri- culture for developing countries; (c) more realism concerning agricultural economies in poor countries, especially in Sub-Saharan Africa and particularly concerning the role of risk; and (d) more attention to regional integration as an important aspect of agricultural policies in Sub-Saharan Africa. Session II, on markets and institutions, looks at institutional mechanisms to allow small farmers and developing countries to participate fully in emerging domestic markets and trade. Benoît Daviron and Isabelle Vagneron examine what institutions are needed to give small farmers access to distant consumers. Distance is an obvious "problem" in tropical trade. It means time and risk. But distance is also a "problem" for many value chains, including domestic ones in developing countries. The authors argue that the "old" system of grades and standards played a central role in giving small farmers access to distant consumers. They did so by allowing the creation of markets at the farm-gate level or close to the villages. Thereby, the historical system of grades and standards enabled small farmers to "impose" their effectiveness on large farms (plantations). Today new standards are emerging. Most of them are process oriented, unlike the old ones, which just defined some intrinsic characteristics of the products. The new process-oriented standards do not eliminate the possibility for market trans- actions at the farm gate. On the contrary, one of their "functions" is to organize mar- ket transactions (including contracts) at the farm level. More important, they are drastically changing the economics of labor supervision in agricultural production and, thus, the competition between small and large farms. Johan F. M. Swinnen and Miet Maertens argue that food and agricultural com- modity value chains in developing and transition countries have undergone tremen- dous changes in the past decades. For example, companies and property rights have been privatized, markets liberalized, and economies integrated into global food sys- tems. Liberalization and privatization initially caused the collapse of state-controlled vertical integration. More recently, private vertical coordination systems have emerged and are growing rapidly as a response to consumer demand for food qual- ity and safety, on the one hand, and factor market imperfections that constrain farm production, on the other. The authors (a) demonstrate the importance of these 4 | GUDRUN KOCHENDÖRFER-LUCIUS AND BORIS PLESKOVIC changes, (b) discuss the implications for efficiency and equity, and (c) summarize empirical findings on the effects in developing and transition countries. Session III, on agriculture, natural resources management, and the environment, focuses on institutions and technological responses to the growing scarcity of land and water resources, land degradation, loss of biodiversity, environmental pollution from intensification, and climate change. H. Ade Freeman raises serious concerns about the capacity of the natural resource base to sustain current and future levels of agricultural growth, although the use of natural resources has made significant contributions to economic growth and improvements in human well-being. Degradation of the resource base increasingly threatens the effort to reduce poverty, since poor people tend to be highly dependent on natural assets for their livelihood. He examines the links between agriculture, nat- ural resource management, and the environment in developing countries. Lessons from experience in the developing world indicate that effective technologies and insti- tutions for managing natural resources in agriculture need to recognize that natural resource management is an investment choice, take into account the linkages between the farm and nonfarm economy, emphasize participatory processes, recognize that natural resource management practices are knowledge intensive, and ensure that they are profitable. These principles imply that efforts to design and scale up interventions need to shift focus from the farm to a livelihoods framework; emphasize targeted approaches for identifying poverty pathways, strategies, and priorities; increase investments in human capital in rural areas; rethink and strengthen institutional arrangements and enabling environments that support pro-poor natural resource management interventions; and exploit the complementarities between social protec- tion and growth-promoting natural resource management. Gunnar Köhlin argues that there are good reasons why land degradation in Africa should be a focus of public intervention and international development assistance. Land degradation is at the very center of the agriculture-poverty nexus, and sustain- able land management is an important component of both poverty alleviation and sustainable development. Köhlin reviews the literature on the cost of land degrada- tion in Africa as well as the economics literature dealing with the adoption of soil and water conservation and past interventions in the sector. He argues that necessary direct interventions must be designed very carefully to local conditions and be suffi- ciently sensitive to the farmers' need for short-term profitability, while promoting sustainable land management. Laurence Tubiana and Sheila Wertz-Kanounnikoff argue that agriculture assures global food security while producing harmful impacts on the global environment. Therefore, agriculture is addressed by the three Rio Conventions--that is, the United Nations (UN) Framework Convention on Climate Change, the UN Convention on Biological Diversity, and the UN Convention on Desertification. Such international regulatory efforts can have profound effects on local agriculture. Especially the cli- mate negotiations, yet also the biodiversity negotiations, will remain important for the regulation of agriculture-related global environmental externalities in both trop- ical and nontropical countries. To meet future challenges, there is, however, a need to enhance the effectiveness of the existing scheme of global environmental governance. INTRODUCTION | 5 A precondition to this end is overcoming national interests for the sake of collective international action in favor of the global environment. Session IV, on agriculture-rural nonfarm linkages and rural poverty, focuses on policies to stimulate synergies between farm and rural nonfarm areas in reducing rural poverty and managing the transition of populations out of agriculture. Raghav Gaiha and Katsushi Imai analyze the relationships among agricultural productivity, employment, technology, openness of the economy, and inequality in land distribution, drawing on different specifications and methods of panel data esti- mation designed to make efficient use of cross-country samples. Agricultural pro- ductivity varies with technology and employment, but more so with employment. The effect of openness on agricultural productivity is ambiguous: it is positive, nega- tive, or not significant, depending on the definition of openness, specification, and estimation procedure used. A somewhat surprising result is the positive effect that inequality in land distribution has on agricultural productivity. Arguably, when credit markets are incomplete, greater inequality in land distribution may imply a more sig- nificant role for large landowners in agricultural investment through easier access to credit. Furthermore, the growth rate of agricultural employment has a strong (lagged) positive effect on the growth rate of nonagricultural employment. At the same time, agriculture's contribution to overall economic growth and generation of employment is substantial, although the share of agriculture has declined in devel- oping countries. While a case for acceleration of agricultural growth through mod- ernization of its technology, crop diversification, and exploitation of high-value export opportunities rests on more complete credit and insurance markets and infra- structural support, crop diversification is likely to have some direct negative effects on employment. In view of the growing potential for exports of fruits and vegetables and other high-value agricultural products linked to rising global income and rapid urbanization, the overall expansionary effect of high-value agricultural products may, however, outweigh any negative effects. Daniel Bradley outlines the parameters of the debate on farm-nonfarm linkages. He points out that, where agricultural growth has stagnated, the broader transfor- mation of economies has stalled. Although few countries have reduced poverty through agriculture alone, increased agricultural productivity has nonetheless been shown to spur economic development and poverty reduction in both urban and rural areas. This is due to strong linkages that continue to generate a multiplier effect on the rest of the economy despite the new, more challenging context in which agricul- ture takes place today. Agriculture is likely to remain the most effective source of growth in many poor countries, and will retain a comparative advantage to reduce poverty multipliers in transforming economies. This strengthens the case for renewed support to the agriculture sector matched to development context. Approaches that capitalize on promising sources of both farm and nonfarm growth are preferable to an either/or approach. Alexander Schejtman observes that the relationship between temperatures and yields is highly nonlinear and asymmetric: a temperature increase above the optimal growing temperature is much more harmful than a drop of equal magnitude below the optimal growing temperature. Developing countries are especially vulnerable to 6 | GUDRUN KOCHENDÖRFER-LUCIUS AND BORIS PLESKOVIC a potential increase in temperatures, as agriculture still constitutes a large fraction of gross domestic product and these countries are generally located in warmer climates. Temperature increases are more harmful in warmer climates due to the highly asym- metric relationship, which implies that yields are declining at an increasing rate. Schejtman estimates the relationship using a unique micro-level panel data set for the United States, yet the results are robust across climatic regions, which suggests that they might be applied to other parts of the world. For example, there is a highly non- linear and asymmetric relationship between corn or soybean yields and temperatures, where yields increase slowly in temperatures up to about 29°C before they quickly become very harmful. In addition, the United States is a major agricultural producer and exporter, so any effects on the United States will likely exert a powerful influence on world commodity and food prices. However, subsidies and lower food prices can have a beneficial effect if the poorest farmers consume imported food, as a result of subsidies. Whether poor countries are net importers or net exporters of food deter- mines whether they benefit from lower world food prices. This may suggest large shifts in growing regions, but not necessarily significant changes in total production. Thus the large impacts on yield may belie impacts on aggregate welfare. Regardless of how much prices are ultimately affected, it is likely that substantial changes will occur in where food is produced and how much is produced. Session V, on decentralization, local governance, and rural development, discusses the political economy of decentralization, financial systems in support of develop- ment, informal finance, and microfinance. Jean-Louis Arcand poses three key questions related to community-driven and community-based development programs of the World Bank. First, is the participa- tory nature of community-driven development programs a more effective approach to alleviating poverty than conventional, top-down approaches? Second, is elite cap- ture the key concern, or is the geographic proximity of benefits just as critical? Third, is ethnicity a key impediment to the success of community-driven development? Arcand provides answers to these three main questions based on his own work on a major community-driven development program in Senegal. John Farrington argues that, in principle, decentralization and devolution can be beneficial in alleviating poverty. However, in practice, decentralization and devolu- tion are likely to be hampered by several factors: (a) lack of clarity in the assignment of functions between central and local decision-making agencies; (b) lack of account- ability of line department staff to locally elected representatives; (c) continuing heavy reliance of local government on central funds for public services; (d) predisposition of local politicians toward "photogenic" projects; (e) increasing concentration of poverty in fragile states where local capacity is weak; and (f) elite capture to suit the ends of politicians at the federal and state levels. Farrington argues that, to improve the effectiveness of decentralization, these risks need to be safeguarded against, effec- tive local governance requires the support of strong central government and of engaged citizens, and the comparative advantage of central and local governments with respect to (especially) the provision of services needs to be identified. John A. Okidi and Madina Guloba present the experience of Uganda with decen- tralization, observing that the current decentralization in Uganda originated in the INTRODUCTION | 7 late 1980s as part of a broader effort to restore state credibility and deepen democ- racy following several years of political and economic turmoil. Using a detailed legal framework, Uganda entrenched political, administrative, and fiscal decentralization as a strategy for broad-based growth with poverty reduction. The authors highlight several issues, including empowerment of local leaders and residents, experience with local elite capture, improvement of service delivery, promotion of real sector response to improved economic environment, and enhancement of progress toward achieve- ment of the Millennium Development Goals. They also review the achievements and challenges of Uganda's decentralization corresponding to different levels of decen- tralization, which are consistent with certain levels of economic and democratic development. Martin Petrick and Andreas Gramzow argue that, in former socialist countries, development approaches built on local collective action and participation of the civil society are often poorly received among the rural population due to their negative experience with former regimes. They show how the involvement of local govern- ments and private individuals allowed the successful creation of a public-private part- nership for agricultural marketing in southeastern Poland, despite an initial hostile reception to participatory approaches. For example, the partnership buys products from 500 local farmers and delivers them processed to customers in the further ambit. The partnership provides nonfarm jobs for 400 local inhabitants and improves agricultural incomes. While the regional distribution network for agricul- tural products is largely market based, the partnership came into being only as a result of local collective action. Local government officials played a crucial role in this founding process, and part of the seed capital came from public sources. The partnership offers complementary services to farmers and has been created and sup- ported by respected local leaders. However, challenges remain for its sustainabi- lity: there has been little stimulation of broader development activities within the rural civil society, and the involvement of local governments has become unstable over time. Session VI, on emerging policy issues, highlights recent and emerging trends and challenges, including biotechnology, agricultural science in Africa, energy prices, threats of global disease epidemics, and climate change. Oliver Henniges analyzes the competitiveness of bioethanol production from both a national and an international perspective. He observes that, in Germany, as well as in the European Union, the use of bioethanol as an alternative fuel has been demanded for several years by different stakeholders and supported by political deci- sion makers. In his presentation, he compares the costs of producing bioethanol in Australia, Brazil, China, the European Union, Thailand, and the United States in 2005 by standardizing plant capacity and cost allocations. Of these countries, he finds that production costs are by far the highest in the European Union, at approx- imately 45 per hectoliter (hl), using wheat and sugar beet as feedstock and assum- ing a wheat price of 100. He emphasizes that, in the case of a continuously high price for crude oil and further expansion of biofuel production, it can be assumed that the price of crude oil will influence other agricultural commodity markets as well. This would necessarily lead to higher food prices. Thus, Henninges argues that 8 | GUDRUN KOCHENDÖRFER-LUCIUS AND BORIS PLESKOVIC the future competitiveness of bioethanol production in Europe will depend on the price of crude oil and the willingness of petrol companies to accept the blending of biofuels. In order to achieve international competitiveness of German and European bioethanol production, the author argues that further tremendous reductions in the costs of production at all stages of the process as well as creation of a long-term, reli- able legal framework are essential. Wolfram Schlenker and Michael Roberts open their presentaton with Schelling's view on how the impacts of climate change will likely weigh more heavily on the developing world while the costs of curbing climate change will likely fall more heav- ily on developed nations. They argue that, while developing countries might be espe- cially vulnerable to climate change, there are good reasons to focus research on the United States. The first is that the United States produces a large share and is one of the largest exporters of two of the world's most important staple crops: corn and soy- beans, which have a large influence on the world supply and prices of food. A sec- ond reason is data quality. The authors argue that the United States is so large and climatically diverse, the available data traverse a reasonably large range of climates. To illustrate their argument, the authors present new evidence on the relationship between weather and crop yields in the United States by using unique fine-scale weather data and a long history of observed crop yields. They found a highly non- linear and asymmetric relationship between temperature and corn and soybean yields. In all cases, yields increase gradually in temperature up to about 29°C, but above 30°C, they decline sharply. Because developing countries are predominantly located in warmer climates, they are especially vulnerable to temperature increases. Schlenker and Roberts' results suggest substantial changes in where things are pro- duced and in how much is produced. In a global economy with integrated commod- ity and food markets, these changes are of interest to both rich and poor countries. Monty Jones argues that policies have profound effects on the application of sci- ence, and agricultural science policies are not an exception. There is a near universal interest in them by both consumers who want assured low-cost and high-quality food supplies and by the public at-large who are concerned about the accompanying risks to the health and safety of food and the welfare of farm animals and the environ- ment. These different interests present difficulties in formulating policies with the right balance. At the international level, there are divergent priorities between policy makers responsible for relieving the hunger and poverty of the majority of Africans and those responsible for serving communities who are increasingly concerned with safety and welfare issues. Likewise within the continent, there are divergent views on policies governing agricultural research. Given the complex African policy environ- ment, policy makers at different levels need to formulate complementary policies that will collectively enable Africa to make the best use of science and seize the opportu- nities for and remove the constraints on African agriculture's full potential. The com- mon aim of all agricultural science policies should be the ethical and safe application of science to improve the welfare of African people, with appropriate protection for producers, consumers, farm animals, and critical global biodiversity and environ- mental resources. INTRODUCTION | 9 In his closing remarks, Alain de Janvry highlights a few lessons that were learned at the meeting. He discusses five different typologies of countries: (1) the agriculture based economies; (2) transforming economies, in which there is an urgent need to transform the economy to reduce rural poverty; (3) urbanized economies, where investing in agriculture is purely a matter of competitive advantage; (4) countries that have not performed up to their agricultural potential because the right policies have not been in place; and (5) the countries that have had poor performance, but have very weak resources in agriculture and have low potential to do better; Finally, he argues that there is a need to go beyond cross-country comparisons and look at regional differences within countries. Additionally, there is a need for a typology of rural households. He also discusses the issues of political economy, the role of the state and the issues concerning models that simulate the impact of policy reforms. He concludes with the discussion of agricultural poverty in Africa and calls for an analy- sis of a broader picture and creative thinking in order to make a difference in again elevating the role that agriculture can play in development. Welcoming Address GUDRUN KOCHENDÖRFER-LUCIUS Ladies and gentlemen, dear participants! I very much appreciate the fact that agri- culture is to be a main focus of the World Development Report 2008. Agriculture in developing countries has been out of the spotlight for many years among the inter- national donor community. Twenty-five years after the last WDR on this subject, it is absolutely essential that we redefine the framework for agriculture so that this sector can utilize its potential to make a positive contribution toward development. Never- theless, one also has to bear in mind that the context has changed tremendously and new challenges and needs now prevail. When it comes to bridging the gap between the poor and the rich in a globalized world, agriculture is a key factor, even if it cannot work any wonders alone. Rather than trying to offer a comprehensive analysis of the range of views and standpoints toward the role of agriculture in development, I will put forward some ideas and pose some questions, providing some food for thought for our two-day discussion, and in so doing, reflect the discursive notion behind the overall concept of our workshop. I will first deal with global agricultural development and trade issues. One of the extremely odd facts about agriculture and development is that, of those suffering from hunger and malnutrition worldwide, 75 percent--a staggering one billion people--actually live in rural areas, in precisely those places where food is grown. The facts and figures provided by various research institutions on the critical impor- tance of agriculture and the multifunctionality of rural areas for national economies speak for themselves--and this is a topic we will be returning to in detail during our policy workshop. This is, though, one of the reasons why leading institutions have long been calling for us to redirect our focus toward agriculture and the resources allocated both to agriculture and the rural areas. As research findings show, invest- ing development funds in this area provides acceleration on the path to achieving the Millennium Development Goals (MDGs) and even helps us move beyond them. Gudrun Kochendörfer-Lucius is Managing Director of InWEnt--Capacity Building International, Germany. Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 11 12 | GUDRUN KOCHENDÖRFER-LUCIUS Given this context, then, it seems appropriate that the World Bank once again chooses "Agriculture and Development" as the theme for the WDR. Now, we read and hear a lot about what the best approach to agriculture in terms of poverty distribution could be. In our globalized world, we have, on one side of the equation, regions like the European Union (EU) and the United States, with heavily subsidized agriculture and agricultural production, which, together with tariffs and nontariff barriers, is significantly hindering Southern agricultural development by distorting patterns of international trade and depressing prices. On the other side of this equation, we have the claims, or should I rather say hopes, of developing nations, which want to profit from liberalized agricultural markets. They expect that those people who depend on agriculture will benefit from an expansion of the markets and the consequent improvement in the prices of agricultural commodities, and that this will contribute to a more equal balance in the distribution of incomes within their countries and on a wider global level. However, taking into consideration the mechanics of farm price determination and the technological limits to the deploy- ment of capital in agriculture, is this really a comprehensive solution? "Trade, not aid"--that is, trade rather than conventional development assistance-- is a slogan which cropped up in many discussions over the past years. The United States used to advocate an approach to development cooperation that is mainly based on encouraging free trade, while the EU has always tended toward the opinion that aid and trade must be deployed together in order to stimulate economic growth and alleviate poverty. Now "aid for trade" is the new approach to development assis- tance, and its aim is to help developing countries, the least developed in particular, build the supply-side capacity and infrastructure they need to take advantage of trade liberalization and enhance their participation in the world trading system. For us, it will certainly be interesting to discuss the outcomes and effects of those approaches to agricultural development and evaluate which nations or which population groups are actually benefiting. This will help to develop target group-oriented strategies in poverty alleviation and agriculture. Furthermore, the issues the WDR is going to address have a political dimension as well. Solutions for agriculture on a global scale might have to be tackled from a political angle, and the question might arise as to what role the World Trade Oganization (WTO), the Food and Agriculture Organization of the United Nations (FAO), the World Bank, the EU, and other bilateral actors should play. I personally hope that the WDR will debate this political dimension of development issues. I am aware that, on many occasions in the past, there have been limitations to the mandate of the World Bank on that point. However, with respect to agriculture, I think this is something that has to be taken into account, because international organizations, bilateral donors and other nongovernmental organizations involved in developing cooperation are all looking to the WDR as a credible manual for effective policy change. My second point pertains to agriculture and the rural sector: are they an effective engine for growth? The WDR has to examine if growth in gross domestic pro- duct (GDP) stemming from the agricultural sector has a greater impact on income than growth originating from nonagricultural sectors; agriculture's contribution to WELCOMING ADDRESS | 13 poverty reduction is sometimes considered to be small, because its relative economic importance usually decreases when low-income countries develop successfully. If we look at developments in India, it is fair to ask this question: Is agriculture really the growth engine in a country like this? Or, has growth rather been fueled by the liber- alization policies that have allowed the private sector to grow, and is it not rather the private sector with its small and medium-size enterprises that has become the growth sector? If we consider that India increases its middle class every year by about 25 mil- lion people, we can see it is they who are lifting the country out of poverty, and not agriculture. But what happens to the large number of rural poor? Growth has come from urban services and manufacturing employment, but what about agriculture? As such, one could assume that an effective way to secure higher incomes for the major- ity of people in rural areas is to move them from agriculture into manufacturing and services. Or do we need another `Green Revolution'? And yet, the way we look at agriculture should be based on the premise that the importance of agriculture by far exceeds its direct impact on farmers' incomes. Faster growth and more diversity in the agricultural sector can help the economy in many ways. First of all, it will mean an increase in exports. Second, it will generate surplus and, by doing so, support various agri-processing industries, thereby making a posi- tive impact on personal disposable incomes in rural areas. Third, by generating patterns of development that are employment-intensive, it benefits both rural and urban areas; for example, by helping slow down rural-urban migration. Fourth, it can fuel economic development outside the agriculture sector where growth and job creation are faster and wages high. Considering all that, the question is, what is it that agriculture needs in the way of technology, infrastructure, and financial support to become a growth engine? A great deal has happened in recent decades, and many countries have shown tremendous development and growth in agriculture. Chile is one such example, and there are many other positive examples in Africa. For instance, wine in South Africa; flowers, fruit, and vegetable products in Kenya; and so forth. However, in many poor coun- tries, agriculture has stagnated and failed to deliver its potential. Examples in those countries show that there are other problems involved; for instance, abuses of natu- ral resources due to current agricultural techniques, shortage of land and water, and environmental problems. The list of worsening conditions is long, and the list of questions that relate to how agriculture can be turned into a growth engine is also long. I am therefore looking forward to hearing about what new insights and recommendations this WDR will give us. What is more is that the pace of agricultural growth in today's poor countries will probably be slower than during the green revolution, and it will no doubt differ between countries, depending on local conditions. It will also be more difficult for small-scale farmers to join in. And yet, despite this, agriculture's potential can be real- ized; for example, by more effective investment and better policies. These must spread the benefit of new technologies, provide better access to finance for rural households, and tackle not only donor failures but also market and government shortcomings. In the face of this, we have to consider how to formulate and imple- ment tailored agriculture-for-development agendas. 14 | GUDRUN KOCHENDÖRFER-LUCIUS My third and last point focuses on development in connection with capacity build- ing. If we agree that there is some scope for technical solutions for agricultural growth, then what instruments would be needed? Instruments in terms of developing cooperation, or multilateral institutions, or governments of nations? And what would the priorities be for these? I am particularly interested in capacity building, something which I mention for two reasons. First, because InWEnt is the German government institution for inter- national capacity building which, with over 800 highly specialized employees and a budget of 150 million for training purposes only, can reach as many as 57,000 par- ticipants a year. The capacity-building services we offer cover a wide range of topics, from rural development, nutrition, and environmental concerns to sustainable eco- nomic and social development issues, international rules, good governance, and economic policy. InWEnt is one of the leading organizations in this area worldwide. And so this is a relevant topic for us, because, from the very beginning, and even through times of dwindling consideration in the global community toward the agricultural sector, InWEnt has persistently been engaged in capacity building in agriculture and rural development. There are seven specific examples of projects and programs that ought to be mentioned here: · Assessment tools and intervention strategies for food security (Caucasus/Central Asia), · Improving efficiency in the biofuel value chain (southern Africa), · Implementation of standards and International Chemical Convention recommen- dations (worldwide), · Biotechnology and bio-safety management (worldwide), · Market integration and regional branding (southeastern Europe), · Regional policies and forest management options (Amazon region), and · Management of hillside development (Himalaya-Hindukush-Pamir region). Through our work in these sectors and regions, InWEnt is convinced that we can contribute substantially to agricultural development. But, one can always make things better, which is why we are now looking with a lot of interest to the WDR and your answers, or your solutions, of what a training institute like InWEnt could contribute and what we can possibly improve. The second reason for mentioning capacity building has to do with the World Bank's evaluation of its capacity-building activities in Africa. Between 1995 and 2004, the Bank provided some US$9 billion in lending and some US$900 million in grants and administrative budget to support public sector capacity building in Africa, which was a very important contribution in terms of finance. However, according to the Independent Evaluation Group (IEG) Review on Capacity Building in Africa, the evaluation results are unfortunately not so positive. For example, the evaluation states that its tools--notably technical assistance and training--are not being used WELCOMING ADDRESS | 15 effectively, and its range of instruments--notably programmatic support, economic and sector work, and activities by the World Bank Institute--are not being exploited to the full. What is more, most of the activities seem to be deficient in standard quality assurance processes at the design stage, and they could be better routinely tracked, monitored, and evaluated. So, if we all agree that agriculture is relevant as a growth engine, a food-producing engine, or poverty alleviation machine, what would the contribution then be in terms of capacity building? And what would the contributions be in terms of governance, even in the rural sector? If we take these questions into account, we will end up talk- ing about capacity building measures. In any case, I hope that all this is part of the WDR, because, as I was saying, we all look with great expectations to this report, as we all feel that agriculture and development must be tackled. With this idea in mind of using the WDR as an instrument for discussion and for gaining insights for development cooperation, it was way back in 1998 that we started our joint effort of the Berlin Workshop Series in cooperation with the World Bank. We were aiming to draw lessons from successes, to present recent research findings, and, finally, to debate key policy issues. The objective always was, and still is, to widen the scope of inputs into the WDR by integrating German and European views as well as the opinions of well-known and respected experts from the South. The value added when you bring 20 leading experts together is that you all come here with the papers you have prepared; have a good, extensive, and intense exchange of views; and then obtain feedback on the content of these papers and on the first out- lines of the WDR. I also think that, the fact that this has kept going for nine years now, is feedback to us that our idea was not such a bad one. I am very much convinced that it is good to have a platform here in the Development Policy Forum where we can raise controversial views and express our opinions openly. Ladies and gentlemen, let us have a debate. And let us share our views, so that the authors of the WDR profit from these and are provided with more food for thought. Keynote Address Current Thinking on the Scope and Approach of WDR 2008 FRANÇOIS BOURGUIGNON Let me start by thanking you for accepting our invitation to share your thoughts on agriculture and development, the topic of this year'sWorld Development Report (WDR). The WDR is both a synthesis of existing knowledge in a specific area and a reflection on the best development strategy that can be drawn from such synthetic knowledge. It is an enormous advantage to start working on a WDR with the general views of, and real brainstorming that involves, the best specialists in the area being covered. My task today is not easy, given my ignorance of agriculture. I was initially not enthusiastic about this topic, as many people I talked to about it thought it was much too circumscribed. Unlike equity--the topic of World Development Report 2006--they felt that agriculture was too specific and technical a topic, restraining our ability to discuss the various important dimensions of development. But after some discussion with experts, including members of the WDR team, I realized that we were wrong: agriculture encompasses everything that matters in development, which is what makes this year's undertaking as challenging as it was in past years. If we are serious about reducing poverty, we must deal with agriculture, because 70 percent of the world's poor people live in rural areas and depend directly or indi- rectly on agriculture. It is indeed surprising that for a long period of time--certainly more than a decade--agriculture has been a secondary preoccupation for the World Bank. The last WDR on agriculture was in 1982. Of course, the Agriculture and Rural Development Department at the World Bank has produced several reports since then, but the Bank as a whole has not made the kind of huge effort involved in a WDR to reexamine in depth the role of agriculture in development. In today's endeavor, I simply pose a series of questions, hoping to stimulate the discussion that will take place in the next two days. The WDR team already knows many of these questions, but it is always good to rephrase and rethink the questions from a more general perspective. François Bourguignon is Chief Economist and Senior Vice President of the World Bank. Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 17 18 | FRANÇOIS BOURGUIGNON FIGURE 1. Economic Development and Agriculture as a Share of GDP in 143 Countries, 2004 Share of agriculture in GDP (%) 70 60 50 40 30 20 10 0 6 7 8 9 10 11 Log of GDP per capita (PPP) Source: World Development Indicators. The basic question, which might appear paradoxical, as I want it to be, is, Why and how should we invest public resources in agriculture--or implement agriculture- oriented development policies--to reduce poverty? This seems to be a stupid ques- tion, but it might not be irrelevant, given what we could call the "declining share paradox." Figure 1, which we all recognize, illustrates such a paradox. It depicts the share of agriculture in GDP on the vertical axis and the log of GDP per capita on the horizontal axis for 143 countries in 2004. We can see that, overall, agriculture's share of GDP declines with economic development. We also know, and I return to this later, that many countries have, in time, followed this path during their develop- ment process. The paradox became explicit when, inside and outside the Bank, I tried to explain why we were interested in the role of agriculture in development. The reaction was, "Why do you work on agriculture? Eventually, agriculture will only represent a few percentage points of GDP, so it cannot be the engine of growth." In the WDR, we would like to analyze this paradox and find a way to answer those skeptics who believe that development is systematically taking place outside agriculture. There are, in my opinion, three ways to address this question. The first is to look at what I call the "old" development paradigm and the spillover effects of agriculture. Under this paradigm, agriculture truly is an engine of growth, and productivity gains in agriculture have a positive impact on the whole economy as they lower wages and facilitate development in the nonagricultural sector. The second is to look at what I call the "new" development paradigm, which I tend to associate with international trade. Under this paradigm, a comparative CURRENT THINKING ON THE WDR 2008 | 19 advantage in some crops permits the local economy to become more efficient and to generate a surplus that can be invested elsewhere in the economy. And the third, which perhaps should go first, given its obvious reality, is the poverty argument: agriculture is the sector that encompasses the largest number of the poor. Old Development Paradigm Let us first consider the old development paradigm. As we all know, the paradigm goes along the following lines. In order for an economy to take off, productivity gains in agriculture are needed to generate various types of surpluses. Food will be avail- able for the nonagricultural sector because some people in agriculture will manage to exceed the subsistence level of production and sell the surplus to the rest of the economy; some labor will be freed from agriculture and employed in other activities; resources will be made available for investment there if some rural households get truly out of the subsistence constraint and can start saving. In turn, the process of industrialization is fostered through a number of mecha- nisms, which the development models I was teaching 20 years ago were insisting upon. The thrust is that the decline in food prices made possible by productivity gains in agriculture lowers wages in the nonagricultural sector (or, at least, increases the ratio of nonagricultural prices to wages) and reduces the relative cost of labor out- side agriculture. It simultaneously raises farmers' incomes and their demand for industrial products. This is a kind of fundamental view of development, which anticipates that the industrialization process will absorb labor coming from the agricultural sector, pre- sumably the least productive labor. This last point is very important when you think of poverty: thanks to this process, the poorest people are able to move to more pro- ductive jobs in the rest of the economy. This view of poverty reduction essentially rebalances sectors of the economy, with agriculture going down and manufacturing, and possibly services, going up. This mechanism may be combined with the Engel curve mechanism, which anti- cipates that, as income increases, the population spends less on food and agricultural products in relative terms. This generates a decline in agriculture's share of GDP. Yet, total agricultural output still increases because households consume more agricul- tural goods in absolute terms. And finally, with increasing productivity in agriculture and larger output, countries progressively improve their food security. They become less subject to famines. What kind of evidence do we have for this old paradigm? Of course, the first evidence is history. I will not develop what we know about Britain and the fact that its industrial revolution was made possible by productivity gains in agriculture, which were, for many authors, due to the enclosure of the commons. But the model of Britain was not necessarily observed later in other European countries. For exam- ple, in countries that started their industrial revolution later, like Germany, low food 20 | FRANÇOIS BOURGUIGNON FIGURE 2. Economic Development and Agriculture as a Share of GDP in China, 1965­2004 Share of agriculture in GDP (%) 45 40 1978 35 30 25 20 15 10 0 300 600 900 1,200 1,500 Real GDP per capita (index 100 1965) Source: World Development Indicators. prices were very much due to the availability of imports from the American continent (South and North), Australia, and elsewhere. So, the development of these countries initially incorporated an important element of international trade. We have the same kind of contemporary evidence, and China is a fascinating example. Figure 2 shows the evolution in China of the share of agriculture in GDP on the vertical axis and GDP per capita on the horizontal axis. The various points correspond to the years between 1965 and 2004. The line labeled 1978 marks the separation between the years before (on the left) and after (on the right) 1978, the beginning of the reform period in China. Between 1965 and 1978, economic devel- opment, as measured on the horizontal axis, was slow, especially in comparison with the fantastic acceleration that took place afterward. The first and probably most important reform of 1978 consisted of implementing the "household responsibility system," which gave farmers the right to sell part of their output on markets. This led to a rapid increase in farms' total output, as shown by the dots just after 1978. And then China entered the standard economic evolution, with increasing agricul- tural production but a declining share of agriculture in GDP. Now we may be a little more technical. Some economists have tried to gauge the existence of spillover effects by running causality tests from agricultural growth to nonagricultural growth. Recently, Tiffin and Irz (2006) ran such statistical tests on periods spanning between 30 and 35 years and found, for certain countries, some positive causality going from agriculture to the rest of the economy. Yet, I am uncer- tain about this kind of evidence. I am not sure that causality tests based on year-by- year analysis can capture fundamentally structural effects. CURRENT THINKING ON THE WDR 2008 | 21 Another issue is heterogeneity across countries. I played around with some data on the share of agriculture in GDP and the log of GDP per capita for some countries. Fig- ure 3 presents the GDP per capita relationship for a small set of countries over the period 1975­2003, with the long, solid black line corresponding to the cross section of countries in 2004 and the other lines corresponding to specific countries. Such lines are trend lines in the face of large yearly fluctuations in agricultural output in many of the countries. This figure is interesting because it shows very different trends across countries. China, India, Indonesia, and Tunisia are growing quickly. The rate of growth of those economies is shown by the horizontal distance between the first and last points on the curve. In these countries, the share of agriculture is declining, more or less at the pace observed in the cross section of countries in 2004. Brazil, Ghana, and Mexico have sluggish growth, but the share of agriculture is declining more quickly than in the cross-section. Other countries have the opposite evolution. Chile is growing rather quickly, and the share of agriculture is increasing over time. This is in large part due to the production and exportation of new items such as wine and fish. Peru and Senegal did not grow as rapidly, and they also have an increasing share of agriculture in GDP. Finally, Kenya shows an ambiguous evolution, with a very slow rate of growth and a share of agriculture that follows a U-curve. It first declined and then, over the last couple of years, began to increase again. It is difficult not to relate this evolution to the recent export boom of horticulture and flowers in that country. Figure 3 suggests a huge heterogeneity across countries. We have very much to learn from comparing and trying to synthesize this variety of experiences. True, the FIGURE 3. Share of Agriculture and Development in GDP in Select Countries, 1975­2003 Percent 60 Cross section 2004 Ghana 50 China 40 Kenya 30 India 20 Indonesia Senegal Tunisia 10 Peru Chile Brazil Mexico 0 5 6 7 8 9 10 11 Log of GDP per capita (PPP US$) Source: Author's compilation. 22 | FRANÇOIS BOURGUIGNON share of agriculture in GDP tends to go down in fast-growing economies. But, in other countries growth seems to be associated with an increase in the agricultural share. We have much to learn from this kind of comparison, which I hope will appear somewhere in the WDR. Finally, concerning productivity gains in agriculture as an engine of growth, exist- ing data lead to a rather prudent view. For the 91 countries for which data exist (excluding countries of the Organisation for Economic Co-operation and Develop- ment), annual average growth of agricultural labor productivity was below 1 percent over the period 1981­2001; it is difficult to believe that there is a strong force behind such weak productivity gains. This does not mean that it is impossible to accelerate the growth of agricultural productivity or that some countries have done consider- ably better or are able to do better than others. This is certainly one of the challenges that the WDR has to study. New Development Paradigm Let me say a few things about the new paradigm, with the caveat that this is subject to debate. The new paradigm sees agriculture not so much as a key sector--the development of which is a necessary condition for development of the rest of the economy--but rather as a sector to be developed for its own sake, given its compar- ative advantage. This view is very much linked to globalization. The old paradigm applied more often to closed than to open economies. We know the story of Britain. We know the stories of China and India, which were closed economies in their early phases of development. We also know that agriculture had little to do with the successful devel- opment of Hong Kong (China) or Singapore. The real point is that those economies were completely open and did not need to develop their agricultural sector to feed their population. This remark leads us to consider the role of globalization and inter- national trade in the relationship between agriculture and growth. Free trade, logically, would make some of the agriculture-led stages of develop- ment mentioned above irrelevant, although perhaps not all of them. With open trade, lowering the price of food can take place without generating productivity gains in domestic food crop production. It is sufficient to rely, at the margin, more on food imports than on local production, provided the country can exploit some compara- tive advantage outside agriculture or within particular exportable crops. Further- more, free trade considerably weakens the Engel curve argument and the link between agriculture and overall economic growth. Basically, this is the same story as that of investment and savings. In a fully open world there should be no correlation between savings and investment in a country; similarly, in a fully open world, there should be no direct relationship between agricultural consumption and production. Production should be as efficient as it can be, while consumption should simply reflect households' preferences. Now, if we were to adhere to this view of trade, the adequacy of climatic and geographic conditions for specific crops becomes the key factor in analyzing the CURRENT THINKING ON THE WDR 2008 | 23 agricultural potential of a given economy. This adds, or maybe introduces, a new dimension to our understanding of the role of agriculture in development. And, of course, in an increasingly globalized world, issues such as supply chains, biotech- nologies, or infrastructure that can help agriculture to integrate into global markets also become very important. This should remind us of the importance of the World Trade Organization agenda--whether progressing or not today--and the definitive need for freer global trade, which is certainly related to this new paradigm of looking at agriculture essentially under the angle of comparative advantage in an open world. I offer one last point on this new paradigm. I am not sure that food insecurity can simply disappear with globalization. If a country specializes in a few crops, and if the prices of those crops are highly volatile, then food insecurity remains an issue. Poverty Argument Let me finish with the poverty argument. As I said at the outset, we know that poverty tends to concentrate in rural areas, where agriculture is a dominant activity. Even when we talk about rural nonfarm jobs, we know that these jobs are in many cases generated by agriculture through backward and forward linkages or simply by farmers' consumption. Hence rural poverty is very much linked to the state of agri- culture. The old paradigm thus anticipates some kind of automatic poverty reduction through the growth of essentially nonagricultural, nonrural sectors. This view applies to the new paradigm as well. As long as some sectors inside or outside agriculture are growing strongly, poverty will automatically be absorbed or poor people will move progressively to those sectors. The profession has held this view for a long time. My intuition is that the relative neglect of agricultural issues in the development com- munity is due to this understanding of development as being essentially characterized by the absorption of poverty and agriculture by the rest of the economy. However, things are not that simple, even in a successful country like China. We know that it is increasingly difficult to absorb workers from lagging regions and from rural sectors in lagging regions. It is relatively easy to identify the conditions under which this automatic absorption will not take place. One obvious condition is the labor intensity of the rest of the economy. Biased technical change (favoring the demand for skilled workers or physical capital at the expense of unskilled workers) might reduce the labor intensity in the rest of the economy, preventing the absorption of workers from agriculture. This means that there could be, for quite some time, pockets of poverty, even in rapidly growing economies. What can be done to reduce that poverty? Assuming that all the possibilities offered by migrating to other sectors have been exploited, one would be tempted to say, "Fine. As this is a poverty issue, let's handle it with the simplest instrument that may be mobi- lized: income transfers." But it may be very difficult, not to say impossible, to make transfers to the poor in a lump-sum way as postulated in orthodox economic theory. What does this mean? This means that if we want to alleviate poverty and cannot make lump-sum transfers, then we have to find other ways. One of them may be pre- cisely to design agricultural policies with the primary objective of reducing poverty 24 | FRANÇOIS BOURGUIGNON in the less favored areas of developing countries. In other words, agricultural policies could become second-best policies to reduce poverty, possibly complemented with other policies that may reduce poverty in the longer run, such as education or health policies. Providing subsistence farmers in lagging regions with access to cheaper credit, inputs, or agricultural techniques will not suddenly create comparative advantage in these areas and will probably not make agriculture an engine of growth. But it will improve the conditions in which people live. Moreover, these policies may be win- win, because most markets in the rural areas of developing countries are very imper- fect. Therefore, making credit available to people who do not have access to it today may allow them to exploit agricultural projects with a rate of return much higher than that observed today, and providing insurance to small farmers may encourage them to specialize their production (instead of diversifying it to reduce various types of risks) and thus raise their expected and therefore their average rate of return. Thus, in some instances, redistribution policies associated with the correction of market imperfections may generate productivity gains. Let me conclude with some final remarks, which summarize what I just said. The way in which I presented things seems to suggest that the policies needed to develop spillovers from the agricultural sector, create comparative advantages, or eliminate pockets of poverty are essentially different. But the last examples I gave show that this is not true. One of the difficulties in trying to understand the role of policies in agriculture is to know whether it is possible to kill two birds, maybe three, with the same stone or whether we should have one stone for each bird. Thus it is important to map policy instruments with objectives. What is the main consequence, the main impact, of those policies, and do they simultaneously address those three dimensions of the agricultural sector? To what extent do the policies and investments needed to exploit comparative advantage in agriculture overlap with those needed to foster the spillover effects of agriculture on the rest of the economy? To what extent can redis- tribution policies foster productivity gains? These are the questions we should ask. Finally, cross-country heterogeneity is enormous, as we have seen, but hetero- geneity is also large within countries. In most countries, there is often a comparative advantage in some crop in some area and, at the same time, subsistence farming in some other area where people are living in conditions of extreme poverty. So the dis- tinction among the three perspectives on agricultural policies does apply within many countries as well as across countries. This is an argument in favor of not separating them completely, but it also logically reinforces the need to be clear about what issues are to be addressed by policies. As keynote speaker, I know that it is easy to list questions without offering any answers. I hope that I have given you some work to do for the next two days and have contributed to that goal in an efficient way. Reference Tiffin, Richard, and Xavier Irz. 2006. "Is Agriculture the Engine of Growth?" Agricultural Economics 35 (1): 79­89. Opening Address ASTRID KÜHL We have a strenuous working day ahead of us with difficult and, for the future of One World, decisive topics, so I am not simply resorting to a cliché when I say that I hope you all slept well last night. For our tasks today, we all need to have had a good night's rest. "Strom kommt aus der Dose" ("Electricity comes from the socket") is the slogan, as our participants from Germany will remember, that electricity companies used to make fun of their critics in the debate over atomic power, claiming that they had no idea of the problems involved in electricity production. Of course, that was not true. One could hardly miss the power stations and high-tension cables in a modern indus- trial landscape. Food supply is even more fundamental than the supply of energy and electricity. Yet over the years of the European Wirtschaftswunder in the developed industrial states of Europe, agriculture, with its achievements and problems, has taken a back seat. In the European Union (EU), agriculture only registered as a bureaucratic prob- lem of budgets and subsidies, a situation hardly altered by the agricultural policy reforms attempted in Europe. Agriculture is marginalized. Many children and young people in the major cities have never seen a real cow. This is not a fable; it is statisti- cally proven. Milk seems to come out of a tetra pack from the supermarket. And you can't know that a bunch of imported flowers is the product of forced child labor simply by looking at it. In the developed countries, the gap between agricultural production and the con- sumption of agriculturally produced goods has grown so markedly that, at the store counter, the traces of a product's agricultural origins have vanished almost entirely. Agriculture and development are a highly complex issue. On the one hand, urban and industrial development depend entirely on sufficient multiproduct agriculture; on the other hand, for a long time development as a concept remained closely linked to Astrid Kühl is Director of the Development Policy Forum, InWEnt (Capacity Building International, Germany). Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 25 26 | ASTRID KÜHL urbanization and the growth of industrial production. Development was not meas- ured by agricultural production, but by the achievements built on it. Such a perspective can lead to development taking a fatal wrong turn. For exam- ple, the Soviet Union's faith in progress was based entirely on eroding agriculture. And this, not least, led to its downfall. In some developing countries, the Soviet Union had long been sponsoring a type of development that took place at the expense of agriculture. Only recently have people begun to see that development involves, first and fore- most, developing agriculture. The development surges in European agriculture pro- vided a fundamental basis for Europe's development as a whole. And especially in the farming economy, advances cannot be achieved without the support of the producers. In this case, their demands, and the development of their skills and expertise, are the basic prerequisite for development. And since women frequently bear the lion's share of the workload in agricultural production, development requires, above all, support for women. In many modern industrial states, the percentage of the working population involved in agricultural production has fallen to under 5 percent. In money terms, agricultural production makes an almost negligible contribution to gross national product. And yet autarkic memories of Europe's agricultural self-sufficiency and the work of farming association lobbyists have protected the EU budget for agriculture and agricultural subsidies. While the European agricultural market remains pro- tected, market access liberalization in developing countries functions as a gateway for subsidized products from the United States and Europe. This practice remains a stumbling block in the World Trade Organization negotiations. Given such a con- stellation, competition does not provide a development impulse to agriculture in developing countries. Instead, it endangers its existence. This issue is the subject of our first panel. Representation of the farming population's interests is especially poor where most of the general population is involved in agriculture. Apart from population growth and weak productivity, this is no doubt one reason for poverty in rural areas and the migration that it causes. We will be discussing these problems today as well. InWEnt has taken on the task of contributing to international capacity building. As I see it, rural areas and the farming population offer a much suppressed and neg- lected potential. Undoubtedly, apart from pointing to problematic issues and con- cerns, the World Development Report 2008 will highlight such potential as well. I am convinced that our workshop will contribute to this debate and look forward to two very productive days. Introductory Remarks MICHAEL HOFMANN I would like to welcome you on behalf of our Minister Heidemarie Wieczorek-Zeul. I am happy that the sun is shining, as the weather has been quite dreadful. But as we are all interested in farming, we know that this is the way it is. Nature plays a role and, it is hoped, a positive one. I confess that I have an urban bias, but my colleagues have reminded me of the importance of agriculture and rural development. Since I am also governor of the Inter- national Fund for Agricultural Development (IFAD), I remember well its report of 2001, which noted that the majority of the poor live in rural areas. I am well aware that agricultural development has received declining attention in recent years. In development policy and cooperation, as well as in other fields, if the pendulum swings too far in one direction, there is a correction, and I suppose we are now swinging in the other direction. In this sense, the World Development Report (WDR) 2008 on agriculture is timely. Even though poverty issues have played a prominent role in recent WDRs, the last one with a focus on agriculture dates back quite some time, to 1982. In 1986, agriculture was dealt with in the context of trade. It is high time for another report on agriculture, and I urge you to use this opportunity! It is time to gain new insight, to correct what has been neglected in recent years. It is not a coincidence that we have not been more active in this regard. Earlier efforts were quite costly and not sufficiently effective. So it was not by chance that there was little interest--be it in the World Bank or in institutions like mine--in investing heavily in agriculture. Nonetheless, we have to look at the opportunities and chances that arise and reflect on recent trends. In this context, we have to take into account what will happen if we do not have a successful Doha Round or if it is postponed until after the U.S. elections in 2008. Even then, the outcome would be uncertain, as a new Michael Hofmann is Director General of Global and Sectoral Tasks, European and Multilateral Development Policy for Africa and the Middle East, in Germany's Federal Ministry for Economic Cooperation and Development (BMZ). Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 27 28 | MICHAEL HOFMANN administration might need another year to get up to speed. The importance of mak- ing significant progress in terms of agricultural markets and subsidies becomes all the more obvious if we remember how many subsidies are given to a European cow in comparison with the support given to poor people in developing countries. A strong argument has been made in favor of the Doha obligations, but even though this is wise and important, we should be prepared to face a situation without an interna- tional trade environment that really benefits the developing countries. We have to ask ourselves who would benefit from a trade solution and from more investment in agriculture. Parliaments tend to point to Brazil, because they fear that only a few countries like Brazil and the developing countries in the Cairns Group would benefit. Would that really benefit small farmers? What does it mean for Africa? These are certainly questions you will address in your report. You will also have to show best practices, which teach us what lessons can be learned. There is no one-size-fits-all formula that one has to follow, but it is important to learn from these experiences. We also have to examine how urban and rural areas are mutually dependent. I can see how agriculture has prospered since reunification due to the opportunities created by Berlin alone. This is true in many developing countries as well: distance from and access to markets play a tremendous role. We also know that migration is a tremendously fierce problem. With a new empha- sis on agriculture, people might stay in some places and leave others. The question is, then, how much can you invest in certain places to retain people there, and what are the off-farm opportunities in these rural areas apart from agriculture? All of that is extremely important for us to know more about, and I hope that you will find the right track. What I have seen so far of the proposed storyline makes me and others confident that your report will be a very important one. I hope very much that the discussion will be candid. I heard that you invited some people who are not known for being politically correct. I think this is important. Otherwise, there might be a danger of replicating stereotypes, which is not helpful. I hope that you will reflect controversy, that you will avoid saying, "This is the only way to go," and that you will show that there are options around the world. There are alternatives in different places; there are different types of agriculture. And it is not only institutions that matter, which is a lesson we have known for more than a decade and which has played a very important role in recent World Bank reports. Institutions are more than market incentives and rule of law, because we know, apart from institutions, that culture and nature also matter. Perhaps some of you might look again at chapter 1 of David Landes's famous book on the wealth and poverty of nations. It focuses on nature: "nature's inequalities," as it is called. I have looked into it again, and even though it might be crude in its argumentation, one has to remember that not everything is possible everywhere. If I remember rightly, Kuan Yew, when he was asked why the wonder of Singapore was possible, answered bluntly "air conditioning." Since there is no air conditioning for agriculture, the question is, what can you do in its place? Development, in my understanding, is not the rule but the exception. It has always been made possible by circumstances--a complex set of very specific ingredients--that INTRODUCTORY REMARKS | 29 might follow the pathfinders but never can be simply copied. Rather, one has to adopt these ingredients to the given cultural, institutional, and natural fabric. In that sense, I am a big admirer of those who try to be strong in words but modest in perspectives. Our institutions tend to give the message, "We tell you what to do." Thus if we adopt a more modest style and show options, if we say what is possible and what is not, and if we are aware that there might be setbacks, we give perhaps a better reading. If you do not pretend to have "quick-fix solutions" and try not to bore the reader with too many cross-country regression analyses, which are wonderful for computer freaks but produce findings that are of minor relevance, you will produce something of utmost importance for all of us. I wish you much success in this endeavor. References IFAD (International Fund for Agricultural Development). 2001. Rural Poverty Report 2001: The Challenge of Ending Rural Poverty. Rome: IFAD. Landes, David. 1998. The Wealth and Poverty of Nations: Why Some Are So Rich and Some So Poor? New York: W. W. Norton. Part I: Agricultural Policy, Subsidies, and Trade Agriculture and Development: The Perspective of a Think Tank on the WDR 2008 Outline MICHAEL BRÜNTRUP The German Development Institute is a research and advisory body attached to the Federal Ministry for Cooperation and Development. Why should the perspective of a donor government think tank on the subject of the World Development Report (WDR) 2008--agriculture and development--be of particular interest to the organ- izers of the report? Two answers might be worth consideration. First, as an interdisciplinary organization, the German Development Institute is a microcosm of very different, often diverging views on development. It reflects the continuing enlargement of the development policy agenda and, in turn, is an appropriate way to look at sectoral perspectives such as the WDR 2008. In a small think tank, the intense exchange and appropriation of views and standpoints are certainly easier than in large, departmentally fractioned organizations or in smaller specialized ones. Second, we work as an intermediary body between the world of abstract, sophis- ticated scientific approaches and models and the more down-to-earth world of devel- opment cooperation, where the major vehicles of action are projects, programs, and concerted efforts to influence wider policy agendas that have to be established in the real world. Needless to say, as an agricultural economist, I welcome the fact that agriculture is again at the center of a WDR. What is particularly welcome is the very broad agenda of the WDR, which reflects the fact that agriculture plays a much greater role in the economies and societies of poor countries than economic textbooks, national accounts, and the development community tell, a role that has been neglected for some time. It is a healthy development that this role is no longer taken for granted by simply reiterating it or pointing to the status quo of developing countries or to the history Michael Brüntrup is Senior Researcher with the German Development Institute. These comments are based on the World Development Report 2008 prospectus dated July 17, 2006. Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 33 34 | MICHAEL BRÜNTRUP of industrial countries--a standpoint that is mentioned somewhere in the prospectus as "fundamentalist." Agriculture should be regarded as one--sometimes more, some- times less important--component in a combination of means toward achieving several ends. This process of widening the scope of the development policy agenda has many reasons. The objectives of development policy are reflected, for instance, in the grow- ing and accumulated discourses about macroeconomic stability and growth (as suf- ficient conditions for development), sustainability, the Millennium Development Goals (MDGs), good political governance, and human security. Regarding the means in development policy, some of the issues discussed are the importance of macroeco- nomic policy for growth and poverty alleviation, the private sector as the engine of wealth, the redefinition of food security as an access issue, decentralization, public service delivery, ownership of development planning, and budget support, among others. These discourses have rightly taken the lead in discussions within the overall development community on the "why and where," as well as on the larger principles of the "what and how" of development policy and cooperation. No one discipline can claim or play the lead in this arena. Disciplinary approaches should work in a world of checks and balances to achieve the best results and prevent biased, often unstable short-term solutions for complex real-world problems. Modern societies are governed and govern themselves in this way. In such a multidisciplinary arena, agriculture has to offer compelling means to contribute to these ends. However, in the past the agricultural community did not always perform well. One important reason is that checks and balances hardly exist in development assistance, with its unbalanced distribution of power and funds and badly working agricultural and rural policies in developing countries, where the rural populations have generally little voice. Another important reason why agriculture has faded from the attention of international development policy seems to be that the agricultural community did not participate fully in the discourses of the international development community. The main expectation for the WDR 2008 is that it should fill these gaps and deliver the why, what, and how agriculture can contribute to development. Against these expectations of the general profile of the WDR, what follows are some issues of particular importance in the broad agenda that seem to be at risk of remaining underdeveloped, at least in the early outline version. Sound Derivation of the Role of Agriculture and Agricultural Policy in Development The role of agriculture and agricultural policy should be soundly derived from over- all goals and the available instruments of development policy. This must include his- toric analysis and evaluation of asset endowments, normative values (for instance, enshrined in the MDGs), other goals of development cooperation, sound hypotheses on the development opportunities and threats, as well as scenarios on alternatives to achieve the goals. More particularly, emphasis could be given to the following points. AGRICULTURE AND DEVELOPMENT | 35 Although different versions and a background paper of the WDR 2008 already provide a wealth of insights, the initial section could be strengthened even further by taking readers from the different development subcommunities into the diverse aspects of agriculture. For instance, strong emphasis should be given to the MDGs: not only to poverty, food security, and water but also to health, education, environ- ment, partnership, security, and global environmental aspects of agriculture. A clear acknowledgment of the many potential sources of market failure, derived from the particularities of agriculture, would be welcome. This includes the wish for a more critical distance from structural adjustment and its main use for "getting prices and the macroenvironment right." However, this critique should not neglect the risks of government failure and the primacy of private sector principles in agri- culture. A more explicit balancing of public versus private sector strengths, weak- nesses, opportunities, and threats in an early chapter of the WDR would strengthen the clarity and credibility of further chapters, where ample claims are made for pol- icy interventions but against subsidies. Differentiating agriculture's role in the possible general development pathways is a good approach to relate the sector's contribution to various stages of the economic macroenvironment. Agriculture's role in nondevelopment pathways also needs to be considered. There are a number of developing countries in which the overall eco- nomic, social, or political situation is not going to improve in the foreseeable future. These analyses would show that, in many cases, agricultural development and its support are inevitable. In some situations, the growth potential will be highlighted; in others, the social aspects. This cannot be clearly separated for countries or even subregions. Who would have thought 40 years ago that in Western Africa the wealth- iest farmers would be living in some very remote and undeveloped areas, thanks to cotton production, a marginal product with high transport value? Or that coffee would afford remote highlands of Ethiopia or other remote areas of Eastern Africa a new competitive growth pole. In addition, competitiveness is not only an issue of factor endowment. It has to be achieved through long-term investments in technology, institutions, and market development. These differentiated views could lead to the best thing that could happen to agri- culture in developing countries--its "mainstreaming" in locally specific development strategies, coordinated at the national and even regional level. It seems logical that, under such circumstances, the three paradigms of the role of agriculture can coexist. However, agriculture sector specialists, organizations, and ministries are poorly suited and mostly too weak to design such differentiated and comprehensive policies. More Explicit Guidance on the Role of the Public Sector While analyzing the role of agriculture in development, several issues point to the fact that the role of the public sector in agriculture is particular in various ways, caught between the typical private and public sectors. The particularities stem from various sources, including the type of actors involved (many small, illiterate farmers and 36 | MICHAEL BRÜNTRUP households and informal organizations, on the one hand, and different levels of gov- ernment, private traders, and international enterprises, on the other hand), the cir- cumstances under which they work and interact (rural areas with many types of high transaction costs including transport, communication, and overlapping formal and informal institutions), and the markets that they serve and on which they depend (often highly imperfect local, national, and international markets for food, inputs, and financial services). It may be useful to distinguish between the three fundamental areas of political analysis: polity (the structure and organization of society), politics (the search for political power in competition with other actors), and policy (the content of political decisions in understanding and promoting agricultural development in developing countries). The first two issues need to be addressed more clearly in the WDR 2008 outline: · The sphere of "polity" is probably (but not clearly) included in the outline when dealing with the "institutional settings of agriculture," which are mentioned repeatedly, for instance, when discussing the roles of decentralized governments or farmer associations. The material, institutional, and cultural basis of social and economic actors must be considered when developing the institutional innovations and support measures required and feasible in agriculture. · The fact that "politics" matters is addressed directly in the context of the political economy of agricultural price policy. However, other political issues are hardly addressed, for instance, the consequences of democratic changes or decentraliza- tion in many developing countries for the options for agricultural policy making. Also, governance issues are not taken explicitly into account as a decisive deter- minant of--and for--agricultural policies. Agricultural ministries, for instance, are often very weak, having very few compe- tences even for core agricultural policy issues such as agricultural trade, livestock, land rights, forests, and irrigation. They are hardly able to make themselves heard in the other minor ministries dealing with these issues, not to mention the "big" min- istries, particularly finance, and the presidency. How to make agricultural issues heard and enshrined in the overall political debates and strategies would be a useful subject of the report. Another point stemming from a sociopolitical approach is the relation between quality of governance and agricultural policies. There is at present a large discourse on good governance, failed or weak states, and the consequences for development assistance, including budget aid. Classifying countries according to their governance and its consequences for agricultural development and assistance would add to the relevance of the allocation decisions of donors. Finally, it would be useful to discuss "inverse coherence" of donors concerning agriculture. The neglect of agriculture in recent decades probably stems from the absurdities of agricultural policies in industrial countries, which not only spoil agri- cultural markets worldwide but also make development agencies and their personnel reluctant to engage in this sector. AGRICULTURE AND DEVELOPMENT | 37 Realism Concerning Agricultural Economies in Poor Countries, Especially in Sub-Saharan Africa As stated, strong emphasis must be placed on realistically assessing the potentials and weaknesses of agriculture in poor countries and the potential for poverty alleviation by supporting agriculture (or alternatives). This often means placing at the center of analysis the risks and vulnerability of the poor and rural economies (through general volatility of agricultural and food markets, market failure, government failure, and external shocks of many kinds, including markets, policies, weather, disasters, and diseases, among others). In industrial countries and, thus, in conventional agricul- tural economics, the risk mitigation effects of functioning insurance markets and of agricultural policies with a high insurance component are highly underestimated: there are very few instances of substantial uninsured risks and many relatively low- risk exit strategies for farmers. In developing countries, particularly the poorest ones in Sub-Saharan Africa, such mechanisms do not exist, and analysis must take this into account. A point in that regard concerns the analysis of agricultural trade policies, particu- larly in the context of multilateral World Trade Organization (WTO) liberalization. The treatment of Sub-Saharan Africa in neoclassic-style trade models was, and still is, dubious. The weaknesses include lack of data for many African countries, over- simplification of market functioning, disregard of risks, use of developed-country elasticities and transport costs, disregard of adjustment costs, or inability to adjust in a timely manner. In the absence of further empirical research, infant industry and sen- sitive product arguments risk being ignored or dispelled as harmful to developing countries simply for not being captured in the models. This ignores the experience of several industrial and developing countries with the differentiated use of protection in time and across products, ignores the sometimes harmful effects of fast liberaliza- tion under structural adjustment, and is eventually counterproductive for convincing developing countries and rural stakeholders to engage in trade liberalization policies in cases where this is really reasonable. Regional Integration as an Important Aspect of Agricultural Policies in Sub-Saharan Africa Regional integration could be the main driving force for agricultural markets from the perspective of trade policy in the next decade, at least for Sub-Saharan Africa. (In other regions of the world where there is a dynamic development of activities other than agriculture and higher incomes, the following arguments probably do not hold.) This will be particularly relevant if the economic partnership agreements (EPAs) between blocks of African countries and the European Union materialize. Multilateral trade policy (WTO) has probably little to offer and certainly almost nothing to demand from Sub-Saharan African countries, most of which are least- developed countries (LDCs) that are generally exempted from disciplines. The few Sub-Saharan African countries that are not LDCs mostly have low applied tariffs 38 | MICHAEL BRÜNTRUP compared to the maximum allowed tariffs that they have bound in the WTO and generally have few subsidies from which to cut; their former state enterprises have mostly been privatized. In general, African countries are not reporting to, and are hardly controlled by, WTO partners. The minor increases in world prices that are expected from a moderately ambitious Doha Round will not be substantial enough to improve decisively the position of small net producers or exporting countries in light of heavy risks and low supply capacities. The situation could even deteriorate for net consumers, food-importing countries, or preference losers. At best, minor issues such as trade facilitation will support agriculture, in many cases, to the extent that corresponding aid for trade is provided. In contrast, regional integration (pushed by EPAs, it is hoped) will affect these countries more profoundly. Since about 70 percent of the agricultural products in Sub-Saharan Africa are destined, up to now, for national markets and are often not tradable internationally but at best regionally, regional integration constitutes an important leverage with which to create more agricultural markets and more stable demand for African farm products. Today, in West Africa, for instance, regional trade policy is the relevant framework for tariffs and trade policy. But in many regions, regional integration is stagnant, blocked by many overlapping, often contradictory regimes and by lack of implementation. EPAs aim to strengthen regional integration by reducing uncontrolled proliferation of regional trade agreements and by deepen- ing the remaining ones; they also aim to standardize and achieve peer control of mar- ket interventions. They are intended to cut red tape, bureaucracy, and discretionary policies, which are among the most important impediments to more equitable and reliable regional trade. The EPA negotiation agenda is much wider than the WTO agenda and includes Singapore issues, which are to be negotiated with a focus on development, not on the unilateral advantage of the European Union. The funds to support EPAs are much more important and, at least in principle, better focused and coordinated than multilateral funds. In summary, EPAs potentially have an important role to play in agriculture in Sub- Saharan Africa. Their impact will depend on how they are negotiated, implemented, and supported. The WDR 2008 could stress the role of regional integration for agri- culture in Sub-Saharan Africa and help to define what are development-friendly EPAs. Part II: Markets and Institutions Market Access for Small Farmers: The New Standards Challenge BENOÎT DAVIRON AND ISABELLE VAGNERON The reasoning presented in this paper is based on our past work on international trade in tropical products, in which we address the question, What institutions are needed to give small farmers access to distant consumers? Distance is an obvious "problem" in tropical trade--it means time and risk--but it is also a "problem" for many value chains, including domestic ones in developing countries. In this paper we argue that the "old" system of grades and standards played a central role in giving small farmers access to distant consumers. They did so by allowing the creation of markets at the farm-gate level or close to villages. Thereby, the historical system of grades and standards enabled small farmers to "impose" their effectiveness on large farms (plantations). Today new standards are emerging. Most of them are process oriented, unlike the old ones, which just defined some intrinsic characteristics of the products. The new process-oriented standards do not eliminate the possibility of market transactions at the farm gate. One of their "functions" is to organize market transactions (including contracts) at the farm level. More important, they are drastically changing the eco- nomics of labor supervision in agricultural production and thus the competition between small and large farms. The New Standards and Their Impact: The Current Debate Larry Busch's path-breaking work signaled the new impulse of the standardization process starting in the 1980s (Busch 1997, 2000). After him, many have elaborated a large amount of analysis proposing more detailed accounts of the renewed stan- dardization process and of the diversity of the standards implemented by private and Benoît Daviron, Economist, and Isabelle Vagneron, Economist, are from Cirad (Centre de Cooperation Internationale en Recherche Agronomique pour le Développement) in Paris. Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 41 42 | BENOÎT DAVIRON AND ISABELLE VAGNERON public organizations (Fulponi 2004); explaining the economic incentives behind this process (Fulponi 2004; see also the special issue of Food Policy edited in 2005 by Spencer Henson and Thomas Reardon); and evaluating the impact of the new stan- dards on developing countries (Dolan and Humphrey 2000; Friis Jensen 2004; Jaffee 2005; Otsuki, Wilson, and Sewadeh 2001; Reardon and others 2001). This new wave of standardization is already affecting a large part of the agrofood system, and it will endure. Strong forces are supporting it. Consumer anxiety is cer- tainly one of them. Consumers are demanding that an increasing amount of attrib- utes be incorporated in the food they buy. They are demanding more for themselves (mostly taste, safety, and health) and for others (distant producers and future gener- ations). The growing role of supermarkets in agribusiness is a second potent force. The implementation of specific standards responding to their proper constraints--be they technical or juridical--is one of the changes they introduce in the country where they develop their activities. More generally, the spread of new standards is sup- ported by the need, for enterprises located in the upper part of the value chain and confronted by global competition, to adopt new technologies and logistic devices: zero stocks, faster working speed, automation, continuous process, and so forth. Most of the new standards are part of a voluntary management system. The role of public law and public authority is limited. The adoption of the standards is not mandatory. Buyer choice, and especially consumer choice, is supposed to be the ulti- mate sanction in validating the adoption of a new standard. Beside this voluntary dimension, or maybe because of it, most of the standards (and their setting up) are linked to the emergence of large coalitions involving not only firms from various countries but also a very large variety of actors like nongovernmental organizations, intergovernmental organizations, state agencies, farmer organizations, and so on. As such, new agricultural standards respond perfectly to the participatory dimension of the sustainable development agenda. Finally, and maybe the most salient feature of these new standards, is their focus on the attributes of production processes. Production process attributes are included because they are valued on their own, like organic farming or animal welfare. They can be included as proxies of valued attributes too difficult or too costly to measure (mycotoxins, level of pesticide residues, processability, and so forth). Regarding the impact of the new standards, two separate issues have been addressed. The first deals with the impact of standards on trade. Standards are ana- lyzed here as new barriers imposed by developed countries on the exports of devel- oping countries. This issue is ignored in this paper. The second deals with the impact of standards on the organization of value chains from the field to the consumer. Less work has been done on this second issue. A first package of studies was conducted within the framework of the research on the supermarket carried out by Thomas Reardon and his colleagues (Reardon, Timmer, and Berdegué 2003; Reardon and others 2005). This body of work considers new standards as just one of the changes the "supermarket revolution" brings to the retailer procurement system. Beside standards, three changes are evident: · A trend toward the centralization of procurement (per chain), with the creation of distribution centers serving stores in a larger and larger area, MARKET ACCESS FOR SMALL FARMERS | 43 · A shift from the reliance on spot markets (in particular, traditional wholesale markets and brokers) and toward the growing use of specialized, dedicated wholesalers, and · The adoption of contracts with supermarket suppliers--in particular, via dedi- cated, specialized wholesalers managing a system of preferred suppliers for them. A second package of studies deals directly with the impact of standards on value chains and farmers (Dolan and Humphrey 2000; Friis Jensen 2004). The general conclusion is that new standards generate new constraints for small farmers. Three factors have been considered: · Economies of scale in field and postharvest activities: many new standards, par- ticularly safety standards, call for lumpy investments, such as new equipment (for example, potable water) or new management capacity, · Imperfect markets for labor credit and information, and · Vertical information motivated by increased information asymmetries related to product attributes. The Participation of Small Farmers in Long-Distance Trade: Historical Evidence and Theoretical Discussion The predominance of family farms in the production of agricultural goods for distant consumers is a historical event. The transition occurred, more or less, at the end of the nineteenth century, when the large farm (or plantation) was replaced by the family farm in long-distance trade for agricultural products (Daviron 2002; Koning 1994). Hayami (1996) summarizes the essence of the interpretation developed by econo- mists to explain the success of small farms in the face of large plantations. Any eco- nomic activity that requires the use of paid workers is confronted with the problem of controlling the activities of its labor to ensure that its behavior conforms with the objectives of the enterprise. In the case of farming, this problem of control is aggra- vated by the following features: · The spatial dispersion of activities, increasing the cost of control operations, · The low predictability of the results of tasks (such as no control of biological phe- nomena, climatic risks, and heterogeneous agroecological conditions), making it impossible to assess effort simply by measuring input and output. This makes close supervision essential. Because of these two characteristics of agriculture, family farms have lower super- visory costs than large farms; in other words, technological economies of scale are defeated by diseconomies of scale in labor supervision. On the basis of this interpre- tation, the historical shift from large to family farming, when it is considered, is explained by the end of slavery or the agrarian reform, which gave small farms access to land. But what about access to markets? 44 | BENOÎT DAVIRON AND ISABELLE VAGNERON It is important to underline that the change in the organization of agricultural pro- duction accompanied a major change in the organization of the whole value chain, giving small farmers access to distant consumers. Until the end of the nineteenth cen- tury, the operation of bringing the product from the plantation to the consumer was centrally organized by a specific actor: the factor. The factorage system had its ori- gins in the West Indies sugar economy: "The factor was the home agent of the colo- nial planter. He was at once his merchant and banker. He bought the goods which the planter has to purchase at home and sold for him the product return in exchange" (Holt Stone 1915: 557). In practice, the factor was much more than an agent of the planter in the European market. The factor dealt with (1) transportation of the prod- uct, by contracting with the railway company and the shipper; (2) storage of the product, by contracting with the owners of warehouse facilities in the country of des- tination; (3) insurance and the payment of taxes and harbor fees; (4) sorting of the product in grades; and (5) the broker in charge of the sale. The factor could also arrange the supply of new slaves for the plantation, provide equipment and consumer goods to the planters, and even act as a guardian of the planter's children while they were schooled in England. Later, in the south of United States, the factor also kept the account book of the cotton plantation. The factor did not own the product. He or she received the product, sold it in auc- tion markets on behalf of the plantation owner, and received a commission. The pro- vision of credit to the planter was another important activity of the factor. Initially conceived as an advance on consignment, credit was a way to secure product supply (the factor provided credit well before he received the product and even before the beginning of the harvest). Moreover, most of the goods supplied by the factor to the planter were provided on credit. Such an organization, with no market transaction before the arrival of the product in the consuming country, can be interpreted in rela- tion to the existence of two risks created by the distance and the time necessary to bring the product to the selling place: the price risk and the quality risk. Without spe- cific institutional devices dealing with these risks, traders' activities will be limited to brokering and, of course, financing.1 At the end of the nineteenth and beginning of the twentieth century, this organi- zation changed radically. In a few decades, the function of merchants was trans- formed. Merchants became traders, buying in the countryside, storing and selling the products hundreds or thousands of miles away. "Suddenly" the first market transac- tion was displaced from the place of consumption to the farm gate or the close village, a decisive change in the access of small farmers to distant consumers. The merchants' new activities were related largely to two interlinked institutional innovations: · The creation of futures markets, which give merchants the ability to cope with the price risk,2 and · The creation of grades and standards that permit (a) the operation of futures mar- kets, (b) the ability to buy a product at a long distance without any previous direct physical contact with it, and (c) the stabilization and homogenization of the quality demanded by manufacturers.3 MARKET ACCESS FOR SMALL FARMERS | 45 The newly created grades and standards were able to play these roles because they were based on very simple criteria limited to some intrinsic attributes (mostly mois- ture content, cleanness, absence of damage, and size) and on very simple measure- ment methods (mostly visual inspection: cut test, pulling). With these characteristics, the "old" system of grades and standards could organ- ize the incorporation of "traditional" agriculture into "modern" value chains, even in remote areas, even with limited access to information and infrastructure. At the same time, these grades and standards acted as a curtain or a screen between the farmers and the upper part of the value chain, the former being a sort of "black box" for the latter. New Standards, Certification, and Supervision in Agricultural Production The process-oriented standards are drastically changing this situation. One of their aims is clearly to open the "black box" of the agricultural production process. Some kind of labor supervision must be organized to provide information to the rest of the value chain. Third-party certification (TPC) is the "solution" adopted by many actors to implement the new standards. With TPC, an independent and skillful organization assesses and verifies the farmer's compliance with the standard and its technical specifications (Hatanaka, Bain, and Busch 2005; see also the special issue of the Journal of Rural Studies edited in 2005 by Tad Mutersbaugh, Daniel Klooster, Marie-Christine Renard, and Peter Taylor). The very function of TPC is to permit market transactions at the farm level--in other words, to avoid vertical integration, even with process-oriented standards. But certification can be both an important constraint for small farmers and a strong limit on their ability to compete with large farms in agricultural production. For small farmers, TPC is a constraint first because of its costs and the economies of scale associated with certification activities. More important, TPC is a constraint because it drastically changes the economics of labor supervision at the field level. Even if TPC appears to be completely external to the farm, it inevitably implies an increase of internal supervision and monitoring. An interesting comparison with audit practices is useful here. Like TPC, audit implies information gathering and supervision by an external agent. However, as argued by Michael Power (1997), to work, audit needs "things" to be made auditable. With TPC, farms need to get "inspectability," that is, the ability to moni- tor and record the activities realized on the farm and to produce a certain number of documents usable by the certifier. To deal with this problem, many small farmers act collectively and participate in procedures of group certification. Laura Gomez Tovar and her colleagues (2005) have carried out one of the few comparisons existing between the certification practices in large and small producers. They show how different certification pro- cedures are for these two categories of farms: "Samples, receipts, security arrange- ments, access to capital, and use of international corporate expertise of northern 46 | BENOÎT DAVIRON AND ISABELLE VAGNERON agribusiness organic certifications stand in sharp contrast to document-based, process-oriented, labor-intensive communal-land inspection and use of regional (peasant associations) networks of smallholders certifications" (Gomez Tovar and others 2005: 462). Endnotes 1. The role of brokers is also well informed by several studies dealing with "domestic" long- distance trade. See the work of Gabre-Madhin (1999) on the grain trade in Ethiopia. 2. For a fascinating historical account of the creation of futures markets, see Cronon (1991). 3. The stabilization policies implemented by government in most of the countries from the 1930s played a very important role in providing many agents not able to use the futures markets with access to domestic trading activities. References Busch, Lawrence. 1997. "Grades and Standards in the Social Construction of Safe Food." Invited paper presented at a conference on the Social Construction of Safe Food, Trondheim, Norway. ------. 2000. "The Moral Economy of Grades and Standards." Journal of Rural Studies 16 (3): 273­83. Cronon, William. 1991. Nature's Metropolis: Chicago and the Great West. New York: W. W. Norton. Daviron, Benoit. 2002. "Small Farm Production and the Standardization of Tropical Products." Journal of Agrarian Change 2 (2): 162­84. Dolan, Catherine, and John Humphrey. 2000. "Governance and Trade in Fresh Vegetables: The Impact of the U.K. Supermarkets on the African Horticulture Industry." Journal of Development Studies 37 (2): 147­76. Friis Jensen, Michael. 2004. Developing New Exports from Developing Countries: New Opportunities and New Constraints. Ph.D. dissertation, Copenhagen, Royal Veterinary and Agriculture University. Fulponi, Linda. 2004. Private Standards and the Shaping of the Agro-Food System. Paris: OECD, Directorate for Food Agriculture and Fisheries. Gabre-Madhin, Elini Z. 1999. Of Markets and Middlemen: The Role of Brokers in Ethiopia. IFPRI-MSSD Discussion Paper (November). Washington, DC: International Food Policy Research Institute. Gomez Tovar, Laura, and others. 2005. "Certified Organic Agriculture in Mexico: Market Connections and Certifications Practices in Large and Small Producers." Journal of Rural Studies 21 (4): 461­74. Hatanaka, Maki, Carmen Bain, and Lawrence Busch. 2005. "Third-Party Certification in the Global Agrifood System." Food Policy 30 (3): 354­69. Hayami, Yujiro. 1996. "The Peasant in Economic Modernization." American Journal of Agricultural Economics 78 (December): 1157­67. Henson, Spencer, and Thomas Reardon, eds. 2005. Food Policy 30 (3). MARKET ACCESS FOR SMALL FARMERS | 47 Holt Stone, Alfred. 1915. "The Cotton Factorage System of the Southern States." American Historical Review 20 (3, April): 557­65. Jaffee, Steven M., ed. 2005. Food Safety and Agricultural Health Standards: Challenges and Opportunities for Developing Country Exports. Report 31207. Washington, DC: World Bank. Koning, Niek. 1994. The Failure of Agrarian Capitalism: Agrarian Politics in the UK, Germany, the Netherlands, and the USA, 1846­1919. London: Routledge. Mutersbaugh, Tad, Daniel Klooster, Marie-Christine Renard, and Peter Taylor, eds. 2005. Journal of Rural Studies 21 (4). Otsuki, Tsunehiro, John Wilson, and Mirvat Sewadeh, 2001. "Saving Two in a Billion: Quan- tifying the Trade Effect of European Food Safety Standards on African Exports." Food Policy 26 (5): 495­514. Power, Michael. 1997. The Audit Society: Rituals of Verification. Oxford: Oxford University Press. Reardon, Thomas, C. Peter Timmer, and Julio A. Berdgué. 2003. "The Rise of Supermarkets and Private Standards in Developing Countries: Illustrations from the Produce Sector and Hypothesized Implications for Trade." Paper presented at the International Conference "Agricultural Policy Reform and the WTO: Where Are We Heading?" Capri, June 23­26. Reardon, Thomas, and others. 2001. "Global Change in Agrifood Grades and Standards: Agribusinesss Strategic Responses in Developing Countries." International Food and Agribusiness Management Review 2 (3­4): 421­35. ------. 2005. "Links among Supermarkets, Wholesalers, and Small Farmers in Developing Countries: Conceptualization and Emerging Evidence." Communications to the research workshop "The Future of Small Farm," Wye. Globalization, Privatization, and Vertical Coordination in Food Value Chains JOHAN F. M. SWINNEN AND MIET MAERTENS Twenty-five years ago, a vast share of the poor and middle-income countries, cover- ing a large share of the world's agricultural areas and farmers, were characterized by state-controlled supply chains for agricultural and food commodities. This was most extreme in the Communist world, spreading from Central Europe to East Asia, where the entire agrofood system was under strict control of the state. However, as in many African, Latin American, and South Asian countries, the state played a very impor- tant role in the agrofood chain. For example, in Brazil and Mexico, wholesale mar- kets were run by the state; in South Asia, the state heavily regulated food markets, and many African commodity markets and trade regimes were controlled by (para- statal) state organizations. In many of these countries, the state played an important role in agricultural production and marketing in the decades after independence from colonial power. Governments in Sub-Saharan Africa and South Asia were heavily involved in agricultural marketing and food processing through the creation of mar- keting boards, government-controlled cooperatives, and parastatal processing units. These government institutions were often monopoly buyers of agricultural products, especially for basic food crops and important export crops. This system of state intervention and control underwent tremendous changes in the 1980s and 1990s as a global process of liberalization induced dramatic changes in many of these regions. In the transition world, the liberalization of prices, trade, and exchanges, the privatization of state enterprises, and so forth removed much of the state control over commodity chains as well as the vertical coordination in the chains. Similar processes of privatization and liberalization of domestic and interna- tional commodity and financial markets reduced the control of the state over the food and agricultural chains in many developing and emerging economies. Johan F. M. Swinnen is Director and Miet Maertens is Senior Economist at the LICOS Centre for Institutions and Economic Performance and the Department of Economics, University of Leuven. This note summarizes the conclusions and arguments in Swinnen and Maertens (2007). Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 49 50 | JOHAN F. M. SWINNEN AND MIET MAERTENS Globalization Globalization of the food chains in transition and developing countries has been driven by several factors. Here we focus on four factors that are of special importance. First, trade liberalization caused major changes in the trade of agrofood products. For example, in Central and Eastern Europe, it caused a major reorientation of the agrofood trade from "east to west" (that is, from trade with the former Soviet coun- tries to trade with Western Europe) and a shift in the position of agrofood trade from net exporters to net importers. Second, liberalization of the investment regimes induced foreign investments in agribusiness, food industry, and further down the chain, with major implications for farmers (Dries and Swinnen 2004). Several food sectors in Eastern Europe, such as sugar, dairy, and retail, have received massive amounts of foreign investment, which now holds dominant market shares. An example is the rapid growth of modern retail chains ("supermarkets") in transition and developing countries, which was triggered by the reform process in former state-controlled economies (Reardon and Swinnen 2004). Third, in addition to increasing trade, the structure of this trade also changed con- siderably. There has been an increase in the share of high-value products--mainly fish and fishery products and fruits and vegetables--in world agricultural trade. Devel- oping countries experienced a particularly sharp increase in such high-value exports and a decrease in the importance of traditional tropical export commodities, such as coffee, cocoa, and tea. Fourth, associated with these changes is the spread of (private and public) food standards. Consumers are increasingly demanding specific quality attributes of processed and fresh food products and are increasingly aware of food safety issues. These food quality and safety demands are most pronounced in western markets (and increasingly in urban markets of low-income countries) but also affect traders and producers in transition and developing countries through international trade. The Fall and Rise of Vertical Coordination Vertical coordination (VC) was widespread in state-controlled food supply chains. Again, this was most extreme in the Communist system, where production at vari- ous stages and the exchange of inputs and outputs along the chain were coordinated and determined by the central command system (Rozelle and Swinnen 2004). However, vertical coordination also was widespread in other regions where the state played an important role in food chains. For example, many African govern- ment marketing organizations and parastatal processing companies used VC systems with upstream suppliers. The dominant form of state-controlled VC was the provi- sion of seasonal inputs and credit to small farmers in return for supplies of primary produce. Most analyses point at the deficiencies and inefficiencies of these systems. They are considered one of the primary causes of the inefficiency of the Soviet farming complex. GLOBALIZATION, PRIVATIZATION, AND VERTICAL COORDINATION | 51 In Africa, several studies conclude that state-controlled outgrower schemes were inef- ficient and poorly managed, which manifested itself, among other things, in low rates of credit repayment. Liberalization, Privatization, and the Breakdown of Vertical Coordination The system of vertical coordination underwent tremendous changes in the 1980s and 1990s. In the transition world, the liberalization of exchanges and prices and the pri- vatization of farms and enterprises caused the collapse of vertical coordination and caused major disruptions in the food chain. The disruptions in relationships of farms with input suppliers and food companies resulted in many farms facing serious constraints in accessing essential inputs (feed, fertilizer, seeds, capital). As government marketing boards and cooperatives ceased to play a major role in the procurement of agricultural produce, so did the provision of credit and agricultural inputs through state-controlled vertical coordination. The Emergence of Private Vertical Coordination Following privatization and liberalization, new forms of vertical coordination have emerged and are growing (Swinnen 2007; World Bank 2005). New forms of vertical coordination no longer are controlled by the state; instead they are introduced by pri- vate companies. Private traders, retailers, agribusinesses, and food-processing com- panies increasingly contract with farms and rural households to whom they provide inputs and services in return for guaranteed supplies of a certain quality. This process of interlinked contracts is growing rapidly in the transition and developing world. The emergence and spread of private VC is caused by the combination of, on the one hand, growing demand for products of high quality and safety standards, as a result of private sector investments and rising consumer income and demand (both domestically and through trade) and, on the other hand, the problems facing farms seeking to supply such products reliably, consistently, and in a timely manner to processors and traders due to a variety of market imperfections and poor public institutions. Effects of Private Vertical Coordination The emergence of private VC is often mentioned as a new engine for economic growth, rural development, and poverty reduction. In this section, we summarize the empirical evidence on the impact of vertical coordination in transition and develop- ing countries, distinguishing between efficiency effects and equity effects. Efficiency Effects The impact of private VC systems on productivity is difficult to quantify, as sev- eral other factors affect output simultaneously and as company-level information 52 | JOHAN F. M. SWINNEN AND MIET MAERTENS is difficult to obtain. Still, the evidence suggests that successful private VC has impor- tant positive effects, both direct and indirect. The direct impact is on the output and productivity of the processing company that initiates vertical contracting and of its suppliers involved in VC schemes. Supplying farmers have experienced beneficial effects on output, productivity, and product quality--and ultimately on incomes--through better access to inputs, timely payments, and improved productivity as a result of new investments. Case studies indicate that private VC programs can lead to strong growth in output, quality, and productivity. While successful state-controlled VC programs exist, case studies point out that state-controlled VC is generally less effective in fostering farm productivity than private VC. Indirect effects emerge through (a) cross-company spillover effects and (b) house- hold and farm spillover effects. Household and farm spillovers occur as households' risk declines, their access to capital increases, and the productivity of noncontracted activities increases. First, VC not only implies the provision of inputs, working capi- tal, and technical assistance to farmers, it also implies guaranteed sales, often at guar- anteed prices. This reduces the marketing risk for farmers. In addition, coordinating firms share in the production risk of farmers through ex ante provision of inputs and credit. Reduced production and marketing risks improve the stability of farmers' income, which is an important benefit for farmers operating in high-risk environ- ments and in the absence of insurance markets. Second, credit arrangements and prompt cash payments after harvest in VC programs improves farmers' cash flow and access to capital. This might ease farmers' financial constraints and benefit investment in other farm and nonfarm activities. This effect is particularly important in the presence of capital market imperfections. Third, contract farming can lead to productivity spillovers on other crops, resulting from management advice, access to improved technologies, better use of inputs, and so forth. Equity Effects There are two potential equity issues with VC processes. The first concerns the dis- tribution of rents in vertically coordinated food supply chains. The second con- cerns the participation in and exclusion of smallholders and poorer farmers from contract farming. Contract farming has often been criticized as being a tool for agro-industrial firms and food multinationals to exploit unequal power relationships with farmers and extract rents from the chain. However, our review of empirical evidence on the effects of VC indicates that farmers do share importantly in the benefits of contract farming and VC. The capacity of emerging VC in agrofood supply chains to serve as an engine of pro-poor economic growth critically depends on the type of farmers who are included in contract schemes. If agro-industrial firms prefer to contract with wealth- ier farmers, then poorer households will be excluded from direct benefits. There are three reasons why this might be so. First, transaction costs favor larger farms in supply GLOBALIZATION, PRIVATIZATION, AND VERTICAL COORDINATION | 53 chains. Second, when some amount of investment is needed in order to contract with or supply the company, small farms are often more constrained in their financial means for making necessary investments. Third, small farms typically require more assistance from the company per unit of output. However, there are also reasons why agro-industrial firms do contract with small- holders and poorer farmers. First, the most straightforward reason is that companies have no choice. In some cases, small farmers represent the vast majority of the poten- tial suppliers. Second, while processors may prefer to deal with large farms because of lower transaction costs in, for example, collection and administration, contract enforcement may be more problematic, and hence costly, with larger farms. Proces- sors repeatedly emphasize that farms' willingness to learn and a professional attitude are more important than size in establishing fruitful farm-processor relationships. Third, in some cases small farms may have substantial cost advantages. This is par- ticularly the case in labor-intensive, high-maintenance production activities with rel- atively small economies of scale. Fourth, processors may prefer a mix of suppliers in order not to become too dependent on a few large suppliers. Empirical observations show a very mixed picture of actual contracting, with many more small farms being contracted than predicted based on the arguments given above. Studies in Madagascar by Minten, Randrianarison, and Swinnen (2008) and in Senegal by Maertens and Swinnen (2006) find that high-value supply chains have important effects on poverty reduction. The effects on income, from both con- tract farming and agro-industrial employment, are significantly positive. This sug- gests that, as smallholder contract farming and large-scale industrial farming reach different groups of the poor, mixed VC systems can have major effects on poverty reduction. The Role of Competition Competition will affect both equity and efficiency in supply chains. First, competi- tion induces VC spillover effects across the sector as other processors are forced to introduce similar supplier assistance programs since suppliers may not want to deliver unless they get similar conditions. As such, competition is a key factor in encouraging innovation, productivity, and technological development. Second, farmers benefit from competition among processing firms. More competition leads to more equal rent sharing, which is reflected in higher producer prices and more services to farmers. However, there is a danger that if competition becomes too vigorous in the inter- linked input and credit market, coordination may break down. Several case studies report on input programs that collapsed due to competition. Still other studies show that input programs have remained sustainable under competition as a result of special insti- tutional arrangements, such as frequent monitoring, buyer coordination, or local infor- mation networks. An important area for further research is to analyze the conditions under which competition leads to beneficial outcomes, while avoiding VC failure. 54 | JOHAN F. M. SWINNEN AND MIET MAERTENS References Dries, Liesbeth, and Johan Swinnen. 2004. "Foreign Direct Investment, Vertical Integration, and Local Suppliers: Evidence from the Polish Dairy Sector." World Development 32 (9): 1525­44. Maertens, Miet, and Johan Swinnen. 2006. Trade, Standards, and Poverty: The Case of Vegetable Exports from Senegal. LICOS Discussion Paper 17706. Leuven: University of Leuven, Centre for Institutions and Economic Performance. Minten, Bart, Lalaina Randrianarison, and Johan Swinnen. 2008. "Global Retail Chains and Poor Farmers: Evidence from Madagascar." Forthcoming in World Development. Reardon, Thomas, and Johan Swinnen. 2004. "Agrifood Sector Liberalization and the Rise of Supermarkets in Former State-Controlled Economies: Comparison with Other Developing Countries." Development Policy Review 22 (4): 317­34. Rozelle, Scott, and Johan Swinnen. 2004. "Success and Failure of Reforms: Insights from Transition Agriculture." Journal of Economic Literature 42 (2): 404­56. Swinnen, Johan F. M., ed. 2007. Global Supply Chains, Standards, and the Poor. Oxford: Oxford University Press. Swinnen, Johan, and Miet Maertens. 2007. "Globalization, Privatization, and Vertical Coor- dination in Food Value Chains." Agricultural Economics 37 (1): 89­102. World Bank. 2005. The Dynamics of Vertical Coordination in Agrifood Chains in Eastern Europe and Central Asia: Implications for Policy Making and World Bank Operations. Washington, DC: World Bank. Part III: Agriculture, Natural Resources Management, and the Environment Designing Improved NRM Interventions in Agriculture for Poverty Reduction and Environmental Sustainability in Developing Countries H. ADE FREEMAN Agricultural production worldwide depends largely on natural resources such as land, water, pasture, fish, forests, and biodiversity. More than 55 percent of nonfor- est land and 80 percent of freshwater is used for agricultural production (World Bank 2004). Ecosystems such as cultivated lands, rangelands, forests, genetic resources, and oceans provide multiple uses and direct and indirect services that help to meet the growing demand for food, freshwater, fiber, and energy.1 The use of natural resources has contributed significantly to economic growth and human well-being in many parts of the world. Properly managed natural resources provide an essential foundation for reducing poverty and promoting sustainable economic growth. Yet there are serious concerns about the capacity of the base of natural resources to sus- tain current and future levels of agricultural growth in ways that improve the well- being and health of poor people. The objective of natural resource management (NRM) in agriculture is to enable efficient and sustainable use of ecosystem services that provide food and other direct and indirect benefits for human well-being.2 Of the 1.2 billion poor people across the developing world, more than 70 percent live in rural areas (Dixon, Gibbon, and Gulliver 2001; IFAD 2001). Natural assets, such as land, water, pasture, and trees, often make significant contributions to the livelihood of rural households. Degradation of natural resources is particularly costly for the poor, whose high level of dependence on natural resources for their liveli- hoods makes them extremely vulnerable when these resources lose their productive potential. Given projected increases in the demand for food, limited opportunities to expand natural resources (such as unexploited cultivable land and water), and grow- ing concerns about degradation of the natural resource base, there is considerable interest in promoting sustainable agricultural intensification that improves factor productivity in agriculture without further degrading the natural resource base. H. Ade Freeman is Director, Targeting R&D Opportunities Theme, International Livestock Research Institute in Nairobi, Kenya. Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 57 58 | H. ADE FREEMAN Sustainable agricultural intensification should also contribute directly to reducing poverty and vulnerability, particularly in rural areas where most poor people depend directly or indirectly on agriculture and other ecosystem services for their livelihood. This paper examines the links between agriculture, NRM, and the environment in developing countries. It addresses critical issues for understanding these links to iden- tify key implications and priorities for designing investments that stimulate the adop- tion of NRM interventions that reduce poverty. The paper reviews broad trends in agricultural production and human use of ecosystems in the provision of food and other services. The focus is at the household and farm level, but consideration is given to local and community issues that directly affect the choice of NRM investments. The next section summarizes key trends in agricultural productivity, human use of ecosystem services, and the state of degradation of ecosystem services, noting impor- tant differences across regions and agricultural production systems. This is followed by a discussion of key principles that need to be considered in designing improved NRM interventions, based largely on lessons learned from different agroecological and socioeconomic contexts in the developing world. The paper concludes with a summary of key implications for designing investments in sustainable NRM in agri- culture and challenges for scaling up promising technical and institutional innova- tions to enhance the poverty impacts of sustainable NRM investments. Trends in Agricultural Production and Use of Ecosystem Services Over the past 40 years, global food production has grown faster than overall popu- lation growth. Consequently, the per capita production of food and major agricul- tural commodities such as cereals and livestock has been steadily increasing in most parts of the developing world, except in Sub-Saharan Africa (see figure 1). In Sub- Saharan Africa, the main reasons for poor agricultural performance are rapid popu- lation growth and severe land degradation (Sánchez and others 1997). On the demand side, rising incomes, shifting consumption patterns, and popula- tion growth have raised per capita consumption of food in most areas of the world. The demand for cereals, which provide about half of the calories consumed by humans, has increased worldwide, although this growth has slowed in recent years. In the developing world, the annual growth in demand for cereal declined from 3.8 percent in 1967­82 to 2.7 percent in 1990­97 (Rosegrant and others 2001). Worldwide, consumption of livestock and livestock products has risen dramatically in recent decades. In the developing world, consumption of meat increased by 70 mil- lion tons and milk by 105 million tons between 1971 and 1995 (Delgado and others 1999). Correspondingly, the demand for meat in developing countries grew from 2.0 percent a year in 1967­82 to 5.2 percent a year in 1990­97. The recent trends in livestock consumption reflect an increasing shift in consumption from the developed to the developing world. Per capita consumption of fish is growing rapidly in the developing world, and total fish consumption has doubled since 1973 (Hassan, Scholes, and Ash 2005). These trends are expected to continue. Demand for cereal is projected DESIGNING IMPROVED NRM INTERVENTIONS | 59 FIGURE 1. Per Capita Agricultural Production Index in Select Regions, 1970­2000 Crops Livestock Per capita net crop production index Per capita net livestock production index 140 190 130 170 120 150 110 130 100 110 90 80 90 70 70 1970 1975 1980 1985 1990 1995 2000 1970 1975 1980 1985 1990 1995 2000 Sub-Saharan Africa Latin America and the Caribbean South Asia Source: FAO (2005). to increase from 1.9 billion tons in 1997 to 2.5 billion tons in 2020, and demand for meat is projected to increase from 209 million to 327 million tons (Rosegrant and others 2001). The projected increase in the demand for food and fiber has important implications for the management of natural resources. Hassan, Scholes, and Ash (2005) comprehensively analyze trends in the human use of ecosystem services and implications for the enhancement or degradation of the services these systems provide (see table 1). A key message from their assessment is that the rapid growth of the agricultural sector has contributed significantly to economic growth and human well-being, but that the gains in human well-being have, in some cases, come from the unsustainable use of natural resources or have degraded the ecosystem and its underlying capacity to provide services. Climate change remains a continuing threat to livelihoods in the developing world, and its impact on vulnerability and poverty is likely to be considerable in places where populations are highly dependent on natural resource activities, such as agriculture, for their livelihood. Many countries in the developing world, particularly in Africa, also have limited financial and institutional ability and ineffective social protection programs with which to cope with and adapt to patterns of highly variable temper- ature and extreme weather that result from the complex and highly uncertain impacts of climate change (Thornton and others 2006). This presents considerable scientific and development challenges for many societies and poor people in the developing world. The remarkable growth in agricultural productivity in much of the developing world has taken place in more favored agricultural systems, both irrigated and rainfed, with high production potential, good market access, and adequate provision of key public goods such as roads and agricultural support services. In these systems agri- cultural growth has been driven largely by intensive use of land, irrigation, fertilizer, 60 | H. ADE FREEMAN TABLE 1. Trends in the Human Use of Ecosystem Services Ecosystem Human Enhanced service use or degraded Notes Crops Food provision has grown faster than overall population. The primary source of growth is the increase in production per unit area but also the significant expansion in cropland. Areas of low productivity and areas of rapid expansion persist, for example, in Sub-Saharan Africa and parts of Latin America. Livestock The area devoted to livestock has increased significantly in some regions, but a major source of growth has been more intensive, confined production of chicken, pigs, and cattle. Fisheries Marine fish harvest increased until the late 1980s and has been declining since then. Currently, one-quarter of marine fish stocks are overexploited or significantly depleted. Freshwater capture fisheries have also declined. Human use of capture fisheries has declined because of reduced supply, not reduced demand. Timber / Global timber production has increased 60 percent in the last four decades. Plantations provide an increasing volume of harvested roundwood, amounting to 35 percent of the global harvest in 2000. Roughly 40 percent of forest areas has been lost during the industrial era, and forests continue to be lost in many regions (and the service has been degraded), although forests have been recovering in some temperate countries (and the service has been enhanced, albeit from a lower baseline) in recent decades. Genetic Traditional crop breeding has relied on a relatively narrow resources range of germ plasm for the major species, although molecular genetics and biotechnology provide new tools to quantify and expand genetic diversity in these crops. Use of genetic resources also is growing in connection with new industries based on biotechnology. Genetic resources have been lost through the loss of traditional cultivars of crop species (due in part to the adoption of modern farming practices and varieties) and through species extinctions. Source: Hassan, Scholes, and Ash (2005). and other agrochemicals and is often supported by strong government policies. Resource-intensive agricultural development in more favored areas has often resulted in environmental degradation, particularly where incentives encourage overex- ploitation or mismanagement of natural resources. Environmental degradation also occurs in more favored areas where policy and institutional support mechanisms to DESIGNING IMPROVED NRM INTERVENTIONS | 61 internalize the costs of negative externalities are weak or nonexistent. In Pakistan, for example, mismanagement of irrigation costs more than $200 million a year in lower food yields (Bass and Steele 2006). The intensification of rice-wheat systems in the Indo-Gangetic plains of South Asia has been linked to the degradation of natural resources, such as salinity and waterlogging, depletion of groundwater resources, and deficiencies in soil nutrients (Pingali and Rosegrant 1998). In the live- stock sector, the recent doubling of meat and milk production in the developing world is associated with significant negative impacts on the environment, such as water pollution, greenhouse gas emissions, and loss of biodiversity (World Bank 2005b). In many areas with medium and low agroecological potential, weak market inte- gration, and limited investments in public goods, agricultural productivity is low and natural resources are often highly degraded. Rapid rates of population growth have considerably weakened the capacity of traditional self-sustaining systems to cope, leading to downward spirals into poverty, rapid depletion of the natural resource base, and growing conflict over access to land and water in many pastoral and agropastoral systems in Sub-Saharan Africa and Asia (ILRI 2006). In large parts of Sub-Saharan Africa, soils have highly negative balances of nutrients--nitrogen, phos- phorus, and potassium--and soil quality is so depleted that the adoption of improved crop germ plasm brings limited gains in productivity. By 2010, fallow lands will have disappeared altogether in 20 African countries (Crawford, Jayne, and Kelley 2006; Sánchez and others 1997). The challenge of meeting future demand for food and other environmental serv- ices in ways that sustain the resource base is a key issue on the development agenda. This agenda takes into account not only the need to increase productivity and incomes for current generations but also the equally important need to maintain and improve the stock and quality of resources to meet the needs and enhance the well-being of future generations. Key Principles Guiding Sustainable NRM Interventions Empirical evidence from the developing world shows that NRM innovations that conserve and productively manage natural resources can make significant contribu- tions to agricultural productivity, environmental sustainability, poverty reduction, and human welfare (Barrett and others 2002; Bass and Steele 2006). In cases where public and private investments in NRM in agriculture are limited, the resulting degra- dation of natural resources comes with real economic, environmental, and social costs. The growing consensus that the objectives of poverty reduction, food security, and sustainable management of natural resources are highly interdependent explains, in part, the resurgence of development lending to NRM and environmental projects over the past two decades (Barrett 2003; Kelley and Gregersen 2005; World Bank 2003). Lessons from a range of agricultural production systems and socioeconomic and public goods contexts suggest the following key principles for designing improved 62 | H. ADE FREEMAN NRM practices in agriculture. The focus here is on medium- to low-potential agri- cultural systems, which characterize large areas of Sub-Saharan Africa and Asia. NRM Is an Investment Choice Natural assets, such as land and water, are key household assets that should provide a return. Investments in improved NRM sustain natural assets, enhance their pro- ductivity, and involve real opportunity costs. Rural households evaluate the relative returns and risks of alternative activities when making investment decisions. Improved land productivity, for example, typically involves increased investment in NRM prac- tices, such as enhanced soil fertility or water management techniques. But some rural households may not adopt these technologies if the returns to them accrue at a later date or are lower than the returns to investments in alternative off-farm or nonfarm activities. The design of improved NRM techniques therefore needs to consider several important characteristics that influence investment choices, such as prices, property rights, risk preferences, externalities, and market failures (Barrett and others 2002; Freeman and Ellis 2005). NRM Investments Must Recognize the Linkages between Agriculture and the Nonfarm Economy Recent empirical evidence from Africa, Asia, and Latin America shows that rural households obtain their livelihoods from a diverse portfolio of income sources (Barrett, Reardon, and Webb 2001; Ellis 2000; Ellis and Freeman 2004; López and Valdes 2000; Reardon 1997; Thanh, Anh, and Tacoli 2005). In many developing countries rural households pursue multiple locational and occupational strategies as they seek to improve their livelihoods (Ashley and Maxwell 2001). Households' asset position and the quality of the context in which those assets are used need to be given much more attention when designing NRM interventions aimed at increas- ing the productivity of natural assets, particularly those that are used by poor people (de Janvry and Sadoulet 2005). The design of NRM interventions needs to consider the broader livelihood strate- gies of households. Diversification into the rural nonfarm economy can have positive and negative impacts on the adoption of improved NRM practices (Barrett and others 2002; Ellis and Freeman 2004; Thanh, Anh, and Tacoli 2005). In the LADDER studies,3 land productivity, measured by mean net agricultural output per hectare, is positively correlated with income. Nonfarm income enables the household to hire labor to undertake timely cultivation practices and provides cash to buy farm inputs. Conversely, when poor households hire out labor, their own farm productivity stag- nates or declines (Ellis and Freeman 2004). Participatory Processes Enhance the Likelihood of Success Client participation is an essential ingredient in successful NRM investment. In recent years, technology development has emphasized farmer participatory research due, in DESIGNING IMPROVED NRM INTERVENTIONS | 63 part, to the high location specificity of NRM issues. Farmer and community involve- ment in the design, testing, development, and evaluation of improved NRM prac- tices encourages experimentation and rapid adoption and adaptation of promising technologies (Freeman 2001). Successful water resource management projects in Southern Africa and watershed management projects in India have shown that com- munities are more likely to implement NRM projects and maintain them when user groups have been effectively represented in NRM institutions and communities have participated in planning and decision making (Hirji, Mackay, and Maro 2002; Kerr and Chung 2005). NRM Technologies Are Knowledge Intensive NRM technologies are inherently knowledge intensive. Unlike crop production tech- nologies, improved NRM technologies involve management practices that are not easily embodied in discrete inputs such as seeds, agrochemicals, or machinery (Barrett and others 2002). Improved NRM practices such as integrated pest management, conservation agriculture, and integrated nutrient management offer promising oppor- tunities for increasing factor productivity and the efficiency of resource use, but they are knowledge intensive. This is partly because agricultural production practices are interrelated. Applying inorganic fertilizer to enhance soil fertility and raise crop yields requires paying proper attention to improved cultivation practices such as appro- priate placement and timely planting and weeding. Similarly, high payoffs to improv- ing water control require effective application of fertilizer as well as timely planting and weeding (Poulton, Kydd, and Dorward 2006). Farmers therefore need to man- age multiple resources (and information) effectively to achieve high payoffs from NRM technologies. Improved NRM Practices Must Be Profitable Rural households in developing countries allocate their scarce resources within farm- ing and between farming and nonfarm activities according to the relative returns to resource use and subjective assessment of the riskiness of alternative investments. Indeed, many rural households are part-time farmers, part-time workers, and microentrepreneurs (de Janvry and Sadoulet 2000). Lessons from across the devel- oping world indicate that farmers do not invest in improved NRM practices when the returns are lower than those to alternative activities, both on and off farm (Barrett and others 2002). Farmers who cultivate high-value crops or link into dynamic food chains, such as supermarkets or agro-industry, are more likely to adopt improved NRM practices than farmers who grow low-value food crops. Institutional innovations that make natural resource inputs more affordable also help to address cash constraints that limit their use. In the case of inorganic fertilizer, such innova- tions include packaging fertilizer in small amounts, offering seasonal credit, and dis- tributing vouchers (Freeman and Omiti 2003; Omamo and Mose 2001; Poulton, Kydd, and Dorward 2006).4 Although there seems to be a resurgence of the case for subsidies that make fertilizer more affordable for smallholder farmers,5 there is little 64 | H. ADE FREEMAN empirical evidence on the cost and benefits of fertilizer subsidies compared with alter- natives such as small packs and seasonal credit. The returns to NRM investments also require institutional innovations, including marketing arrangements that reduce transaction costs in trade; clear and secure property rights that encourage long-term investments; affordable arrangements for the implementation of grades and stan- dards and the enforcement of contracts; collective action for the management of common property and the pooling of resources to make NRM investments attrac- tive; and financial market innovations to reduce the impact of cash constraints and provide alternative instruments for saving (Barrett and others 2002; Kelley and Byerlee 2003). Implications for NRM Investments Understanding the nexus of agriculture, NRM, the environment, and its poverty dimensions leads to the identification of guiding principles for designing NRM inter- ventions. Several implications can be inferred for designing and scaling up sustain- able NRM interventions for poverty reduction. The Shift from a Farm Focus to a Livelihoods Focus The multiple and diverse character of rural livelihoods highlights the importance of taking a broad livelihoods approach that explicitly considers the heterogeneity of household asset positions in designing improved NRM interventions. The multidi- mensionality of livelihoods implies that each livelihood activity generates opportuni- ties for some and places constraints on others. Households engaged in wage labor and without the resources to hire labor may record lower crop yields because they cannot undertake timely cultivation on their farms. NRM interventions that exploit complementarities in livelihood activities and minimize competition among the liveli- hood objectives of households are likely to be more attractive. The design of NRM interventions therefore needs to pay more attention to the opportunities and constraints implied by household behavior. Targeted NRM Interventions Although there is considerable evidence of strong links between agricultural growth and poverty reduction, the heterogeneity of household assets and differences in the quality of contexts--markets, governance, policy, and public goods--in which assets are used imply that NRM interventions have to be targeted to increase agricultural and asset productivity and enhance the impact on the poor. It is increasingly recog- nized that one-size-fits-all strategies do not work in the complex economic, social, and agroecological contexts in which poor people operate. Development strategies need to be tailored to capture the diversity that exists in agricultural production sys- tems (Byerlee and Alex 2003; Dixon, Gibbon, and Gulliver 2001; World Bank DESIGNING IMPROVED NRM INTERVENTIONS | 65 2005a). Key elements that define comparative advantage, including agricultural potential, access to markets, infrastructure, and population density, have been used spatially to characterize agricultural systems and identify location-specific development strategies (Chamberlin, Pender, and Bingxin 2006; Ehui and Pender 2005; Omamo and others 2005). Multiple pathways out of poverty or household strategies to improve livelihoods have been identified for different development contexts. Context-specific strategies and priorities for policy, institutional change, and public investments for poverty reduction have been identified for specific domains or farming systems (Chamberlin, Pender, and Bingxin 2006; Dixon, Gibbon, and Gulliver 2001; Ehui and Pender 2005; Kelley and Byerlee 2003; World Bank 2005a). Spatial analytical tools, such as geographic information systems, and databases, such as satellite-based imagery, are greatly facilitating the use of spatial characterization in guiding the planning and implementation of NRM investment. Investment in Human Capital in Rural Areas The knowledge intensity of improved NRM interventions implies that there are sig- nificant payoffs to investments that build human capital in rural areas. Such invest- ments need learning processes that build the capacity of rural people to learn about the complex ecological and biophysical contexts on their farms, provide improved access to information and practices that foster ecologically sustainable NRM, and support farmer experimentation and evaluation of improved NRM practices. Given that the NRM inputs will increasingly be delivered by the private sector and civil society organizations such as nongovernmental organizations (NGOs), it becomes increasingly important to improve the technical knowledge of input suppliers and NGO staff so that they can advise farmers on input use. This changing context implies a different role for public extension systems, most of which face severe budg- etary constraints and barely function in many rural areas, particularly in Sub-Saharan Africa. Public extension systems need to consider shifting their limited resources and undertaking efforts to improve the capacity of other players so that they can provide effective advisory services for farmers. Institutional Arrangements for NRM Innovations Technological change in NRM takes place through institutions, whether for the development and dissemination of research results or for the adoption of improved practices. To foster change, it is critical to understand the nature and functioning of institutions and how they could be strengthened. One approach is to form broad public sector partnerships with both traditional and new partners, including NGOs and the private sector (World Bank 2005a). The partnership mode that supports and delivers services effectively to farmers will differ according to the context in which household assets are used. The private sector is likely to have strong incentives to deliver services and support farmers where population density is high, markets are well developed, and risks are low. In areas where private sector incentives are low, 66 | H. ADE FREEMAN NGOs and other civil society organizations may partner with the public sector to deliver services to farmers. Where the benefits derived from collective action are greater than the benefits from individual action, group-based approaches through producer or farmer organizations can be used to improve bargaining power, access to resources, information, and markets for smallholder producers as well as to empower those with limited resources or political voice. Another approach is to shift the focus to more adaptive institutional frameworks that support institutional learn- ing and innovation. Rather than pursuing a single "ideal" model of market structure and management, institutions need to be constantly evolving, searching out new alliances and partners as new opportunities arise. Institutional learning processes have to be made more explicit. Institutions have to learn to acknowledge and diag- nose their problems. They also have to pay more attention to monitoring and evalu- ation, not simply to justify past work, but to learn, make adjustments, and identify future opportunities. The Enabling Environment Demand-led approaches such as community-driven development and decentral- ization emphasize the need to organize NRM research and development as par- ticipatory processes. This shifts the focus to local and district-level planning and implementation. Decentralized, demand-led processes should, however, not be seen as an act of faith but as a parallel process of promoting efficient organiza- tions that are based on effective leadership and build accountable rural institu- tions and service delivery. A key implication is that decentralization cannot be separated from broader political and governance issues in developing countries. Legal processes and institutional innovations for property rights need to ensure that they do not discriminate against the poor and women, particularly in societies where social relationships, such as kinship and marital norms, take precedence over legal rights. Programs for Social Protection The importance of risk and vulnerability and their links to natural resource degra- dation in many areas with poor agricultural potential, underdeveloped markets, and poor infrastructure implies that there may be high payoffs to social protec- tion programs that improve the capacity of households to manage risk at low opportunity cost. Social protection programs that promote NRM interventions and enhance the impacts on poverty reduction should be seen as more than mech- anisms for transferring resources to the poor. There are good opportunities for com- bining strategies that focus on growth with social protection in less favored areas (Farrington and Gill 2002). The key challenge is to identify synergies between growth-promoting strategies, such as NRM interventions, and social protection strategies that promote growth. This will require rethinking the approaches, strate- gies, and options for managing risk and vulnerability to focus on the complementar- ities between social protection and the promotion of pro-poor growth (Farrington, Slater, and Holmes 2004). DESIGNING IMPROVED NRM INTERVENTIONS | 67 Conclusions This paper has examined the links between agriculture, natural resource manage- ment, and the environment, noting that the effective design and scaling up of NRM interventions in agriculture should contribute to reducing poverty and vulnerability, particularly in rural areas where the majority of poor people in the developing world live. Increasing food production and consumption has, in many cases, weakened the ability of ecosystems to deliver key services that contribute to improved human well- being. This trend needs to be reversed in order to support economic growth and the well-being of current and future generations. Improved NRM interventions, includ- ing technologies and institutions, are critical to advancing the global goal of reducing poverty in ways that sustain natural resources. Smallholder farmers in the developing world use improved NRM practices when the context of incentives and public goods supports their use. There are many successful cases where NRM institutions have benefited poor people and communities. These experiences need to be replicated over several millions of households in the developing world if they are to contribute meaningfully to sustainable poverty reduction. The discussion of key principles and implications of lessons learned in the developing world suggests that the process of reducing poverty through NRM interventions in agriculture is complex and there are multiple entry points for development interventions. With agriculture back on the development agenda, development institutions have exciting new opportunities to enhance the poverty impacts of NRM investments in agriculture. Endnotes 1. An ecosystem is a dynamic complex of plant, animal, and microorganism communities and nonliving environment interacting as a functional unit. Humans are an integral part of the ecosystem. Ecosystem services are defined as the benefits people obtain from ecosystems (Hassan, Scholes, and Ash 2005). 2. NRM in agriculture refers to human administration and sustainable use of biophysical resources for the production of food, feed, fiber, and fuel. 3. LADDER (Livelihoods and Diversification Directions Explored by Research) is a rural livelihoods research program conducted in Kenya, Malawi, Tanzania, and Uganda from 2000 to 2003. 4. In some cases vouchers have been referred to as smart subsidies because they can have positive spin-offs on market development. 5. See, for example, the recommendations of the United Nations Millennium Project Hunger Task Force. References Ashley, Caroline, and Simon Maxwell. 2001. "Rethinking Rural Development." Development Policy Review 19 (4): 395­425. Barrett, Christopher B. 2003. Natural Resource Management in the CGIAR: A Meta-Evaluation in the CGIAR at 31. Washington, DC: World Bank. 68 | H. ADE FREEMAN Barrett, Christopher B., John Lynam, Frank Place, Thomas Reardon, and Abdillahi A. Aboud. 2002. "Towards Improved Natural Resource Management in African Agriculture." In Natural Resources Management in African Agriculture: Understanding and Improving Current Practices, ed. Christopher B. Barrett, Frank Place, and Abdillahi A. Aboud. Wallingford: CAB International. Barrett, Christopher B., Thomas Reardon, and Patrick Webb. 2001. "Nonfarm Income Diver- sification and Household Livelihood Strategies in Rural Africa: Concepts, Dynamics, and Policy Implications." Food Policy 26 (5): 315­31. Bass, Stephen, and Paul Steele. 2006. "Managing the Environment for Development and to Sustain Pro-Poor Growth." IDS Bulletin 37 (3, May): 7­16. Byerlee, Derek, and Gary Alex. 2003. "Designing Investments in Agricultural Research for Enhanced Poverty Impacts." Agriculture and Rural Development Working Paper 6, World Bank, Washington, DC. Chamberlin, Jordan, John Pender, and Yu Bingxin. 2006. "Development Domains for Ethiopia: Capturing the Geographical Context of Smallholder Development Options." Paper presented at Earth System Science Partnership (ESSP) Policy Conference, Addis Ababa, June 6­8. Crawford, Eric W., Thomas S. Jayne, and Valerie A. Kelley. 2006. Alternative Approaches for Promoting Fertilizer Use in Africa. Agriculture and Rural Development Discussion Paper 22. Washington, DC: World Bank. de Janvry, Alain, and Elisabeth Sadoulet. 2000. "Rural Poverty in Latin America: Determinants and Exit Paths." Food Policy 25 (4): 389­409. ------. 2005. "Achieving Success in Rural Development: Toward Implementation of an Inte- gral Approach." In Reshaping Agriculture's Contributions to Society, ed. David Colman and Nick Vink. Proceedings of the twenty-fifth International Conference of Agricultural Economists, Durban, South Africa, August 16­23, 2003. Delgado, Christopher, Mark Rosegrant, Henning Steinfeld, Simeon Ehui, and Claude Curbois. 1999. Livestock to 2020: The Next Food Revolution. Washington, DC: International Food Policy Research Institute. Dixon, John A., David P. Gibbon, and Aidan Gulliver. 2001. Farming Systems and Poverty: Improving Farmers' Livelihoods in a Changing World. Rome: Food and Agriculture Organization. Ehui, Simeon, and John Pender. 2005. "Resource Degradation, Low Agricultural Productivity, and Poverty in Sub-Saharan Africa: Pathways out of the Spiral." In Reshaping Agricul- ture's Contributions to Society, ed. David Colman and Nick Vink. Proceedings of the twenty-fifth International Conference of Agricultural Economists, Durban, South Africa, August 16­23, 2003. Ellis, Frank. 2000. Rural Livelihoods and Diversity in Developing Countries. Oxford: Oxford University Press. Ellis, Frank, and H. Ade Freeman. 2004. "Rural Livelihoods and Poverty Reduction Strategies in Four African Countries." Journal of Development Studies 40 (4): 1­30. FAO (Food and Agriculture Organization). 2005. FAOSTAT. Rome. Farrington, John, and Gerard J. Gill. 2002. Combining Growth and Social Protection in Weakly Integrated Rural Areas. Natural Resource Perspectives 79. London: Overseas Development Institute. Farrington, John, Rachel Slater, and Rebecca Holmes. 2004. Social Protection and Pro-poor Agricultural Growth: What Scope for Synergies? Natural Resource Perspectives 91. London: Overseas Development Institute. DESIGNING IMPROVED NRM INTERVENTIONS | 69 Freeman, H. Ade. 2001. Comparison of Farmer-Participatory Research Methodologies: Case Studies in Malawi and Zimbabwe. Patancheru: International Crops Research Institute for the Semi-Arid Tropics. Freeman, H. Ade, and Frank Ellis. 2005. "Implications of Livelihood Strategies for Agri- cultural Research: A Kenya Case Study." In Rural Livelihoods and Poverty Reduction Policies, ed. Frank Ellis and H. Ade Freeman, 198­212. London: Routledge. Freeman, H. Ade, and John M. Omiti. 2003. "Fertilizer Use in Semi-Arid Areas of Kenya: Analysis of Smallholder Farmers' Adoption Behavior under Liberalized Markets." Nutrient Cycling in Agroecosystems 66 (1): 23­31. Hassan, Rashid., Robert Scholes, and Neville Ash. 2005. Ecosystems and Human Well-Being. Vol. 1: Current State and Trends. Millennium Ecosystem Assessment. Washington, DC: Island Press. Hirji, Rafik, Heather Mackay, and Paul Maro. 2002. Defining and Mainstreaming Environ- mental Sustainability in Water Resources Management in Southern Africa: A Summary. Washington, DC: World Bank. IFAD (International Fund for Agricultural Development). 2001. Rural Poverty Report, 2001. Rome: IFAD. ILRI (International Livestock Research Institute). 2006. Lessons from a Changing World: Implications for Livestock Research and Development. Nairobi: ILRI. Kelley, T. G., and Dereck Byerlee. 2003. Surviving on the Margin: Agricultural Research and Development Strategies for Poverty Reduction in Marginal Areas. Draft paper, Agriculture and Rural Development Department, World Bank, Washington, DC. Kelley, T. G., and H. M. Gregersen. 2005. "NRM Impact Assessment in the CGIAR: Meeting the Challenge and Implications for CGIAR Centers." In Natural Resource Management in Agriculture: Methods for Assessing Economic and Environmental Impacts, ed. Bekele Shiferaw, H. Ade Freeman, and Scott M. Swinton. Wallingford: CAB International. Kerr, John M., and Kimberly R. Chung. 2005. "Evaluating the Impacts of Watershed Man- agement Projects: A Practical Econometric Approach." In Natural Resource Management in Agriculture: Methods for Assessing Economic and Environmental Impacts, ed. Bekele Shiferaw, H. Ade Freeman, and Scott M. Swinton. Wallingford: CAB International. López, Ramón, and Alberto Valdes. 2000. "Fighting Rural Poverty in Latin America: New Evidence and Policy." In Rural Poverty in Latin America, ed. Ramón López and Alberto Valdes. London: Palgrave Macmillan. Omamo, Steven, Xinshen Diao, Stanley Word, Jordan Chamberlin, Liang You, Sam Benin, and Peter Hazell. 2005. Strategic Priorities for Agricultural Development and Agricultural Research-for-Development in Eastern and Central Africa. Entebbe: Association for Strengthening Agricultural Research in Eastern and Central Africa; Washington, DC: International Food Policy Research Institute. Omamo, Steven W., and Lawrence O. Mose. 2001. "Fertilizer Trade under Market Liberal- ization: Evidence from Kenya." Food Policy 26 (1): 1­10. Pingali, Prabhu L., and Mark Rosegrant. 1998. "Supplying Wheat for Asia's Increasingly Westernized Diets." American Journal of Agricultural Economics 80 (5): 954­59. Poulton, Colin, Jonathan Kydd, and Andrew Dorward. 2006. Increasing Fertilizer Use in Africa: What Have We Learned? Agriculture and Rural Development Discussion Paper 25. Washington, DC: World Bank. Reardon, Thomas. 1997. "Using Evidence of Household Income Diversification to Inform the Study of the Rural Nonfarm Labor Market in Africa." World Development 25 (5): 735­48. 70 | H. ADE FREEMAN Rosegrant, Mark W., Michael S. Paisner, Siet Meijer, and Julie Witcover. 2001. Global Food Projections to 2020. Washington, DC: International Food Policy Research Institute. Sánchez, Pedro A., K. D. Shepherd, M. J. Soule, F. M. Place, R. J. Buresh, A. N. Izac, A. U. Mokwunye, F. R. Kwesiga, G. G. Ndiritu, and P. L. Woomer. 1997. "Soil Fertility Replen- ishment in Africa: An Investment in Natural Resource Capital." In Replenishing Soil Fertility in Africa, ed. R. J. Buresh, Pedro A. Sánchez, and Frank G. Calhoun. Madison, WI: Soil Science Society of America. Thanh, Hoang X., Dang N. Anh, and Cecilia Tacoli. 2005. Livelihood Diversification and Rural-Urban Linkages in Vietnam's Red River Delta. FCND Discussion Paper 193. Washington, DC: International Food Policy Research Institute. Thornton, Philip K., P. G. Jones, T. M. Owiyo, R. L. Kruska, M. Herrero, P. Kristjanson, N. Bekele, and A. Omolo. 2006. Mapping Climate Vulnerability and Poverty in Africa. Nairobi: International Livestock Research Institute. World Bank. 2003. Environment Matters: Annual Review, July 2002­June 2003. Washington, DC: World Bank. ------. 2004. Agriculture Investment Sourcebook. Washington, DC: World Bank. ------. 2005a. Agricultural Growth for the Poor: An Agenda for Development. Washington, DC: World Bank. ------. 2005b. Managing the Livestock Revolution: Policy and Technology to Address the Negative Impacts of a Fast-Growing Sector. Washington, DC: World Bank. Aspects of Land Degradation in Lagging Regions: Extent, Driving Forces, Responses, and Further Research with Special Reference to Ethiopia GUNNAR KÖHLIN Land degradation is a major factor behind the low agricultural productivity of Sub- Saharan Africa. Poverty in Africa is still predominantly rural, and a majority of the rural population is dependent on the agricultural sector. This puts land degradation at the center of an agricultural-poverty nexus. Dealing with land degradation is cen- tral to addressing both poverty alleviation and sustainable development. This paper reviews some of the estimates that have been made of the cost of land degradation in various African countries. Although the estimates differ in their cover- age and methodology, it is evident that land degradation constitutes a drain on the sector, at least on the order of 2 to 3 percent loss in agricultural GDP.1 Although land degradation is recognized as a major development issue, sustainable land management (SLM) has not received commensurate attention in the development agenda of Sub- Saharan African countries. There are many reasons for this. Particularly relevant for this paper is the current attention given by the TerrAfrica initiative at the World Bank to the fact that there is an unorganized body of knowledge and weak analytical under- pinnings to support such decision-making processes. As a result, the scale and scope of investments in SLM programs continue to be limited. In Ethiopia during the 1990s, for example, only $15 million of government spending was allocated to the agricul- tural sector, including natural resources conservation. This amounts to less than 3 per- cent of government spending during this period (EPA 2004). The situation is almost as bleak for overseas development assistance. According to the Development Assis- tance Group, only about 7 percent of overseas development assistance currently goes to the agricultural sector in Ethiopia, and only a fraction of this is devoted to SLM. Gunnar Köhlin is Associate Professor with the Environmental Economics Unit, Department of Economics, Göteborg University. The content of this paper draws heavily on work carried out in the following projects: Sida (Swedish Inter- national Development Cooperation Agency)-supported Ph.D. program in environmental economics at the Department of Economics, Göteborg University; Sida-supported Sustainable Land Management Research Project in Ethiopia; World Bank economic sector work on poverty and land degradation in Ethiopia (in particular, Yesuf and Pender 2005; Yesuf and others 2005); the World Bank and Global Environment Facility, "TerrAfrica Project: Cost Benefit Framework for Pro-Sustainable Land Management Decision Making in Sub-Saharan Africa." Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 71 72 | GUNNAR KÖHLIN In support of such a hands-off policy, it could be argued that agriculture is an intrinsically private undertaking and that the farmer, as a rational entrepreneur, will invest in SLM technologies as long as the net present value of such investments is positive. However, public sector involvement in this sector is warranted for a number of reasons, offering another rationale for why the WDR 2008 is dealing with this issue. First, these are extremely long-term investments with intergenerational implica- tions. Given that large parts of Sub-Saharan Africa have had negative growth over the last decades, it is obvious that private and social discount rates, and subsequently investments, will differ. This situation is aggravated by the fact that the sector is char- acterized by missing or imperfect markets, for example, for labor, credit, and land. There are also a number of important externalities related to land degradation, includ- ing sedimentation of irrigation facilities, freshwater, and hydro dams, with long-term implications for economic development. Finally, land degradation can lead to damages that are irreversible, thus indefinitely decreasing the set of choices available for the optimal use of land and violating most definitions of sustainable development. Government and international involvement in dealing with land degradation is therefore warranted and badly needed. It is less clear how much should be invested in this sector and how. In this paper we review part of the adoption literature deal- ing with SLM technologies. The review shows that the results from past interventions are very mixed and it is not even clear what are the most important factors affecting adoption. The paper therefore argues for more careful designs of future interventions and recommends that such designs make better use of research stemming from soil science, agriculture, and economics. The Extent of Land Degradation in Africa Land degradation is most often associated with soil erosion.2 However, the concept is much broader than that. The United Nations Convention to Combat Desertification defines land as "the terrestrial bio-productive system that comprises soil, vegetation, other biota, and the ecological and hydrological processes that operate within the system" and land degradation as "reduction or loss . . . of the biological or economic productivity and complexity of rainfed cropland, irrigated cropland, or range, pasture, forest, and woodlands resulting from . . . processes . . . such as (i) soil erosion caused by wind and/or water; (ii) deterioration of the physical, biological, or economic prop- erties of the soil; and (iii) long-term loss of natural vegetation" (Pagiola 1999). Soil degradation is a narrower term for declining soil quality, encompassing the deterioration in physical, chemical, and biological attributes of the soil (Enters 1998). Soil erosion is a particular physical process that causes land and soil degradation and refers to the wearing away of the land surface by water or wind as well as the reduc- tion in soil productivity due to physical loss of topsoil, reduction of rooting depth, removal of plant nutrients, and loss of water (Enters 1998). Other forms of soil degradation include other types of physical degradation, such as compaction, surface sealing and crusting, waterlogging, and aridification; chemical degradation, including depletion of soil nutrients, acidification, salinization, and pollution; and biological ASPECTS OF LAND DEGRADATION IN LAGGING REGIONS | 73 degradation, including loss of soil organic matter (which also affects physical and chemical properties of the soil), flora and fauna populations, or species in the soil, such as earthworms, termites, and microorganisms (Scherr 1999). Land degradation is thus a complex phenomenon influenced by natural and socio- economic factors. Most cost estimates of land degradation do not distinguish between soil erosion, soil degradation, and land degradation. In many economic analyses, there is a tendency to attribute a decline in soil fertility exclusively to soil erosion. Erosion is treated as the sole factor contributing to soil or land degradation and yield declines, as the impacts of nutrient depletion on crop yields are underestimated or completely neglected. In fact, soil erosion could be a major component of on-site costs, particu- larly on steeper slopes (Kerr and Pender 2005). However, in many developing coun- tries, especially in Africa, soil nutrient mining is also a very serious problem (Bojö and Cassells 1995; Stoorvogel, Smaling, and Janssen 1993). Research on soil nutrient balances in Africa highlights that soil fertility is reduced not only by erosion but also by the removal of harvested crop parts and residues, leaching, and volatilization or denitrification losses (Smaling, Fresco, and De Jager 1996; Stoorvogel, Smaling, and Janssen 1993). The fact that different studies estimate different components of land degradation makes comparison of cost estimates across studies a difficult task. Table 1 summarizes a selection of studies that deal with various aspects of land degradation in Africa. In Yesuf and others (2005), a careful comparison is made TABLE 1. Cost of Soil Erosion Estimates from Sub-Saharan African Countries On-site cost of soil erosion Study Country (% agricultural GDP) Hurni (1988) Ethiopia 2 Sutcliffe (1993) Ethiopia 7 Bojö and Cassells (1995) Ethiopia 3 Sonneveld (2002) Ethiopia 3 Convery and Tutu (1990) Ghana 5 Pagiola (1993) Kenya -- Bojö (1991) Lesotho 5 World Bank (1988) Madagascar 1 World Bank (1992) Malawi 18 Bishop (1995) Malawi 17­55 Eaton (1996) Malawi 3 Bishop and Allen (1989) Mali 4­16 Bishop (1995) Mali 3­13 Ehui, Hertel, and Preckel (1990) Western Nigeria -- McKenzie (1994) South Africa 4 Stocking (1986) Zimbabwe 9 Norse and Saigal (1992) Zimbabwe 8 Grohs (1994) Zimbabwe 0.4 Source: Adapted from Yesuf and others (2005). Note: -- not available. 74 | GUNNAR KÖHLIN between these studies. The review reveals that the studies differ in a number of aspects. First, they look at different aspects of land degradation, with a predominance of stud- ies focusing on on-site soil erosion. Second, they use different approaches to assess the physical aspects of the problem. In the case of soil erosion, the universal soil loss equation (USLE) is the most commonly used approach. The next step is to quantify the impact of degradation on agriculture. A dose-response relationship between degra- dation and agricultural productivity is needed for this step. Finally, the economic impact is estimated. The most commonly used approaches for this are the change in productivity approach and the replacement cost approach. The economic calculations are also sensitive to the discount rates and time horizons chosen. Since almost all of the studies differ in most of these steps, comparisons between studies are difficult to make. Still, the order of magnitude is most likely a couple of percentage points of agricultural GDP per year. Given that not all costs are included in any of the studies, the total discounted economic value can be substantially higher (lost soil and decreased dam capacity lead to accumulative costs over long periods of time). Many of the studies explicitly seek to guide policy making and interventions. Although the studies contain a wealth of information, they are not necessarily useful for this purpose. Ideally, public investments should be allocated to sectors and interventions that give the highest possible return to society. In order to do that, the analyst would need to indicate both the cost of and the returns to the intervention. However, to be fair, the fact that a nonmarginal proportion of agricultural GDP is lost every year in some of the poorest and most resource-dependent countries should give a strong impetus to carrying out such an analysis. In the following sections, we sketch such an analysis by discussing the literature that has studied the adoption of soil conservation, the results from past interventions, as well as a recent initiative to take a more comprehensive approach to such interventions that also integrates the off-site effects of erosion. Driving Forces behind Land Degradation Of course, a number of general factors determine the extent of land degradation. The most fundamental geographic factors, such as topography, erosivity, and precipita- tion, are well represented in all erosion models such as the USLE. Erosion models typically also include land cover and management components that depend on the policy environment, including legislation, land administration, and the extension service. Finally, population density has been a contentious factor in natural resource management ever since the writings of Thomas Robert Malthus, and one can be sure to find evidence and proponents of both the view that higher population density will lead to greater land degradation and the view that it will not. Lessons from the Soil and Water Conservation Adoption Literature In order to discuss potential interventions to address land degradation, it is worth- while to review the economics literature on the adoption of SLM measures.3 This ASPECTS OF LAND DEGRADATION IN LAGGING REGIONS | 75 literature is particularly rich with regard to soil and water conservation (SWC), but a general theoretical literature is also applicable to other agricultural technologies, such as fertilizer, crop choice, and tree plantation. The theoretical literature is by now quite voluminous, with a number of strands applying either static or dynamic models, assuming perfect or imperfect factor markets, and particularly modeling adoption under uncertainty, for example, tenure insecurity. For a recent review of this literature, see Yesuf and Pender (2005). From this theoretical literature follows a number of predicted determinants of adoption: the factors affecting the profitability of SWC (see box 1). Empirical Evidence of Factors Affecting Adoption These theoretical predictions have been the focus of a number of empirical studies. As is evident from the following paragraphs, the evidence is mixed, and the picture that emerges is far from clear cut. BOX 1. THEORETICAL PREDICTIONS OF THE DETERMINANTS OF SWC ADOPTION In perfect markets, the following factors affect the profitability of soil conservation: · Prices of inputs (including opportunity cost of labor), · Prices of outputs (affected by market access, for example), · Productivity impacts of the investment, · Agroecological conditions, · Population density (since it affects the price of labor and land), · Capital (land quality, human, physical), · Access to improved technologies, and · Interest rate. In imperfect markets, the same factors exist as in the case of perfect markets plus the following: · Property rights and land tenure relationships, · Household endowments of labor, land, physical capital, financial capital, and social capital, · Implicit discount rate, and · Risk aversion. 76 | GUNNAR KÖHLIN Tenure Insecurity. The proxy used for tenure insecurity differs between authors. The following are the most commonly used: existence of past land redistribution, frequency of land redistribution, number of years since last redistribution, future expectations of own farm size (or expectation of change in own farm size), land- lessness in community, year of landholding, type of landholding, and right to transfer land. The empirical evidence on the impact of tenure insecurity on the adoption of medium- and long-term land management technologies and land-use practices is not conclusive, but it is clearly indicative. The expected detrimental impact of tenure inse- curity on adoption is found in a number of cases, such as Alemu (1999), Gebremedhin and Swinton (2003), and Gebremedhin, Pender, and Ehui (2003). Still, in a number of cases, the analysts do not find any significant impact (Benin and Pender 2001; Hagos 2003; Shiferaw and Holden 1998; Yesuf 2004). Finally, the standard expectation that increased security has a positive impact on investments is substantiated in a number of studies showing that perceived transfer right is significant for land improvements (Deininger and others 2003) and for trees and perennial crops (Ayalew, Dercon, and Gautam 2005). Households' Endowments of Physical and Human Capital. Given labor mar- ket imperfections, household endowment of labor is expected to matter. Availability of adult labor is also positive and significant for the construction of stone terraces both in Alemu (1999) and in Gebremedhin and Swinton (2003). It is also positive in Deininger and others (2003) for terraces and trees. However, when family labor is used as a proxy, the results are mixed. Land size has a positive impact on physical SWC structures in Alemu (1999); Demeke (2003); Hagos (2003); Kassie and Holden (2005); Shiferaw and Holden (1998); Teklewold (2004); and Yesuf (2004). It is therefore critical that fragmenta- tion of land is still the norm in many countries. In Ethiopia the government policy is only to stop fragmentation below 0.25 hectare per household, which by any standard is very low. The impact on SWC also depends on the type of conservation practices promoted. Pender and Gebremedhin (2004) find a positive relationship between livestock and contour plowing, manure or composting, and intercropping and a negative relation- ship for reduced tillage. However, in many studies, no significant relationship is observed between livestock ownership and adoption of physical structures. Similarly, male-headed households and technology adoption are positive, if they are not insignificant, while education and SWC show mixed results. Access to Markets, Roads, and Off-farm Opportunities. The predictions from the theoretical models are inconclusive with respect to market access. The reason is that market access affects the opportunity cost of time, which makes SWC more expensive to undertake, although it might also give a higher return. The empirical literature also indicates that this is not a clear-cut case. In low-potential areas of Ethiopia, access to roads is found to be negatively correlated with SWC adoption ASPECTS OF LAND DEGRADATION IN LAGGING REGIONS | 77 (Hagos 2003; Yesuf 2004). It is also found that off-farm opportunities and the adoption of SWC technology are negatively correlated (Alemu 1999 in Amhara; Ersado, Amacher, and Alwang 2003 in Tigray; and Shiferaw and Holden 1998 in low-potential Amhara, all in Ethiopia). Population Pressure. As indicated, some hypotheses predict both a positive and a negative relationship between population pressure and the adoption of SWC. The empirical literature shows a similar span. Grepperud (1996) finds a strong and positive relationship between population pressure and soil erosion, while Pender and Gebremedhin (2004) argue that the relationship depends on the type of land management being promoted. They find more use of fertilizer, manure, compost, burning, and intercropping in areas where there is high population density, but no significant relationship for contour plowing and reduced tillage. Pender and others (2001) find a negative relationship between population pressure and adoption of manure and fallow, but no significant relationship for compost, fertilizer, stone and soil terraces, gully checks, trees, and live fences. Risk Aversion and Discount Rates. The expected sign of risk aversion on adoption depends on whether the investment is expected to be risk increasing or risk decreasing. The empirical literature from Ethiopia shows that risk aversion and fertilizer adoption are negatively correlated in Amhara (Yesuf 2004), but positive in Tigray (Hagos 2003), implying that fertilizer is risk decreasing in Tigray but risk increasing in Amhara. Furthermore, Teklewold (2004) finds that risk-averse farmers invest more in SWC, which indicates that SWC is expected to reduce risk. The subjective discount rate is clearly negatively correlated with SWC investments, given their long payback time. That a high rate of time preference reduces SWC adoption is also found empirically by Shiferaw and Holden (1998), Teklewold (2004), and Yesuf (2004). Summing Up the Empirical Evidence The first conclusion to be drawn from the divergent empirical literature on SLM adoption is that the impacts of many factors depend on the context, making gen- eralizations difficult. However, if we maintain the view that the adoption decision is primarily a private investment decision, then we can see that almost all of the factors that are generally expected to improve the investment environment would be conducive to SLM investments as well. For example, tenure security generally improves adoption, although some studies have found insignificant results. Simi- larly, creating safety nets that reduce risk and discount rates is positive, both in its own right and for the potential effect it has on land management. The fact that increased access to credit is only important when the technology is prof- itable reminds us of the important role of profitability as a threshold criterion for any adoption. 78 | GUNNAR KÖHLIN The only caution that needs to be made regarding a "conducive environment approach" concerns the role of market access, roads, and off-farm opportunities. The expectation is that improved market access will lower input prices and raise output prices, leading to increased profitability in agriculture and thus warranting more investments in SLM. However, increased market access and, in particular, off-farm opportunities simultaneously lead to higher opportunity costs for labor. For labor- intensive activities such as soil conservation, this could decrease the profitability of SLM investments. This effect seems to dominate in the low-potential areas in Ethiopia reported on above. Experiences from Past SLM Interventions Although greater attention recently has been given to the importance of a conducive policy environment to stimulate SLM interventions, this has not always been the case. Land management interventions have instead been dominated by soil scientists attempting to maximize the conservation, rather than the welfare, of farmers. As a result, specific technologies have been promoted without due consideration to local circumstances. An emerging literature is evaluating SLM measures from an economic point of view as well (see, for example, Nyangena 2006 for a review and a recent application). Unfortunately, many interventions prove to be unprofitable, with no significant positive effect on productivity. Yesuf and Pender (2005) therefore argue that SLM interventions need to be care- fully targeted and that profitability should be a necessary first hurdle to pass in order to be considered for implementation. Unfortunately, given the widely varying physi- cal conditions in many countries, this implies a detailed matching and targeting process. In Ethiopia, for example, quite a few evaluations of existing structures have been conducted. The results have been very mixed, but if the observations are organ- ized according to their agroecological characteristics, it can be shown that physical SWC structures are positive in moisture-stressed areas of Tigray and negative in high- rainfall areas of Amhara: · Positive returns to SWC in Tigray (up to 46 percent), as reported by Gebremedhin, Swinton, and Tilahun (1998) and Pender and Gebremedhin (2004), · Positive returns to fertilizer and SWC in drought-prone eastern Amhara (15 per- cent to stone terraces and 51 percent to fertilizer), as reported by Benin (2004), · High returns to fertilizer (up to 70 percent), but no significant returns to SWC in high-rainfall Amhara, as reported by Benin (2004), and · Negative returns to SWC (fanya juu) in West Amhara, as reported by Kassie and Holden 2005). Agricultural Extension and Credit Programs More indirect interventions could also be expected to have an impact on the adop- tion of SWC. Agricultural extension by the Bureau of Agriculture in Ethiopia is ASPECTS OF LAND DEGRADATION IN LAGGING REGIONS | 79 negatively associated with fallow, manure, and compost but positively associated with tree planting (Pender and others 2001). Credit from the Relief Society of Tigray is associated with greater use of compost, soil bunds, tree plantation, and live fences (Pender and others 2001). Similarly, credit from the Amhara Credit and Saving Institution is associated with less use of fallow and tree plantation, but more invest- ment in soil bunds and live fences (Pender and others 2001). Availability of more fertilizer credit could lead to substitution of fertilizer for lost nutrients and conse- quently discourage soil conservation (Shiferaw and Holden 1998). The authors therefore argue for a cross-compliance strategy in which fertilizer credit is combined with SWC activities. A general lesson to be learned is that extension drives particular technologies. If that continues to be the case in the future, these technologies will have to be well adapted and profitable to the farmers. One would also like to see extension that is more sensitive to local conditions and needs and less oriented to technology. Toward a Framework for SLM Interventions This paper has argued that it is important to create a conducive environment for sustainable management that increases profitability from SLM. This could be done by creating credible institutions for managing the resource, reducing risk and vul- nerability for farmers, achieving lower subjective discount rates, increasing prod- uct market access, and reducing transaction costs. However, given the dismal state of many lands and the externalities involved, it is probably not enough just to improve the policy environment. Well-targeted interventions are also needed, not least on public lands that would not be affected by improvements in the private investment climate. TerrAfrica has been launched by the World Bank and the Global Environment Facility as a program to address this lack of investments in African agriculture. Based on the experiences in Ethiopia and Ghana, a framework has been developed to guide such investments. The rationale for the framework is the same as has been presented in this paper: land degradation leads to substantial welfare losses, and there are many promising SLM practices, but mixed economic returns. Scaling-up of promising practices is also difficult, and it is nontrivial to base extension and up- scaling on research case studies. Still, there is a need to assess the overall invest- ment needs for the sector and prioritize interventions: both what should be done and where. The proposed framework will compile relevant physical and economic informa- tion and make it available for policy analysis in a consistent manner. This will increase the reliability and availability of information on land degradation as well as its costs, both on agricultural lands and downstream. It will also make use of exist- ing research on the effectiveness of various SLM treatments, and future research can be incorporated fairly easily, so that this information can be used to calculate the returns from such treatments, taking into consideration both physical characteristics and economic factors. This is expected to be a valuable support to up-scaling of 80 | GUNNAR KÖHLIN BOX 2. MAIN STEPS OF THE COST-BENEFIT FRAMEWORK · Identify recommendation domains that share the same conditions for the application of treatments, · Estimate the soil erosion and nutrient depletion rates for each recommenda- tion domain based on explanatory factors proposed in the USLE, · Estimate production responses to erosion for various crops and treatments in each recommendation domain, · Translate soil erosion and nutrient depletion into income losses using different economic valuation techniques, · Compute the net present value of each treatment for each development domain (mapping unit), · Map on- and off-site returns from treatment, · Prioritize areas with highest return according to budget constraints, and · Disseminate this information to the relevant users, particularly land-use planners and extension staff at the regional level and below. promising SLM technologies. Finally, since the cost-benefit framework will be able to provide simulations of on- and off-site returns from treatments, this information can be used both to scale the size of investments in SLM and to prioritize the use of these investments to areas with the highest expected returns to society (see box 2). Concluding Remarks The paper started out by arguing for public intervention and international develop- ment assistance to deal with land degradation in Africa. Poor people in Africa have extremely high subjective discount rates. Therefore, they cannot be expected to manage the land in a fully sustainable fashion. Poverty alleviation objectives and the need to deal with externalities and irreversibilities together build a strong case for intervention. The review of the literature on the cost of land degradation in Africa shows that it is very difficult to compare different studies primarily for methodological reasons. Such studies are of limited use for guiding the size and allocation of funds to the sector. In the search for appropriate interventions, highlights were given from the economics literature dealing with the adoption of soil and water conservation and past interventions in the sector. Both the theoretical and the empirical literature point to the importance of decision rules for the farmer that fulfill the need for both short- and long-term profitability ASPECTS OF LAND DEGRADATION IN LAGGING REGIONS | 81 from the SLM investment. The paper therefore argues that an improved policy envi- ronment is very important in order to improve land management. Most institutional reforms in the agricultural sector have the potential to improve the profitability from SLM investments, although they do not always show up as significant in empirical studies. The exception might be improved market access since it also increases the opportunity cost of time, thus acting as a disincentive to labor-intensive sustainable land management. Unfortunately, a conducive investment environment will probably not be enough to achieve truly sustainable land management. However, past interven- tions have often failed, and good examples have proved difficult to scale up. When direct interventions are considered, they must be tailored very carefully to local con- ditions and be sufficiently sensitive to the farmers' need for short-term profitability. An example of such a cost-benefit framework for SLM was also given. Endnotes 1. This estimate is based on a recent review of the cost of land degradation in Ethiopia (Yesuf and others 2005). This is discussed further in the next section. 2. This section draws heavily from Yesuf and others (2005). 3. This section draws heavily on Yesuf and Pender (2005). References Alemu, Tekie. 1999. Land Tenure and Soil Conservation: Evidence from Ethiopia. Ph.D. dissertation, Department of Economics, Gothenburg University, Sweden. Ayalew, Daniel, Stefan Dercon, and Madhur Gautam. 2005. "Land Tenure Insecurity and Invest- ment in Ethiopian Agriculture: Evidence from Panel Data." World Bank, Washington, DC. Benin, Samuel. 2004. "Increasing Land Productivity in High Versus Low Agricultural Potential Areas: The Case of the Ethiopian Highlands." Unpublished mss., International Food Policy Research Institute, Washington, DC. Paper under review for Food Policy. Benin, Samuel, and John Pender. 2001. "Impacts of Land Redistribution on Land Management and Productivity in the Ethiopian Highlands." Land Degradation and Development 12 (6): 555­68. Bojö, Jan. 1991. "The Economics of Land Degradation: Theory and Applications to Lesotho." Stockholm School of Economics, Stockholm. Bojö, Jan, and David Cassells. 1995. "Land Degradation and Rehabilitation in Ethiopia: A Reassessment." AFTES Working Paper 17, World Bank, Washington, DC. Convery, F. J., and K. A. Tutu. 1990. "Evaluating the Costs of Environmental Degradation in Ghana: Applications of Economics in the Environmental Action Planning Process in Africa." University College, Dublin. Deininger, Klaus, Songqing Jin, Berhanu Adenew, Samuel Gebre-Selassie, and Berhanu Nega. 2003. "Tenure Security and Land-Related Investment: Evidence from Ethiopia." Policy Research Working Paper 2991, World Bank, Washington, DC. Demeke, A. B. 2003. Factors Influencing the Adoption of Soil Conservation Practices in Northwestern Ethiopia. Discussion Paper 37. Goethingen: University of Goethingen, Institute of Rural Development. 82 | GUNNAR KÖHLIN Enters, Thomas. 1998. Method for Economic Assessment of the On- and Off-Site Impacts of Soil Erosion. Issues in Sustainable Land Management 2. Bangkok: International Board for Soil Research and Management (IBSRAM). EPA (Environmental Protection Authority). 2004. The Third National Report on the Implemen- tation of the UNCCD/NAP in Ethiopia. Addis Ababa: EPA, Federal Republic of Ethiopia. Ersado, Lire, Gregory Amacher, and Jeffrey Alwang. 2003. Productivity and Land Enhancing Technologies in Northern Ethiopia: Health, Public Investments, and Sequential Adoption. Environment and Production Technology Division Discussion Paper 102. Washington, DC: International Food Policy Research Institute. Gebremedhin, Berhanu, John Pender, and Simeon Ehui. 2003. "Land Tenure and Land Man- agement in the Highlands of Northern Ethiopia." Ethiopian Journal of Economics 8 (2): 47­63. Gebremedhin, Berhanu, and Scott M. Swinton. 2003. "Investment in Soil Conservation in Northern Ethiopia: The Role of Land Tenure Security and Public Programs." Agricultural Economics 29 (1): 69­84. Gebremedhin, Berhanu, Scott M. Swinton, and Yibebe Tilahun. 1998. "Effects of Stone Terraces on Crop Yields and Farm Profitability: Results of On-Farm Research in Tigray, Northern Ethiopia." Journal of Soil and Water Conservation 54 (3): 568­73. Grepperud, Sverre. 1996. "Population Pressure and Land Degradation: The Case of Ethiopia." Journal of Environmental Economics and Management 30 (1): 18­33. Hagos, Fitsum. 2003. Poverty, Institutions, Peasant Behavior, and Conservation Investment in Northern Ethiopia. Ph.D. dissertation, Department of Economics and Social Sciences, Agricultural University of Norway, Aas. Kassie, Menale, and Stein Holden. 2005. "Parametric and Non-parametric Estimation of Soil Conservation Adoption Impact on Yield in the Ethiopian Highlands." Unpublished mss., Department of Economics and Resources Management, Norwegian University of Life Science, Norway. Kerr, John, and John Pender. 2005. "Farmers' Perceptions of Soil Erosion and Its Consequences in India's Semiarid Tropics." Land Degradation and Development 16 (33): 257­71. McKenzie, C. 1994. "Degradation of Arable Land Resources: Policy Options and Considera- tions within the Context of Rural Restructuring in South Africa." Paper presented at the LAPC workshop, Land and Agricultural Policy Centre, Johannesburg, March 30­31. Norse, D., and R. Saigal. 1992. "National Economic Cost of Soil Erosion: The Case of Zimbabwe." Paper prepared for the CIDIE workshop on Environmental Economics and Natural Resource Management in Developing Countries, World Bank, Washington, DC, January 22­24. Nyangena, Wilfred. 2006. Essays on Soil Conservation, Social Capital, and Technology Adop- tion. Ph.D. dissertation, Department of Economics, Gothenburg University, Sweden. Pagiola, Stefano. 1999. "The Global Environmental Benefits of Land Degradation Control on Agricultural Land." World Bank Environment Paper 16, World Bank, Washington, DC. Pender, John, and Berhanu Gebremedhin. 2004. "Impacts of Policies and Technologies in Dryland Agriculture: Evidence from Northern Ethiopia." In Challenges and Strategies for Dryland Agriculture, ed. S. C. Rao. CSSA Special Publication 32. Madison, WI: American Society of Agronomy and Crop Science Society of America. Pender, John, Berhanu Gebremedhin, Samuel Benin, and Simeon Ehui. 2001. "Strategies for Sustainable Development in the Ethiopian Highlands." American Journal of Agricultural Economics 83 (5): 1231­40. ASPECTS OF LAND DEGRADATION IN LAGGING REGIONS | 83 Scherr, Sara. 1999. Soil Degradation: A Threat to Developing-Country Food Security by 2020? Food, Agriculture, and the Environment Discussion Paper 27. Washington, DC: Interna- tional Food Policy Research Institute. Shiferaw, Bekele, and Stein T. Holden. 1998. "Resource Degradation and Adoption of Land Conservation Technologies in the Ethiopian Highlands: A Case Study in Andit Tid, North Shewa." Agricultural Economics 18 (3): 233­47. Smaling, Eric, Louise O. Fresco, and A. De Jager. 1996. "Classifying, Monitoring, and Improv- ing Soil Nutrient Stocks and Flows in African Agriculture." Ambio 25 (8): 492­96. Stoorvogel, Jetse, Eric Smaling, and B. H. Janssen. 1993. "Calculating Soil Nutrient Balances in Africa at Different Scales." Fertilizer Research 35 (3): 227­35. Teklewold, Hailemariam. 2004. "Risk and Time Preferences on Soil Conservation Decision in the Central Ethiopian Highlands of Ethiopia." M.Sc. thesis, Addis Ababa University. Yesuf, Mahmud. 2004. Risk, Time, and Land Management under Market Imperfection: Appli- cations to Ethiopia. Ph.D. dissertation, Department of Economics, Gothenburg University, Sweden. Yesuf, Mahmud, Alemu Mekonnen, Menale Kassie, and John Pender. 2005. "Cost of Land Degradation in Ethiopia: A Critical Review of Past Studies." Ethiopian Development Research Institute, Environmental Economic Policy Forum for Ethiopia (EDRI/EEPFE), Addis Ababa, Ethiopia. Yesuf, Mahmud, and John Pender. 2005. "Determinants and Impacts of Land Management Technologies in the Ethiopian Highlands: A Literature Review." EDRI/EEPFE, Addis Ababa, Ethiopia. Agriculture and the Rio Conventions LAURENCE TUBIANA AND SHEILA WERTZ-KANOUNNIKOFF Although agriculture assures global food security, it also has profound impacts on the environment. It occupies about 40­50 percent of the Earth's land surface and is the largest consumer of water and the main source of nitrate pollution of ground and surface water as well as an important source of ammonia and phosphate pollution. It is also a major contributor to the release of greenhouse gases. But agriculture can also provide positive environmental services and amenities such as water storage and purification, carbon sequestration, and the maintenance of rural landscapes. The existence of global environmental externalities from agricultural activities calls for international regulatory action. Multilateral negotiations and agreements are nation-state efforts to regulate issues of global importance, such as the provision of global public goods (Severino and Tubiana 2002). The most important environmen- tal negotiations are associated with the three "Rio conventions" on climate, biodi- versity, and desertification adopted at the 1992 World Environment Summit in Rio de Janeiro: · The United Nations Framework Convention on Climate Change (UNFCCC) seeks to stabilize greenhouse gas (GHG) concentrations in the atmosphere at a level that would prevent dangerous anthropogenic interference with the climate system. · The UN Convention on Biological Diversity (CBD) aims to promote the conser- vation of biodiversity, the sustainable use of its components, and the fair and equi- table sharing of benefits arising from the use of genetic resources. · The UN Convention to Combat Desertification (CCD) aims to promote effective action against desertification and land degradation. Laurence Tubiana is Director and Sheila Wertz-Kanounnikoff is Research Associate with the Institut du Développement Durable et des Relations Internationales (IDDRI, Institute of Sustainable Development and International Relations) in France. Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 85 86 | LAURENCE TUBIANA AND SHEILA WERTZ-KANOUNNIKOFF However, an intrinsic difficulty of collective global governance is that negotiators often compete for national interests rather than cooperate for global public interests. Because decisions from international negotiations can profoundly shape agriculture at the local level, nation-states seek to maximize their economic and political bene- fits. In addition to the international environmental negotiations, agriculture is also shaped by other international negotiations, notably those within the framework of the World Trade Organization. This paper focuses on the three Rio conventions to examine the role of agriculture in international negotiations and to illustrate the chal- lenges for global environmental governance. Agriculture and the UNFCCC Agriculture can be both positive and negative for the global climate. On the one hand, agriculture contributes to the emission of greenhouse gases and thus to global warming. On the other hand, agriculture plays an important role in the mitigation of global climate change by producing agrofuels as a substitution for fossil fuels and by maintaining forests and soils that serve as natural carbon sinks. For these reasons, agriculture is a subject of international negotiations, notably the UN Framework Convention on Climate Change. Greenhouse Gas Emissions from Agriculture Agriculture contributes to the emission of greenhouse gases and thus to global warm- ing. According to the Intergovernmental Panel on Climate Change (IPCC 2007), agri- culture accounted for an estimated 13.5 percent and forestry for another 17.4 per- cent of global anthropogenic emissions of greenhouse gases in 2004. Principal GHG emissions from agriculture are carbon dioxide (CO2), methane (CH4), and nitrous oxide (N2O), whereby the latter two have a high warming potential of about 20 times (methane) and 300 times (nitrous oxide) more powerful than carbon diox- ide. In addition, of global anthropogenic emissions in 2005, agriculture accounted for about 58 percent of N2O and about 47 percent of CH4 (IPCC 2007). Important sources of agriculture-related emissions include changes in land use (deforestation), livestock, and fertilizer applications. According to estimates from the Food and Agriculture Organization (FAO) of the United Nations, the livestock sec- tor alone is responsible for 9 percent of anthropogenic CO2 emissions, the largest share coming from land-use changes, especially deforestation (Steinfeld and others 2006). This calls for the adoption of more sustainable agricultural practices to reduce GHG emissions from agriculture. The IPCC (2007) identifies the following measures: improved management of crop and grazing land to increase soil carbon storage; restoration of cultivated peaty soils and degraded lands; improved rice cultivation techniques and livestock and manure management to reduce CH4 emissions; improved techniques for applying nitrogen fertilizer to reduce N2O emissions; and use of dedicated energy crops to replace fossil fuels. Applied in the livestock sector, AGRICULTURE AND THE RIO CONVENTIONS | 87 such measures could sequester up to 1.3 tons of carbon per hectare per year, with additional amounts available through the restoration of desertified pastures (Steinfeld and others 2006). Their adoption is also promoted by the Kyoto Protocol of the UNFCCC,1 either through direct emissions reductions from agriculture in the developed countries or through the compensation of GHG emissions in developed countries by financing agriculture-related projects in developing countries through the Clean Development Mechanism (CDM). The CDM is the protocol's mechanism for attracting additional investments in resource conservation practices that reduce GHG emissions in developing countries. Most current policy frameworks at the national and international level provide few incentives for sustainable agriculture. In the Kyoto Protocol, for instance, the rules for CDM projects are little adapted to the agricultural sector. In turn, there are few such projects. Out of 2,260 CDM projects, only 177 are in the agricultural sec- tor (not counting the 123 that are related to biogas).2 But even beyond the CDM framework, there is little support. A well-known example refers to the European and U.S. agricultural subsidies that promote unsustainable agricultural practices in the European Union or in the United States against any economic or moral logic. Such policies are often the result of powerful agricultural lobby groups that prefer short- term over long-term financial benefits and influence public policy such that local preferences (local economy) often outrank international collective preferences (organic agriculture for the associated environmental benefits, equity aspects regard- ing developing countries). Despite some modest reform efforts, these policies con- tinue to undermine valuable opportunities for organic agricultural practices (in both developed and developing countries) and for economic growth in developing coun- tries. With the increasing role of agriculture in the global combat against climate change, one can also expect that the eligibility rules for agriculture-CDM projects (or adequate alternative measures) will become more adapted to sector needs. Still, the topic of agriculture shows how local interests (domestic economic growth) compete with global collective interests (global climate). Production of Renewable Energy (Agrofuels) The substitution of fossil fuels by agrofuels as a means to reduce greenhouse gas emissions is another avenue where agriculture can help to combat climate change. While the production of agrofuels has long been prohibitively costly, the increase in global oil prices since 2004 and the enhanced recognition of the negative externali- ties associated with the use of traditional fossil fuels on global climate have renewed the interest in agrofuels: ethanol and biodiesel. Due to substantial public subsidies since the 1970s, Brazil is the global leader in the production of agrofuels. However, the economic and environmental effectiveness of agrofuels depends largely on the pri- mary resources applied and the associated methods of production. The reduction of GHG emissions realized by substituting ethanol for fossil fuels can, for instance, vary between 10 percent for ethanol from wheat and 90 percent for ethanol from sugar- cane in the Brazilian context (that is, using cogeneration in the factory). Use of veg- etable oils, such as from palm oil, soybeans, rape seed, or sunflowers, reduces GHG 88 | LAURENCE TUBIANA AND SHEILA WERTZ-KANOUNNIKOFF emissions, in some cases, about 50 to 80 percent. These differences arise from the energy returns, which differ strongly between agricultural sectors. The production of 1 liter of ethanol from wheat, for example, can imply the burning of 1 liter of petrol. Within the Kyoto Protocol of the UNFCCC, the implementation of agriculture- related CDM projects suffers from the stringency of rules imposed. This is part of the reason why the direct production of agrofuels is not represented at the registration level within the CDM of the Kyoto Protocol. Only the use of agro-industrial by-products is eligible for the production of bioenergy such as the recycling of molasses from the production of sugar to obtain ethanol. Still, given the rising interest in agrofuels as a source of renewable energy to combat climate change but also as a new business opportunity for the agricultural sector, it is very likely that these policies will change. The production of agrofuels can also affect biodiversity and human well-being. Adverse impacts on biodiversity and human well-being include, among others, frag- mentation and degradation of habitats, land-use competition between production objectives (cash crops versus food crops) with associated effects on food prices (for example, the recent "tortilla crises" in Mexico or food riots in a number of African countries) or on land-use stakeholders (commercial farmers, smallholders, indige- nous communities), increased water consumption, application of agrochemicals, and uncontrolled cultivation, introduction, and spread of genetically modified organisms (GMOs) or invasive alien species. In addition, even though it is believed that emis- sions can be reduced through the use of agrofuels, it is also feared that large increases in GHG emissions can result from deforestation, land-use changes, and the loss of major carbon sinks, such as peat lands. Nevertheless, depending on the land-use change and biomass considered, agrofuels can also have beneficial impacts on biodi- versity and human well-being. For instance, displacing annual crops with perennial grassy crops or restoring degraded lands with tree plantations could lead to greater animal biodiversity and reduce pesticide and net fertilizer use. For these reasons, and the growing interest in agrofuels within the climate negotiations, the production of agrofuel has become an emerging topic within the biodiversity negotiations, first at the Paris meeting of the CBD scientific advisory body in July 2007 and subsequently at the upcoming Conference of the Parties to the CBD in Germany in May 2008.3 The challenge of international environmental regulation regarding agrofuels is thus to reduce the tradeoffs between climate change mitigation and biodiversity con- servation. Hence, in addition to dealing with tensions between international and local preferences, the case of agrofuels illustrates how topics of pure international preference may compete with each other. Biological Carbon Sequestration Biological carbon sequestration refers to the natural removal of carbon from the atmosphere by forests, soils, and oceans. It can be enhanced through specific agri- cultural practices such as crop rotations, improved land management practices, or restoration of degraded lands. Forests present an important carbon sink. Within the CDM framework, forestry- related carbon sequestration has so far been limited to reforestation and afforestation. AGRICULTURE AND THE RIO CONVENTIONS | 89 Yet only eight reforestation or afforestation CDM projects are currently registered.4 A main reason lies in the technical difficulties associated with forest carbon sinks (monitoring, leakage, permanency). Other reasons may refer to associated opportu- nity costs, which sometimes can be very high (that is, commercial agriculture having greater returns than reforestation projects). Another variant of biological carbon sequestration lies with primary forests or "avoided deforestation." Although deforestation can spur climate change through the release of about 3.8 billion tons of CO2 per year (Achard and others 2004), it has been excluded from CDM eligibility. While technical difficulties (calculation of base- line scenarios, monitoring, permanency, leakage) certainly have played a role, politi- cal aspects are probably the main underlying reasons. Political reasons include mis- trust of international regulatory action on national territory, which could result in a loss of control over the nation's natural resources. Such fears are probably part of the reason why Brazil continues to oppose any international efforts to "internationalize the Amazon." In addition, powerful environmental nongovernmental organizations have long feared that efforts to mitigate climate change in developed countries (that is, improve energy efficiency, adopt clean energy) will be replaced by efforts in devel- oping countries (that is, pay countries to avoid tropical deforestation). However, given the increasing political recognition that deforestation accounts for about 25 percent of total anthropogenic CO2 emissions and the growing urgency to act (Stern 2006), new cost-effective opportunities to reduce global GHG emissions are seen in the establishment of financial incentives for "avoided deforestation" in tropical countries (Chomitz and others 2006; Stern 2006). The proposition to consider "emission reduction from avoided deforestation and degradation (REDD)" in the international climate negotiations was put back on the table by Papua New Guinea and Costa Rica at the Conference of Parties (COP11) of the UNFCCC in Montreal in 2005; since then, a vibrant debate has emerged on the subject. Even countries that were long opposed to any international measures that may affect local resource hegemony (for example, Brazil) seem more open to discussing international financial efforts to com- bat tropical deforestation. At the recent summit of the G-8, official support was expressed for the World Bank­led effort on a Forest Carbon Partnership Facility to explore avoided deforestation in practice (in five pilot countries) and draw lessons for the post-2012 Kyoto negotiations in the UNFCCC. Decisive steps in this direction are expected to occur at the upcoming Conference of Parties (COP13) to the UNFCCC in Bali in 2007. Agriculture and the UN Convention on Biological Diversity According to the Millennium Ecosystem Assessment (2005), current extinction rates of biodiversity are about 100 times the rate they were before humans existed and could increase by another 10­100 times. There are two principal ways in which agri- culture affects biodiversity. One refers to agricultural expansion into pristine areas, the other to agricultural intensification and associated use of pesticides and fertiliz- ers as well as genetically modified organisms. There is a large debate on the value of 90 | LAURENCE TUBIANA AND SHEILA WERTZ-KANOUNNIKOFF GMOs: some say that they will improve global food security with fast-growing and resistant species; others fear that their use may profoundly change the natural eco- logical system. To overcome the adverse effects and risks of commercially intensive agriculture on biodiversity, the CBD seeks to valorize agricultural biodiversity in agri- cultural production methods such as in agroforestry or no-tillage/conservation agriculture. Agricultural Intensification Agricultural intensification can also lead to severe environmental problems, espe- cially in developed countries, but increasingly also in developing countries. Intensifi- cation is a result of technical progress, characterized by monoculture, mechanization, intensive use of fertilizers and pesticides, and massive water consumption. Several developing countries already have average fertilizer and pesticide application rates exceeding those causing major environmental problems in developed countries. Future environmental pressure from agriculture is expected to come primarily from intensification rather than from land-use change or deforestation. About 80 percent of the projected growth in crop production in developing countries will come from intensification, primarily through higher yields (67 percent) and higher cropping intensities (12 percent), especially in the land-scarce regions of the Near East, North Africa, and South Asia (Bruinsma 2003). The effects of intensive agriculture on biodiversity can be positive and negative, depending on the situation. On the one hand, crop intensification can lead to a reduc- tion in the quantity and quality of species composition due to monoculture or other biodiversity-harming agricultural practices and tends to increase water pollution as a result of fertilizer and pesticide use. On the other hand, it reduces the pressure to con- vert new lands to agriculture, and agricultural landscapes can function as crucial habitats for species. In Norway, for example, around half of the threatened species depend on agricultural landscapes and therefore on the conservation of biodiversity (Bruinsma 2003). A similar picture is valid for intensive livestock. While the switch from extensive to intensive livestock lowers grazing damage to rangelands, it may, for example, increase water pollution as a result of poorly managed storage of manure. However, intensification also allows the abandonment of erosion-prone marginal lands and the more efficient use of fertilizer and integrated pest management, which, together with the expansion of organic farming, are projected to slow the growth in use of mineral fertilizers and pesticides. Moreover, in an increasing number of cases, the tradeoffs between agricultural production and environmental conservation are becoming less serious, such as for no-tilling/conservation agriculture, which seeks to valorize agricultural biodiversity and thus reduce the application of pesticides, soil erosion, and vulnerability to drought, while raising carbon sequestration, natural soil nutrients, and farm income. Recognition of the value of agricultural biodiversity has led to efforts to its con- servation and sustainable use. Generally speaking, "agricultural biodiversity" is a broad term that includes all components of biological diversity of relevance to food and agriculture, including all components of biological diversity that support the AGRICULTURE AND THE RIO CONVENTIONS | 91 ecosystems of which agriculture is a part (agroecosystems): the variety and variability of animals, plants, and microorganisms at the genetic, species, and ecosystem levels, which are necessary to sustain key functions, structure, and processes of the agro- ecosystem. Specifically, the CBD adopted in 1996 a multiyear program on "conser- vation and sustainable use of agricultural biological diversity," which promotes (a) the positive effects and mitigates the negative impacts of agricultural practices on biolog- ical diversity in agroecosystems and their interface with other ecosystems (for exam- ple, adaptive management practices), (b) the conservation and sustainable use of genetic resources of actual or potential value for food and agriculture, and (c) the fair and equitable sharing of benefits arising from the use of genetic resources. Much of the work on agricultural biological diversity under the convention has been under- taken in cooperation with the FAO. Yet despite these multiyear efforts to achieve equitable access to and share the benefits from the use of genetic resources and the existence of the potentially powerful cosmetic and pharmaceutical industry as a source of financing, there are only scattered successes: interest in bioprospecting is limited or even running dry, as expressed by the small amount of financing flowing into this activity. This is explained largely by the complexity and lack of transparency of regulations. The insufficient institutional and legal certainty for bioprospecting at the international level can thus result in substantial economic losses at the local level. Agricultural Expansion and Deforestation Expansion of agricultural land, especially into pristine tropical areas where biodiver- sity is concentrated, can harm biodiversity through the degradation of ecosystems and the loss of species; the level of harm depends largely on the method of produc- tion (monoculture, fertilizer application, erosion control measures, and so forth). Especially in developing countries, biodiversity is threatened by deforestation and agricultural expansion. Tropical forests shrink at about 5 percent per decade. Defor- estation has never been as widespread as during the second half of the twentieth cen- tury, and pressures on tropical forests will not disappear soon (Chomitz and others 2006). Over the next 30­50 years, croplands, pastures, and plantations are expected to expand into natural forests. Expansion of livestock is a key factor in deforestation, especially in Latin America: 70 percent of previously forested land in the Amazon is occupied by pastures (Steinfeld and others 2006). And the impact of agriculture and livestock on biodiversity will continue. According to estimates from the FAO, crop- lands in the developing world (where biodiversity is concentrated) will expand by 3.8 million hectares a year for the next 30 years; gross expansion will be even greater because of farmland abandonment (Bruinsma 2003). Measures to reduce biodiversity loss from land-use change and deforestation are highly context specific and depend on the underlying causes (Chomitz and others 2006). Important steps toward improved forest governance include effective enforce- ment of forest policy. In the Brazilian Amazon, for example, an estimated 80 percent of deforestation is illegal. The CBD recognizes the need to combat desertification through an expanded program of work on forest biodiversity, which was adopted in 2002 at COP6. This forest work program promotes the conservation and sustainable 92 | LAURENCE TUBIANA AND SHEILA WERTZ-KANOUNNIKOFF use of forest biodiversity and the fair and equitable use of the benefits arising from the use of forest genetic resources. An important impetus for the conservation of forest biodiversity may arise with the potential to finance "avoided deforestation" to combat climate change. Yet to take full advantage of this opportunity, the biodiver- sity community needs to be far more active and cooperative with the ongoing dis- cussions in the climate negotiations than is currently the case. In addition to greater recognition of the value of biodiversity at the local level, this could also reduce the potential tradeoffs--at the local level--between the objectives of combating climate change and conserving biodiversity. Agriculture and the UN Convention on Desertification Desertification is the degradation of land in arid, semiarid, and dry subhumid areas. About 70 percent of the 5.2 billion hectares of drylands used for agriculture around the world are already degraded. Desertification is estimated to damage almost 30 percent of the total land around the world, especially in Africa and Asia, but also in the United States, Europe, and the former Soviet Union.5 Desertification is caused primarily by human activity and climatic variation. Although droughts often aggravate or trigger it, inappropriate agricultural activities (overgrazing, overcultivation, deforestation, and poorly drained irrigation systems) are usually the most immediate causes of desertification. Yet the main underlying cause is poverty. Poverty drives the people residing in drylands to exploit their lands as much as possible to feed their families in the short term, even though they are fore- closing their long-term futures. Still, climate change is projected to worsen the natu- ral conditions, that is, to increase the frequency and severity of droughts in already water-scarce countries. This imposes an additional challenge for local agriculture, which poor countries might not be prepared to face. Land degradation and desertification can threaten the livelihoods of the affected population and food security more generally. With degrading agricultural opportu- nities, rural populations often see no alternative than to migrate to urban centers, pri- marily into slums with few opportunities to improve their living conditions. More than 135 million people may be in danger of being driven from their land. These include many of the world's poorest, most marginalized, and politically weak. While the United Nations estimates that desertification costs $42 billion a year, the annual cost to prevent land degradation is estimated at only $2.4 billion. Desertification is the core subject of the Convention to Combat Desertification. Adopted in 1994, the CCD recognizes the physical, biological, and socioeconomic aspects of desertification. To combat desertification, the convention addresses not only the direct causes (agricultural practices) but also the underlying causes (poverty). To maintain long-term productivity of inhabited drylands, it promotes improved management of natural resources (for example, no-tillage/conservation agricultural practices that maintain soil cover and reduce erosion) and the creation of alternative sources of income for rural populations (off-farm employment opportunities) to reduce the pressure on agriculture (Giraud and Loyer 2007). Yet past efforts to AGRICULTURE AND THE RIO CONVENTIONS | 93 combat desertification have proven difficult, and the problem of desertification con- tinues to worsen. In addition, compared to the other two Rio conventions, the CCD has serious funding difficulties, as it was long excluded from any funding from the Global Environment Facility. It is often referred to as the most neglected convention among the three, probably because TerrAfrica is perceived to concern poor countries only. Steps in a new direction include the TerrAfrica initiative. Launched during the last Conference of the Parties in 2005, it is a multidonor initiative that seeks to com- bat land degradation in Sub-Saharan Africa. Outlook Today, about 40­50 percent of the global land area is under some kind of agricul- tural practices, and this share is expected to grow in the future. Future agroenviron- mental impacts are expected to be shaped by two countervailing forces. On the one hand, environmental pressures will tend to rise as a result of growing demand for food and agricultural products, mainly because of population and income growth, but also because of climate change. On the other hand, environmental pressures may be reduced by technological change and institutional responses to the environmental degradation caused by agriculture. The Rio conventions present important measures for global environmental gover- nance, in which agriculture plays an important role. Especially the climate negotia- tions and, to a certain extent, the biodiversity negotiations can be expected to exert a great deal of influence on both tropical and other countries. To meet future chal- lenges, global environmental governance needs to be more effective, notably regard- ing the coordination (a) among international environmental negotiations (climate, biodiversity, desertification), (b) between environmental negotiations and other rele- vant negotiations, especially international trade, and (c) among policy levels (local, international). The latter implies the recognition that poorer countries might need assistance with implementing and enforcing demanding international efforts to regu- late global agriculture-related problems at the local level. This often requires over- coming national interests for the sake of collective international action (Jacquet and Tubiana 2007). Endnotes 1. The Kyoto Protocol, adopted in 1997 and coming into effect in 2005, presents the conven- tion's application protocol, which quantifies the commitments of GHG emissions reduction for developed countries and provides the basis for a global carbon market. 2. CDM Project Database (http://www.cdmpipeline.org/cdm-projects-type.htm#3), accessed July 23, 2007. 3. http://www.cbd.int/default.shtml. 4. CDM Project Database (http://www.cdmpipeline.org/cdm-projects-type.htm#3), accessed July 23, 2007. 5. http://www.unccd.int/. 94 | LAURENCE TUBIANA AND SHEILA WERTZ-KANOUNNIKOFF References Achard, Frédéric, Hugh D. Eva, Philippe Mayaux, Hans-Jurgen Stibig, and Alan Belward. 2004. "Improved Estimates of Net Carbon Emission from Land Cover Change in the Tropics for 1990s." Global Biogeochemical Cycles 18 (2): GB2008, doi:10.1029/2003GB002142. Bruinsma, Jelle. 2003. World Agriculture: Towards 2015/2030; An FAO Perspective. London: Earthscan Publications. Chomitz, Kenneth, with Piet Buys, Giacomo di Luca, Timothy S. Thomas, and Sheila Wertz- Kanounnikoff. 2006. At Loggerheads? Agricultural Expansion, Poverty Reduction, and the Environment in the Tropical Forests. World Bank Policy Research Report. Washington, DC: World Bank. IPCC (Intergovernmental Panel on Climate Change). 2007. "Mitigation of Climate Change: Fourth Assessment Report." IPCC Working Group III. Final draft (http://www.ipcc.ch/). Giraud, Pierre-Noël, and Denis Loyer. 2007. "Les enjeux de la `révolution doublement verte' en Afrique." In Regards sur la terre: L'annuel du développement durable 2008, ed. Pierre Jacquet and Laurence Tubiana. Paris: Les Presses de Sciences Po. Jacquet, Pierre, and Laurence Tubiana. 2007. "Prise de conscience, et crise de confiance." In Regards sur la terre: L'annuel du développement durable 2007, ed. Pierre Jacquet and Laurence Tubiana. Paris: Les Presses de Sciences Po. Millennium Ecosystem Assessment. 2005. Ecosystems and Human Well-being: Synthesis. Washington DC: Island Press. (www.maweb.org/en/products.aspx.) Severino, J.-M., and Laurence Tubiana. 2002. "La question des biens publics globaux." In Gouvernance mondiale: Rapport de Synthèse, ed. Pierre Jacquet, Jean Pisani-Ferry, and Laurence Tubiana. Paris: Conseil d'Analyse Economique. (www.pisani-ferry.net/base/re02- gouvernance-contributions.pdf.) Stern, Nicholas. 2006. The Economics of Climate Change. Cambridge, U.K.: Cambridge Uni- versity Press. Steinfeld, Henning, Pierre Gerber, Tom Wassenaar, Vincent Castel, Mauricio Rosales, and Cees de Haan. 2006. Livestock's Long Shadow: Environmental Issues and Options. Rome: United Nations, Food and Agriculture Organization (www.virtualcentre.org/en/ library/key_pub/longshad/A0701E00.pdf). Part IV: Agriculture-Rural Nonfarm Linkages and Rural Poverty Agricultural Growth, Employment, and Wage Rates in Developing Countries RAGHAV GAIHA AND KATSUSHI IMAI This paper systematically assesses the relationships among agricultural production, employment, and wage rates, based on a detailed econometric analysis of cross- country panel data. Although agricultural growth is central to economic growth and poverty reduction in developing countries, there have been relatively few attempts to analyze the determinants of agricultural growth per se and to link the latter to growth of employment in agriculture and elsewhere.1 Much of the recent literature on agricultural growth consists of country case studies. This is partly because growth of agricultural production or productivity is specific to each country's endowments of physical and human capital. Besides, agricultural growth is affected by development of new labor-intensive technologies (Lipton 1977; Thirtle, Lin, and Piesse 2003). More important, greater openness to international markets and crop diversification are transforming agriculture and livelihoods in many ways. Accordingly, this paper seeks to assess (a) the determinants of agricultural growth, (b) the determinants of agricultural employment, and (c) the role of agricultural employment in stimulating nonfarm employment through backward and forward linkages with the rest of the economy (Mellor 1976; Mellor and Lele 1972).2 In view of the potential of high- value agricultural crops such as fruits, vegetables, and dairy products in boosting agricultural trade and overall growth, it also conducts a supplementary analysis of their export prospects. The rest of the paper is organized as follows. The next section describes the data sources, followed by a section devoted to a discussion of changes in agricultural pro- ductivity, employment, and crop diversification, first for select regions and then for Raghav Gaiha is Professor of Public Policy, Faculty of Management Studies, University of Delhi, India. Katsushi Imai is a Lecturer in Development Economics, School of Social Sciences, University of Manchester, United Kingdom. This study was carried out under the supervision and support of Thomas Elhaut, Director, and Ganesh Thapa, Regional Econo- mist in the Asia and the Pacific Division of the International Fund for Agricultural Development, Rome. We also bene- fited from discussions with Pasquale Scandizzo, Raghbendra Jha, and Anil Deolalikar. An earlier version was presented at a World Development Report (WDR) 2008 workshop in Berlin, and the incisive comments of Alain de Janvry were valuable in carrying out some revisions. Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 97 98 | RAGHAV GAIHA AND KATSUSHI IMAI a few countries, a section devoted to the specification and estimation of econometric models, a section focusing on the econometric results, and a section discussing simulations, based on the regression results, for China and India. A final section offers some concluding remarks. Data We construct panel data from various sources, such as the World Development Indi- cators (WDI), FAOSTAT, and LABORSTA.3 The data on research and development (R&D) are from Thirtle, Lin, and Piesse (2003). A major constraint is that agricul- tural wage series and R&D data are available only for a limited number of develop- ing countries. A related difficulty is that the overlap between different sources is limited. For these two reasons, the best we could do was to construct and analyze small panel data sets. Diversification, Productivity, and Employment Table 1 focuses on changes in the average share of noncereal crops, agricultural employment, productivity, and wages by region.4 The first three sets of averages are compared for the 1980s and 1990s, and the last is compared for the periods 1990­96 and 1997­2004, as wage data are only available after 1990. In East Asia and the Pacific, the share of area devoted to noncereal crops increased during the period. Agricultural employment per hectare decreased, while agricultural value added increased. East Europe and Central Asia also saw an increase in the share of noncereal crops, while agricultural employment per hectare increased and agricultural value added decreased. Real wages did not change much. In South Asia, the share of noncereal crops decreased, while both employment and agricultural value added increased.5 The share of noncereal crops, agricultural employment, and productivity was more or less stable between the 1980s and 1990s in Latin America and the Caribbean. Real wages, however, doubled between 1990­96 and 1997­2004. In the Middle East and North Africa, the share of noncereal crops rose slightly, while agricultural employment dropped and agricultural value added increased. In Sub-Saharan Africa, the crop diversification index decreased slightly, while agricul- tural value added increased. Employment did not change much. Real wages--lowest among all regions in both periods--rose markedly during 1990­96 to 1997­2004. Annex A contains graphs of changes in the crop diversification index (that is, the share of area devoted to noncereal crops in total arable land), agricultural employ- ment per hectare of arable land, and agricultural value added per hectare of arable land for select countries. These variables are scaled in the range of 0­100 to facilitate comparison. It is difficult to find a single pattern in these graphs largely because (a) the causality among these variables is complex and (b) different countries face different conditions, depending, for example, on their stage of agricultural and indus- trial development. The various patterns are delineated below. TABLE 1. Changes in Crop Diversification, Agricultural Employment, Production, and Wages, by Region Averages of crop mix, agricultural employment, and agricultural production Averages of agricultural wages and productiona 1980s 1990s 1990­96 1997­2004 East Asia and Pacific East Asia and Pacific Crop mix (the share of area for noncereal production 57.25 65.86 Real wage (constant US$ in 2000) 955.73 NAb in the total area of land, %) Agricultural value added 12.9 14.7 L (agricultural employment per arable land, no. of people) 2.42 2.08 Agricultural value added (constant billion US$ in 2000) 10.70 13.30 East Europe and Central Asia East Europe and Central Asia Crop mix (the share of area for noncereal production 45.25 54.70 Real wage (constant US$ in 2000) 330.37 203.44 in the total area of land, %) Agricultural value added 3.17 2.95 L (agricultural employment per arable land, no. of people) 0.34 0.44 Agricultural value added (constant billion US$ in 2000) 4.34 3.08 Latin America and the Caribbean Latin America and the Caribbean Crop mix (the share of area for noncereal production 74.41 73.10 Real wage (constant US$ in 2000) 2,286.05 2,200.92 in the total area of land, %) Agricultural value added 3.24 3.79 L (agricultural employment per arable land, no. of people) 0.44 0.46 Agricultural value added (constant billion US$ in 2000) 2.74 3.31 Middle East and North Africa Middle East and North Africa Crop mix (the share of area for noncereal production 68.39 69.62 Real wage (constant US$ in 2000) NAb NAb in the total area of land, %) Agricultural value added 4.49 5.18 L (agricultural employment per arable land, no. of people) 1.52 0.98 Agricultural value added (constant billion US$ in 2000) 4.14 4.58 South Asia South Asia Crop mix (the share of area for noncereal production 49.44 44.32 Real wage (constant US$ in 2000) NAb NAb in the total area of land, %) Agricultural value added 19.2 21.5 L (agricultural employment per arable land, no. of people) 1.73 2.46 Agricultural value added (constant billion US$ in 2000) 14.10 19.80 Sub-Saharan Africa Sub-Saharan Africa Crop mix (the share of area for noncereal production 58.32 56.55 Real wage (constant US$ in 2000) 30.41 46.51 in the total area of land, %) Agricultural value added 1.01 1.24 L (agricultural employment per arable land, no. of people) 1.27 1.22 Agricultural value added (constant billion US$ in 2000) 0.83 1.05 99 a. Agricultural wage data are unavailable before 1990. b. Real wages cannot be computed as appropriate exchange rates are unavailable. 100 | RAGHAV GAIHA AND KATSUSHI IMAI Brazil saw a simultaneous increase in the share of noncereal crops and in agricul- tural productivity and a decrease in agricultural employment. Bolivia experienced an increase in agricultural employment, while the share of noncereal crops and agricul- tural productivity remained unchanged. Sri Lanka's agricultural productivity gradually improved over the years. However, the share of noncereal crops increased from 1990 to 1996, followed by a decline until 2000. In Thailand the share of noncereal crops fell over time, while agricultural pro- ductivity rose. Employment gradually decreased. Indonesia is an interesting case because a decline in the share of noncereal crops coincided with a rise in agricultural productivity until 1995­96, when the pattern reversed. Mongolia experienced a sig- nificant increase in the share of noncereal crops and employment, but no noticeable improvement in agricultural productivity. Other countries show no clear pattern among these variables, except Iran, where both noncereal crop share and agricultural productivity rose during 1990­2004, while agricultural employment decreased dur- ing 1990­95, a pattern similar to Brazil's. Some robust relationships are, however, obtained from the econometric analysis. Econometric Models: Specification and Estimation Two models are estimated. Model 1 analyzes the relationship among agricultural pro- duction, employment, new technologies, and wage rates. Model 2 focuses on growth rates of farm and nonfarm employment and the relationship between them. Model 3 analyzes the determinants of exports of various noncereal agricultural products. Model 1 Model 1 is formulated to assess the determinants (such as technology, R&D, agri- cultural employment, inequality in land distribution, openness) of agricultural pro- duction per hectare of arable land (hereafter referred to as agricultural productivity). More specifically, equations 1, 2, and 3, are estimated. (1) logYa ,it 1log Amit 2 log Irrit 3log Fertit 4 log Lit 5log R& Di 6 log GINI(Land)i 7 Openi t i it , where logYa,it is log of agricultural value added per hectare of arable land (in constant 2000 U.S. dollars, taken from the World Development Indicators), i denotes country, and t denotes year. logYa measures a country's agricultural productivity. The ,it explanatory variables include technology comprising log of agricultural machinery or tractors per hectare of arable land (log Amit), log of share of irrigated land in total arable land (log Irrit), and log of fertilizer consumption per hectare of arable land (log Ferit). log Lit refers to log of total employment in the agricultural sector per hectare of arable land (based on WDI); log R&Di denotes the log of agricultural R&D expenditure per hectare (in 1995 U.S. dollars), available for select countries in Sub-Saharan Africa, East and Southeast Asia, and Latin America (Thirtle, Lin, and AGRICULTURAL GROWTH, EMPLOYMENT, AND WAGE RATES | 101 Piesse 2003: 1970); log GINI(Land)i is the log of the Gini coefficient of land distribution (based on FAOSTAT). Recent growth theory literature emphasizes gen- erally positive effects on economic growth of a reduction in income inequality (which is related to asset inequality), say, due to political economy factors in fiscal policies (for example, the median voter's preference for redistribution).6 Lower inequality in land distribution is likely to be associated with better incentives as well as higher effi- ciency of small farmers or laborers in agricultural production or activities. This raises total agricultural productivity and production. However, to the extent that ability to invest in agriculture is determined by easy access to credit, large landowners are in an advantageous position in a context of incomplete credit markets. It is difficult to say a priori whether the degree of inequality in land distribution will have a positive or a negative effect on agricultural productivity. Openi refers to openness of an economy to the rest of the world or the degree of integration with global markets. Different measures have been proposed in recent studies.7 We have experimented with a few. The first is the instrumental variable (IV) estimate of openness, where trade share in GDP is estimated by a measure of institu- tional quality and two exogenous variables: a measure of physical isolation8 and country size (that is, surface area; Gaiha and Imai 2008). Other measures of open- ness used are the Frankel-Romer and the Sachs-Warner indexes.9 tis a time-effect constant for all countries for a particular year, iis an unobservable individual (coun- try) effect, and itis an independent and identically distributed error term.10 The maximum set of observations for equation 1 covers 33 countries over the period 1980 to 2002.11 As an extension of equation 1, we include the interaction of log GINI(Land)i and a dummy variable on whether the observations are after 1990, D1990s, to check if there was a shift in the effect of land inequality on agricultural productivity between the 1980s and the 1990s and after. (1) log Ya,it 1log Amit 2 log Irrit 3 log Fertit 4log Lit 5log R & Di 6 log GINI(Land)i 7Openi 8log GINI(Land)i* D1990s t i it As the relationship between land inequality or concentration and agricultural pro- ductivity is complex and the data on the former are sparse, we have also experi- mented with the specification without land Gini. Equations 2 and 3 specify the determinants of agricultural employment and wage rates, respectively. (2) log Lit 2 log Wageit t i it, where log of agricultural employment per hectare, log La , is estimated by log of ,it monthly agricultural wage, logWageit. Monthly agricultural wages are taken from LABORSTA, the International Labour Organisation (ILO) data set. We use both nominal agricultural wages and real agricultural wages.12 is a constant, tis a time effect, iis an individual (country) effect, and itis an error term. As ILO wage data have limited coverage, the maximum set of observations for equation 2 covers 102 | RAGHAV GAIHA AND KATSUSHI IMAI 47 countries over the period 1995 to 2003 for nominal wages and 26 countries for real wages. The number of countries for the latter is smaller due to limited availabil- ity of exchange rate data in WDI. Agricultural wages are estimated by equation 3: (3) log Wageit 1 log Food Priceit t i it where log Wageit refers to log of nominal or real agricultural wage rate, and log Food Priceit is log of consumer food price index. is a constant,t is a time effect, iis an individual (country) effect, and itis an error term. The maximum set of observations for equation 3 covers 58 countries over the period 1996 to 2003 for nominal wages and 29 countries for real wages. As noted earlier, simultaneous-equations estimation is ruled out because of the small overlap between samples for different variables: in particular, there is a very small overlap among the sets of countries with wage, R&D, or openness data. For effi- cient use of these samples, a compromise is made. First, equations 1­3 are estimated separately as fixed-effects or random-effects specifications. While this strategy is some- what ad hoc, as it overlooks the endogeneity of some of the explanatory variables, it yields approximations to the coefficient estimates or elasticities from larger samples. Second, we estimate real agricultural wages using fixed- or random-effects specifica- tions of equation 3 to obtain the out-of-sample predictions for the sample with other variables, that is, log Food Priceit and log Ya . We are thus able to derive predicted ,it wages for a sample larger than the actual. Using the predicted values, we use fixed- effects or random-effects IV or G2SLS (generalized two-stage least squares) specifica- tions for equations 1 and 2 simultaneously. As a supplementary analysis, we estimate equations 2 and 3 as a system using the IV (G2SLS) random-effects estimator. Model 2 In an alternative specification, we focus on growth rates, rather than levels, of agri- cultural employment as a function of lagged growth rates of agricultural value added, lagged agricultural wage rates, and an index of crop diversification (that is, log of share of land used for crops other than cereal crops in total arable land). This is a proxy for agricultural diversification toward high-value commodities, such as fruits and vegetables.13 Growth rates can be expressed as the first differences of logarith- mic transformation of a variable. Growth rates of nonagricultural employment are then estimated as a function of the predicted growth rates of agricultural employ- ment, as shown in equation 5. (4) D log Lit 1D log Lit 1 2 D log Lit 2 3D log Yait 1 4D log Yait 2 5 D log Wageit 1 6D log Wageit 2 7D log NonCerealit 1 it (5) D log LNit 1D log Lit 1 i t it where LNit denotes nonagricultural employment and LNit is defined as the total number of people employed in sectors other than agriculture. Note that the lagged AGRICULTURAL GROWTH, EMPLOYMENT, AND WAGE RATES | 103 predicted value of D log Lit 1 is used as one of the explanatory variables in equation 5 to examine the effects of growth of agricultural employment on growth of non- agricultural employment over time. Estimating equation 4 as a static panel data model may cause bias in coefficient estimates. Accordingly, we use the Arellano-Bond estimator (Arellano and Bond 1991) to estimate equation 4. Ideally, equations 4 and 5 should be estimated as a sys- tem of dynamic equations. However, to avoid complication, we use actual values of DlogLit 1 in the fixed-effects version of equation 5 when the Arellano-Bond estimator is applied to equation 4. In other specifications, equations 4 and 5 are simultaneously estimated as the static version of the fixed-effects IV panel model. Model 3 As the importance of high-value agricultural commodities has grown rapidly in recent years, especially in agricultural trade, and their effects on the rest of the econ- omy are likely to be substantial, we focus here on their exports.14 Accordingly, in model 3, we estimate export demand functions of various noncereal agricultural products, posited to depend on world income, real exchange rate, and share of urban population in the total world population. (6) Eit 0 1Wt 2Rit 3 D log Ut oi t eit where Eit is export of various components of noncereal agricultural products, namely, (a) food and vegetables, (b) dairy products, (c) meat, (d) coffee, tea, and cocoa, and (e) sugar. These variables are constructed from FAOSTAT. Wt is a detrended measure of world income (deviations from a time trend); similarly Rit is a detrended measure of the real exchange rate; and D log Ut is the growth rate of Ut, the share of urban population in total world population. We also consider the case without D logUt. These measures are designed to circumvent collinearity and serial correlation. Results In this section we present the results for models 1, 2, and 3. Model 1 Table 2 shows elasticities of agricultural value added per hectare to technological and other explanatory variables in equation 1, for fixed- and random-effects specifica- tions. Note that all explanatory variables (including employment per hectare of arable land) are taken as given. These results are contrasted with those in table 5, where the endogeneity of log Lit is taken into account. Six cases are presented in table 2, based on different specifications (including dif- ferent measures of openness). In case A (second and third columns), an IV estimate of trade openness and log of Gini coefficient of land distribution are used as explana- tory variables, among others.15 In case B, land Gini is dropped. In case C, land Gini 104 TABLE 2. Determinants of Agricultural Value Added per Hectare (Model 1, Single-Equation Estimation of Equation 1) Case C Case B IV Estimate IV Estimate (With an interaction Case A (without of land Gini & Case D Case E Case F IV Estimate land Gini) 1990s Dummy) Trade share Frankel-Romer Sachs-Warner Openness: Coef. Z Coef. Z Coef. Z Coef. Z Coef. Z Coef. Z Fixed-effects model logAm 0.0709 (3.36)** -- -- -- -- 0.0845 (4.04)** 0.0702 (3.33)** 0.0709 (3.36)** logIrr 0.6196 (13.29)** -- -- -- -- 0.5946 (12.94)** 0.6216 (13.33)** 0.6196 (13.29)** logFert 0.0961 (4.31)** -- -- -- -- 0.1024 (4.68)** 0.0924 (4.10)** 0.0961 (4.31)** logL 0.0398 (2.55)* -- -- -- -- 0.0427 (2.81)** 0.0402 (2.58)* 0.0398 (2.55)* Openness -- -- -- -- -- -- 0.1629 ( 3.90)** -- -- Constant 4.8278 (30.81) -- -- -- -- 5.5012 (23.83)** 4.8373 (30.69)** 4.8278 (30.81)** No. of observations 347 -- -- -- -- 347 345 347 No. of countries 33 -- -- -- -- 33 32 33 Joint Significance F(26,288)= 35.88** -- -- -- -- F(27,287) 39.54** F(26,287) 38.64** F(26,288) 38.59** Overall R2 0.1702 -- -- -- -- 0.1481 0.1678 0.1702 Random--effects model logAm 0.0557 (2.34)* 0.0489 (2.41)* 0.0730 (2.88)** 0.0741 (3.10)** 0.0586 (2.54)* 0.0590 (2.49)* logIrr 0.3211 (8.25)** 0.4793 (12.05)** 0.2625 (7.09)** 0.3017 (7.75)** 0.3809 (9.57)** 0.3270 (8.31)** logFert 0.1122 (4.49)** 0.1031 (5.11)** 0.1170 (4.62)** 0.1243 (5.03)** 0.1046 (4.28)** 0.1155 (4.65)** logL 0.0589 (3.39)** 0.0411 (2.69)** 0.0612 (3.47)** 0.0617 (3.59)** 0.0523 (3.09)** 0.0596 (3.44)** logR&D 0.1514 (2.33)* 0.1410 (1.34) 0.1554 (2.73)** 0.1615 (2.47)* 0.2102 (3.27)** 0.1447 (2.09)* Gini Coef. of land distribution 0.4671 (2.20)* -- -- 0.4023 (2.10)* 0.4085 (1.91)+ 0.2662 (1.25) 0.3450 (1.51) Land Gini*Dummy(whether 1990s) -- -- -- -- 0.2095 (3.05)** -- -- -- -- Openness 0.3348 (2.16)* 0.3754 (1.43) 0.3041 (2.25)* 0.1484 ( 3.17)** 0.0535 (4.82)** 0.1989 (1.09) Constant 4.8104 (7.29)** 3.6618 (3.37)** 5.1104 (8.80)** -- -- -- -- -- No. of observations 347 389 347 347 345 347 No. of countries 33 38 33 33 32 33 Joint Significance Wald Chi2(29) = Wald Chi2(29) = Wald Chi2(29) = Wald Chi2(29) = Wald Chi2(29) = Wald Chi2(29) = 769.478** 1052.71** 762.58** 792.97** 8517.33** 7550.51** Overall R2 0.4588 0.3317 0.5013 0.3414 0.4612 0.3923 Hausman Test Chi2(25) 77.73** -- -- -- -- Chi2(26) 201.15** Chi2(27) 781.45** Chi2(26) 7550.51** In favour of Fixed -- -- -- -- In favour of Fixed In favour of Fixed In favour of Fixed effects model effects model effects model effects model Breusch Pagan Test Chi2(1) 1395.31** -- -- -- -- Chi2(1) 1259.73** Chi2(1) 1435.92** Chi2(1) 1698.20** In favour of Random -- -- -- -- In favour of Random In favour of Random In favour of Random effects model effects model effects model effects model Note: Single equation: equation (1), Fixed or Random Effects Model. Dep. Variable: logYa. Results for year dummies are omitted from the table. ** significant at 1% level. * significant at 5% level. + significant at 10% level. 105 106 | RAGHAV GAIHA AND KATSUSHI IMAI and the interaction of land Gini and D1990s (the dummy variable on whether after the year 1990) are included. In case D, the IV estimate of openness is replaced by log of trade share from WDI (log of sum of imports and exports as a percent of GDP). Although not endogenized, this measure has the advantage of varying over time. Case E uses the Frankel-Romer index as a measure of openness and log of land Gini. In case F, the Sachs-Warner index replaces the Frankel-Romer index. Each case in table 2 has two sets of regression results, one for fixed effects and another for random effects. As the choice between the two is far from straightfor- ward, we present results based on both specifications.16 First, we comment on the sig- nificance of explanatory variables. The magnitude of elasticities is discussed later in conjunction with the results in table 5. As expected, all technological variables, syn- onymous with modernization of agricultural production, have positive and signifi- cant roles in explaining differences in agricultural value added per hectare of arable land, our proxy for agricultural productivity. Specifically, machinery or tractors per hectare of arable land has a positive and significant coefficient at the 1 percent level. Irrigation share also has a positive and highly significant coefficient (at the 1 percent level), confirming the importance of irrigation in raising agricultural productivity. Fertilizer use has a positive and significant coefficient at the 1 percent level, as does R&D expenditure per hectare.17 The positive and significant coefficient of land Gini suggests that the greater is the inequality, the higher is agricultural productivity. If land Gini is dropped in case B, log R&D and openness cease to be significant. The positive and significant coefficient of the interaction of land Gini and the 1990s dummy implies that the positive effect of land inequality on agricultural productivity was much larger in the 1990s. The elasticity of agricultural productivity with respect to land Gini was 0.4 in the 1980s and 0.61 in the 1990s. The results on openness are mixed. Positive and significant coefficients are obtained with the IV estimate of openness (case A) and the Frankel-Romer index (case E). How- ever, the coefficient of trade share is negative and significant (case D). The Sachs- Warner index has a positive but not significant coefficient (case F). The reason for the negative coefficients of trade share in case D is far from obvious.18 So, under certain conditions, a favorable effect of openness on agricultural productivity cannot be ruled out.19 This is plausible, as greater openness implies more competitive markets and eas- ier access to better technology. However, in specific cases of incomplete credit and insurance markets and weak infrastructure, the potential gains from trade liberaliza- tion may not be fully realized or may accrue to a small subset of large landowners. The choice of a specification is usually based on a specific test, but this is limiting in some ways. With this caveat, it may be noted that while the Hausman test favors the fixed-effects specification, the Breusch-Pagan test favors the random-effects one.20 The Shapiro-Wilk tests for normality show that the null hypothesis that resid- uals are normally distributed is rejected at the 5 percent level in a few cases (fixed effects in case F; random effects in cases A, B, C, and D). Thus the results have to be interpreted with caution. Table 3 contains results on determinants of employment per hectare of arable land (that is, equation 2 of model 1). Three cases are shown: case A, where log of nominal TABLE 3. Elasticity Estimates of Agricultural Employment per Hectare Case C Case A Case B Nominal wage Nominal wage Real wage (sample is same as Case B) Random-effects model Random-effects model Random-effects model Wage: Coef. Z Coef. Z Coef. Z log Wage 0.0234 ( 1.20) 0.0216 ( 0.73) 0.0249 ( 1.00) constant 0.6469 ( 2.28) -- -- -- -- No. of observations 226 135 135 No. of countries 47 26 26 Joint Significance Wald Chi2(8) 6.01 Wald Chi2(8) 2.60 Wald Chi2(8) 2.76 Overall R2 0.0027 0.0000 0.0001 Hausman Test Chi2(7) 0.32 Chi2(7) 0.06 Chi2(7) 0.08 In favour of Random effects model In favour of Random effects model In favour of Random effects model Breusch Pagan Test Chi2(1) 455.43** Chi2(1) 286.21** Chi2(1) 286.43** In favour of Random effects model In favour of Random effects model In favour of Random effects model Note: Single equation: equation (2), Fixed or Random Effects Model. Dependent variable: logL ** significant at 1% level. * significant at 5% level. significant at 10% level. 107 108 | RAGHAV GAIHA AND KATSUSHI IMAI monthly wage is used as an explanatory variable in a sample of 47 countries; case B, where log of real wages replaces log of nominal wages in a sample of 26 countries; and case C, where log of nominal wages is used in the same sample of 26 countries to facilitate comparison with case B. Since the random-effects version is favored by both the Hausman and Breusch-Pagan tests in all of these cases, we confine our com- ments to this case. Only wage is included as an explanatory variable in each case.21 Neither nominal nor real wage is significant, and thus the results have to be interpreted with caution. However, the negative coefficient of wage rate in all cases (with t-values exceeding 1) is consistent with the hypothesized labor demand function. The results on determinants of agricultural wage rates are shown in table 4. Three cases are shown: case A, random-effects specification for nominal monthly wages in a sample of 58 countries; case B, fixed-effects specification for real wages in a sam- ple of 29 countries; and case C, random-effects specification for nominal wages in the same sample of 29 countries as in case B. As expected, log of food price has a negative and significant coefficient, but only in case B. Several year dummies have positive coefficients in case B, implying a rising real wage rate. Somewhat surprising, log Yait does not have a significant coefficient in any of the three cases in question. The predicted wage from this regression is used in the IV estimation for equations 1 and 2 in table 6. Table 5 shows the case where equations 2 and 3 are estimated simultaneously using a random-effects IV technique. This shows that the results are generally simi- lar to those in case B in table 3 and case B in table 4, in which each equation is esti- mated separately. The procedure used for estimating the results in table 6 is summarized first. In the first stage of IV estimation, equation 2 is estimated and, in the second stage, equa- tion 1 is estimated. The results are shown in the same format as in tables 2 and 3 to facilitate comparison. There are some notable differences between the results in tables 2 and 6. In table 6, log of fertilizer use ceases to be significant, while agricultural machinery or tractor and irrigation have positive and significant coefficients. Log of R&D expenditure also ceases to have a significant effect on agricultural productivity. Openness indica- tors are generally not significant except in case D of fixed effects and case F of ran- dom effects in table 6. As trade share is not instrumented, we are inclined to be skeptical of the negative effect. The interaction of land Gini and the 1990s dummy is positive, but not significant. Elasticities of agricultural production with respect to factors related to modern- ization of agriculture also vary. Table 2 shows the elasticities of 0.06­0.08 for machinery or tractor, 0.32­0.62 for irrigation, 0.09­0.12 for fertilizer use, and 0.15­0.21 for R&D expenditure. In table 6, however, the elasticities are 0.27­0.34 for machinery or tractor and 0.33­0.45 for irrigation. As we are inclined to rely more on the results in table 6, it follows that modernization of agricultural technology has the potential to raise agricultural productivity in a substantial way.22 TABLE 4. Estimates of Log Agricultural Wage (Model 1, Single-Equation Estimation of Equation 3) Case A Case B Case C Nominal wage Real wage Nominal wage(same sample as B) Random effects model Random effects model Random effects model Dep. Variable: Coef. Z Coef. Z Coef. Z logFoodPrice 0.0269 (0.29) 0.8861 ( 4.49)** 0.2192 (1.15) Year 1996 0.2005 (1.32) 0.2603 (1.17) 0.6635 ( 2.43)** Year 1997 0.3860 (2.57)** 0.4174 (1.81)** 0.3597 ( 1.43) Year 1998 0.5592 (3.68)** 0.5220 (2.22)** 0.1717 ( 0.73) Year 1999 0.5558 (3.62)** 0.3144 (1.32) 0.0675 ( 0.30) Year 2000 0.5054 (3.18)** 0.3780 (1.52) 0.2650 ( 1.18) Year 2001 0.6095 (3.68)** 0.4360 (1.67)+ 0.2013 ( 0.91) Year 2002 0.6679 (3.98)** 0.5282 (2.00)* 0.1391 ( 0.62) Year 2003 0.7263 (4.03)** 0.6512 (2.30)** 0.0865 ( 0.39) Constant 8.0370 (16.33) 7.7745 (7.99) 7.8644 (7.40) No. of observations 398 202 202 No. of countries 58 29 29 Joint Significance Wald Chi2(9) 38.12** Wald Chi2(9) 25.32** Wald Chi2(9) 20.64** Overall R2 0.0069 0.0211 0.0035 Hausman Test Chi2(9) 35.38 Chi2(8) 0.38 Chi2(9) 0.20 In favour of Random effects model In favour of Random effects model In favour of Random effects model Breusch Pagan Test Chi2(1) 1154.76** Chi2(1) 615.19** Chi2(1) 621.19** In favour of Random effects model In favour of Random effects model In favour of Random effects model Note: Single equation: equation (3), Static Model (Random Effects Model). Dependent variable: log Wage. ** significant at 1% level. * significant at 5% level. significant at 10% level. 109 110 | RAGHAV GAIHA AND KATSUSHI IMAI TABLE 5. Elasticity Estimates for Employment per Hectare and Real Wage (Model 1, IV Estimation of Equations 2 and 3) Equation (2) Equation (3) (Second stage) (First stage) logL Real wage Dep. variable: Coef. Z Coef. Z log Real Wage 0.0100 (0.14) -- -- logFoodPrice -- -- 1.0023 ( 3.95)** Year 1995 1.2583 ( 3.07)** 8.5963 (6.39)** Year 1996 1.2939 ( 3.09)** 8.8547 (6.52)** Year 1997 1.2100 ( 2.90)** 9.0806 (6.44)** Year 1998 1.2372 ( 2.96)** 9.2175 (6.44)** Year 1999 1.2458 ( 3.14)** 8.8968 (6.20)** Year 2000 1.2667 ( 3.21)** 8.9527 (6.14)** Year 2001 1.2345 ( 3.14)** 8.9925 (6.08)** Year 2002 1.2325 ( 2.59)* 9.2381 (5.64)** Year 2003 -- Constant -- constant 0.6469 ( 2.28) -- -- No. of observations 126 126 No. of countries 24 24 Joint Significance Wald Chi2(8) 30.43 Wald Chi2(9) 50.00 Overall R2 0.0034 Davidson-MacKinnon test for consistency of OLS Ho: OLS is consistent F(1,93) 0.03 Prob z 0.87 Hausman Test Chi2(8) 0.19 In favour of Random effects model Note: System equation: G2SLS random effects IV regression for equations (2) and (3). ** significant at 1% level. * significant at 5% level. significant at 10% level. Model 2 Table 7 shows two sets of results for model 2, which focuses on determinants of agri- cultural and nonagricultural employment, including the link between them. Consider first the case of agricultural employment. The first two cases--case A where nominal wage is used as an explanatory variable and case B where real wage replaces nomi- nal wage--are based on the Arellano-Bond estimator for equation 4 and the fixed- effects specification of equation 5. The next two cases--case C for nominal wage and case D for real wage--are based on fixed-effects IV panel data specifications. As the samples are small, the results have to be interpreted with caution. Dlog L( 1), or one-period lagged agricultural employment growth, has a positive and significant effect only in case C, while D log L( 2), or two-period lagged agri- cultural growth, has a positive and significant coefficient in cases A and B in the TABLE 6. Determinants of Agricultural Value Added per Hectare and Agricultural Employment per Hectare (Model 1, IV Estimation of Equations 1 and 2) Case C Case B IV estimate IV estimate (With an interaction Case A (Without of land Gini & Case D Case E Case F IV estimate Land Gini) 1990s Dummy) Trade share Frankel-romer Sachs-warner Openness: Coef. Z Coef. Z Coef. Z Coef. Z Coef. Z Coef. Z 2nd Stage Fixed-effects IV panel model logAm 0.1974 (1.12) -- -- -- -- 0.2901 (1.06) 0.1974 (1.12) 0.2889 (2.74)** logIrr 0.4823 (3.42)** -- -- -- -- 0.3702 (1.65) 0.4823 (3.42)** 0.5289 (3.49)** logFert 0.0580 ( 0.25) -- -- -- -- 0.1350 ( 0.38) 0.0581 ( 0.25) 0.1748 ( 1.37) logL 0.4611 (0.89) -- -- -- -- 0.7033 (0.89) 0.4611 (0.89) 0.6781 (3.04)** Openness -- -- -- -- -- -- 0.2591 ( 1.83)+ -- -- -- -- Constant 7.6631 (7.18) -- -- -- -- 8.7943 (2.44) 6.7793 (3.14) 7.6631 (7.18) No. of observations 335 -- -- 335 335 335 No. of countries 32 -- -- 32 32 32 Joint Significance Wald Chi2(26) -- -- Wald Chi2(27) Wald Chi2(27) Wald Chi2(26) 386236.39** 188940.46** 386236.37** (386236.37) Overall R2 0.1404 -- -- 0.0749 0.1404 0.1702 Davidson-MacKinnon test for -- -- consistency of OLS Ho: OLS is consistent F(1,276) 2.741 F(1,275) 6.302* F(1,276) 2.74 F(1,276) 2.74 Prob z 0.09 Prob z 0.01 Prob z 0.10 Prob z 0.10 (continued) 111 112 TABLE 6. (continued) Case C Case B IV estimate IV estimate (With an interaction Case A (Without of land Gini & Case D Case E Case F IV estimate Land Gini) 1990s Dummy) Trade share Frankel-romer Sachs-warner Openness: Coef. Z Coef. Z Coef. Z Coef. Z Coef. Z Coef. Z 2nd Stage Random-effects IV panel model logAm 0.1327 (0.62) 0.1896 (1.23) 0.2467 (0.91) 0.1019 ( 0.32) 0.0981 (0.51) 0.1056 (0.55) logIrr 0.1433 (1.70) 0.4318 (3.62)** 0.2271 (2.27)* 0.0233 ( 0.19) 0.1287 (2.40)* 0.1521 (1.89) logFert 0.1011 (0.50) 0.0342 ( 0.19) 0.0357 ( 0.12) 0.3805 (1.40) 0.1576 (0.97) 0.1183 (0.58) logL 0.3642 (0.74)** 0.4454 (0.99) 0.5764 (0.87) 0.1591 ( 0.23) 0.2693 (0.61) 0.2998 (0.64) logR&D 0.0715 (0.50) 0.1169 (0.46) 0.0390 (0.23) 0.2297 (0.91) 0.1140 (0.70) 0.0698 (0.51) Gini Coef. of land distribution 0.9247 (3.81)** -- -- 0.7871 (1.97)* 0.7782 (2.69)** 0.8031 (2.83)** 0.7430 (3.14)** Land Gini*Dummy(whether 1990s) -- -- -- -- 0.1962 (1.31) -- -- -- -- -- -- Openness 0.2140 (1.26) 0.2178 (0.33) 0.1710 (0.65) 0.0250 ( 0.35) 0.0197 (1.12) 0.3788 (2.13)* Constant 5.7660 (2.62) 5.7917 (1.75) -- -- -- -- -- -- -- -- No. of observations 335 362 335 335 335 335 No. of countries 32 36 32 32 32 32 Joint Significance Wald Chi2(29) Wald Chi2(28) Wald Chi2(30) Wald Chi2(29) Wald Chi2(29) Wald Chi2(29) 377.26** 327.25** 3640.86** 27008.06** 20067.91** 11534.06** Overall R2 0.4051 0.2192 0.2453 0.6191 0.4757 0.4440 Hausman Test Chi2(25) 14.52 NA NA Chi2(27) 202.06** Chi2(26) 5310.08** Chi2(26) 635.58** In favour of Random In favour of Fixed In favour of Fixed In favour of Fixed effects model effects model effects model effects model Note: IV estimation, 2nd stage: equation (1), Fixed or Random Effects IV Panel Model. Dep. Variable: logYa. TABLE 6. (continued) 1st Stage Random Effects Model 1st Stage Fixed Effects Model Predicted real wage Predicted real wage G2SLS Wage: Coef. Z Coef. Z Predicted logWage 0.0130 (0.97) 0.0134 (0.95) constant 4.1300 ( 6.91) 4.1254 ( 3.40) No. of observations 335 335 No. of countries 32 30 Joint Significance F(26,277) 3.85** Wald Chi2(29) 101** Overall R2 0.1987 NA Note: IV estimation, 1st stage: equation (2), Fixed or Random Effects IV Panel Model. Dep. variable: logL. Results for year dummies are omitted from the table. The first stage includes vari- ables exogenous to the system (log Am, log Irr, logFert, logR&D, log Gini_Land), but the results are not shown. ** significant at 1% level. * significant at 5% level. significant at 10% level. 113 114 TABLE 7. Determinants of Agricultural and Nonagricultural Employment (Model 2; Equations 4 and 5) Case A Case B Nominal wage Real wage Case C Case D dynamic model dynamic model Nominal wage Real wage Arellano-Bond Arellano-Bond static panel model static panel model (2 step estimator) (2 step estimator) Fixed-effects IV model Fixed-effects IV model Coef. Z Coef. Z Coef. Z Coef. Z D log L( 1) 0.0658 (0.35) 1.2616 ( 1.17) First Stage 0.3328 (1.66) 0.2828 ( 0.86) D log L( 2) 0.2077 (2.07)* 1.1539 (2.78)** 0.1328 (0.63) 0.1079 ( 0.32) D logYa( 1) 0.1670 ( 0.49) 1.2350 (0.81) 0.4625 (1.39) 0.0605 (0.13) D logYa( 2) 0.0542 ( 0.26) 2.4760 ( 1.37) 0.1261 (0.41) 0.2152 (0.56) D logWage( 1) 0.0033 ( 0.23) 0.0351 ( 0.55) 0.0817 ( 1.64) 0.0639 ( 2.10)* D logWage( 2) 0.0019 ( 0.15) 0.0265 ( 0.52) 0.2059 (1.06) 0.4797 ( 1.70) Dlog NonCereal( 1) 0.1721 (1.48) 0.5167 ( 0.75) 0.3840 ( 1.68) 0.5603 ( 2.19)* Constant 0.0006 ( 0.04) 0.0009 (0.02) 0.7698 ( 2.06) 1.9653 ( 2.65) No. of observations 79 45 55 33 No. of countries 26 13 23 12 Joint Significance Wald Chi2(7) 1365** Wald Chi2(7) 31.11** F(7,25) 2.70* F(7,14) 1.21 m2a 0.77 Pr m2 0.4410 0.95 Pr m2 0.3404 -- -- Saran Testb Chi2(93) 11.86 Chi2(93) 3.38 -- -- Note: Dep. Variable: D log L (equation (4)). a. Arellano-Bond Test for second order serial autocorrelation (H0: No autocorrelation). b. Sagarn Test of over identifying restriction (H0: the overidentifying restrictions do not hold in the precence of autocorrelation). ** significant at 1% level. * significant at 5 % level. significant at 10% level. TABLE 7. (continued) Case C Case D Case A Case B Fixed-effects IV model Fixed-effects IV model Fixed-effects model Fixed-effects model second stage second stage D logL( 1)a 0.1063 (1.88)+ 0.1315 (1.57) Second Stage 0.5190 (2.21)* 1.1642 (2.61)* Constant 0.1237 (2.20) 0.1451 (1.75) 0.7102 (2.29) 1.6604 (2.65) No. of observations 140 88 55 33 Joint Significance F(1,100) 3.54 F(1,100) 2.47 Wald Chi2(3) 10.24** Wald Chi2(2) 9.28** Hausman Test Chi2(1) 3.16+ Chi2(1) 2.39 Hausman Test Chi2(1) 4.72* Chi2(1) 6.64** In favour of fixed In favour of random In favour of fixed In favour of fixed effects model effects model effects model effects model Breusch Pagan Test Chi2(1) 0.04 Chi2(1) = 0.15 -- -- In favour of Fixed In favour of Fixed effects model effects model Note: Dep. variable: D log LN (equation (5)). a. D logL( 1) takes the actual value in Cases A and B, while predicted values are used in Cases C and D in the IV estimations. ** significant at 1% level. * significant at 5% level. significant at 10% level. 115 116 | RAGHAV GAIHA AND KATSUSHI IMAI determination of D log Lit or growth of agricultural employment (that is, equation 4), as shown in the first panel of table 7. Only the former seems consistent with our intu- ition, as it is not obvious why the two-period lagged agricultural employment growth matters. Contrary to some evidence, lagged growth of crop diversification or nonce- real production has a negative effect on growth of agricultural employment (cases C and D).23 In the second panel of table 7, the effect of the lagged growth rate of agricultural employment on the growth rate of nonagricultural employment is positive and sig- nificant in all cases. This implies that higher growth of agricultural employment has some positive effects on the growth of nonagricultural employment, for example, through backward and forward linkages with the rest of the economy. It is important to note that the effect of the growth rate of agricultural employment is substantially greater when predicted values of D log Lit 1 are used. This further confirms that endogenous treatment of some of the variables used here can lead to marked differ- ences in corresponding elasticities. An issue of policy significance is whether conditions can be created for small- holders to benefit from the potential benefits of agricultural diversification. Some evidence points to substantially greater profitability of fruits and vegetables. Fruits are eight times more profitable than cereals and other crops, while vegetables are 4.8 times more profitable. But both yields and prices of horticultural commodities tend to be more volatile. So, apart from protection against production and price risks (through, for example, technological choices, provision of crop insurance), there is a strong case for ensuring easy access to markets and credit. As the perisha- bility of fruits, vegetables, and dairy products is also a major concern for small- holders, expansion of storage and refrigeration facilities is an additional priority.24 Finally, food safety standards may be harder to fulfill without some intermediation. While the prospects may not seem bright, it would be somewhat pessimistic to sur- mise that smallholders are unlikely to integrate into rapidly emerging high-value food chains (Gaiha and Thapa 2006). Model 3 Table 8 contains results on exports of noncereal agricultural products. Results in case A suggest that an increase in world income has a positive and significant impact on exports of fruits and vegetables in less-developed countries. It is also found that higher growth rates of share of urban population tend to result in higher exports of fruits and vegetables. As expected, exchange rate depreciation boosts these exports. A similar pattern is observed in case B, where the dependent variable is dairy prod- ucts. However, a somewhat surprising result is that world income has a negative effect on exports of meat (case C) and coffee, tea, and cocoa (case D). Urbanization has a positive and significant effect on export demand for coffee, tea, and cocoa. Export demand for sugar (case F) varies inversely with the real exchange rate and positively with urbanization. TABLE 8. Determinants of Export of Noncereal Agricultural Products (Model 3; Equation 6) Case A: Fruits and Vegetables Case B: Dairy Products Case C: Meat log [export of fruit & log [export of fruit & log [export of dairy log [export of dairy log [export log [export vegetables] vegetables] products] products] of meat] of meat] Random effects Random effects Random effects Random effects Random effects Random effects model model model model model model Dependent Variable: Coef. Z Coef. Z Coef. Z Coef. Z Coef. Z Coef. Z World Income Shocks 0.4472 (2.53)* 0.4716 (2.68)* 2.6030 (9.47)** 2.6373 (9.63)** 0.3756 ( 1.45) 0.3749 ( 1.44) [Defined as a Deviation of Trend of log World Income (GDP)] Exchange Rate Shocks 0.0000047 ( 1.88)* 0.0000047 ( 1.88)* 0.0000139 ( 3.39)** 0.0000140 ( 3.42)** 0.0000074 ( 1.90)* 0.0000074 ( 1.90)* [Defined as a Deviation of Trend of Real Exchange Rte Index] D log [Share of Urban -- -- 43.5752 (5.11)** -- -- 64.8439 (4.74)** -- -- 1.1570 (0.09) Population in Total Population in the World] Constant 1.3930 (0.48) 0.6164 (0.21) 36.4102 ( 8.11) 36.4102 ( 8.11) 12.6455 (2.98) 13.7500 (3.24) No. of observations 4641 4641 3266 3266 3525 3525 No. of countries 139 139 139 139 139 139 Joint Significance Wald Chi2(2) 11.44** Wald Chi2(3) 37.54** Wald Chi2(2) 112.35** Wald Chi2(3) 135.63** Wald Chi2(2) 5.04+ Wald Chi2(3) 5.05 Hausman Test Chi2(1) 0.10 Chi2(1) 0.38 Chi2(2) 0.51 Chi2(2) 0.55 Chi2(2) 0.79 Chi2(2) 3.93 In favour of Random in favour of Random in favour of Random in favour of Random in favour of Random in favour of Random effects model effects model effects model effects model effects model effects model Breusch Pagan Test Chi2(1) 46411.18** Chi2(1) 46504.53** Chi2(1) 19328.83** Chi2(1) 19389.70** Chi2(1) 27497.72** Chi2(1) 27468.31** In favour of Random in favour of Random in favour of Random in favour of Random in favour of Random in favour of Random effects model effects model effects model effects model effects model effects model 117 Note: Equation (6), Fixed or Random Effects Model. Dep. variable: log of export of noncereal (and cereal) agricultural products. (continued) 118 TABLE 8. (continued) Case D: Coffee, Tea and Cocoa Case F: Sugar log [export of coffee, tea, log [export of coffee, tea, and cocoa] and cocoa] log [export of sugar] log [export of sugar] Fixed effects model Fixed effects model Random effects model Random effects model Dependent Variable: Coef. Z Coef. Z Coef. Z Coef. Z World Income Shocks 1.1361 ( 6.19)** 1.1182 ( 6.10)** 0.0079 ( 0.03) 0.0281 (0.12) [Defined as a Deviation of Trend of log World of log World Income (GDP)] Exchange Rate Shocks 0.0000023 ( 0.87) 0.0000023 ( 0.88) 0.0000099 ( 2.91)** 0.0000098 ( 2.91)** [Defined as a Deviation of Trend of Real Exchange Rte Index] D log [Share of Urban -- -- 34.6873 (3.86)** -- -- 51.2649 (4.46)** Population in Total Population in the World] Constant 27.3051 (9.14) 26.7079 (8.94) 7.3340 (1.93) 6.3011 (1.66) No. of observations 4412 4412 3703 3703 No. of countries 139 139 139 139 Joint Significance F(2,4271) 19.18** F(3,4270) 17.80** Wald Chi2(2) 8.60* Wald Chi2(3) 28.51** Hausman Test Chi2(1) 7.36** Chi2(1) 5.41 Chi2(2) 0.12 Chi2(2) 0.08 In favour of Fixed effects model In favour of Fixed effects model In favour of Random effects model In favour of Random effects model Breusch Pagan Test Chi2(1) 44169.64** Chi2(1) 44224.61** Chi2(1) 21493.23** Chi2(1) 21560.94** In favour of Random effects model In favour of Random effects model In favour of Random effects model In favour of Random effects model Note: Dep. variable: log of export of noncereal agricultural products. ** = significant at 1% level. * = significant at 5% level. + = significant at 10% level. AGRICULTURAL GROWTH, EMPLOYMENT, AND WAGE RATES | 119 TABLE 9. Simulations for China and India Based on table 6, equation 5, case A Change in average of growth rates of nonfarm employmenta Indicator China India Change in average of growth rate of agricultural employment per hectare of arable landb 5 percent increase 2.52 1.39 10 percent increase 4.87 2.79 15 percent increase 7.31 4.18 Average of actual growth rate of agricultural employment per hectare of arable land 0.92 3.50 a. Time-series average is calculated by the observations in 1987­2000 for China and in 1991­95 for India. The periods are determined by the availability of explanatory variables for these countries. b. The baseline is the predicted value of time-series average growth rates of nonfarm employment based on the actual values of explanatory variables. Simulation is carried out by increasing the growth rate of agricultural employ- ment per hectare of arable land by 5, 10, or 15 percent with the values of other explanatory variables unchanged. Based on table 5, equation 2, random-effects model Change in average of agricultural employment per hectare of arable landa Indicator China India Share of noncereal crop in total arable land 5 percent increase -- 2.40 10 percent increase -- 4.74 15 percent increase -- 7.03 Average of actual agricultural employment per -- 1.58 hectare of arable land a. Time-series average is calculated by the observations in 1980­2002 for India. China does not have data on noncereal crop share and thus computation is not feasible. These calculations correspond to elasticity estimates. Simulations for China and India As China and India are two important developing countries with large populations and high growth rates, simulations based on regression results have illustrative value. The first panel of table 9, based on case A in table 7 (equation 5), presents the effects of higher rates of growth in agricultural employment on rates of growth in nonfarm employment. The baseline is the average of predicted values based on actual values of all explanatory variables. Three counterfactual cases are shown: 5, 10, and 15 percent higher growth rates of agricultural employment per hectare (or DLit). Our simulations suggest that higher growth of farm employment would accelerate the growth of nonfarm employment. For example, a 10 percent higher growth rate of agricultural employment would raise the growth of nonfarm employment by 4.87 percent in China and 2.79 percent in India. The second panel illustrates the effects of change in the share of noncereal crop areas on agricultural employment per hectare of arable land. Since the share of noncereal 120 | RAGHAV GAIHA AND KATSUSHI IMAI crop area is not available for China, the results are confined to India. It turns out that agricultural employment per hectare is reduced substantially as the share of noncereal crop area increases. This, however, should not be taken to imply that the overall expan- sionary effect of high-value agriculture would be negligible. Concluding Observations Drawing on different specifications and methods of panel data estimation designed to make efficient use of cross-country samples (including panel data), we have ana- lyzed the relationships among agricultural productivity, employment, technology, openness of the economy, and inequality in land distribution. Agricultural produc- tivity varies with technology and employment, but more so with employment. The effect of openness on agricultural productivity is ambiguous--it is positive, negative, or not significant--depending on the definition of openness, specification, and estimation procedure used. In a specific case, openness does contribute to higher agri- cultural productivity. A somewhat surprising result is the positive effect of inequality in land distribu- tion on agricultural productivity. Arguably, when credit markets are imperfect or incomplete, greater inequality in land distribution may imply a more significant role for large landowners in agricultural investment through easier access to credit. In another specification, the determinants of growth rates of agricultural and nonagri- cultural employment and their linkages are examined using both dynamic and static models. The growth rate of agricultural employment has a strong (lagged) positive effect on the growth rate of nonagricultural employment. This is illustrated by our simulations for China and India. Even though the share of agriculture has declined in developing countries, its contribution to overall economic growth and generation of employment is substantial. While a case for acceleration of agricultural growth through modernization of technology, crop diversification, and exploitation of high- value export opportunities rests on more complete credit and insurance markets and infrastructural support, some negative effects of crop diversification on employment are likely. In view of the growing potential for exports of fruits and vegetables and other high-value agricultural products linked to rising global income and rapid urbanization, the overall expansionary effect of high-value agricultural products is likely to outweigh any negative effects. AGRICULTURAL GROWTH, EMPLOYMENT, AND WAGE RATES | 121 ANNEX A. Crop Diversification, Agricultural Employment, and Production in Select Countries Brazil Bolivia NonCereal/Agricultural_ValueAdded_10/ NonCereal/Agricultural_ValueAdded_5/ Agricultural_Employment_200 Agricultural_Employment_2000 75 140 70 120 65 100 60 80 55 60 50 40 45 20 1990 1995 2000 2005 1990 1995 2000 2005 Year Year NonCereal NonCereal Agricultural_ValueAdded_10 Agricultural_ValueAdded_5 Agricultural_Employment_200 Agricultural_Employment_2000 Sri Lanka Thailand NonCereal/Agricultural_ValueAdded_100/ NonCereal/Agricultural_ValueAdded_20/ Agricultural_Employment_10 Agricultural_Employment_30 40 45 30 40 20 35 10 30 0 25 1990 1995 2000 2005 1990 1995 2000 2005 Year Year NonCereal NonCereal Agricultural_ValueAdded_100 Agricultural_ValueAdded_20 Agricultural_Employment_10 Agricultural_Employment_30 122 | RAGHAV GAIHA AND KATSUSHI IMAI Indonesia Malaysia NonCereal/Agricultural_ValueAdded_50/ NonCereal/Agricultural_ValueAdded_66/ Agricultural_Employment_10 Agricultural_Employment_50 35 80 30 70 25 60 20 50 15 10 40 1990 1995 2000 2005 1990 1995 2000 2005 Year Year NonCereal NonCereal Agricultural_ValueAdded_50 Agricultural_ValueAdded_66 Agricultural_Employment_10 Agricultural_Employment_50 Mongolia Kazakhstan NonCereal/Agricultural_ValueAdded_3/ NonCereal/Agricultural_ValueAdded_1/ Agricultural_Employment_200 Agricultural_Employment_100 120 100 80 100 60 80 40 60 20 40 0 1990 1995 2000 2005 1990 1995 2000 2005 Year Year NonCereal NonCereal Agricultural_ValueAdded_3 Agricultural_ValueAdded_1 Agricultural_Employment_200 Agricultural_Employment_100 AGRICULTURAL GROWTH, EMPLOYMENT, AND WAGE RATES | 123 Kyrgyz Republic Tajikistan NonCereal/Agricultural_ValueAdded_10/ NonCereal/Agricultural_ValueAdded_5/ Agricultural_Employment_100 Agricultural_Employment_50 100 150 80 100 60 50 40 20 0 1990 1995 2000 2005 1990 1995 2000 2005 Year Year NonCereal NonCereal Agricultural_ValueAdded_10 Agricultural_ValueAdded_5 Agricultural_Employment_100 Agricultural_Employment_50 Iran Kenya NonCereal/Agricultural_ValueAdded_20/ NonCereal/Agricultural_ValueAdded_10/ Agricultural_Employment_200 Agricultural_Employment_100 60 65 60 50 55 50 40 45 40 30 35 1990 1995 2000 2005 1990 1995 2000 2005 Year Year NonCereal NonCereal Agricultural_ValueAdded_20 Agricultural_ValueAdded_10 Agricultural_Employment_200 Agricultural_Employment_100 124 | RAGHAV GAIHA AND KATSUSHI IMAI Mauritius NonCereal/Agricultural_ValueAdded_50/Agricultural_Employment_100 100 80 60 40 1990 1995 2000 2005 Year NonCereal Agricultural_ValueAdded_50 Agricultural_Employment_100 Note: The number after "Agricultural_Value Added_" or "Agricultural Employment_" is used to rescale "Agricultural value added per hectare of arable land" or "Agricultural Employment per hectare of arable land" in the range of 0­100. For example, "Agricultural_ValueAdded_50" is "Agricultural value added per hectare of arable land" divided by 50, and "Agricultural Employment_100" is "Agricultural employment per hectare of arable land" multiplied by 100. AGRICULTURAL GROWTH, EMPLOYMENT, AND WAGE RATES | 125 ANNEX B. Descriptive Statistics of Variables Number of Standard Variable observations Mean deviation Minimum Maximum log_Wage 505 8.6368 2.4127 4.1320 16.7901 log_RWage 223 4.3252 2.7235 0.5158 10.8494 log_Ya 3,904 6.5770 1.2028 3.9506 12.0010 log_L 1,033 0.7443 1.3986 6.6609 2.1933 log_LN 1,127 14.6967 1.7198 9.5460 19.3180 log_Am 6,043 1.1295 2.0114 8.3798 3.6589 log_Irr 5,045 1.7973 1.6613 3.6243 4.6052 log_Fert 5,179 5.4019 2.0742 2.1671 11.0818 log_RandD 1,845 0.1729 1.2004 4.6052 2.7014 log_GiniLand 3,015 0.6968 0.3658 2.5257 0.2614 OPENNESS 3,915 4.2346 0.4716 2.8375 5.3354 log_Tradeshare 4,914 4.1160 0.6463 0.4257 5.8009 FRANKEL 3,690 15.8535 12.2046 2.3000 68.8300 Sachs & Warner 3,915 0.4138 0.4926 0.0000 1.0000 log_FoodPrice 2,731 3.3237 2.6900 19.9189 8.4860 log_NonCereal 4,903 3.9815 0.7025 3.6836 4.6052 logExport_Fruit-Vegetables 4,641 9.063572 2.941149 0 15.91371 log Export_Dairy Products 3,266 6.765084 3.109535 0 15.39571 log Export Meat 3,525 7.548246 3.29528 0 15.56841 log Export Coffee Tea Cocoa 4,412 8.806393 2.959631 0 15.11659 log Export Sugar 3,703 8.228732 3.028177 0 14.69258 log World Income Shocks 8,280 16.27133 0.1146782 16.00801 16.47742 Real Exchange Rate Shocks 8,280 2,856.184 8,025.843 54.8803 55,167.96 D log Urban Population Share 7,912 0.0087995 0.0023768 0.0061648 0.0203843 126 | RAGHAV GAIHA AND KATSUSHI IMAI ANNEX C. Graphs of Residuals of Equations 1, 2, and 3 and Share of Agricultural GDP in Total GDP (and Openness Indicator for Equation 3) FIGURE C.1. Equation 1 (Based on IV Random-Effects Model): Residuals and Share of Agricultural GDP in Total GDP Residuals-equation (1) 0.4 0.2 0 0.2 0.4 0 20 40 60 80 100 Share of agricultural GDP in total GDP FIGURE C.2. Equation 2 (Based on Random-Effects Model): Residuals and Share of Agricultural GDP in Total GDP Residuals-equation (2) 0.5 0 0.5 1.0 1.5 0 20 40 60 80 100 Share of agricultural GDP in total GDP AGRICULTURAL GROWTH, EMPLOYMENT, AND WAGE RATES | 127 FIGURE C.3. Equation 3 (Based on Random-Effects Model: Real Wages): Residuals and Share of Agricultural GDP in Total GDP Residuals-equation (3) 6 4 2 0 2 4 0 20 40 60 80 100 Share of agricultural GDP in total GDP FIGURE C.4. Equation 3 (Based on Random-Effects Model: Real Wages): Residuals and Openness Indicators (IV Estimates) Residuals-equation (3) 6 4 2 0 2 4 3.0 3.5 4.0 4.5 5.0 5.5 Openness indicator 128 | RAGHAV GAIHA AND KATSUSHI IMAI Endnotes 1. A notable exception is Thirtle, Lin, and Piesse (2003). 2. The linkages between farm and nonfarm activities have been emphasized in the develop- ment literature. There are production linkages, both backward and forward. Backward linkages relate to the demand of farmers for inputs such as plows, engines, and tools, while forward linkages relate to the need to process agricultural commodities, such as spinning, canning, and milling. Moreover, there are consumption linkages. As agricultural income rises, it feeds into higher demand for nonfarm goods produced locally or in neighboring villages or towns. Finally, there are linkages through the supply of labor and capital. As agricultural productivity rises, either labor is released or wages go up. Also, agricultural surpluses could finance expansion of the nonfarm sector. And the latter, in turn, could stimulate agricultural production via lower input costs, technological change, and plow- ing of profits back into farming. 3. World Bank (2005); FAOSTAT is available at http://faostat.fao.org/site/291/default.aspx; LABORSTA is available at http://laborsta.ilo.org/. 4. Measurement of agricultural diversification in terms of share of area devoted to noncereal crops is limiting in two respects: (a) one is neglect of livestock, and (b) the other is the ambiguity of enlargement of noncereal crop share. The noncereal crop share could be larger either because of greater diversity of noncereal crops or because of greater special- ization in fruits or vegetables. For an elaboration, see Joshi and others (2004). 5. As (real) wage data were not available for South Asia, no comment is made. Joshi and others (2004), however, report a slow rise in the Simpson index of crop diversity for South Asia, from 0.59 in the triennium ending 1981-82 to 0.64 in the triennium ending 1999­2000. Bangladesh, Bhutan, and Nepal show less diversity compared to other coun- tries in this region. 6. For a characterization of the median voter's preferences, see Alesina and Rodrik (1994). 7. These include the Sachs-Warner and Frankel-Romer indexes, among others. For a review and details of how we constructed the index of openness used here, see Gaiha and Imai (2008). Briefly, all these measures are confined to trade liberalization. 8. The physical isolation index measures the proportion of a country's population that lives less than 100 kilometers from the coast (McArthur and Sachs 2002). 9. The Frankel-Romer index is the aggregate fitted value of trade share, derived from a bilat- eral trade equation with geographic variables (for example, area, population; see Frankel and Romer 1999). In contrast, the Sachs-Warner measure is binary, based on a series of trade-related indicators: tariffs, quotas, black market premium, social organization, and the existence of export marketing boards (Sachs and Warner 1995). 10. As STATA does not have a ready-made command for a two-error components model to control for both time and individual effects, we use one-error component models (see, for example, Baltagi 2005), where year dummies are included to control for time effects. 11. See annex B for descriptive statistics of the variables. 12. Nominal wages are first converted into real wages in domestic currency and then into dol- lars using the real exchange rates taken from WDI. 13. For example, diversification toward high-value agricultural products is an important fea- ture in recent years in India (Rao, Birthal, and Joshi 2006). Log Noncerealit is not used as an explanatory variable in equations 1 or 3, as it did not have a significant effect. In prin- ciple, the crop diversification index should be treated as an endogenous variable, depend- ing on agroecological conditions, access to markets, and changing dietary patterns, espe- cially in urban areas. For some illustrative evidence, see Joshi and others (2004); Rao, Birthal, and Joshi (2006). AGRICULTURAL GROWTH, EMPLOYMENT, AND WAGE RATES | 129 14. World trade in agriculture has changed since the 1980s, as has the rate of growth of trade overall. Expanding groups include fruits and vegetables (about 19 percent of world exports); fish and seafood (about 12 percent); and alcoholic and nonalcoholic beverages (about 9 percent). While these commodity groups tend to have high income elasticities, they also have low protection in industrial and large developing countries (Aksoy 2005). 15. The IV measure of openness was constructed by Gaiha and Imai (2008). 16. See, for example, the discussion in Greene (2002). 17. A cautious interpretation is, however, necessary, given the relatively high correlations of log Am, log Irr, log Fert, and log R&D. Details of correlations will be furnished on request. 18. In some recent contributions, Rodrik and his collaborators (for example, Rigobon and Rodrik 2004; Rodrik 1999) report a negative contribution of trade liberalization to income growth. In a conversation, Rodrik was emphatic that trade liberalization often involves exploitation of scarce natural resources that could slow growth over a short period. 19. In an earlier study (Gaiha and Imai 2008), openness did not contribute to higher income, controlling for the effects of institutional quality and lagged agricultural output. So the fact that openness has a positive effect on agricultural productivity is not necessarily inconsistent with the earlier result. 20. The Hausman specification test compares the fixed versus random effects under the null hypothesis that the individual effects are not correlated with the other regressors in the model (Hausman 1978). If the effects are correlated or the null is rejected, the random- effects specification produces biased estimators and the fixed-effects version is preferred. The test statistic is as follows: W 2 ^ [K] [bfixed brandom] 1[bfixed brandom], ^ where 1 Var[bfixed brandom] Var(bfixed) Var(brandom) and b fixed or b random is a vector of coefficients for fixed- or random-effects specifica- tion. Under the null, W is asymptotically distributed as chi-square with K degree of free- dom (Greene 2002). We present the results of both fixed and random effects, as (a) our test statistics are based on all the coefficients of fixed effects and a subset of the coeffi- cients of random effects, as time-invariant variables are dropped from the former in the first-differencing; and (b) testing for orthogonality of the individual effects and the regres- sors is based on the comparison of only two specific versions and thus is relative. 21. We also experimented with the proportion of small farmers among all farmers, but doing so did not yield a significant positive coefficient or a significant effect of wage rate. 22. Annex C contains graphs of residuals of equations 1, 2, and 3, plotted against share of agriculture in GDP, and residuals of equation 3, plotted against (instrumented) openness indicator. Although there are more than a few outliers, a general inference is that residu- als are independent of the share of agriculture in GDP and openness. We do not make much of the absence of a significant relationship between predicted wage rates and agri- cultural employment in table 6, given the small sample. 23. Attention may be drawn to high labor intensity of fruits and vegetables. Joshi and others (2004), for example, point out, on the basis of rough and ready comparisons, that a 1-hectare shift in area from wheat to potato would generate 145 additional person-days of employment; similarly, a shift from coarse cereals (sorghum and pearl millet) to onion would generate 70 person-days of additional employment. While these comparisons are illustrative, two caveats must be borne in mind: these differences (a) are not adjusted for changes in technology and agroecological conditions and (b) are commodity specific. Our finding of a negative effect of crop diversification on employment, controlling for other effects on labor intensity, thus is not necessarily contradicted by the notional differ- ences in employment per hectare for specific cereal and noncereal crops. 24. Some of these concerns are voiced by Joshi and others (2004). 130 | RAGHAV GAIHA AND KATSUSHI IMAI References Aksoy, M. Ataman. 2005. "The Evolution of Agricultural Trade Flows." In Global Agricul- tural Trade and Developing Countries, ed. M. Ataman Aksoy and John C. Beghin. New Delhi: Manas Publications. Alesina, Alberto, and Dani Rodrik. 1994. "Distributive Politics and Economic Growth." Quarterly Journal of Economics 109 (2): 465­90. Arellano, Manuel, and Stephen Bond. 1991. "Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations." Review of Eco- nomic Studies 58 (2): 277­97. Baltagi, Badi H. 2005. Econometric Analysis of Panel Data, 3d ed. New York: John Wiley and Sons. Frankel, Jeffrey A., and David Romer. 1999. "Does Trade Cause Growth?" American Eco- nomic Review 89 (3): 379­99. Gaiha, Raghav, and Katsushi Imai. 2008. "Do Institutions Matter in Poverty Reduction? Prospects of Achieving the MDG of Poverty Reduction in Asia." Statistica & Applicazioni 4 (2): 129­60. Gaiha, Raghav, and Ganesh Thapa. 2006. "Supermarkets, Smallholders, and Livelihood Prospects in Selected Asian Countries." Unpublished mss. IFAD, Rome. Greene, William. 2002. Econometric Analysis, 5th ed. Upper Saddle River, NJ: Prentice-Hall. Hausman, Jerry A. 1978. "Specification Tests in Econometrics." Econometrica 46 (6): 1251­71. Joshi, Pramod K., Ashok Gulati, Pratap S. Birthal, and Laxmi Tewari. 2004. "Agriculture Diversification in South Asia: Patterns, Determinants, and Policy Implications." Economic and Political Weekly, June 12, p. 2467. Lipton, Michael. 1977. Why Poor People Stay Poor? Urban Bias in World Development. London: Temple Smith. McArthur, John, and Jeffrey Sachs. 2002. "A Millennium Development Strategy for Achieving Poverty Alleviation and Economic Growth." Unpublished mss. Mellor, John W. 1976. The New Economics of Growth: A Strategy for India and the Devel- oping World. Ithaca, NY: Cornell University Press. Mellor, John W., and Uma Lele. 1972. "Growth Linkages of the New Food Grain Technolo- gies." Indian Journal of Agricultural Economics 19 (1): 35­55. Rao, P. P., Pratap S. Birthal, and Pramod K. Joshi. 2006. "Diversification towards High-Value Agriculture: Role of Urbanisation and Infrastructure." Economic and Political Weekly, June 30, pp. 2457­67. Rigobon, Roberto, and Dani Rodrik. 2004. "Rule of Law, Democracy, Openness, and Income: Estimating the Interrelationships." Unpublished mss. National Bureau of Economic Research, Cambridge, MA. Rodrik, Dani. 1999. The New Global Economy and Developing Countries: Making Openness Work. Washington, DC: Overseas Development Council. Sachs, Jeffrey, and Andrew Warner. 1995. "Economic Reform and the Process of Global Inte- gration." Brooking Papers on Economic Activity 1: 1­118. Thirtle, Colin, Lin Lin, and Jenifer Piesse. 2003. "The Impact of Research-Led Agricultural Productivity Growth on Poverty Reduction in Africa, Asia, and Latin America." World Development 31 (12): 1959­75. World Bank. 2005. World Development Indicators. Washington, DC: Oxford University Press. Are Farm-Nonfarm Linkages Still Strong Enough to Tackle Rural Poverty? DANIEL BRADLEY Where agricultural growth has stagnated, as in large parts of Africa today, the broader transformation of economies has stalled and poor countries have remained trapped in a cycle of slow growth, low labor productivity, and poverty. This underlines the importance of improving agriculture's performance in many of the poorest countries. History shows very few examples of countries that have suc- cessfully reduced poverty through agriculture alone, but almost none have achieved it without first increasing agricultural productivity. Agriculture's importance goes far beyond rural areas. In the past, increasing agri- cultural productivity spurred economic development and poverty reduction outside agriculture, in both urban and rural areas (DFID 2005). This is due to particularly strong linkages that, in sum, generate a strong multiplier effect on the rest of the economy: · Production linkages both forward to, for example, agricultural processing and backward to input suppliers, · Consumption linkages, including spending by farm families on local goods and services,1 · Factor market linkages through labor demands, finance flows, and rising rural wages, and · Productivity linkages at the macro level, via lower food prices that benefit both the rural and the urban poor, who typically spend more than half of their income on food, plus benefits derived from improved food security, political stability, and enhanced knowledge flows. Estimates of agricultural multipliers for Asia run in the range of 1.6 to 1.9, with 1.3 to 1.5 being typical for Latin America and Africa (see table 1).2 This is significantly Daniel Bradley is with the Department for International Development, United Kingdom. Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 131 132 | DANIEL BRADLEY TABLE 1. Evidence of Agricultural Growth Multipliers (US$) Additional farm Additional nonfarm Region or country income created income created Asia 1 0.6­0.9 Africa 1 0.3­0.5 Latin America 1 0.3­0.5 Burkina Faso 1 1.88 Zambia 1 1.50 Niger 1 0.96 Source: Adapted from DFID (2005); Hazell (2005). higher than equivalent multipliers for other sectors, due in part to their comparatively smaller economic base (Block and Timmer 1994; Hazell and Haggblade 1991). Evi- dence for the comparative advantage of agriculture comes from various countries and regions: · Kenya, where models of the economy show that multipliers from agricultural growth are three times greater than those for nonagricultural growth (Block and Timmer 1994), · Republic of Korea, where studies show that agriculture-led approaches outperform export-led industrialization (Adelman 1984), and · Ethiopia, where analysis of the economy shows agriculture creating 0.54 of non- farm growth for every $1 of agricultural income growth compared with 0.22­0.3 for industrial-led growth (Block 1999). While some posited agricultural multipliers appear high (for example, 2.50 for Zambia; 2.88 for Burkina Faso), in other cases the links between farm and nonfarm growth may be underestimated. This is because linkages with nonfarm but agricul- turally dependent activities, including agro-industry, services, and trade, are fre- quently omitted from analyses of agriculture's contribution to economic development (OECD 2006). For example, research in eight Latin American countries shows that, according to official statistics (based on traditional measurements of harvest and sale of raw materials), agriculture constituted only 7 percent of GDP in 1997, whereas "extended agriculture" (incorporating farm and nonfarm agricultural enterprises) contributed about 30 percent (IICA 2004). Can Agriculture Generate Large Multiplier Effects Today? The big question is whether these farm-nonfarm linkages still hold true today. Are recent forces of change like globalization sufficient to shift the prevailing agricultural and rural development paradigm? ARE FARM-NONFARM LINKAGES STILL STRONG ENOUGH? | 133 This question is particularly pertinent given the recent return of agriculture to the development agenda after a long absence, including recent statements by the G-8, the Maputo "10 percent" declaration by African leaders, the emergence of the Compre- hensive African Agricultural Development Programme of the African Union and the New Partnership for Africa's Development, and the forthcoming World Development Report (WDR). Many governments indicate that national growth processes that exclude agriculture are "likely to be of limited scope and duration, pulled back by the increased relative and absolute poverty of the majority rural population" (Brooks 2006). There is some concern that this new focus on agriculture will be a failed under- taking unless it is based on an updated understanding of the changing context in which agriculture takes place. In particular, skeptics question the relevance today of a number of elements of the farm-nonfarm linkages model (for example, Ellis 2005). In this view, · past "evidence" of favorable multipliers and linkages between agriculture and other economic activities is not scientifically rigorous. · globalization and transforming supply chains are weakening traditional rural linkages and income multipliers and creating an environment in which smallholders cannot compete. · terms of trade decline faster than productivity can rise in Africa. · limited domestic markets and unstable prices in free markets constrain the poverty impact of any improvements in productivity. · degradation of natural resources undermines the potential for productivity increases. · declining farm size under inheritance norms limits agricultural prospects; for example, the Department for International Development (DFID)-funded Liveli- hoods and Diversification Directions Explored by Research (LADDER) project showed a decline in average farm size from 20 hectares to just over 1 hectare in Suba District, Kenya, over the last 40 years (Ellis 2005). Pro-agriculturalists are concerned that this view could undermine newly reemerg- ing policy commitments to the sector. But most agri-skeptics qualify their argument by acknowledging that agriculture is still an important, but not the only, tool in the development box (Ellis 2000). A strong role for agriculture within a multisector approach appears to be the best way forward. Agriculture is likely to remain the most significant engine of growth in many of the poorest countries, particularly in Africa, where multipliers are the most favorable. But in transforming economies, notably in Asia, agriculture will retain a comparative advantage for the reduction of poverty. Developments in farm and nonfarm activity must move together (Haggblade 2006b). The value of the skeptical view is to challenge policy makers to think afresh about appropriate, effective, and evidence-based policies, investments, and institutions 134 | DANIEL BRADLEY for unlocking agriculture's potential as part of a broader strategy for achieving growth and reducing poverty. These broader strategies for successful rural growth and transformation need to strengthen farm-nonfarm linkages, not least because a more productive agriculture sheds labor, and this workforce must be absorbed by other sectors. In countries like China, small town and village enterprises are important for absorbing labor, but there are serious concerns as to how China will cope in the future as the transition beyond agriculture continues (Hazell 2006). In many countries, migration will play an important role, which requires policy measures to facilitate rather than exclude movement (OECD 2006). This includes protecting the rights of migrants as they move and enabling remittances through improved rural financial services. Implications for WDR 2008 DFID supports the view that, while multipliers may be weaker in some areas than in the past, they are still significant. Agriculture remains important, particularly · in the poorest countries where agriculture is often stagnant and few other viable sources of growth exist and · as a means to improve the distributional pattern of growth and tackle persistent rural poverty and inequality in both transforming economies and those approach- ing middle-income status. The WDR will need to provide a thorough analysis of the farm-nonfarm linkages model in order to provide convincing and practical guidance on policies, institutions, investments, and governance. Five areas for attention are suggested below. Evaluate the Rigor of Existing Evidence about Agriculture's Multiplier Effect The first is the need to evaluate the rigor of existing evidence about the nature and strength of agriculture's multiplier effect. Particularly, · whether and the extent to which conditions that enabled strong agricultural growth multipliers in the past are present or can be replicated in today's priority areas, · whether benefits from consumption multipliers today will be captured by small- scale, poor farmers or larger enterprises, · whether and where credible alternatives to agriculture really exist, and · whether it is possible to reconcile the view that Africa can emulate Asia's green revolution with the opposing view that today's more challenging context renders obsolete all past evidence on agriculture's role as a stimulus to nonfarm growth. ARE FARM-NONFARM LINKAGES STILL STRONG ENOUGH? | 135 Highlight Innovative Ways to Strengthen Agricultural and Nonfarm Linkages It is important to highlight innovative ways of "making markets work" to strengthen agricultural and nonfarm linkages. Particularly, · make innovative ways to remove constraints on the private sector and improve market access for poor people (for example, Almond and Hainsworth 2005) a central focus of WDR 2008. Some priorities are familiar, like infrastructure (water, rural roads, transportation networks, telecommunications) for connecting people to markets and enhancing productivity, but improved business and investment climates and an improved regulatory environment (public and private standards) are also important. · recognize the importance of domestic markets, including the general agreement that Africa is likely to evolve as one of the largest markets for its own products in the next 10­20 years (Diao and Hazell 2004). Consider Agriculture's Linkages to the Broader Economy It is important to consider agriculture's linkages to the broader economy, not just the rural economy. Livelihoods and economies in poor countries transcend rural-urban distinctions, with increasing interactions between rural, periurban, and urban economies. Diversification and mobility are increasingly important components of resilient and adaptive livelihoods (Bradley and Grainger 2004). The WDR will need to take account of increasingly fluid urban-rural realities rather than adopt a narrow rural nonfarm economy approach. Give Guidelines for Prioritization Guidelines are needed for prioritization, that is, the WDR needs to focus on areas where analysis of economic opportunities for poor people can demonstrate that increasing agricultural productivity is likely to remain the most promising route to poverty reduction and identify where strong existing or potential synergies and mar- ket linkages between farm and nonfarm growth exist. Clarify the Institutional Framework Needed for Rural Development The institutional framework needed for rural development should be clarified, based on integrated and mutually reinforcing agricultural and nonfarm activities. In partic- ular, it is important to weigh, in specific countries, the role of sectoral relative to ter- ritorial or spatial approaches that recognize market linkages ("marketsheds") as well as agroecological potential (Haggblade 2006a). One final observation is that the transformation of agriculture and transition into alternative activities is always a difficult process. The WDR should acknowledge that there will be winners and losers and outline policy measures to manage the difficulties 136 | DANIEL BRADLEY of transformation. Investments in human capital (health and education) help people to prepare for change. Effective social protection can be used to smooth the transition to new activities, insure against the effect of shocks, and act as a safety net for people who find themselves excluded from economic opportunity. Endnotes 1. Consumption linkages are thought to account for some 80 percent of agricultural demand linkages in Africa and Asia (Haggblade, Hazell, and Reardon 2005). Estate-led agricultural growth, notably in Latin America, generates a different pattern of consumption and weaker linkages. 2. Strength of multipliers is influenced by the distribution of initial income gains, density of rural infrastructure and population, available technologies, social and political structure, and physical distribution of urban settlements (Haggblade, Hazell, and Reardon 2005). References Adelman, Irma. 1984. "Beyond Export-Led Growth." World Development 12 (9): 937­49. Almond, Frank R., and Sue D. Hainsworth, eds. 2005. Beyond Agriculture: Making Markets Work for the Poor. Proceedings of an international seminar, Crop Post-Harvest Pro- gramme (CPHP), Natural Resources International Limited, and Practical Action, London, February 28­March 1, 2005. Block, Steven. 1999. "Agriculture and Economic Growth in Ethiopia: Growth Multipliers from a Four-Sector Simulation Model." Agricultural Economics 20 (3): 241­52. Block, Steven, and C. Peter Timmer. 1994. "Agriculture and Economic Growth: Conceptual Issues and the Kenyan Experience." Harvard Institute for International Development, Cambridge, MA. Bradley, Daniel, and Alan Grainger. 2004. "Social Resilience as a Controlling Influence on Desertification." Land Degradation and Development 15 (5): 1­20. Brooks, Karen. 2006. "Agro-Fundamentalism without Apology: The World Bank's Support for Rural Development in Africa Circa 2006." In Beyond Agriculture? The Promise of the Rural Economy for Growth and Poverty Reduction: Workshop Synthesis. Rome: Food and Agricultural Organization of the United Nations, Agricultural and Development Economics Division. DFID (Department for International Development). 2005. Growth and Poverty Reduction: The Role of Agriculture. London: DFID. Diao, Xinshen, and Peter Hazell. 2004. "Exploring Market Opportunities for African Small- holders." Brief prepared by the International Food Policy Research Institute for the con- ference "Assuring Food and Nutrition Security in Africa by 2020: Prioritizing Actions, Strengthening Actors, and Facilitating Partnerships," Kampala, Uganda, April 1­3. Ellis, Frank. 2000. Rural Livelihoods and Diversity in Developing Countries. Oxford: Oxford University Press. ------. 2005. "Small Farms, Livelihood Diversification, and Rural-Urban Transitions: Strategic Issues in Sub-Saharan Africa." In The Future of Small Farms. Proceedings of a research workshop, Wye, U.K., June 26­29. Washington, DC: International Food Policy Research Institute. ARE FARM-NONFARM LINKAGES STILL STRONG ENOUGH? | 137 Haggblade, Steven. 2006a. Comments in Roundtable Discussion. In Beyond Agriculture? The Promise of the Rural Economy for Growth and Poverty Reduction: Workshop Synthesis. Rome: Food and Agricultural Organization of the United Nations, Agricultural and Devel- opment Economics Division. ------. 2006b. "Rural Nonfarm Dynamics." In Beyond Agriculture? The Promise of the Rural Economy for Growth and Poverty Reduction: Workshop Synthesis. Rome: Food and Agricultural Organization of the United Nations, Agricultural and Development Economics Division. Haggblade, Steven, Peter Hazell, and Thomas Reardon. 2005. "The Rural Nonfarm Economy: Pathway out of Poverty or Pathway in?" In The Future of Small Farms. Proceedings of a Research Workshop, Wye, U.K., June 26­29. Washington, DC: International Food Policy Research Institute. Hazell, Peter. 2005. "Agriculture and the Rural Nonfarm Sector: Rivals or Complements?" Presentation at the Future Agricultures Seminar Series "Achieving Pro-poor Growth through Agriculture: The Challenges," Overseas Development Institute, London, November 4. ------. 2006. "Transformation in Agriculture and the Impacts on Rural Development." In Beyond Agriculture? The Promise of the Rural Economy for Growth and Poverty Reduc- tion: Workshop Synthesis. Rome: Food and Agricultural Organization of the United Nations, Agricultural and Development Economics Division. Hazell, Peter, and Steven Haggblade. 1991. "Rural-Urban Growth Linkages in India." Indian Journal of Agricultural Economics 46 (4): 515­29. OECD (Organisation for Economic Co-operation and Development). 2006. Promoting Pro- Poor Growth: Agriculture. Paris: OECD. Agriculture: Rural Nonfarm Linkages and Poverty ALEXANDER SCHEJTMAN Without belaboring the definitional problems of what is "rural" and what is "nonagriculture" and taking at face value the different estimates of its magnitude and trends, nonfarm household income represents roughly 42 percent of household income in Africa, 32 percent in Asia, and 40 percent in Latin America. Further- more, both the relative weight of rural nonfarm employment and incomes in all the regions have been increasing through time, implying that the equation rural development agricultural development no longer holds. Even assuming that agricultural activities continue to be central in the development of rural space, the strict agricultural bias of policies and programs will have to be revised (Reardon and others 1998). Incentives to Participate in the Rural Nonfarm Economy In very schematic terms, the decisions of rural households to undertake rural non- farm activity are motivated by push or pull factors and, although the distinction between them is not always clear, push factors refer to the expectation of higher returns in the nonfarm sector relative to the farm, and pull factors refer to, among other things, inadequate farm income, attempts to diversify income by managing risk and employment cycles, a need for cash in the absence of credit markets for farm inputs, and so forth. The nature of these push and pull factors tends to differ depend- ing on the stage of structural or rural nonfarm transformation, which is closely related to the presence of a dynamic or a stagnant agriculture in the region (Escobar, Reardon, and Berdegué 2001). Alexander Schejtman is Head Researcher, Latin American Center for Rural Development (RIMISP), Santiago, Chile. Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 139 140 | ALEXANDER SCHEJTMAN Trends in the Development of the Rural Nonfarm Economy Many regions in Africa and South Asia are in a "first-stage rural nonfarm transfor- mation," with a stagnant agriculture and little demand for modern inputs and serv- ices, a rural population that is growing more rapidly than the urban population, and a large share of the population engaged in farming activities. Weak rural-urban links and cottage-type labor-intensive manufactures at a subsistence level of income and productivity are the main characteristics. Latin America's rural areas are in a "second stage," with a relatively dynamic agriculture, a much lower proportion of the labor force in agriculture, and stronger upstream and downstream rural-urban linkages for inputs, services, pro- cessing, and marketing. There is also a tendency for rapid "agro-industrialization," both of labor-intensive, small-scale units in the countryside and of medium-large, capital-intensive national and transnational firms in intermediate cities. Tourism and other service activities that grew out of agriculture have become more auton- omous, and other activities have tended to replace agriculture as regional engines of growth. In some regions of East Asia, more advanced forms of business linkages are pres- ent, and agriculture is no longer a driver of the local economy. Subcontracting arrangements and labor commuting are widespread, and an increasing amount of employment arises out of agriculture in various kinds of manufactures for distant markets. Increases in the opportunity cost of labor redefine the production patterns of the areas involved, with the practical disappearance of low-return rural nonfarm activities. In each region and in any country within each region, the presence of dynamic or stagnant agricultural areas will give rise to processes with strong similarities to the ones attributed to the different phases of structural change (Haggblade, Hazell, and Reardon 2005). Rural Nonfarm Activities and Their Impact on Rural Poverty The impact of rural nonfarm activities on the level and distribution of incomes is mixed and somehow related to the phases of the transformation process and the dynamics of agriculture. Meso and micro paradoxes describe the limits of rural nonfarm activities as a path out of poverty. The meso paradox underscores the fact that, in areas where rural nonfarm employment is most needed as a way out of poverty, the returns are lower than in dynamic agricultural regions. The micro paradox points out that, for households with very low returns from agriculture, rural nonfarm options are of the low-productivity, low-return, or "refuge" type. The levels of assets, education, infrastructure, and distance from urban areas, gender, and ethnicity, among other factors, determine the quality of rural nonfarm activities available to each household (Reardon and others 1998). AGRICULTURE: RURAL NONFARM LINKAGES AND POVERTY | 141 Linkages and Multipliers Three kinds of linkages between agriculture and the nonfarm sector are usually considered: production, consumption, and factor market, whose nature tends to differ according to the phase of structural transformation (Haggblade, Hazell, and Reardon 2005). Upstream or backward production linkages to input suppliers of equipment, fuel, and fertilizer and forward or downstream linkages to processors and markets tend to increase in importance and complexity from the first to the third stage. Of particular relevance in this area is the accelerated transformation of the agrofood system of developing countries, where agricultural outputs are destined primarily for super- markets and other large national and foreign direct investment­based industries. The impact of consumption linkages derived from the expenditures of farm fami- lies in the local economy, according to Engel's law, will lead to an increase in the share of nonfarm products as incomes rise and to changes in the product mix as diets diversify (Bennett's law). The impact of this trend on local income and employment will depend on the degree of openness and competitive capacity of the economy and on the relative weight of the modern retail system. Production and consumption linkages also can run from other sectors to agricul- ture, as when an agro-industrial investor "creates" suppliers for his agricultural inputs or when a tourist initiative or the wages paid by a local mine or manufacturer generate demand for locally produced food. As grades and standards are increasingly being defined by the largest retailers, the influence on consumer demand will be one of the dominant links to agriculture. Labor market linkages derived from the seasonal character of agricultural employ- ment can give rise to rural nonfarm activities in slack periods, even in modern flexible manufacturing activities (maquila, putting out). Rural nonfarm incomes can replace the lack of credit for improved inputs or diversification of agricultural production, or surpluses out of agriculture can be invested in rural nonfarm initiatives. These linkages create the agricultural growth multipliers whose magnitude seems to be related to the stages of structural transformation, tending to be stronger in Asia than in Latin America and in Latin America than in Africa. The consumption linkages tend to dominate, and rural services represent the majority of nonfarm linkages. Implications for Rural Development Once the equation agricultural development rural development is abandoned and rural development is conceived as a pro-poor strategy, a series of changes in the con- ceptual framework of traditional programs and policies has to be made: · From the small farmer to the rural household, · From agricultural production to intersectoral linkages, with agriculture-based value chains, 142 | ALEXANDER SCHEJTMAN · From agricultural employment to the different paths of employment out of poverty, · From the agricultural space to the rural territory (agricultural hinterland and its urban nucleus), and therefore, · From agricultural development to territorial rural development, understood as the simultaneous process of changes in local production patterns with social equity of rural areas. The following criteria are assumed (Schejtman and Berdegué 2004): 1. Changes in production patterns (that is, innovations in products, processes, or management) have to be adopted simultaneously with institutional changes to assure equitable access to benefits. 2. Rural development programs must internalize the rural-urban linkages. 3. Rural territory has to be conceived as a space with an identity derived from a concerted project by most stakeholders. 4. Programs should consider the heterogeneity of territories and design-differentiated policies. 5. Programs should be able to involve all stakeholders because some pathways out of poverty require alliances with actors who are not poor. 6. Programs must include options for all the different paths followed by poor house- holds to escape poverty. 7. Programs must be able to integrate sector policies with locally integrated policies. 8. Programs must consider a complex local institutional architecture with inclusive private-public platforms with capacities to decide on the use of public funds. 9. Programs must be able to subordinate short-term demands by different agents to the long-term nature of rural development initiatives. References Escobar, Germán, Thomas Reardon, and Julio Berdegué. 2001. "Best Practices and Strate- gies for Promoting Nonfarm Employment Creation in Rural Development in Latin America: Synthesis of Case Studies." Report of the RIMISP, Santiago, for the Department for International Development (DFID), London, June. Haggblade, Steven, Peter Hazell, and Thomas Reardon 2005. "The Rural Nonfarm Economy: Pathway out of Poverty or Pathway in." In The Future of Small Farms: Proceedings of a Research Workshop. Washington, DC: International Food Policy Research Institute. Reardon, Thomas, María Elena Cruz, and Julio Berdegué. 1998. "Los pobres en el desarrollo del empleo rural no agrícola en América Latina: Paradojas y desafíos." Paper prepared for the Tercer Simposio Latinoamericano de Investigación y Extensión en Sistemas Agropecuarios, Lima, August 19­21. AGRICULTURE: RURAL NONFARM LINKAGES AND POVERTY | 143 Reardon, Thomas, Kostas Stamoulis, María Elena Cruz, Arsenio Balisacan, and Julio Berdegué. 1998. "Rural Nonfarm Income in Developing Countries: Policy Implications and Importance." In The State of Food and Agriculture 1998. Rome: Food and Agriculture Organization. Schejtman, Alexander, and Julio Berdegué. 2004. Desarrollo territorial rural. Debates y Temas Rurales 1. Santiago: RIMISP. http://www.rimisp.org/FCKeditor/UserFiles/File/documentos/ docs/pdf/dtr1_desarrolloterritorialrural.pdf. Part V: Decentralization, Local Governance, and Rural Development Decentralization, Local Governance, and Rural Development JEAN-LOUIS ARCAND This paper poses three questions about community-based (CBD) or community- driven development (CDD) programs. First, does the participatory nature of many decentralized development programs add anything over and above other, more con- ventional, top-down, alternatives? Second, is elite capture the main risk associated with these programs, or should we be more worried about other issues, such as the geographic appropriation of benefits and the implications for local public finance? Third, are ethnicity issues, particularly in the context of Sub-Saharan Africa, a seri- ous impediment to such programs, or should we be more concerned with the form taken by local political institutions? In what follows, I attempt to ascertain what, if anything, we know concerning the answers to these questions. The analytical framework is an extremely simple model of the gains and costs of decentralization. I also illustrate partial answers to the ques- tions, drawing on my own research on a large-scale CDD program in Senegal. Harnessing the Advantages of Decentralization Recent years have seen the emergence of an interesting economics literature on the issue of decentralization in general and the decentralization of antipoverty programs in particular. Key references include work by Bardhan and Mookherjee (2000, 2005, 2006a, 2006b), Besley and Burgess (2001, 2002), Besley and Coate (2003), Besley and others (2004), and Foster and Rosenzweig (2004). Although the structure of the theoretical models developed in this work varies greatly, the models are all based on some sort of tradeoff between the putative gains, in terms of efficiency, of taking decisions in a manner that is closer to the preferences Jean-Louis Arcand is Professor of Economics with the Centre d'Etudes et de Recherches sur le Développement Interna- tional (CERDI) and the Centre National de la Recherche Scientifique (CNRS) at the Université d'Auvergne and the European Union Development Network (EUDN). Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 147 148 | JEAN-LOUIS ARCAND of the beneficiary populations and the gains to centralized decision making that inter- nalizes externalities more effectively. Differential costs, in terms of corruption and capture, at the central and local level are also essential ingredients. The Intuition of These Models Although in no way encompassing the existing literature, the following, extremely stylized model captures some of its key features and is useful in organizing our thoughts. Consider a country constituted by two regions, A and B, which account for proportions and 1 , respectively, of the population. Households are assumed to be characterized by single-peaked preferences, and a household is parameterized by its preferred policy, denoted by . There is heterogeneity in each region, which I model by assuming that is distributed according to a probability density function fi( ), i A, B in each region. For the sake of simplicity, I assume that fi( ) is given by a normal density: fi( ) N( , 2 i i). Let idenote the policy implemented in region i. The cost to a household of hav- ing a policy that does not correspond to its preferences is assumed to be given by a simple quadratic loss function ( i )2. It follows that the average welfare cost in region i of implementing policy i is given by the following: (1) ( i)2N( , 2 i i)d . For simplicity, I assume that the benefits of the program are equal in a one-to-one fashion to total expenditures, denoted by G. The Centralized Solution Consider first the case of a centralized setup for the program. In this case, an omni- scient social planner sets a single value of ifor both regions. Formally speaking, this is obtained by picking C so as to maximize a national social welfare function, where the planner's optimization program is given by the following: (2) max W max G ( C)2N( A, 2) d {C} {C} A (1 ) ( 2 C)2N( B , B) d . Straightforward calculations yield the centralized solution: (3) * C A (1 ) B . The intuition behind this result should be obvious: the centralized solution yields a policy that is a population-weighted average of the mean preferred policies of each re- gion. As a result, welfare in the regions is given byWA C G 2 A (1 )2( A B)2 and WB C (1 )G 2 2 B ( A B)2, whereas overall welfare is equal to the following: (4) WC G 2 2 A (1 ) B (1 )( A B )2. Apart from the benefits of the program at the national level (G), the second term on the right-hand side of equation 4 represents the cost induced by heterogeneity DECENTRALIZATION, LOCAL GOVERNANCE, AND RURAL DEVELOPMENT | 149 within region A, the third term represents the corresponding cost for region B, while the last term represents the costs stemming from differences between regions in terms of the mean preferred policy. Decentralization In the decentralized case, the social planner picks a value of ifor each region, and his optimization problem is given by max W. { , A B} It is then obvious that the solution in this case is given by * * A A, B B and that the welfare in the regions is given by WA C G 2 C 2 A and WB (1 )G B , whereas overall welfare is equal to: (5) WD G 2 2 A (1 ) B. If the mean preferred policy is also the median, then direct participatory democ- racy in each region will also yield i as the solution by dint of a straightforward application of the median voter theorem. The gains to decentralization are therefore given by comparing equations 4 and 5: (6) W WD WC (1 )( A B)2. Once again, this expression is simple to interpret. If there are no interregional dif- ferences in preferences, at least in terms of the mean preferred policies (which would correspond to A B), then there are no gains to decentralization. Moreover, the gains to decentralization are maximized when the regions are of equal size ( 1 2 ). Does Participation Add Anything? A number of recent assessments of CDD and CBD programs have focused on the World Bank's current practices regarding impact evaluation (Rawlings, Sherburne- Benz, and Van Domelen 2004; Wassenich and Whiteside 2003). Although various programs have indeed been evaluated, using either classic difference-in-differences methods or randomized trials, a major question remains, as stressed in the overview by Mansuri and Rao (2004): Does the participative component per se contribute any- thing more than would alternative formulas? The first question--whether such programs do, in fact, have a positive effect on welfare--involves estimating, within the simple theoretical framework sketched above, the magnitude of G, where G is operationalized through various indicators of welfare such as income, expenditures, access to basic services (water, health, educa- tion), or child anthropometrics. The second question involves disentangling, for a given estimate of the magnitude of G, whether any multiplier effects are associated with G that can be attributed directly to the participatory nature of the program. To the best of my knowledge, no such analysis exists, hence the need for experimental work, using randomization, aimed, for example, at assessing the effect on the impact of the program of (a) the 150 | JEAN-LOUIS ARCAND proportion of self-provided funds required of villagers, (b) the degree of adminis- trative capacity of grassroots organizations, and (c) the degree of potential elite capture. I am in the process of setting up just such a randomized design in the con- text of the Fundo de Apoio Social (FAS III), an important World Bank­sponsored social fund in Angola. Who Gets Projects and How Much Is Spent? An influential critique of CDD and CBD programs associated with the work of Plat- teau and Gaspart (2003) is that decentralization leads to a portion of the benefits of the program being appropriated by local elites or "development entrepreneurs." The intuition behind the argument is that decision makers at the center are at an infor- mational disadvantage concerning local conditions with respect to the local elites, allowing the latter to capture part of the benefits from the program. Denoting the benefits captured by local elites by C, the gains to decentralization are then given by the following: (7) W (1 )( A B)2 C. Although it is difficult to test for the presence of elite capture using standard econometric techniques (hence the need for the experimental, randomized approaches alluded to above), table 1 illustrates the positive impact of a major CDD program (the Programme National d'Infrastructure Rurales, PNIR) on child anthro- pometrics. Of particular interest are the instrumental variables (IV), child fixed-effect estimates of the impact of receiving a completed PNIR infrastructure project on wast- ing (weight-for-age) and stunting (height-for-age), two standard indicators of the short- and long-term nutritional status of children (between 0 and 5 years of age here). See Arcand and Bassolle (2006) for further details concerning the PNIR pro- gram itself as well as the empirical strategy. The coefficient of 0.667 associated with the weight-for-age z score (WAZ) indicates that a child living in a village that has received a PNIR project has a WAZ that is 0.667 standard deviations higher than a child in the control group. Given that children in rural Senegal (our sample is repre- sentative of the poorer regions of the country) have WAZ measures that are on aver- age one standard deviation below the acceptable norm, this CDD program has been able to make up for two-thirds of this deficiency in nutritional outcomes. The estimated impact of the program is even more dramatic for a subsample of children from initially poor households (lower portion of the table), where being in a village that receives a PNIR project improves their WAZ by 2.470 standard deviations. While these empirical results indicate that the PNIR program has been extremely suc- cessful in improving the living standards of the beneficiary populations, especially the poor, they do not tell us whether any of this improvement can be attributed to the participatory or decentralized nature of the program per se. While the elite-capture view of decentralized development is intuitively appealing, and most researchers or development practitioners can tell many "horror stories" concerning capture, my personal view is that, while the problem is important in DECENTRALIZATION, LOCAL GOVERNANCE, AND RURAL DEVELOPMENT | 151 TABLE 1. Impact of the PNIR Program on Child Anthropometrics Weight-for-age Height-for-age Weight-for-height Dependent variable (WAZ) (HAZ) (WHZ) Full sample Least squares estimate 0.187 0.197 0.118 (0.12) (0.13) (0.13) IV estimate 0.667 0.922 0.238 (0.31) (0.39) (0.33) Test of the overidentifying restrictions 1.763 1.825 1.211 p value 0.623 0.609 0.750 Balanced sample Least squares estimate by initial expenditure class Poor (bottom quintile) 2.470 1.897 1.518 (0.93) (1.03) (0.90) Middle class (three middle quintiles) 0.481 0.708 0.247 (0.41) (0.52) (0.42) Rich (top quintile) ­0.424 1.440 ­2.012 (1.28) (1.42) (1.39) Source: Author's compilation. Note: Child fixed effects are included in all specifications. Data cover five periods, 36 communautés rurales, 71 vil- lages, 493 households, 974 children, and 1,960 observations (1,032 treated). Numbers in parentheses are standard errors corrected for within­communauté rurale clustering. Instrumental variables estimates are obtained using the power of village and village chief opinion variables as exclusion restrictions in the structural equations. certain contexts, its importance has been slightly exaggerated. To whit, other aspects of decentralization may be more important regarding the benefits that actually reach the population. The type of phenomenon that I have in mind might be termed "vil- lage capture" and involves the interplay of decentralization and competitive politics at the subregional level. The reason for focusing on this interaction is that many decentralizated antipoverty programs involve either the strengthening or the outright creation of an extra layer of government at the local level, which is a key actor in the allocation of funds between competing localities. In a typical CDD program, such as the PNIR in Senegal, the allocation of funds among different villages is decided at the level of the lowest administrative unit in the country--the communauté rurale (each commu- nauté rurale includes roughly 40 villages). This implies that the functioning of local politics can be a key ingredient in determining who gets a project and who does not. This point is illustrated in table 2, which examines (using a simple linear proba- bility model with village-specific fixed effects) the determinants of which villages get PNIR projects and which villages do not. In particular, a village's political power at the local level, measured by (a) its stock of mandates on the conseil rural (elected by universal suffrage in the communauté rurale) and by (b) the number of individuals it sends to the conseil de concertation et de gestion (CCG, the second institution, 152 | JEAN-LOUIS ARCAND TABLE 2. Determinants of Which Villages Receive a PNIR Project and Which Do Not: Linear Probability Model with Village-Specific Fixed Effects Village has received a Dependent variable completed PNIR project Village's stock of mandates (in months) on conseil rural 0.001 (0.0005) Number of villagers on CCG 0.020 (0.002) Number of village chief's priorities that are PNIR-compatible ­0.080 (0.025) Number of women on CCG 0.126 (0.037) Number of villagers willing to contribute money and manpower 0.091 (0.037) Ethnic fractionalization of CCG ­0.015 (0.004) Majority on CCG is not PDS ­0.318 0.216 R2 0.821 Source: Author's compilation. Note: Data cover five periods, 36 communautés rurales, 71 villages, and 341 observations (92 observations corre- spond to villages that received a PNIR project). Numbers in parentheses are standard errors. alongside the conseil rural, which determines which villages get projects and is co- opted by the conseil rural president), is a statistically significant determinant of the likelihood of a village receiving a PNIR project.1 Table 2 also illustrates that, at least as far as the PNIR program is concerned, the institutions created by CDD can run orthogonal to traditional power structures and perhaps reduce elite capture. This is indicated by the fact that when the village chief identifies types of projects that are, in his opinion, the key priorities of the village (and which are PNIR-compatible), this reduces the likelihood of the village receiving a project. On the gender front, more highly feminine CCGs are associated with a greater likelihood, ceteris paribus, of a village receiving a project. Some indication that the participatory nature of the process does matter is given by the result that when villagers are, on average, willing to contribute money and manpower to the project, this increases the likelihood of their receiving a project, although this does not imply that the participatory nature of the process increases the benefits accruing to the villagers. Can the importance of local politics, which translates geographic loyalties, vitiate the attempt to decentralize? If we consider the same model as above, but let A and B denote two villages within the same region, and let represent the political power of village A, then the problem becomes one of a zero-sum game between the villages, rather than a problem of "leakage" to local elites. If the proportion of the total pie G that goes to village A (B) is affected positively (negatively) by , with village A DECENTRALIZATION, LOCAL GOVERNANCE, AND RURAL DEVELOPMENT | 153 receiving a proportion and village B receiving a proportion 1 , then the simple model implies that the welfare in each village will be given by the following: (8) WC 2 2 A ( )G A , WB C (1 )G B. The important point here is that such a situation may ultimately lead to village B rejecting decentralized solutions if its loss of share of funds because of its weak polit- ical power at the local level is greater than its gains from decentralization. In the con- text of the model, this will be the case when: (9) G 2( A B)2. Another issue that does not seem to have captured the interest of practitioners working on CDD, but which is well known to researchers in applied political econ- omy, is that a proliferation of representatives can, ceteris paribus, increase expendi- tures. The key theoretical result, known since the seminal work of Weingast, Shepsle, and Johnsen (1981) as the "Law of 1 n," is that total expenditures in a given set of constituencies will be an increasing function of the total number of representatives that make up its governing council. Whether the political structures set up by CDD will have the perverse effect of leading to fiscal profligacy has yet to be seen, although my own experience in Senegal suggests that this problem may be partially compen- sated by a reduction in "tax evasion": in the case of the PNIR program, for example, local authorities were extremely surprised to find that a significant portion of the pop- ulation who had never paid the local taxe rurale began to do so once they began to participate in decision making through the political structures created by the PNIR. Ethnicity or Local Competitive Politics? A good deal has been made, especially in the African context, of the deleterious impact of ethnic diversity on the provision of public goods. The approach is spelled out in its most common form in Alesina, Baqir, and Easterly (1999). Table 2 provides an empirical illustration of this in the Senegalese context: the greater the degree of ethnic fractionalization in the CCG in a given communauté rurale (measured by 1 minus the usual Herfindahl index), the lower the probability of all villages in the cor- responding communauté rurale receiving a completed project, ceteris paribus. Ethnic heterogeneity therefore does result in the underprovision of completed projects in Senegal. While ethnic diversity may indeed be a concern in many situations, I believe that, just like elite capture, its importance has been somewhat overblown. In con- trast, very little attention has been paid to the interplay of competitive politics with the local institutions set up alongside CDD. Tables 2 and 3 illustrate the importance of local politics in the Senegalese context, as well as the relatively limited role played by ethnicity (apart from the ethnic frac- tionalization effect presented above). First, as shown in table 2, villages with a CCG in which the majority is not PDS (Senegalese Democratic Party, the ruling party of President Abdoulaye Wade) are less likely, ceteris paribus, to receive a PNIR project. 154 | JEAN-LOUIS ARCAND TABLE 3. Determinants of Who Becomes the Conseil Rural President and Vice President: Linear Probability Model Characteristic President Vice president Personal characteristics Log age 0.064 0.018 (0.025) (0.035) Female 0.043 0.048 (0.020) (0.028) Educational attainment (excluded category: no schooling) Primary education 0.004 0.064 (0.018) (0.024) Secondary education 0.042 0.089 (0.020) (0.027) Higher education 0.138 0.019 (0.032) (0.043) Literate in "national language" 0.035 0.050 (0.021) (0.029) Koranic schooling 0.007 0.019 (0.018) (0.025) Ethnicity Member of ethnic majority 0.005 0.022 (0.015) (0.020) Professional activity Member of largest professional group 0.029 0.027 (0.018) (0.024) Politics and political experience Member of majority political party 0.060 0.044 (0.014) (0.019) Number of terms on conseil rural 0.001 0.018 (0.006) (0.009) Geographic loyalty Number of members from representative's village 0.004 _0.001 (0.001) (0.001) Joint significance of Ethnic group dummies: p value 0.509 0.482 Principal activity dummies: p value 0.005 0.275 Political affiliation dummies: p value 0.926 0.423 0.181 0.244 R2 0.135 0.087 Source: Author's compilation. Note: Conseil rural fixed effects are in all specifications. Data cover 47 conseil ruraux, 617 villages of origin, and 1,209 representatives. Numbers in parentheses are standard errors. This suggests that the additional layer of local political institutions created by CDD may, at least in the Senegalese case, translate national-level political clout directly down to the local level. Second, as shown in table 3, which examines the determinants of who becomes conseil rural president and vice president, being a member of the ethnic majority on DECENTRALIZATION, LOCAL GOVERNANCE, AND RURAL DEVELOPMENT | 155 the conseil rural has no effect on one's probability of becoming president or vice pres- ident, whereas belonging to the majority political party does.2 In the context of the political institutions set up by the PNIR at least, party politics trumps ethnicity. Third, geographic loyalty, already evidenced in table 2 as being a significant deter- minant of who receives PNIR projects and who does not, is at work here as well: the larger the delegation from one's village, the greater the likelihood of being elected conseil rural president. Other significant determinants of the presidency include age, not being female, and a relatively high level of educational attainment. Vice presi- dents, in contrast, are slightly less educated and are characterized by a particularly high degree of political experience regarding the terms served on the conseil rural. The geographic loyalty of members does not appear to play a role in determining the vice presidency, as if there were an explicit attempt to compensate for the president's coming from villages that, on average, have larger delegations. Concluding Remarks In this paper I have considered a number of issues raised by CDD and CBD programs and have attempted to summarize what little evidence there is concerning the three big questions that should at least be raised by the 2008 World Development Report. To the first question--Does decentralization or the participatory nature of CDD add anything?--the answer must surely be that we do not know; much more research, probably using experimental methods, is needed. To the second question--Is elite capture the main problem?--my answer is that, yes, elite capture may be a problem, but the geographic appropriation of benefits may be just as, if not more, important. To the third question--Is ethnicity or are the local political institutions created along- side CDD the key?--my answer is that the importance of ethnicity has probably been overblown and that much more attention needs to be paid to the consequences of competitive politics at the local level. Endnotes 1. A village's stock of mandates on the conseil rural is expressed in months; that is, two rep- resentatives from the village for 12 months each represents a stock of 24 months of politi- cal power. Using alternative measures of political power, such as Banzhaf indexes, yields the same result. 2. See Arcand and others (2006) for further details. References Alesina, Alberto, Reza Baqir, and William Easterly. 1999. "Public Goods and Ethnic Divi- sions." Quarterly Journal of Economics 114 (4): 1243­84. Arcand, Jean-Louis, and Leandré Bassolle. 2006. "Does Community-Driven Development Work? Evidence from Senegal." Unpublished mss., CERDI-CNRS, Université d'Auvergne, March. 156 | JEAN-LOUIS ARCAND Arcand, Jean-Louis, Leandré Bassolle, Grégoire Rota-Graziosi, and Jean-Pierre Tranchant. 2006. "The Making of a President: Political Party, Ethnicity, or Village?" Unpublished mss., CERDI-CNRS, Université d'Auvergne, July. Bardhan, Pranab, and Dilip Mookherjee. 2000. "Capture and Governance at Local and National Levels." American Economic Review, Papers and Proceedings 90 (2): 135­39. ------. 2005. "Decentralizing Antipoverty Program Delivery in Developing Countries." Jour- nal of Public Economics 89 (4): 675­704. ------. 2006a. "Decentralisation and Accountability in Infrastructure Delivery in Developing Countries." Economic Journal 116 (508): 101­27. ------. 2006b. "Pro-Poor Targeting and Accountability of Local Governments in West Ben- gal." Journal of Development Economics 79 (2, April): 303­27. Besley, Timothy, and Robin Burgess. 2001. "Political Agency, Government Responsiveness, and the Role of the Media." European Economic Review 45 (4­6): 629­40. ------. 2002. "The Political Economy of Government Responsiveness: Theory and Evidence from India." Quarterly Journal of Economics 117 (4): 1415­51. Besley, Timothy, and Stephen Coate. 2003. "Centralized versus Decentralized Provision of Local Public Goods: A Political Economy Approach." Journal of Public Economics 87 (12): 2611­37. Besley, Timothy, Rohini Pande, Lupin Rahman, and Vijayendra Rao. 2004. "The Politics of Public Good Provision: Evidence from Indian Local Governments." Journal of the Euro- pean Economic Association 2 (2­3): 416­26. Foster, Andrew D., and Mark R. Rosenzweig. 2004. "Democratization and the Distribution of Local Public Goods in a Poor Rural Economy." Unpublished mss., Brown University and Harvard University, August. Mansuri, Ghazala, and Vijayendra Rao. 2004. "Community-Based and -Driven Develop- ment." World Bank Research Observer 19 (1): 1­40. Platteau, Jean-Philippe, and Fédéric Gaspart. 2003. "The Risk of Resource Misappropriation in Community-Driven Development." World Development 31 (10): 1687­703. Rawlings, Laura, Lynne Darling Sherburne-Benz, and Julie Van Domelen. 2004. Evaluating Social Funds: A Cross-Country Analysis of Community Investments. Regional and Sectoral Studies. Washington, DC: World Bank. Wassenich, Katherine, and Paul Whiteside. 2003. "CDD Impact Assessments Study: Optimizing Evaluation Design under Constraints." Final draft, December 19. World Bank, Washington, DC. Weingast, Barry R., Kenneth A. Shepsle, and Christopher Johnsen. 1981. "The Political Econ- omy of Benefits and Costs: A Neoclassical Approach to Distributive Politics." Journal of Political Economy 89 (4): 642­64. Decentralization, Agriculture, and Poverty Reduction: Harsh Realities? JOHN FARRINGTON Political decentralization1 and administrative devolution (here, also "decentralization" for convenience) have been promoted not only as a means of engaging citizens in democratic processes and in recognizing and claiming their rights, but also as a means of tailoring public investment and service delivery to suit local needs. Progress, espe- cially in relation to service delivery, has been slow. Against a context in which low agricultural productivity and associated poverty are increasingly concentrated in the more remote and more "difficult" areas (difficult on account of climate, topography, hydrology, and soils) and in "fragile states," this paper makes six broad arguments. First, many of the public functions essential to the promotion of pro-poor growth in agriculture are, by nature, central. This applies to major infrastructure investment, import-export agreements, quality standards, labor standards, export credits, and so on. It also applies to environmental standards and regulatory regimes governing, for example, the release of plant varieties and the approval of agrochemicals. Other services (for example, levels of research higher than the adaptive) lend themselves to a division between central and decentralized management, while others (such as extension) can potentially be almost wholly decentralized. This is not to say that these functions cannot be performed at levels other than central government--many can--but costs, whether direct financial costs or, for example, the costs of coordina- tion failure are likely to be higher. Second, there are difficulties that appear to be proving intractable in making the staff of decentralized services responsive and accountable to local political bodies. For as long as these difficulties persist, one of the potential advantages of decentral- ized bodies particularly relevant to agriculture--namely, that of making services more relevant to local needs--will remain unrealized. The reasons for weak account- ability are well known and center on staff perceptions that their career paths are John Farrington is a Research Fellow with the Overseas Development Institute, London. Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 157 158 | JOHN FARRINGTON upward (that is, within the administrative or professional hierarchy) and not in local government. In addition there are (often justified) fears that the budgets for emolu- ments as well as for related expenses and investments are likely to be more volatile at the local than the central government levels.2 Third, many of these fears are rooted in the very small share of local government expenditure that is self-financed and the widespread reliance on central government for funding. Low self-financing is attributable not only to the low revenue base (espe- cially in remote and difficult areas), but also to the unwillingness of local politicians to make themselves unpopular by levying taxes (or ensuring that existing taxes are more than mere tokenism). Fourth, the nature of local government politics is not merely that politicians are unwilling to risk unpopularity by imposing local taxes. It is also that they wish to be associated with activities that are high profile, "modern," and photogenic. Agricul- ture meets these criteria much more weakly than urban-based investments, such as urban street lighting projects.3 Fifth, and as yet relatively neglected in the literature, there is growing evidence that poverty, and possibly more persistent types of poverty, are disproportionately present in fragile states. These contain 15 percent of the world's population, but one- third of the dollar-a-day poor. "Fragile states" have been defined to include weak states, conflict areas, postconflict environments, and states that have strong capacity but are not responsive to the needs of their citizens or to the international commu- nity.4 At a subnational level, those parts of countries having exceptionally adverse biophysical conditions (such as the mountain districts of Nepal) can be included in this definition, and in federal systems, there are several examples of states within the federation (such as in the eastern and northeastern parts of India) and districts within states where natural resource endowments are good, but weak governments mean that public investment and service delivery are so weak as to severely constrain the prospects for growth among small farmers. Sixth, constitutional provisions in favor of democratic decentralization are rarely enough to protect it from pressures by central governments, which often are disin- clined toward decentralization (especially given the fact that most local governments receive most of their funding from the central government). In federal systems, as in India, national constitutional provisions have, on the positive side, permitted decen- tralization at the district and subdistrict levels to be adapted to local circumstances but, on the less positive side (and partly because of the political difficulty of enforcing them), have allowed some states to distort provisions to suit their own party-political ends or to relegate decentralization to a low priority.5 For citizens, administrators, and local politicians themselves to be faced with a local government that they can regard as robust is rare. The overriding sense is one of fragility. The implications for government in general and for agriculture in particular are several: · Strong decentralized government, as found in some Organisation for Economic Co-operation and Development (OECD) countries, relies on highly educated and articulate electorates deeply interested in local improvement. With rare exceptions, DECENTRALIZATION, AGRICULTURE, AND POVERTY REDUCTION | 159 capacities such as these are not found in poor countries, particularly in the more difficult rural areas of poor countries. Even within the OECD, as the erosion of local powers in Britain over the last two decades has shown, there can be no guarantee that the status quo, or what it is being changed into, is either stable or "right." · Strong government appears essential if decentralization is to work well. This may appear paradoxical, but local government will only work well if the processes of representation, accountability, and public administration are designed well, les- sons are learned and incorporated in redesign, and adherence to design is enforced. Likewise, the ability to raise funds locally needs to be guaranteed, as does the provision of central funds. Strong central government can and should play the role of guarantor in all of these processes.6 · There needs to be a clearer recognition of the comparative advantage of central and local government in the various support roles essential to agricultural growth, including public investment, service provision, and facilitation or regulation. The policy challenge identified by Hans Binswanger at the recent International Agri- cultural Economics Association conference--namely, to improve roads, education, and communications in order to offset the inherent problems of spatial dispersion, seasonality, and covariant risk--is, on current performances, something that cen- tral government can do better than local governments. · There needs to be regular review within central ministries responsible for agriculture-related investment and service provision of how agreed divisions of responsibility between central and decentralized bodies are working and whether pressure to change the balance is necessary. At present there appears little evidence that local government is performing agriculture-related services well, and the prospects of it doing so appear distant. · The provision of social protection geared to agriculture, which is increasingly seen as necessary to allow small, low-resource farmers to take entrepreneurial risk, is likely to remain a responsibility of the center. Finally, although few would dispute Binswanger's recommendation of the need for better education, especially in more remote and difficult areas, it is by no means clear that heavy investment in physical infrastructure, whether by central or local governments, represents the most cost-effective solution to poverty in the more remote and difficult areas. First, especially where topography is difficult, this will be expensive; and, second, commodities transported over large distances to major centers of consumption, no matter how good the infrastructure, are unlikely to be price-competitive with more locally produced commodities. In such conditions there is at least an argument for providing no more than basic levels of infrastruc- ture and instead investing public resources in the alternative of helping people to move into more buoyant parts of the economy where jobs are available. Such a policy has (not without controversy) been explicitly practised in southern Chile, parts of which have been declared "nonviable" as far as commercial agricultural 160 | JOHN FARRINGTON production is concerned. Less explicitly, it has also underpinned the migration of some 125 million workers from difficult parts of the western and central provinces of China to the faster-growing coastal provinces. Endnotes 1. Although federal systems (such as in India) may be seen as a form of decentralization, the interest here is in decentralization at the more local levels, such as subprovinces or districts. 2. For a review of the quality of service provision under differing approaches to local govern- ment in India, see Johnson, Deshingkar, and Start (2005) and Johnson (2006). For the views of extension department staff on local government in Latin America, see Christoplos and Farrington (2004). 3. This point was made explicitly by local government staff and others in the preparation of background material for Christoplos and Farrington (2004). 4. For instance, by the U.K. Department for International Development's Governance and Social Development Resource Centre (http://www.gsdrc.org/go/topic-guides/fragile-states/ terms-and-definitions#frag). See also Moreno Torres and Anderson (2004); Farrington and others (2006). 5. Johnson, Deshingkar, and Start (2005) contrast the Indian states of Andhra Pradesh and Madhya Pradesh in this regard. 6. In his keynote presentation to the International Agricultural Economics Association con- ference in Australia, August 14­18, 2006, Hans Binswanger cited a World Bank study on the extent to which decision taking was decentralized, which found that China rated top of the countries surveyed. This provides some evidence in support of the "strong central government" premise. References Christoplos, Ian, and John Farrington, eds. 2004. Agricultural Extension, Poverty, and Vulner- ability. New Delhi and Oxford: Oxford University Press. Farrington, John, Priya Deshingkar, Craig Johnson, and Daniel Start, eds. 2006. Policy Windows and Livelihood Futures: Prospects for Poverty Reduction in Rural India. New Delhi: Oxford University Press. Johnson, Craig, ed. 2006. "Politics, Livelihoods, and Poverty." In Policy Windows and Liveli- hood Futures: Prospects for Poverty Reduction in Rural India, ed. John Farrington, Priya Deshingkar, Craig Johnson, and Daniel Start. New Delhi: Oxford University Press. Johnson, Craig, Priya Deshingkar, and Daniel Start. 2005. "Grounding the State: Devolution and Development in India's Panchayats." Journal of Development Studies 41 (5): 937­70. Moreno Torres, Magüi, and Michael Anderson. 2004. "Fragile States: Defining Difficult Envi- ronments for Poverty Reduction." PRDE Working Paper, 1, DFID, London. (http://www. gsdrc.org/go/display/document/legacyid/1343.) Decentralization and Development: Emerging Issues from Uganda's Experience JOHN A. OKIDI AND MADINA GULOBA Decentralization entails the transfer of planning, decision making, and administrative authority from the central to local government. The term can be used to mean a sys- tem of government in which power is granted to local authorities or a process by which governance is moved from a centralized to decentralized system (Prud'homme 2003). The main forms of decentralization discussed in the literature are deconcen- tration and devolution. Under deconcentration, the central government transfers responsibilities and functions to local levels for implementation by representatives or employees of the center. In contrast, devolution entails the transfer from the center to local governments of the power to plan, budget, mobilize resources, and implement development programs. Uganda has experienced decentralization both as a system and as a process of devo- lution of power from the central to local authorities. Historically, three main phases of decentralization can be distinguished (MoLG 2006a). In the first phase (1955 to 1964) a strong system of local administration was created in the context of a weak center. In the second phase (1964 to 1985), power was recentralized, severely weakening local administration. In the third phase (1987 to date), much power has been devolved to local authorities through programs to strengthen local government systems. Upon ascending to power in 1986 against a background of political turmoil that had undermined the confidence of citizens in government, the Uganda National Resistance Movement initiated the current system of decentralization as part of a broader strategy to restore state credibility and deepen democracy. The decentral- ization system emanated from the creation of local councils as instruments for political and economic empowerment through programs to improve service delivery and accountability. Upon sufficient consolidation of central political power by the mid-1990s, the government scaled up the local council system and transformed it into a mechanism for political, administrative, and fiscal decentralization. John A. Okidi is Executive Director and Madina Guloba is Assistant Research Fellow with the Economic Policy Research Centre, Kampala, Uganda. Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 161 162 | JOHN A. OKIDI AND MADINA GULOBA To date, Uganda's decentralization has been reasonably entrenched as a system and process of local governance. Its contributions to improving service delivery, promoting welfare and growth, and enhancing overall rural development have been nationally and internationally recognized. Nevertheless, numerous challenges have emerged, accompanied by questions regarding the extent to which a small country like Uganda may decentralize. To illustrate the role of decentralization in develop- ment, we highlight some key background information about Uganda's decentraliza- tion, discuss its achievements and challenges, and conclude with a few questions arising from the challenges. Uganda's Decentralization The legislative foundation of the decentralization system was provided by the Local Government Statute of 1993, which devolved some political and administrative pow- ers to local governments through the local council system. Later, most of the elements of the local council system were incorporated and strengthened as provisions for decentralization in the 1995 Constitution. To operationalize the constitutional requirement for decentralized governance, the Local Government Act of 1997 was passed, devolving several political, administrative, and fiscal powers to local govern- ments, which comprise districts (the majority), city councils, and municipalities. The main objective of the decentralization was to promote good governance by strength- ening local institutions and improving service delivery. Administrative Decentralization The transfer of authority for planning, administration, and financial management was central to Uganda's decentralization process. The central government retained responsibility for security, national planning, immigration, foreign affairs, and national projects. All other activities became the responsibility of the local govern- ment councils. Basically, the central government line ministries are responsible for issuing regulations, policies, advice, and benchmark standards and providing super- visory and inspectorate services to local governments. In terms of division of labor, the chairman of the local government council is the political head of the jurisdiction, and the chief administrative officer is the execu- tive head: the accounting officer. All plans, budgets, and action programs are approved by the council with immediate accountability. The eyes and ears of the central government at the local level is the resident district commissioner. The local government council conducts business through standing committees such as the district service commission for personnel matters, the tender board for procure- ment functions, and the local government accounts committee for financial accountability. DECENTRALIZATION AND DEVELOPMENT | 163 Fiscal Decentralization Uganda's decentralization system empowers local governments to access revenues for the financing of devolved responsibilities. The local governments execute their functions using resources transferred from the center, mobilized locally, and received directly from donors. The transfer of resources from the center to local governments takes the form of conditional, unconditional, or equalization grants. Conditional grants (about 80 percent of transfers from the center) are largely from the Poverty Action Fund (PAF), which is to be spent on centrally determined priorities. Chan- neling local development grants through the PAF and protecting the PAF from any emergency budget cuts have enabled resources to reach local levels without many central bureaucratic hurdles. To fulfill their obligation to finance up to 10 percent of their own budget, local governments exercise their powers to raise revenue locally from the cities, munici- palities, town councils, and rural areas. In rural areas, local government revenue is collected by subcounty officials, who retain 65 percent of the revenue and remit the rest to their local government headquarters. Regarding donor funds, local govern- ments are limited from contracting excessive donor support by operating within the sectoral ceilings established in the medium-term expenditure framework. Decentralization Implementation Support Because of the country's history of centralized governance since the mid-1960s, most areas were incapable of operating decentralized governance systems. A District Development Program (DDP) was launched to provide development and capacity- building grants for local governance. The DDP was scaled up in 2000 to become the Local Government Development Program (LGDP). Basically, decentralization has been established and operated via implementation of the LGDP. The program, which was operated as LGDP I from October 2000 to June 2003 and thereafter as LGDP II, was designed to achieve the medium- and long-term goals and objectives of the coun- try's holistic development framework: the Poverty Eradication Action Plan. The LGDP was designed such that the amount of grant resources received depended on the population and land area of the local government jurisdiction, with 85 and 15 percent as the respective weights. Besides, access to the development grant under the LGDP was conditional on the local government meeting several criteria, including the capacity to procure and manage local government programs and the ability to cofinance 10 percent of the LGDP grant value. However, the LGDP imposed so many conditions that, in 1999/2000, for example, less than 10 percent of the intended grant was disbursed (Onyach-Olaa 2003). At that time there were 27 different transfers from the center to local governments, each of which required separate accounts and accountability reports to the center. The tight fiscal relationships between the two levels of government illustrate that, although decentralization is viewed as synonymous with autonomy, it instead embroils local governments in a wider system of intergovernmental relations. In an 164 | JOHN A. OKIDI AND MADINA GULOBA effort to harness such complex relations, the center could instead be seen by local authorities as interfering in the affairs of local governments. In a newly decentralized system, the problem is further complicated by the rigidity of central authorities to adjust to their new roles as mentors and supervisors of local governments (Onyach- Olaa 2003). Continuous monitoring and evaluation of decentralization prompted a series of refinements of the LGDP, leading to substantial improvements in local government planning, allocation, investment, and management of development resources, partic- ularly as a result of increased transparency in decision making (Onyach-Olaa 2003). To Decentralize or Not to Decentralize As reviewed by Martinez-Vazquez and McNab (2003), the literature recognizes that, although decentralization is like a superior good, the demand for which is higher with higher levels of income per capita, there exist some optimal levels of decentral- ization that are consistent with certain levels of development. Wondering why public expenditure at the subnational level should enhance growth more than the same expenditure at the national level, Martinez-Vazquez and McNab (2003) argue that a critical issue is whether governments at whatever level are welfare maximizers or self- seeking entities. Nevertheless, it is generally accepted in the theory and practice of decentralization that, under certain conditions and for certain services, greater allocative and pro- ducer efficiencies are realized through decentralized expenditure (Martinez-Vazquez and McNab 2003). This is because local governments can better discern local pref- erences and needs and can produce the same services at lower costs, leading to greater economic efficiency and growth. This is the so-called Oates's decentralization theorem (Iimi 2005). But must decentralization be the solution to central government failures? If in a centralized system self-interest drives public officials to behave to the detriment of social development, then the solution to bad governance and poor service delivery is not the proliferation of government to the lower levels. In such circumstances, decen- tralization may simply manifest the ripple effects of the central government system. In other words, decentralization could merely transfer and replicate several-fold the point at which the inhibiting factors operate. However, if a country must decentralize, then it should proceed gradually, while focusing on other approaches to tackling the high-priority binding constraints on administrative and fiscal performance. In the context of rampant poverty, decentralization could be isolationist, diver- sionary, and counterproductive. For example, economic opportunities outside the local jurisdiction may be much more important than local development programs, such that the regional rather than the local linkage is more important for overall development (Romeo 2003). In a decentralized system, the urge to increase local revenue by collecting levies on the movement of goods across jurisdictions yields counterproductive outcomes for integrated development. DECENTRALIZATION AND DEVELOPMENT | 165 Whereas certain fiscal decisions in response to local development demands and priorities may maximize the welfare of local residents, it is not obvious that they can maximize economic growth for sustainable rural development and linkages to global opportunities. In other words, in this era of globalization, the complex forces that govern effective participation in global development processes may be beyond the cognitive reach of local residents and their leaders. In such circumstances, caution must be exercised when deciding to decentralize or when flexing decentralized political powers to make economic choices. Achievements of Uganda's Decentralization As has been the case with economic reforms, Uganda is considered a forerunner in Africa with respect to decentralization. After successful stabilization of the macro- economic environment by the early 1990s, the country embarked on a poverty eradication program that was designed to be delivered through a decentralized system of governance. Although the relationship between decentralization and economic growth remains mixed and controversial (Iimi 2005; Martinez-Vazquez and McNab 2003), decentralization has positive impacts on the efficient allocation of resources, democratic governance, and enhanced accountability (Martinez- Vazquez and McNab 2003). To illustrate these relationships, we summarize some of the key achievements of Uganda's decentralization, but with greater illumination of the employment effects, particularly in the sectors where outstanding improvements in service delivery have occurred. Institutional Strengthening Institutionally, Uganda has made significant progress in the implementation of decen- tralization (MoLG 2006a). First, the legal framework is well defined in the constitu- tion and detailed in the Local Government Act of 1997. Second, the structure of political decentralization is well entrenched and fully operational, as evidenced in the several local elections that have been successfully conducted. Third, a functional administrative system at the local level has been established, with 75 percent of the public service workforce composed of local government employees. Fourth, substan- tial progress has been made in implementing fiscal decentralization, with 38 percent of the national budget being spent through the local government system. Empowerment and Participatory Development Decentralization has empowered the citizens, heightened their awareness of the dif- ferent custodians of responsibilities, delivered coordinated services closer to the peo- ple, promoted creative local resource mobilization, and improved the responsiveness of public investment to local popular demands (Emorut 2006). The administrative hierarchy in the decentralized system of governance has promoted the development 166 | JOHN A. OKIDI AND MADINA GULOBA of channels of communication between the population, on the one hand, and local and central government leaders, on the other. The local council system has enhanced local-level political participation and accountability. In particular, the devolved political powers have enabled citizens to elect local leaders and to exert increasing pressure on them for performance and accountability. This is attributed largely to local residents' participation in planning and other decision-making processes. In several cases community project committees have played valuable supervisory roles. The system of contracting private firms to implement local government development programs has enhanced the growth of the local-area private sector. Improvements in Service Delivery A local government development grant focused on primary health care, primary edu- cation, water and sanitation, feeder and access roads, agricultural extension, street lighting, and market infrastructure was aimed at promoting service delivery both as a means and as an end to decentralization. As a result, service delivery greatly improved under the decentralization system, particularly with respect to access to primary education, health care, and water and sanitation services. In the LGDP resource allocation for 2003/04, the bulk of the development grant financed roads and drainage (37 percent), education (24 percent), health (15 percent), and water and sanitation (12 percent). The resource allocations correspond to the findings of an assessment survey that the Ministry of Local Government conducted in 2005/06, which show that 66 percent of the sampled households were within 2 kilometers of a health facility, 54 percent were within 2 kilometers of a primary school, and 51 percent were within half a kilometer of a water source (MoLG 2006b). Under the universal primary education program, school enrollment expanded from 5.6 million in 1998 to 7.6 million in 2003, and the gender differen- tial in schooling was wiped out, although rural-urban enrollment gaps persist. Rural water coverage increased from 55 to 60 percent of the population between 2003 and 2004 alone. Access to health services increased when new health centers were con- structed and old ones were rehabilitated at both the parish and subcounty levels.1 Employment Effects The creation of new districts and the implementation of local government political and administrative structures provided employment opportunities. Numerous jobs were also created in the construction and furnishing of facilities and staff quarters in the health and education sectors, and community road construction and maintenance provided important local area job opportunities. The public-private-community part- nership in the implementation of decentralized governance generated rapid growth in local nongovernmental organizations and private sector companies, which, in turn, created jobs for local residents. Specifically, the creation of new districts expanded the number of political and administrative jobs from 6,036 in 1991 to 12,948 in 2006. In the education sector, DECENTRALIZATION AND DEVELOPMENT | 167 implementation of the Universal Primary Education Program necessitated the con- struction of new primary schools, which generated increased demand for secondary education facilities. As a result, the total number of primary schools doubled to about 14,000 from 1990 to 2004. During the same period, the number of secondary schools increased more than sevenfold to 3,645. The expansion in educational facil- ities culminated in the growth of teaching jobs from about 80,000 in 1990 to about 147,000 in 2004 at the primary school level and from about 11,000 to about 37,000 at the secondary school level. Although the rapid growth in primary and secondary school enrollment poses employment challenges, it is, nevertheless, an opportunity to accumulate human capital for future investment, which would eventually lead to further employment creation. The other sector of improved service delivery under decentralization is health. When user fees were abolished in 2001 in government health facilities (except for private wings in hospitals), the health system was reorganized into a hierarchy similar to the political structure of the country. The system now comprises national and regional referral hospitals and health centers that are categorized as Health Center IV, Health Center III, Health Center II, and Health Center I. The categorization is essen- tially determined by the extent of services that are designed to be delivered at a given level of facility. The operationalization of this structure required the construction of several new facilities, especially at the Health Center II level, which almost tripled in number between 2001 and 2004. Challenges Onyach-Olaa (2003) summarizes some of the challenges experienced by Uganda's decentralization, the most notable being the local versus national development interests, sectorwide approach versus local government­specific approach to national development, and capacity to plan, implement, and account for development programs that are funded from various government and donor grants. Structural Conflicts Donor assistance to Uganda has taken the form of general budget support and project aid. General budget support is often earmarked by donors to finance specific sectoral programs. Where development finance is delivered as project aid, it has been channeled directly to local governments or directly to local communities, bypassing the local government system. This has often fragmented local development struc- tures, exacerbating the problem of coordinating and monitoring local development activities. Further, it has undermined the institutional growth of local government systems and weakened community­local government linkages. Another source of institutional weakening is the sectorwide approach to development, which sometimes runs counter to decentralization when it bypasses local government planning processes and competes for resources with decentralization (Romeo 2003). 168 | JOHN A. OKIDI AND MADINA GULOBA In a system of decentralization where deconcentration and devolution coexist, conflicts emerge. A clash between the resident district commissioner and the district chairman is a case in point. These authorities often belong to opposing political groups, resulting in tension and inefficiency in public sector management and devel- opment. This challenge is presumably more pronounced at this time because Uganda is rather underdeveloped democratically. Another dimension of conflict concerns salary disparities, which are skewed in favor of local government politicians and against civil servants. Capacity Constraints The capacity to implement decentralized systems is severely limited in the majority of local governments. This has perpetuated the problem of coordination between the central authorities and local governments. In addition, capacity differentials, espe- cially in the initial stages of decentralization, have caused and perpetuated inequality in levels of development between districts. The capacity problem is further exacer- bated by the lack of statistical databases for planning, monitoring, and evaluation at the local levels. The magnitude of the capacity problem is muffled by the general tendency to over- rate the ability of local government and community leaders to prioritize, plan, and implement local area development programs. In some cases, poorly implemented decentralization has led to degradation of natural resources (Ouedraogo 2003), as was the case with the overexploitation of forest resources in the early stages of Uganda's decentralization, a reality that prompted a recentralization of the country's forest management. The capacity problem is also manifested in the vivid contrast between the success of the political and the fiscal decentralization. Whereas systems of local elections have worked reasonably well, inadequate capacity and mechanisms to hold leaders accountable have undermined downward accountability, which is a fundamental attribute of successful decentralization. Further, the conditionality of central govern- ment transfers closes off grassroots control, leaving local leaders and residents with limited voice (Francis and James 2003). The outlook is even grimmer when one con- siders the fact that the financing of devolved responsibilities and the operation and maintenance of local government investments are increasingly reliant on central gov- ernment transfers. The reason for this is that locally generated revenue is not only insufficient (less than 10 percent of total revenue) but declining over time, exacer- bated by the political decision in 2005 to suspend the graduated tax, which consti- tuted 60­70 percent of locally generated revenue. Reliance on the center and the small amount of funds under the control of village councils, coupled with weak plan- ning skills at this level, means that lower-level political and administrative units command very limited influence over district and subcounty budgets and programs (Francis and James 2003; Kamanyi 2004). In its early phases, Uganda's decentralization had a negative impact on service delivery. Using data on the flow of grants to primary schools during 1991 to 1995, DECENTRALIZATION AND DEVELOPMENT | 169 Reinikka (2001) finds that decentralization initially entailed significant costs of adjustments in service delivery; schools in decentralized districts received significantly fewer grants than schools in nondecentralized districts during the first half of the 1990s, when decentralization was rolled out by blocks of districts. Subsequent refine- ments to the system, however, contributed to improvements in service delivery. But Francis and James (2003) argue that improvements in the delivery of social services are, in fact, attributable to increases in central government conditional grants rather than to the decentralized system of decision making. Emphasis on the Easier Part Comparing the social sector and the productive sector, we observe a stark contrast. According to the assessment survey that was conducted in 2005/06, respondents expressed satisfaction with improvements in the proximity and quality of services, but limited awareness of, and satisfaction with, LGDP production projects such as crop and livestock multiplication initiatives. Further, only 36 percent of the respon- dents confirmed having participated in LGDP II project planning. From the decen- tralization perspective, the poor performance of the agricultural sector is attributed to the small budgetary allocation for productive sectors, confusion over responsibil- ities for management and operational funding, limited appropriateness of the advisory services offered to farmers, and the transitional status of agricultural policy pending full implementation of the Plan for Modernization of Agriculture (Francis and James 2003). Other economic constraints facing the agricultural sector include lack of skilled labor, poor technology, lack of purchased inputs, low capital, limited access to credit, poor infrastructure, small landholdings, short fallow period, high rates of soil erosion, systems of land tenure that are inconsistent with agricultural modernization, reliance on natural weather conditions, and limited use of fertilizers (World Bank 2005). Most of these adverse conditions are not necessarily correctible by decentralization. Although access to social services has improved significantly in rural areas, improvements in the quality of services have lagged behind the quantitative indica- tors. For example, primary school dropout rates are still very high (especially due to cost factors), infant and maternal mortality rates have not fallen, and public facilities are used predominantly by the poor, as the better-off people opt for superior privately provided services. Regarding infrastructure, only limited progress has been made. For example, despite spending more than 8 percent of the budget on roads and public works since 1999, access to rural transportation and electrification remains very poor in contrast to the significant improvements realized in health, education, water, and communications facilities (World Bank 2005). According to World Bank databases, the development of Uganda's infrastructure is very low compared to the Sub- Saharan African average. Only 9 percent of Ugandan households have access to electricity compared to 27 percent for the average Sub-Saharan African coun- try. The proportion is 11 percent compared to 32 percent for piped water. The 170 | JOHN A. OKIDI AND MADINA GULOBA telecommunications sector, which has grown significantly, also lags, with 33 per- cent of Ugandans subscribing to telephone services compared to the 99 percent average for Sub-Saharan Africa. In contrast, Uganda's road density compares well with the Sub-Saharan African average, especially national and district roads, but not urban and rural roads. Other glaring mismatches in the outcomes of decentralization are observed with respect to the extent to which service delivery has improved vis-à-vis the capabilities of most local residents to capitalize on the services for welfare improvement. Poverty in Uganda remains a predominantly rural phenomenon, particularly pro- nounced among crop farmers. In terms of headcount, rural poverty declined from 60 percent in 1992 to 37 percent in 2000, before rising to 42 percent in 2003. The corresponding figures for urban areas are 28, 10, and 12 percent. The dispropor- tionate contribution of rural areas to national poverty has remained unchanged at about 96 percent.2 Accountability Uganda's decentralization has also been characterized by an imbalance between upward and downward accountability. Emphasis has been given to upward accountability, with the objective of ensuring that resources released from the center are properly accounted for, but not necessarily properly deployed. In this prioritization context, a strict mechanism for upward accountability was estab- lished and has been implemented at the expense of efficiency in resource use. Com- pounded by the fact that local residents are not clear on how to hold leaders accountable,3 local government authorities have incentives to concern themselves with the rate of fiscal absorption rather than with fiscal efficiency and effective- ness. Because of fungibility that allows local governments to spend on alternative services by means of creative accounting practices (Bardhan and Mookherjee 2006), designing central transfers as conditional grants does not guarantee efficient and effective use of revenue at the local level. This could be a consequence of several factors, including the reality that about 90 percent of local government budgets are funded using transfers from the center and the literacy rate of the population is low. Public-Private Partnership For several services, after development plans and designs are approved by the local government council, the local government tender board contracts the private sector to provide the service. Through downward accountability mechanisms, community members are expected to ensure that there is value for money for every publicly funded development program. But the generation of a public-private-community partnership is a complex process that, in some cases, has resulted in parasitic capture of public resources by local private interests and led to direct local government DECENTRALIZATION AND DEVELOPMENT | 171 involvement in private entrepreneurial activities (Romeo 2003). In some cases, the public-private partnership has weakened institutions of accountability when the people in influential public positions are the very ones who participate actively on the private sector side of the partnership. Uganda's local government revenue collection provides interesting lessons for private-public partnership. Other than the graduated tax and licenses, local govern- ment revenue is collected by private sector contractors. But firms often have to pay bribes to be awarded contracts. Besides, some private collectors pass on to govern- ment only one-tenth of what they collect (Francis and James 2003). In an environ- ment of weak institutional capacity for checks and balances, such inherent flaws can promote inequality among local firms and the broader population. Simply put, decentralization can increase corruption unless it limits monopoly power and makes government more accountable to local constituencies (Martinez-Vazquez and McNab 2003). Political Interests In a state of incomplete political development, the structure and process of decen- tralization may be vulnerable to undue political influence for short-term gains. In the early 2000s as the political leadership that came to power in 1986 was preparing the country for multiparty competitive politics, decentralization drifted from being a vehicle for local democracy and efficient service delivery to being an instrument for advancing central-level political motives. As a result, several seemingly unviable districts were created, driven by the desire to create administrative jobs and fulfill political campaign promises. By creating so many political districts, Uganda runs the risk of excessive decentral- ization, which could lower economic growth at the local level. Basically, allocating budgetary resources to less productive levels of government is harmful to economic efficiency and could curb overall growth (Iimi 2005). Even though decentralization eliminates the high transaction and administrative costs implied by centralized systems (Martinez-Vazquez and McNab 2003), the downside of excessive political freedom is that it inhibits internalization of economies of scale and externalities in the provision of public goods (Iimi 2005). Greater political freedom and power at the local level have also affected the revenue base of local governments. For example, tax assessments reportedly have come under the undue influence of some political leaders who have placed political supporters in lower graduated tax brackets (Bahiigwa and others 2004). In response to inefficiency and corruption in tax administration, local governments have opted to privatize the collection of certain categories of revenue. However, survey evidence shows that the procedure of awarding tax collection contracts is riddled with the very shortcomings that privatization was intended to circumvent, leading to poor local revenue performance (Bahiigwa and others 2004). Suspension of the graduated tax based on political rather than economic determination of its burden to citizens has further deteriorated local governments' revenue 172 | JOHN A. OKIDI AND MADINA GULOBA performance. Besides, the fragmentation of districts from a total number of 39 in 1991 to 83 in 2006 is grossly inconsistent with the need to strengthen local governments' revenue, technical capacity, and autonomy. Conclusions On the whole, the system of decentralization has been fairly well established and has brought improvements in service delivery and accountability. Numerous jobs have been created in the process. However, it is highly conceivable that the achievements could have been much more pronounced both in terms of quantity and quality had the process been sequenced, with capacity building leading the way. Although the previous central government representatives were integrated into the decentralized government structure, the capacity problems have persisted, partially because decen- tralized governance has promoted excessive emphasis on employing the indigenous residents of the districts. Decentralization also has fostered participatory planning, heightened the sense of local ownership, and improved accountability. However, daunting challenges have emerged. In particular, the benefits of decen- tralization are conditional on the political and economic environment of the country in question, and Uganda might have achieved the maximum benefits from decentral- ization that are attainable given its political and economic circumstances. The bind- ing constraints on Uganda's economic growth may be beyond the ability of either centralization or decentralization to address. In a nutshell, Uganda's experience with decentralization poses several questions, including the following: (a) If central government failures are driven by attitudinal or behavioral factors, why would we expect individuals from the same population to perform differently at lower levels of government? (b) Must decentralization be the solution whenever a centralized system does not perform for whatever reason? (c) Should a small country like Uganda decentralize and to what extent? (d) What levels of democratic and economic development and societal enlightenment are consistent with what degree of decentralization? (e) Is there an appropriate pace or sequence of decentralization? These and other questions that may arise from the dis- cussion in this paper provide interesting areas for policy debate and country-specific local area development. Endnotes 1. For more detailed statistics on access and quality, see World Bank (2005). Lawson (2003) also discusses evidence of the poor quality of services, especially in government facilities such as health centers. 2. For in-depth analysis of the growth, poverty, and distributional impacts of Uganda's reforms, see Okidi and others (2004). 3. A survey conducted in 2005 and 2006 as part of a medium-term review of LGDP II in a sample of local government jurisdictions revealed that respondents knew very little about the procedures for holding leaders accountable. DECENTRALIZATION AND DEVELOPMENT | 173 References Bahiigwa, Godfrey, Frank Ellis, Odd-Heldge Fjeldstad, and Vegard Iversen. 2004. Rural Taxation in Uganda: Implications for Growth, Income Distribution, Local Government Revenue, and Poverty Reduction. Research Series 35. Kampala: Economic Policy Research Centre. Bardhan, Pranab, and Dilip Mookherjee. 2006. "Decentralization and Accountability in Infra- structure Delivery in Developing Countries." Economic Journal 116 (January): 101­27. Emorut, Sam. 2006. "Decentralization and Good Governance: Innovative Approaches to Public Service Management and Service Delivery." Unpublished mss. Ministry of Local Government, Kampala. Francis, Paul, and Robert James. 2003. "Balancing Rural Poverty Reduction and Citizen Participation: The Contradiction of Uganda's Decentralization Program." World Devel- opment 31 (2): 325­37. Iimi, Atsushi. 2005. "Decentralization and Economic Growth Revisited: An Empirical Note." Journal of Urban Economic Growth 57 (3): 449­61. Kamanyi, Judy. 2004. "Social Inclusion: A Service Delivery and Accountability Dilemma; A Review of Social in Select Poor Groups in the Planning, Implementation, and Monitoring of Policies and Programs in Uganda." Bukoto, Uganda. Lawson, David. 2003. "Gender Analysis of the Ugandan National Household Surveys, 1992­2003." A report prepared for the 2003­04 review of Uganda's Poverty Eradication Action Plan, Ministry of Finance, Planning, and Economic Development. Martinez-Vazquez, Jorge, and Robert McNab. 2003. "Fiscal Decentralization and Economic Growth." World Development 31 (9): 1597­616. MoLG (Ministry of Local Government). 2006a. "Decentralization Policy Strategic Frame- work." Kampala. ------. 2006b. "Social Impact Assessment Report: A Survey Report." Kampala. Okidi, John A., Sarah Ssewanyana, Lawrence Bategeka, and Fred Muhumuza. 2004. "Opera- tionalizing Pro-Poor Growth: A Country Case Study on Uganda." DFID, London. (http://www.dfid.gov.uk/pubs/files/oppguganda.pdf.) Onyach-Olaa, Martin. 2003. "The Challenges of Implementing Decentralization: Recent Experiences in Uganda." Public Administration and Development 23 (1): 105­14. Ouedraogo, M. G. Hubert. 2003. "Decentralization and Local Governance: Experiences from Francophone West Africa." Public Administration and Development 23 (1): 97­103. Prud'homme, Remy. 2003. "Fiscal Decentralization in Africa: A Framework for Considering Reform." Public Administration and Development 23 (1): 17­27. Reinikka, Ritva. 2001. "Recovery in Service Delivery: Evidence from Schools and Health Centers." In Uganda's Recovery: The Role of Farms, Firms, and Government, ed. Ritva Reinikka and Paul Coulier. Washington, DC: World Bank. Romeo, G. Leonardo. 2003. "The Role of External Assistance in Supporting Decentraliza- tion." Public Administration and Development 23 (1): 7­16. World Bank. 2005. "Uganda Poverty Assessment Report." Unpublished mss., World Bank, Washington, DC. Decentralized Rural Governance in a Post-Socialist Economy: A Case of Community-Based Agricultural Marketing in Poland MARTIN PETRICK AND ANDREAS GRAMZOW In many former socialist countries, decentralized governance mechanisms are diffi- cult to implement due to the traditional strongly centralized administrative system. Similarly, approaches built on local collective action and participation of the civil society often meet with disapproval because they have been ideologically abused by former regimes. For example, Balint (2004) recently investigated the extent to which local cooperative arrangements in Romania could reduce the barriers to development such as inadequate restructuring of input and output markets in agri- culture. Her findings reveal that most farmers reject cooperation, due to a lack of true cooperative tradition and negative experiences with forced cooperation during socialism. In this paper we show how the involvement of local governments and private indi- viduals allowed the successful creation of a marketing association in southeastern Poland, despite a hostile environment for decentralized governance similar to that in other post-socialist countries. In contrast to traditional service cooperatives, the mar- keting organization we study has the form of a public-private partnership with only a small number of members; this has distinct consequences for risk allocation, deci- sion making, and funding. The paper is organized as follows. First we introduce our case study region, and then we summarize a number of typical barriers to successful collective action on the local level, present the public-private partnership Chmielnik Zdroj (Well of Chmielnik), show which factors contributed to its successful operation, and discuss some of its limitations. A final section concludes. Martin Petrick is the Deputy Head of the External Environment for Agriculture and Policy Analysis Department of the Leibniz Institut für Agrarentwicklung in Mittel und Osteuropa (IAMO, Leibniz Institute of Agricultural Development in Central and Eastern Europe). Andreas Gramzow is a Doctoral candidate at IAMO. A detailed presentation of the case study results is given in Gramzow (2005). Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 175 176 | MARTIN PETRICK AND ANDREAS GRAMZOW Regional Background Rural areas in southeastern Poland are dominated by small-scale farms that success- fully resisted several attempts at collectivization during the socialist period. Peasant farms have only been slightly affected by the Polish transition to a market system. Deteriorating agricultural terms of trade and hidden unemployment have conserved farm structures and led to an increasing income gap between booming urban centers and lagging rural areas. This makes rural development one of the top political prior- ities in the region (see Petrick and Tyran 2003 for a more detailed analysis). The region of our case study, Dolina Strugu, covers an area of 300 square kilo- meters in which approximately 38,000 inhabitants are living. The region's economy is strongly related to agriculture. About two-thirds of the total working population work on the 6,200 individual farms located in Dolina Strugu, whereas younger farm- ers often have a second job in Rzeszow, which is the next bigger city. The majority of farmers are between 50 and 70 years old and worked in industrial companies (mainly mechanic industry) in Rzeszow until the early 1990s. The average farm size is slightly above 3 hectares. Soil and climate conditions are moderate, but the culti- vation of land is limited by the hilly relief of the landscape. As a result of the latter, nearly the whole area is classified by the European Union (EU) as a "less favored area." Agricultural production is dominated by plant production, especially soft fruits, cereals, feed, and root plants. Animal production plays an inferior role, with only a few farmers specializing in the production of pig or beef. The unemployment rate in Dolina Strugu is about 18.3 percent (PUP 2005). In 2002 the Podkarpackie Voivodship was Poland's poorest region, with an average income per person equal to 71 percent of the Polish average (Eurostat 2005). Due to the lack of job opportuni- ties and low income, many young people have migrated to bigger cities or abroad. Gramzow (2005) has carried out an in-depth case study of Dolina Strugu to gain specific and detailed information about the principal problems and prospects for agri- culture, nonagricultural income opportunities, investment plans of farmers, their con- tractual arrangements with wholesalers, and their ambitions to act collectively on the market. The following analysis is based on case study research according to Yin (2003) as well as on active, participant observations and 33 qualitative interviews with farm- ers, local leaders, representatives of local authorities, and local entrepreneurs. Typical Constraints to Rural Collective Action in a Post-Socialist Context A common solution to improve income in rural agricultural areas is the creation of farmer cooperatives or producer groups. According to Balint (2004: 247), the advan- tages of cooperation include "the reduction of transaction costs in accessing input and output markets and strengthening of the negotiation power of the farmers." In economies characterized by incomplete markets and a lack of technical knowledge, DECENTRALIZED RURAL GOVERNANCE IN A POST-SOCIALIST ECONOMY | 177 service cooperatives are supposed to be most beneficial (Deininger 1995). Escobal, Agreda, and Reardon (2000) note, "Small farmers lack human and organizational capital embodied in management skills, [and] that lack creates both production and marketing inefficiencies." The authors regard lower input prices, lower interest rates as a result of group borrowing, higher producer prices, and lower transport costs as important advantages of farmers' service cooperatives. In Dolina Strugu, the question emerged whether traditional cooperative arrange- ments between farmers could have similar advantages. However, the results of our case study show that various constraints are likely to prevent successful cooperative action at the local level. The following concerns were expressed by local respondents or emerged as a result of our data collection: 1. Individual reservations about cooperation. This is locally called the "Polish men- tality" and was characterized by a local farmer as follows: "If the one who cooper- ates with me will earn higher profits from this cooperation than I do, I don't see any reasons to cooperate. And how can I trust that he does not take advantage of me?" 2. Principal-agent problems between management and members. If the management is employed by a producer group, members have to be aware of opportunistic behavior of the manager. Therefore, a control system or a particular incentive sys- tem has to be established, which leads to higher transaction costs. 3. Free-riding among members. Investment in the producer group and in its pro- cessing and marketing facilities is a process with a long time perspective. Hence, the question arises: How can members be sure that, if they pay their fees and invest in the producer group, other members will do the same? According to Olson (1965), this problem increases significantly with the number of members of an organization. 4. Inflexible decision-making procedures. It is estimated that producer groups in Dolina Strugu need more than 1,000 members to achieve some bargaining power. But the need to find a consensus and pass decisions in such a big group creates high transaction costs and paralyzes decision making. 5. Highly uncertain business environment. Specialized producer groups face perma- nent threats from crop failure or unexpected changes in the market and legal envi- ronment. In Dolina Strugu in the middle of the 1990s, a soft fruit producer group collapsed after a few years due to a dramatic decrease in prices for soft fruits. 6. Lack of finance. Producer groups in Dolina Strugu will only succeed if they process and market their products on their own. This requires founding capital that small peasant farms are not able to provide due to a lack of suitable assets with which to borrow money and very conservative lending practices among rural banks (Petrick 2004). Even if a bank would agree to give credits, a suitable guar- antor would be hard to find. If guarantors are group members, a moral hazard problem appears between guarantor and the remaining members. 178 | MARTIN PETRICK AND ANDREAS GRAMZOW As our discussion shows, cooperation between farmers based on a producer group in the region of Dolina Strugu is fraught with difficulties. The following section pres- ents the public-private partnership Chmielnik Zdroj, which is working as a commercial entity based on a bottom-up initiative of local authorities, local leaders, and local businesses. We attempt to demonstrate why it was successful despite the previously mentioned obstacles. Chmielnik Zdroj: A Public-Private Marketing Partnership in Southeastern Poland Chmielnik Zdroj is a joint-stock company located in Chmielnik, one of the four com- munes of Dolina Strugu. The most important associates are the four communes of Dolina Strugu and the members of a local nonprofit organization also called Dolina Strugu, which was the main initiator of the public-private partnership. Farmers' rela- tion to the partnership is market based: they usually are not members and do not pay any fees. When Chmielnik Zdroj started in 1994, its main objectives were to create new jobs for people from the Dolina Strugu region, to give farmers the opportunity to sell their products, to provide people from the region with local, traditionally made products, and to keep the environment and fields in good and environmentally friendly conditions. In 2005 the partnership employed 400 people and bought prod- ucts from more than 500 small farms of Dolina Strugu. In the region around the cities of Krakow, Lublin, Rzeszow, and Tarnow, they use 170 vehicles to provide 70,000 households with processed products (vegetables, fruits, honey, bread, soups, and pasta). Customers place their orders by phone or the Internet, and their purchases are delivered to their homes. Farmers have short- or medium-term contracts with the part- nership, which enables them to sell their products without having to deliver them to a distant wholesale company. The partnership provides them with a basic source of income and certainty about next year's market situation. Furthermore, the partnership processes farmers' products and sells them directly to its customers. In addition, it sup- ports farmers in applying for ecological certification. In the future, farmers will also receive assistance in planning production. Local farmers receive information about EU programs and free assistance in applying for direct payments or other funds from an office that is financed mainly by the local nonprofit organization Dolina Strugu. Chmielnik Zdroj has had a widely perceived impact on broader rural development goals. It provides nonfarm jobs, thereby improving the economic situation of rural households and counteracting the outmigration of young people. In addition, it meets consumer demands for regional and environmentally beneficial products. An indirect impact is the larger number of businesses in Dolina Strugu compared to other rural communes in the region. In the four communes of Dolina Strugu, the number of businesses per 100 inhabitants increased between 1995 and 2001 from 2.2 to 5.1, whereas in other rural communes near Rzeszow, this number increased only slightly, from 2.2 to 2.3 per 100 inhabitants in the same period (UNDP 2003). DECENTRALIZED RURAL GOVERNANCE IN A POST-SOCIALIST ECONOMY | 179 Steps on the Way to Chmielnik Zdroj The roots of this initiative go back to a telephone cooperative that was founded in 1991 by five representatives of the local government together with 20 local business administrators. In 1991 these four communes only had access to 800 telephone sub- scribers, and the state-owned telephone company was not willing to invest in this region any further. The founding capital of the cooperative consisted of fees that every inhabitant had to pay to join this telephone company. Additional capital was taken from bank credits, with the communes and some local leaders acting as guarantor. With the help of this founding capital, a new telephone network was set up, and the number of subscribers in Dolina Strugu grew to about 9,600. For the 6,500 members of the telephone cooperative, local calls were free. Since 1993, the telephone cooper- ative has been able to operate without any additional financial support. On the basis of the successful experience of cooperation, leading members of the telephone coop- erative as well as of local authorities founded a nonprofit organization called Dolina Strugu, which seeks to solve communal infrastructure problems, support the envi- ronmental protection of the area, and stimulate economic activity in the region. As one of the first initiatives, Chmielnik Zdroj was founded with start-up funds partly from the communes (especially the Chmielnik commune) and partly from a loan secured using local leaders' houses as collateral. Initially, the partnership provided the regional population with mineral water, to be ordered by telephone and delivered to homes. In the following years, further products were added, and production capaci- ties grew very quickly. Further important initiatives of the nonprofit organization Dolina Strugu are the support of local business start-ups and the formation of a local employment agency. Success Factors The key factors that made Chmielnik Zdroj a success were its specific organizational setup as a public-private partnership, its market-oriented way of sharing risk, as well as the fact that it was built on local needs and thus emerged as a true bottom-up self- help organization. Most associates had been politically active in the local government and had an eye for local problems. As local politicians, they had an incentive to respond to local needs. Because of their detailed knowledge of local conditions, the founding members of the partnership were able to create an organization that pro- vides locally viable solutions for the current problems in agriculture and the rural economy. It is a prime example for the rule that successful cooperation critically depends on the initiative of the affected individuals and their willingness to con- tribute to their development (Müller 1994). However, the actively participating mem- bers are only a small number of local government representatives and local leaders. This makes Chmielnik Zdroj a public-private partnership distinct from traditional farmers' cooperatives. The approach overcomes a number of obstacles to coopera- tion but also creates problems for broader participation. 180 | MARTIN PETRICK AND ANDREAS GRAMZOW In the following, we summarize the major factors in Chmielnik Zdroj's success: · The market-based approach limits the risk of long-term involvement for farmers. The partnership uses its own distribution network and has a regional marketing label. It successfully developed a regional niche market and is independent from the big retail companies. Even so, farmers do not have to become shareholders or invest their own funds, and they do not have to take the risk of setting up the organization. This eliminates the free-rider and moral hazard problems. Due to farmers' market-based relation to the partnership, there is little danger of a hold- up for them. Furthermore, a contract between the commune Chmielnik and the partnership guarantees local farmers that they can sell their products to the organ- ization before it can buy products from other regions. The main risk is taken by the communes as well as by local leaders who manage the partnership. Further- more, the wide array of products provided by Chmielnik Zdroj limits the risk of a product failure, such as that experienced by farmer producer groups in Dolina Strugu during the mid-1990s. · The setup as a public-private partnership with a limited number of members eases decision making and allows professional management. The principal-agent prob- lem between management and the associates of the partnership was minimized because the executive manager, who grew up in Dolina Strugu, is subject to super- vision by a board of trustees and was already involved in the telephone coopera- tive. The associates are a small group of local inhabitants who had cooperated suc- cessfully within the telephone cooperative. This enabled professional management and circumvented the problem of individual reservations against cooperation so prevalent in the rural society. The engagement of local governments as guarantors greatly eased the funding problem of the public-private partnership. · Offering complementary services to farmers provides incentives for a long-term business relationship and serves as a quality-control mechanism. The nonprofit organization Dolina Strugu provides farmers with complementary assistance in applying for EU programs, which is particularly important for the large number of part-time farmers. From Chmielnik Zdroj, farmers receive assistance in apply- ing for ecological certification, which also works as an additional quality-control system for the marketing partnership. · A true bottom-up approach secures support and trust of the local population. The engagement of associates of Chmielnik Zdroj in the telephone cooperative strengthened their reputation within the local community, which particularly appreciated the fact that the telephone company gave a share of its profits to the local community by offering them free local calls. Moreover, 10 percent of the profits from the partnership go to local welfare organizations. This is stated on all product labels and strengthens local trust as well. This trust can be recognized by the fact that vehicle drivers of the partnership have the key to many houses or flats of their clients in Dolina Strugu and other regions, which allows them to deliver goods when the client is absent. The trust in the local leaders and the nonprofit organization also was evident in our case study interviews. Table 1 summarizes the DECENTRALIZED RURAL GOVERNANCE IN A POST-SOCIALIST ECONOMY | 181 TABLE 1. Trust of Local Inhabitants in National or Regional Institutions (percent) Number of Institution A lot A bit Not much Not at all respondents President 4 50 32 14 28 Government 0 25 40 35 28 Local administration 0 39 54 7 28 Members of the nonprofit organizationa 12 65 24 0 17 Mayor 0 63 26 11 27 Source: Author's compilation. a. The remaining 12 respondents said they do not have enough knowledge about the nonprofit organization to answer this question. results of a small survey conducted in Dolina Strugu, where we asked people to answer a questionnaire concerning trust in different national and local institutions. It shows that the trust in representatives of the local institutions (members of the nonprofit organization and the mayor) is much higher than in national institutions. Limitations The public-private partnership had little positive impact in the following areas: · Vitalization of the rural civil society has been limited. Due to its slim management structure, the partnership has had little effect on harnessing the broader civil soci- ety for rural development. Besides some rural women's circles in small villages, whose members tend to be older, associations or clubs contributing to regional identification or development are practically absent. This situation is worsened by a perceived myopia of local inhabitants as a result of the poor social and economic situation. People still do not trust their own ability to become agents of change and generally expect the government to initiate any development process. "Everything depends on the mayor, and people here have no money and no power, so they can- not do anything. The people here will not be against new activities, but if they are in committees without any power and any money, what should they do?" as one interviewee said. It appears there also have been deficits in the transmission of information concerning national and European programs for rural development. · Interest of local governments has been eroding. The periodic change of local gov- ernments in three of the four communes of Dolina Strugu led to a decline in coop- eration between the nonprofit organization and representatives of the local authorities. This is primarily due to the frequent turnover of local authority rep- resentatives. As some interview partners mentioned, the relationship between the local governments seems to be characterized more by competition than by coop- eration. Due to the success of the organization and its high acceptance among local inhabitants, local government authorities apparently fear a loss of power. This, in turn, results in coordination problems between the local authorities and 182 | MARTIN PETRICK AND ANDREAS GRAMZOW the organization's initiatives. For example, there is no common strategy for regional tourism development. Actions undertaken by the local authorities, such as creation of an artificial lake, are not coordinated with Dolina Strugu. · Marketing problems of bigger farms remain unsolved. Despite its success in devel- oping outlets for regional products, Chmielnik Zdroj has been unable to solve the marketing problems of bigger commercial farms. One farmer reported, "They consolidate mostly small farms which have a half hectare of potatoes. I remember, there were some situations where farmers had very good yields, and then they had problems selling their crops because they [the marketing cooperative] did not need that much." Another said, "Farmers here are waiting for another place to sell their crops; a small processing plant is really needed in here." However, the partnership might be a positive example and a promising incentive for similar cooperative ini- tiatives, such as establishing a processing plant. Furthermore, the advisory agency of Dolina Strugu eases farmers' access to additional financial sources, which could lead to further farm investments and improved competitiveness. Conclusions The public-private partnership described here is an example of a decentralized rural development approach that successfully used institutional complementarities among government, the community, and the market. While the regional distribution net- work for agricultural products is largely market based, the partnership came into being only as a result of local collective action. Local government officials played a crucial role in this process, and part of the seed capital came from public sources. However, the market-based strategy comes at the cost of only limited involvement of the broader civil society. The majority of the rural population perceives the mar- keting partnership from the perspective of the producer or consumer, but not as a stimulus to become personally involved in rural development activities. The develop- ment barriers emanating from a widespread reluctance to become involved in collec- tive action have thus been circumvented, but not overcome. Furthermore, the involvement of local governments has become unstable over time, which has led to a lack of coordination within the partnership. Finally, while a marketing network for peasant farmers could be established, the potential demand in the region has its lim- its. For this reason, bigger farms must still look for marketing or processing outlets beyond the regional network. This true bottom-up initiative has had an impressive impact since its foundation as a telephone cooperative in the early 1990s. However, it is important to recognize the extent to which this success is due to specific local circumstances. In particular, the availability of (a few) leaders and their willingness to contribute to public devel- opment goals has been of pivotal importance in Dolina Strugu. This factor of success is often crucial, but hard to transplant to other regions. DECENTRALIZED RURAL GOVERNANCE IN A POST-SOCIALIST ECONOMY | 183 References Balint, Borbala. 2004. "Institutional Factors Influencing Agricultural Sales of the Individual Farmers in Romania." In The Role of Agriculture in Central and Eastern European Rural Development: Engine of Change or Social Buffer? ed. Martin Petrick and Peter Weingarten, 238­56. Studies on the Agricultural and Food Sector in Central and Eastern Europe 25. Halle, Saale: Institute of Agricultural Development in Central and Eastern Europe. (http://www.iamo.de/dok/sr_vol25.pdf.) Deininger, Klaus. 1995. "Collective Agricultural Production: A Solution for Transition Economies?" World Development 23 (8): 1317­34. Escobal, Javier, Victor Agreda, and Thomas Reardon. 2000. "Endogenous Institutional Inno- vation and Agro-Industrialization on the Peruvian Coast." Agricultural Economics 23 (3): 267­77. Eurostat. 2005. "Regional Statistics from the Eurostat Database." http://epp.eurostat.cec. eu.int/portal/page?_pageid=1090,30070682,1090_33076576&_dad=portal&_schema= PORTAL. Gramzow, Andreas. 2005. Experience with Endogenous Rural Development Initiatives and the Prospects for Leaders in the Region "Dolina Strugu" Poland. IAMO Discussion Paper 89. Halle, Saale: Institute of Agricultural Development in Central and Eastern Europe. http://www.iamo.de/fileadmin/institute/pub/dp89.pdf. Müller, J. O. 1994. "Strategies Employed When Establishing Co-operatives." In International Handbook of Cooperative Organizations, ed. Eberhard Dülfer and Juhani Laurinkari, 841­46. Göttingen: Vandenhoek and Ruprecht. Olson, Mancur. 1965. The Logic of Collective Action: Public Goods and the Theory of Groups. Cambridge, MA: Harvard University Press. Petrick, Martin. 2004. "Farm Investment, Credit Rationing, and Governmentally Promoted Credit Access in Poland: A Cross-Sectional Analysis." Food Policy 29 (3): 275­94. Petrick, Martin, and Ewan Tyran. 2003. "Development Perspectives of Subsistence Farms in Southeastern Poland: Social Buffer Stock or Commercial Agriculture?" In Subsistence Agriculture in Central and Eastern Europe: How to Break the Vicious Circle? ed. Steffen Abele and Klaus Frohberg, 106­23. Studies on the Agricultural and Food Sector in Central and Eastern Europe 19. Halle, Saale: IAMO. http://www.iamo.de/fileadmin/institute/ pub/sr_vol22.pdf. PUP (Powiatowy Urzad Pracy). 2005. "Labour Agency of the District Rzeszow [Powiatowy Urzad Pracy w Rzeszowie]." Rzeszowie: PUP. http://www.pup.rzeszow.pl. UNDP (United Nations Development Programme). 2003. "Development of the Strug Valley in Result of the Application of Information and Communication Technologies and Innova- tions." UNDP, Warsaw. http://www.undp.org.pl/en/center.php?nr=196. Yin, Robert K. 2003. Case Study Research: Design and Methods, 3d ed. Applied Social Research Methods Series 5. Thousand Oaks, CA: Sage. Part VI: Emerging Policy Issues Opportunities for European Bioethanol in a Global Environment (2006) OLIVER HENNIGES In Germany, as well as in the European Union (EU), the use of bioethanol as an alter- native fuel has been demanded by different stakeholders and supported by political decision makers for several years. In this paper, the competitiveness of bioethanol production is analyzed from both a national and an international perspective. Bioethanol from starch-, sugar-, and cellulose-containing plants can be blended with gasoline at any ratio, although engine adjustments may be required. In this con- text, the advantages of a higher octane number and higher oxygen content, as well as the disadvantages of lower energy density and higher vapor pressure in low blend- ing ratios, are important. The main objective of this paper is to compare the costs of producing bioethanol in Australia, Brazil, China, the European Union, Thailand, and the United States in 2005. In this comparison, plant capacity and cost allocations are standardized. Of these countries, production costs are by far the highest in the European Union, at approxi- mately 45 per hectoliter (hl), using wheat and sugar beet as feedstock and assuming a wheat price of 100. This can be traced back to relatively high feedstock and utility costs (see figure 1). In Brazil, bioethanol could be produced for the lowest costs (approximately 17 per hl) due to favorable climatic and economic conditions. In addition, Brazil has 30 years of experience in bioethanol production as well as a large potential for the expansion of sugarcane production, making it a long-term dominant competitor in the worldwide market for sugar and ethanol (see figure 2). Since 2005, due to plenty of state aid during every stage of production and processing, the United States has been the largest producer of bioethanol worldwide, followed by Brazil. In contrast to Brazil, corn-based production is not targeted as an export commodity but rather as a substitute for previous octane enhancers. The Oliver Henniges is Research Associate with the University of Hohenheim, Germany. This article was written in the sum- mer of 2006, before the controversial discussion about biofuels had started. The article reflects the author's position at the time it was written and must not be considered an official statement of any institution at which he worked. Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 187 188 | OLIVER HENNIGES FIGURE 1. International Comparison of the Costs of Bioethanol Production in 2005 Costs (-ct/I) 60 50 40 30 20 10 0 Europe Brazil United States Australia Thailand China (wheat/beet) (cane) (corn) (cane) (cane) (corn) Country (feedstock) gross production costs net production costs feedstock costs Source: Author's compilation. FIGURE 2. Global Supply Curve of Bioethanol, 2005 Price (-ct/I) 60 50 40 30 20 10 0 ica Brazil India Asia Africa China Union . L.AmericaRem.Amer Oceania States t Europe C. pean United Res Rem Euro Source: Author's compilation. production costs of about 26 per hl are much lower than those in the European Union as a consequence of an exchange rate in 2005 of only $1.20 per 1, among other reasons. Using sugarcane, Australia and Thailand have cheap feedstock, but so far these two countries produce only small amounts of bioethanol. In particular, the price of sugar, OPPORTUNITIES FOR EUROPEAN BIOETHANOL IN A GLOBAL ENVIRONMENT | 189 which increased in 2005, reduces the willingness of both of these sugar-exporting countries to take the risks entailed in entering bioethanol production. The production costs of about 21 per hl in Thailand are far lower than those in Australia, which amount to 27 per hl. In China, where the largest processing plants in the world are situated, bioethanol production from corn costs about 31 per hl in 2005, due to the relatively high price of feedstock. In light of the rising demand for food in China, the long-term, compet- itive production of bioethanol cannot be assumed. Another focus of this presentation is the analysis of the political and economic framework of bioethanol production in Germany. The petrol tax exemption granted for bioethanol in Germany had until 2005 enabled bioethanol to compete with fossil fuels. This, however, also applies to cheaper imported bioethanol, whose competitiveness has been restricted by an EU import tariff as well as by a high world market price for bioethanol. Figure 3 shows the different possible scenarios for the import of Brazilian bio- ethanol to a European petrol refinery in order to blend it with gasoline. Based on pure costs and tariffs (scenario 1), Brazilian bioethanol could be delivered to Europe for the same price as European domestic ethanol (scenario 7). Given the high price of gasoline in 2006 and the high market price for domestic ethanol in Brazil, the real- istic situation is now (2006) scenario 2, which yields a high import price. In the beginning of 2004, the price of bioethanol in Brazil was lower than its production costs. In that case (scenario 4), the import price would have been lower than the EU costs (scenario 7). If it is assumed that, in addition, the import tariff of 19.2 per hl FIGURE 3. Costs of Brazilian Ethanol for an EU Refinery Costs (-ct/I) 80 70 60 50 40 30 20 10 0 (1) (2) (3) (4) (5) (6) (7) EU 2006 2006 2004 2004 tariff tariff in w/o Costs Prices Costs Prices + lower (4) Costs (4) Scenario (Nr) Source: Author's compilation. 190 | OLIVER HENNIGES is reduced (scenario 5) or even abolished (scenario 6), EU bioethanol production would lose a great deal of competitiveness on the domestic market. It becomes apparent that, under the present conditions, farmers could receive higher than market prices for wheat and sugar beets. Without a general increase in the demand for agricultural commodities in the future, these higher prices will not be realized, since in the bioenergy market, as in the food market, feedstock farmers have little influence on prices. The same applies to rapeseed, whose cultivation since the implementation of nonfood production on reserved land in the EU in 1992 has turned Germany into the largest biodiesel producer worldwide. Thus, due to low prices for rapeseed until now, biodiesel can be produced for 52 per hl, whereas in 2006, freed from petrol tax but including value added tax, it was sold to consumers for 100 per hl at the pump. Despite these large profit margins for processors and distributors, the price of rapeseed so far has increased only slightly. With this, the fed- eral government detected an antinomic overcompensation of the difference between the production costs of biodiesel and fossil diesel by the petrol tax exemption. There- fore, biodiesel was taxed at 9 per hl beginning in August 2006. Similarly, in 2008, an overcompensation assessment will be carried out for bioethanol, which is currently untaxed. From the demand side for biofuels, the sales opportunities for biodiesel are better than for bioethanol, since the EU tends to have a shortage of diesel and a surplus of gasoline. The increasing transportation of cargo, as well as more efficient and attractive diesel engines in the automotive sector, supports this development. When surveyed, the petrol and the car industries argue against the use of bioethanol. At the same time, technical disadvantages are considered the main obstacles. As a compro- mise, the use of 47 percent bioethanol containing ETBE (ethyl tertiary butyl ether) is proposed, since it does not have these disadvantages. Since Germany, as well as the EU, is a net importer of vegetable oils and net exporter of sugar and starch, bioethanol production should be preferred to biodiesel production from an agricultural point of view. It can theoretically be based on agri- cultural surpluses that so far have been exported with the help of refunds. In this con- text, taking into account the demography in 2010, the agricultural land of the EU and Germany would suffice to meet the domestic biofuel need of 5.75 percent, as reg- ulated by EU law, without necessitating food imports. In addition, the reform of the EU common market regulation of sugar leads to lower sugar prices, so the future use of sugar beet as a feedstock for bioethanol production might result in higher revenues, assuming short transport distances and high productivity in the field. Regarding energy and carbon dioxide (CO2 equivalents) efficiency, the production of biofuels in Europe is inferior not only to imported bioethanol from sugarcane but also to alternative domestic climate protection measures such as organic solid fuels. Compared to the direct combustion of wood, the CO2 mitigation costs are so high that, from an environmental point of view, the use of bioethanol is less efficient. The third focus of this presentation is an analysis of the relevant agricultural and biofuel markets. There is a strong correlation between the price of bioethanol and the price of gasoline in Brazil and the United States, the most important countries OPPORTUNITIES FOR EUROPEAN BIOETHANOL IN A GLOBAL ENVIRONMENT | 191 worldwide with respect to bioethanol production. Due to Brazil's dominant position in both markets, the simultaneous use of Brazilian sugarcane for producing ethanol as well as sugar means that the price of crude oil has an indirect influence on the price of sugar, since bioethanol production can be extended rapidly on account of sugar production. In the case of a continuously high price for crude oil and further expan- sion of biofuel production, it can be assumed that this influence will be noticeable in other agricultural commodity markets as well. This would necessarily lead to higher food prices. The future competitiveness of bioethanol production in Europe will depend on the price of crude oil and the willingness of petrol companies to accept the blending of biofuels at all. So far, the political and taxation framework will continue to have a much larger impact on the financial profitability of bioethanol production. In order to achieve international competitiveness of German and European bioethanol pro- duction, further tremendous reductions in the costs of production at all stages of the process as well as creation of a long-term, reliable legal framework are essential. Nonlinear Effects of Weather on U.S. Crop Yields, Implications for Climate Change, and Why These Effects Matter for Developing Countries WOLFRAM SCHLENKER AND MICHAEL ROBERTS In developing countries, . . . as much as a third of GNP and half the population currently depends on agriculture. They may still be vulnerable to climate change for many years to come. [Schelling 1997] In developing nations, the share of agricultural goods constitutes a significant share of domestic income. Thus, Schelling argues, the impacts of climate change will likely weigh more heavily on the developing world, while the costs of curbing climate change will likely fall more heavily on developed nations like the United States, engendering a difficult bargaining problem. Although Shelling's broad view about the split between relative costs and benefits is likely correct, active study and debate con- tinue about the likely impacts of climate change on agricultural production, with a key focus being the United States. While developing countries might be especially vulnerable to climate change, there are good reasons to focus research on the United States. The first reason is that the United States produces a large share and is one of the largest exporters of two of the world's most important staple crops: corn and soybeans. Because these commodities form the caloric basis for most food, their prices drive food prices more broadly. U.S. production of these crops hence has considerable influence on world prices and, via market forces, other key commodities like wheat and, to a lesser extent, rice. Large climatic impacts on the United States, if not offset by production gains in other parts of the world, would have a large influence on the world supply and prices of food. These impacts may be of modest importance to rich countries like the United States, but they are much more important for the developing world. A second reason to focus research on the United States is data quality. What we learn from studying agri- culture in the United States, at least in some instances, might be extrapolated to parts of the world where less or poorer data exist. And because the United States is so large Wolfram Schlenker is Assistant Professor in the Department of Economics and the School of International and Public Affairs at Columbia University, and Michael Roberts is Assistant Professor in the Department of Agricultural and Resource Economics at North Carolina State University. Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 193 194 | WOLFRAM SCHLENKER AND MICHAEL ROBERTS and climatically diverse, the available data traverse a reasonably large range of climates. To illustrate these two points, we present new evidence on the relationship between weather and crop yields in the United States. Using unique fine-scale weather data and a long history of observed crop yields, we are able to accurately discern a highly nonlinear and asymmetric relationship between temperature and corn and soybean yields. The estimated relationship is stable over crops and regions. In all cases, yields increase gradually in temperature up to about 29°C. Above 30°C, yields decline sharply. The stability of this pattern over space and crops suggests that it may be applicable to other parts of the world, and perhaps other crops, which illus- trates why a study of U.S. crop yields can give important insights for developing countries. Moreover, the latest climate change models predict a significant increase in harmful temperatures, which implies that yields will decline sharply by the end of the century. If these declines are not offset by commensurate gains elsewhere (for exam- ple, Canada), food prices will likely rise considerably and have large welfare impacts on consumers in developing countries. Previous Agronomic and Econometric Studies Previous studies linking weather to crop yields can be roughly divided into two cat- egories: agronomic studies and econometric studies. Agronomic studies use crop simulation models to predict plant growth. Initial soil conditions, daily nutrient application, temperatures, and rainfall are fed into simula- tion models to derive continuous measures of evapotranspiration, soil moisture, plant growth, seed formation, and ultimately yields at the end of the growing season. For a nice overview, see the article by Adams and others (1995). Current versions of models developed for many crops are maintained by the Decision Support System for Agrotechnology Transfer.1 Given the number of parameters and complex dynamic structure of these models, researchers must rely on "reasonable parameter assump- tions" or calibrate parameters one at a time, while keeping others fixed. The advantage of these models is that they are based on a rich physiological theory. A drawback is that results are a direct consequence of assumed functional forms and parameter assumptions that can be difficult to verify. Econometric studies have been used to estimate statistically the link between weather and crop yields using observed historical yields for various geographic regions. The advantage of these studies is that parameters can be estimated from large data sets. (Lobell and Asner 2003; Mendelsohn, Nordhaus, and Shaw 1994; Schlenker, Hanemann, and Fisher 2006; Timmins 2006). A disadvantage is the rela- tive simplicity of the models: they cannot incorporate the rich physiological structure of agronomic models and generally use relatively crude aggregate measures of weather, such as average temperature and cumulative precipitation for a few months of the growing season. NONLINEAR EFFECTS OF WEATHER ON U.S. CROP YIELDS | 195 A common feature of both approaches is that they typically use linear or quadratic functional forms. The rationale behind the quadratic form is that too little or too much heat or precipitation is harmful for a plant. Freezes or heat waves will kill a plant, as will droughts or floods. The optimal temperature and rainfall should hence be found for moderate outcomes, while extreme conditions are damaging. At first glance, a quadratic functional form might appear appropriate, as it captures the fact that there is an optimum for moderate values. The quadratic form, however, also has a fairly restrictive implicit assumption: symmetry: 1°C, 2°C, or 5°C above the opti- mum is assumed to be equally as bad as 1°C, 2°C, or 5°C below the optimum. As shown below, this can be a very misleading assumption, as temperatures above the optimum growing temperature decrease yields much more than temperatures below the optimum growing temperature. This asymmetry has strong implications for agri- culture in developing countries that are already experiencing warm temperatures. Further rises in temperature would significantly increase the occurrence of highly damaging extreme temperatures. Our analysis combines the strength of the agronomic approach (a micro-level data set of daily weather outcomes) with the strength of econometric studies (statistical inference). We also use a more flexible functional form between temperature and crop yields, which subsumes quadratic and more complex nonlinearities. Data Data development and sources are outlined in more detail in Schlenker and Roberts (2006). We briefly review them here. We compile a panel data set of corn and soy- bean yields for counties in the eastern United States for the years 1950­2004. Since we have around 2,000 counties for each of the 55 years, our data set has more than 100,000 observations.2 Crop yields in each year are paired with a micro-level weather data set that incorporates the entire temperature distribution between the minimum and maximum for each day of the growing season. From these data, we calculate the fraction of each day a crop is exposed to each 1°C interval and then sum them for the months March through August. The underlying weather data set is con- structed on a 4 kilometer by 4 kilometer grid system and averaged over all grid cells in a county that have agricultural area in a satellite scan. Empirical Results Our empirical model relates the annual log of crop yield to a fifth-order polynomial of the total time a crop is exposed to each temperature during the growing season from March through August as well as a quadratic in total precipitation. We use Chebyshev polynomials for the temperature variable, as they are more flexible than traditional polynomials. In a sensitive check we use a dummy variable for each 1°C interval and obtain similar results, indicating that our results are not driven by the 196 | WOLFRAM SCHLENKER AND MICHAEL ROBERTS assumed functional form. We focus our discussion on the polynomials, as the smooth shape of the estimated relationship is easier to display and interpret. Because the dependent variable is log yield, the functional form implies a multiplicative interaction between temperature and precipitation effects, that is, temperature effects get scaled up or down depending on available precipitation. While agronomic simulation mod- els have more elaborate interactions between daily temperatures and precipitation, our set of daily weather data reveals a very low correlation between temperatures and precipitation (less than 0.1), so our simpler model gives nearly unbiased estimates of average temperature effects. Other controls include quadratic yield trends by state to account for technological progress that has increased average yields almost threefold in our 55-year study period, as well as county fixed effects that capture all time-invariant effects in a county, such as soil quality, access to markets and population centers, and so on. Regression results for corn and soybeans are displayed in figure 1. The solid black line gives the effect of a one-day (24-hour) exposure to a certain temperature on annual crop yield. For example, for corn (left panel), changing the temperature distribution such that the crop would be exposed one less day to 8°C and one more day to 25°C would increase yields by roughly 1 percent (the difference in height of the black line at the two x-values 8°C and 25°C). However, changing the tempera- ture distribution such that the crop would be exposed one less day to 8°C and one more day to 39°C would decrease yields by roughly 5 percent. A very similar pattern holds for soybeans (right panel). In each case, a 95 percent confidence band around the estimated relationship is shown as black dashed lines. We allow the error terms to be spatially correlated within a given year, which on average increases our stan- dard error by a factor of 6. FIGURE 1. Relationship between Yields and Temperature a. Corn yields b. Soybean yields Log yield (bushels) Exposure in days Log yield (bushels) Exposure in days 0.02 0.02 30 30 0.01 0.01 25 25 0 0 20 20 0.01 0.01 0.02 15 0.02 15 0.03 10 0.03 10 0.04 5 0.04 5 0.05 0 0.05 0 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 Temperature (Celsius) Temperature (Celsius) Note: The black line depicts the effect of a one-day exposure to a certain temperature on yearly log yields. The graph is relative to a temperature of 8°C; that is, it depicts the increase or decrease in yields of a one-day exposure to a cer- tain temperature compared to a one-day exposure at 8°C. The 95 percent confidence band is added as dashed lines. The gray histogram shows the number of days a country is exposed to each 1°C interval, on average. NONLINEAR EFFECTS OF WEATHER ON U.S. CROP YIELDS | 197 The gray histogram at the bottom of each graph shows the average number of days in the 184-day growing season from March through August that a crop is exposed to each 1°C interval. The frequencies reported in the histograms are for all counties and years. Since we record the fraction of a day a crop is exposed to each 1°C interval between the minimum and maximum temperature for each day, there is significant time in which a crop is exposed to temperatures above 30°C. This is due to the fact that maximum temperatures in the summer frequently exceed 30°C. The average time that temperatures exceed 30°C in our sample of counties in the eastern United States with corn yields is 16.1 days, enough to identify the sharply downward- sloping portion of our curve. The confidence bands are fairly narrow and only widen slightly for the highest temperatures, again suggesting that the sharply decreasing portion of our curve is well identified. Temperatures above 30°C quickly become very harmful for both corn and soybeans. The graphs look very similar for both crops. Using the distribution of temperatures within each day is crucial, as average tem- peratures rarely exceed 30°C, which would disguise the true underlying relationship. Moreover, the estimated relationship is highly asymmetric: a one-day exposure to temperatures 10°C above the optimum are much worse than temperatures 10°C below it, as annual yields drop by more than 4 percent in the former case and only 0.5 percent in the latter case. Such a relationship is not well approximated by a quad- ratic functional form. The asymmetric relationship suggests that yields would be severely affected by climate change, which is predicted to shift the temperature distribution in the United States upward between 5°C and 10°C by the end of the century under the newest Hadley HadCM3 model.3 The harmful effects of more exposure to high temperatures would far outnumber the mitigating effects of less exposure to cooler temperatures. Warmer places are expected to suffer higher relative reductions in yield, as damages are increasing at an accelerating rate. One might wonder whether the estimated relationship differs for warmer climates, as farmers in hotter climates adapt to higher average temperatures. We therefore split the corn sample into two subgroups, one composed of cooler northern counties and the other of warmer southern counties. In the northern subgroup, temperatures exceed 30°C an average of 6.4 days in the growing season.4 In the southern sub- group, temperatures exceed 30°C an average of 26.5 days, more than four times the number for northern counties.5 Figure 2 compares the results. Results for each subgroup are displayed in black, and results for the pooled results from the left panel of figure 1 are added in gray for comparison. The point estimates (solid lines) are comparable across the pooled model and those for each subgroup. The only difference is that the confidence band becomes wider for the southern counties. This might be explained by the fact that we exogenously fix the growing period to be March through August, but it is shorter for the hotter southern counties. This induces noise, as we include temperatures after the corn is harvested that have no impact on yields. However, because weather is ran- dom, it only induces noise, that is, the point estimate remains unchanged, but the confidence interval widens. 198 | WOLFRAM SCHLENKER AND MICHAEL ROBERTS FIGURE 2. Robustness of Results for Corn for Various Geographic Subsets a. Northern counties b. Southern counties Log yield (bushels) Exposure in days Log yield (bushels) Exposure in days 0.02 0.03 30 0.01 0.02 25 30 0 0.01 25 20 0 0.01 20 0.01 0.02 15 0.02 15 0.03 10 0.03 10 0.04 5 0.04 5 0.05 0 0.05 0 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 Temperature (Celsius) Temperature (Celsius) Note: The black line depicts the effect of a one-day exposure to a certain temperature on yearly log yields. The graph is relative to a temperature of 8°C, that is, it depicts the increase/decrease in yields of a one-day exposure to a cer- tain temperature compared to a one-day exposure at 8°C. The 95 percent confidence band is added as dashed lines. The gray histogram shows the number of days a country is exposed to each 1°C interval, on average. Figures 1 and 2 suggest limited potential to adapt to higher temperatures as coun- ties in warmer regions show the same sensitivity to these extreme temperatures as counties in cooler regions. Furthermore, the stability of the estimates across regions suggests that the estimated relationship is fundamental to plant physiology and hence might be applied to other regions of the world, even though we have only estimated it for the United States. Probably the largest caveat to our analysis is that it relies on a panel of past obser- vations and hence is incapable of incorporating carbon dioxide (CO2) fertilization. However, this effect is still highly debated (Long and others 2006). Implications of Climate Change for Agriculture in Developing Countries Our results may be important for developing countries for several reasons. First, as noted, agriculture still constitutes a large share of income in these countries. A change in agricultural output translates into a significant change in GDP and might affect the overall economy. Furthermore, a large share of the population is employed in agri- culture, and hence changes in agricultural output also have direct consequences for the labor market. Second, weather is highly spatially correlated, and a heat wave in a given year might affect an entire country. For smaller countries, weather shocks are not FIGURE 3. Time in Days Temperatures Exceed 30°C during the Entire Year Note: Figure displays the total time in days temperatures exceed 30°C during an entire year around the globe. Consecutive boundaries in the legend increase by 50 percent. 199 200 | WOLFRAM SCHLENKER AND MICHAEL ROBERTS idiosyncratic and are difficult to insure against, as they will affect everybody. If cli- mate change shifts the temperature distribution upward and makes heat waves more frequent in a predominantly agricultural society, a sizable fraction of the overall economy will become more susceptible to the same shock. Countrywide shocks will lead to possibly larger year-to-year variation in GDP, which might destabilize a coun- try unless it enters an insurance contract with other countries that have uncorrelated weather. Third, figures 1 and 2 show that increasing temperatures become more and more harmful the hotter a country becomes. This is due to the highly asymmetric relationship--countries that are already warm will be hit the hardest from further temperature increases. While Canada, Northern Europe, and Northern Asia may see yield gains from climate change, developing countries are predominantly located in warmer climates and hence especially vulnerable to further increases in temperatures. To illustrate this point, we derive the total time in which temperature readings exceed 30°C around the globe in a year and display it in figure 3. We obtain all global stations of the National Climatic Data Center of the National Oceanic and Atmos- pheric Administration for the years 1960­2005.6 The data set includes 18,380 loca- tions that have on average 4,876 daily records for the 56-year period (some locations only have records for part of our period or have days with missing values). We fit a Gaussian smoothing spline over all locations for each month of the calendar year and derive the fraction of the day in which temperatures exceed 30°C.7 Using the num- ber of records at a station, our smoothing procedure gives larger weights to stations with a complete history. Since the growing season varies by region, especially between the northern and southern hemisphere, the daily values are summed over all months of a calendar year. The distribution of times above 30°C is shown in figure 3. Consecutive bounds increase by 50 percent. Conclusions We find that the relationship between temperatures and yields is highly nonlinear and asymmetric, where a temperature increase above the optimal growing temperature is much more harmful than a drop of equal magnitude below the optimum growing temperature. This estimated relationship is highly robust across various specification checks and even consistent between cooler and warmer climates. The highly non- linear relationship has strong implications for climate change, which will shift the temperature distribution upward. Due to the sharp asymmetry, the increased fre- quency of harmful temperatures above the optimum growing temperature far out- weigh the benefits of fewer temperatures below the optimum growing temperature. Our findings for the United States are relevant to the developing world for two rea- sons. First, we were able to discern this highly nonlinear and asymmetric relationship using a uniquely rich data set that is not available in similar detail for developing coun- tries. At the same time the estimated relationship is stable over both crops and regions and suggests that similar patterns may be applicable to other parts of the world, including developing countries. We show that developing countries have warmer NONLINEAR EFFECTS OF WEATHER ON U.S. CROP YIELDS | 201 climates to start with and that the decrease in yields is accelerating for warmer temperatures. Additional reductions in yield from further temperature increases would be larger in warm climates compared to similar temperature increases in moderate climates. Second, the United States is a major agricultural producer and exporter, so any effects on the United Sates will likely also have a powerful influence on commodity and food prices. However, as Bhagwati (2005) points out, subsidies and lower food prices "can actually have a beneficial effect if the poorest farmers consume imported food (since the subsidies lower the world prices)." Whether poor countries are net importers or net exporters of food determines whether they benefit from lower world food prices. Finally, how these changes in yields translate into changes in welfare depends on general equilibrium effects. Most general equilibrium models suggest large shifts in growing regions, but not necessarily significant changes in total production (Rosenzweig and Hillel 1998). Thus the large impacts on yield may belie aggregate impacts on welfare. But regardless of how much prices are ultimately affected, our results suggest substantial changes in where things are produced and in how much is produced. In a global economy with integrated commodity and food markets, these changes are of interest to both rich and poor countries. Endnotes 1. See http://www.icasa.net/dssat/. 2. We include all counties in the United States that are east of the 100 degree meridian, that is, that have a longitude larger than 100. 3. http://www.met-office.gov.uk/research/hadleycentre/models/modeldata.html. 4. Northern counties include all counties east of 100 degree meridian in the states of Illinois, Indiana, Iowa, Michigan, Minnesota, New Jersey, New York, North Dakota, Ohio, Pennsylvania, South Dakota, and Wisconsin. 5. Southern counties include all counties east of 100 degree meridian in the states of Alabama, Arkansas, Georgia, Louisiana, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, and Texas. 6. http://www.ncdc.noaa.gov/cgi-bin/res40.pl?page=gsod.html. 7. Both the spatial and temporal coverage is better for Europe, Japan, and the United States. We include this figure to give a rough estimate of the frequency of temperatures above 30°C and make it comparable to our estimates for the United States. References Adams, Richard M., Ronald A. Flemming, Ching-Chang Chang, Bruce A. McCarl, and Cynthia Rosenzweig. 1995. "A Reassessment of the Economic Effects of Global Climate Change on U.S. Agriculture." Climatic Change 30 (2): 147­67. Bhagwati, Jagdish. 2005. "From Seattle to Hong Kong." Foreign Affairs, WTO Special Edition 84 (7). 202 | WOLFRAM SCHLENKER AND MICHAEL ROBERTS Lobell, David B., and Gregory P. Asner. 2003. "Climate and Management Contributions to Recent Trends in U.S. Agricultural Yields." Science 299 (5609): 1032. Long, Stephen P., Elizabeth A. Ainsworth, Andrew D. B. Leakey, Josef Nosberger, and Donald R. Ort. 2006. "Food for Thought: Lower-Than-Expected Crop Yield Simulation with Rising CO2 Concentrations." Science 312 (5782): 1918­21. Mendelsohn, Robert, William D. Nordhaus, and Daigee Shaw. 1994. "The Impact of Global Warming on Agriculture: A Ricardian Analysis." American Economic Review 84 (4, September): 753­71. Rosenzweig, Cynthia, and Daniel Hillel. 1998. Climate Change and the Global Harvest: Potential Impacts of the Greenhouse Effect on Agriculture. New York: Oxford University Press. Schelling, Thomas C. 1997. "The Cost of Combating Global Warming." Foreign Affairs 76 (6): 8­14. Schlenker, Wolfram, W. Michael Hanemann, and Anthony C. Fisher. 2006. "The Impact of Global Warming on U.S. Agriculture: An Econometric Analysis of Optimal Growing Con- ditions." Review of Economics and Statistics 88 (1): 113­25. Schlenker, Wolfram, and Michael J. Roberts. 2006. "Non-linear Effects of Weather on Corn Yields." Review of Agricultural Economics 28 (3): 391­98. Timmins, Christopher. 2006. "Endogenous Land Use and the Ricardian Valuation of Climate Change." Environmental and Resource Economics 33 (1): 119­42. Emerging Policy Issues within the Realm of Science Policy for Africa MONTY JONES Much as scientists may wish otherwise, policies have profound effects on the appli- cation of science. This is especially true of agricultural science policies, where there is near universal interest vested in them by both consumers who want assured low- cost and high-quality food supplies and by the public at-large who are concerned about the accompanying risks to the health and safety of food and the welfare of farm animals and the environment. These different interests present difficulties in formulating policies with the right balance between enabling the application of science and ensuring that it is applied safely and equitably. At the international level, there are divergent priorities between policy makers responsible for relieving the hunger and poverty of the majority of Africans and those responsible for serving well-fed communities who are increasingly concerned with health, safety, and welfare issues. Likewise within the continent, there are divergent views on the priorities and content of policies governing agricultural research, depending on factors such as the available resources and capacities for research. Africa has multiple layers of policy making, from the African Union, through the regional economic communities and national governments, down to the boards of agricultural research institutes. There are other international standard-setting and regional bodies such as the International Standard Setting Organization and institu- tions such as the United Nations Economic Commission for Africa that influence science policies. Policy makers at each level have different responsibilities and points of view, and they are influenced by interest groups, which can cause them to frame policies that have unintended consequences for third parties. In view of their limited resources, African countries have established a range of regional and subregional institutions to promote sharing of tasks and building of critical mass for agricultural research through collaboration; at the same time, they maintain strong national aspirations. In these circumstances, a mix of policies is Monty Jones is Executive Secretary of the Forum for Agricultural Research in Africa (FARA). Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 203 204 | MONTY JONES needed to facilitate international task sharing and cross-border transactions without suppressing national aspirations for acquiring new and additional human and institutional capacities. Bioscience and biotechnology have great potential to help Africa, but many issues are inherently contentious and made more so by the influence of interest groups pursuing limited and partial agendas. For biotechnology to flourish especially with the limited means of African nations, the bioscience and biosafety policies of differ- ent players should be harmonized and consistent with each other. In summary, African agricultural science policies must be framed within the prevailing contexts with the aim of enabling Africa to take best advantage of the potential agricultural science. This paper outlines the agricultural science policy environment and highlights three of the most critical policy issues that must be addressed: that is, the need to (a) harmonize policies, (b) frame policies that promote collaboration between countries and regions and strengthen capacity, and (c) enable Africa to judiciously harness bioscience and biotechnology. The common aim of all agricultural science policies should be the ethical and safe application of science to improve the welfare of African people with appropriate protection for producers, con- sumers, farm animals, and critical global biodiversity and environmental resources. The Agricultural Science Policy Environment Everyone is affected by agricultural science policies, but there are vast differences in the level of awareness and in policy priorities. The foremost priorities of the major- ity of Africans are bound to be policies that will lead to secure supplies and lower costs of food. These goals are increasingly urgent in the face of less predictable weather patterns, with shortening cycles and greater extremes of drought and inun- dation and the ever-present threat of pests and diseases. The biblical locust plagues are, for example, a severe present-day reality for many Africans, and their continued occurrence, as in Niger in 2005, is a clear indictment of policy failures at national and international levels. Agricultural science offers exciting prospects for improving the yields of crops and animals and their ability to withstand drought, salinity, pests, and diseases. There are also good prospects for getting more out of the harvests by reducing post-harvest losses and improving the nutritional quality of grains and tubers. In addition to the publicized famines, millions of Africans endure constant hunger, grinding poverty, and preventable diseases such as kwashiorkor and blindness caused by vitamin A deficiencies for which research has potential solutions. These are important policy objectives, but the application of science has to be tempered with ethical and con- sumer health and food safety protections as well as considerations for the welfare of farm animals and the environment, especially in regard to preserving soil, water resources, and biodiversity. The conflict that this mix of objectives can cause is illustrated by the ban imposed on International Livestock Center for Africa scientists from working with the work oxen of farmers in Mali because, under the prevailing animal welfare policies, the oxen were deemed to be in too poor condition. EMERGING POLICY ISSUES | 205 In this environment, policies have to balance the interests of many different groups. However, that is not easily achieved because the complex issues are often taken up by different parties independently and frequently with more emotion than science, thereby denying policy makers the benefit of impartial, thoroughly researched information and advice on which to base their policies. The outcome of such skewed advice is exemplified by the contrasting policies adopted by different African governments with respect to genetically modified crops, even when they have to contend with urgent demands for food relief. Agricultural science policies are influenced not only by scientific issues: the weight given to different factors can also be swayed by, for example, trade relations and other nonscientific motives, which can lead to very different recommendations. This is exemplified by the difference in American and European policies on genetically modified crops. Such policy differences can have important, often unforeseen, knock- on consequences for third parties. For example, African policies with regard to importing American genetically modified seeds are affected by their wish to export to China, which has import policies that are influenced by policies in European markets. Reconciling the different interests is further complicated by the different levels of policy making that affect African agricultural science. Policies are made at institute, national, subregional, continental, international, and global bodies. It is known that these policies affect each other, but the policy makers are not necessarily aware of how or to what extent. This can have serious negative implications. For example, agreed smallholder rehabilitation programs have been frustrated by restrictions on the import of vitally needed seed because of the failure to reconcile seed certification criteria and processes in neighboring countries. In view of the prevailing dearth of crop breeders, there would be much to be gained from having common policies for seed production and transfer that are consistent with the policies intended to promote collaboration. In summary, in the complex African policy environment, policy makers at different levels must take care to formulate complementary policies that will collectively enable Africa to make the best use of science and to seize the opportunities for and remove the constraints on African agriculture's full potential for improving livelihoods and driving development. Harmonizing Policies The problems caused by lack of coordination and discontinuities are not unique to science policies. The need to resolve such problems was a prime motive for the inten- sive effort to develop the Framework for African Agricultural Productivity (FAAP), which is intended to be an advocacy tool available to all stakeholders in the dissem- ination and adoption of African agricultural technology as well as research. It provides principles and guidelines formulated to improve the complementarity and value-adding outcomes from the contributions of all actors. FAAP promotes policies that put farmers at the center of agricultural innovation systems and empower them to be active players in improving agricultural productivity, 206 | MONTY JONES not just in terms of increasing their yields, but also in making decisions on how pro- grams and policies are shaped. This includes making policy makers and researchers more accountable to the farmers. Developing the modalities for achieving this and get- ting compliance will be neither easy nor quick; it will require persistent commitment by all actors to devise policies that are relevant to the diverse contexts in which they will be applied. For example, with their more advanced and better resources and insti- tutions, South African policy makers are bound to have very different perspectives than their Mozambican counterparts, yet they have many common interests within the Southern African Development Community (SADC). In summary, policy makers' and advisors' compliance with agreed criteria and guidelines as set out in FAAP would go a long way toward harmonizing agricultural science policy in favor of the public good. Policies Related to Capacity Utilization and Strengthening Despite the importance of agriculture in their economies and the evidence of good returns to investment in agricultural research (Haggblade and others 2004), African governments have underinvested in agricultural research. Under the strictures of structural adjustment programs, they also have failed to recruit sufficient young scientists. The resultant skewed age distribution has resulted in "the missing genera- tion of African agricultural scientists" (Ståhl and Hall 2003). African countries face many problems in attempting to close the enormous disparities in new science knowledge, skills, and technologies that currently divide Africa from the rest of the world. This will require policies that encourage investment and the creation of innovative institutional frameworks in which the scientific inves- tigations are undertaken and in which the required human capacity is built. This includes not only capacity in the traditional scientific disciplines but also in the hard and soft systems skills that scientists, and indeed all professionals, require to put their disciplines and contributions in perspective, understand the contributions of other dis- ciplines, negotiate their contributions, and resolve conflicts that arise in the imple- mentation of modern multidisciplinary, multi-institutional, and multistakeholder participatory innovation systems. Among the critical policy issues are intellectual property rights in which there is a tension between efforts to make institutions more self-financing through commercial exploitation of their products and efforts to encourage collaboration, establish criti- cal mass, and share expensive facilities. There is also a need for common agreement on science-based biosafety and regulatory systems as advocated by the African Biotechnology and Biosafety Programme (ABBP) hosted at the Forum for Agricul- tural Research in Africa (FARA) Secretariat. New policies are required for building African scientific and institutional capacity that addresses the full extent of the institutional needs and facilitates the input of all actors in the value chains in the most effective and cost-efficient ways. Implementation of such policies has been facilitated by the convergence of several significant institu- tional evolutions. These include the emergence of a vigorous African Union out of the EMERGING POLICY ISSUES | 207 former Organisation of African Unity, the articulation of the Comprehensive Africa Agriculture Development Programme (CAADP) by the New Partnership for Africa's Development (NEPAD 2003), which is now underpinned by FAAP for CAADP Pillar IV, and similar frameworks for the three other CAADP pillars. The strengthening of the Regional Economic Communities and the emergence of FARA and the subregional organizations--that is, the Association for Strengthening Agricultural Research in East- ern and Central Africa (ASARECA), the West and Central African Council for Agri- cultural Research and Development (CORAF/WECARD), the emerging Centre for Agricultural Research and Development in Southern Africa (CARDESA) to take over from SADC/Food, Agriculture, and Natural Resources (FANR), and the North African Sub-Regional Organisation (NA SRO)--have formed complete chains through which all stakeholders can align their contributions to Africa's agricultural research and devel- opment based on their comparative and collaborative advantages. However, the past underinvestment in agricultural development also affected invest- ment in capacity building, especially in tertiary education. Reports of the Commission for Africa and the Inter Academy Council (Commission for Africa 2005; IAC 2004a, 2004b; World Bank 2007) are among many that have highlighted the urgent need for capacity strengthening. The extent of the underinvestment is somewhat hidden by the dramatic expansion in enrollment of agricultural students. However, that expansion in enrollment has not been accompanied by equivalent increases in funding for faculty staffing, equipment, and facilities, and this has led to the following: · Inadequate teaching and research resources, · Low staff levels and skewed age distribution (few young scientists), · Erosion of talent due to brain drain, exacerbated by HIV/AIDS, · Teaching staff whose efforts are spread too thinly on too many activities, some not related to their mandates, and · Inadequate facilities for practical training. New policies are urgently required to redress the imbalances by promoting invest- ment in upgrading undergraduate and postgraduate teaching and training in their pedagogy and also in the context and content of courses. This must be supported by revised staffing policies that restore age and gender balances that have accumulated under the restricted recruitment imposed by structural adjustment policies. If such policies are not enacted soon, the International Foundation for Science prediction of a "missing generation of African scientists" will be fulfilled with disastrous consequences (Ståhl and Hall 2003). The required policy changes extend to the policies of Africa's development part- ners, who have for too long neglected Africa's need for capacity strengthening despite promoting policies, programs, and projects that are increasingly demanding of Africa's overstretched human and institutional capacity. Even in the limited extent to which they have supported capacity strengthening, it has been mostly in respect of building particular capacities through their own institutions or in relation to staffing the development projects that they are supporting. Where they have 208 | MONTY JONES supported capacity-strengthening initiatives, these have tended to be programs and networks that are supply pushed and focused on particular needs, which cannot give sufficient consideration to the big picture. This has resulted in both duplications and critical gaps in African tertiary education. The development partners whose policies have allowed them to support wider African-driven initiatives should be commended for their significant achievements. Examples of these are the African Network for Agriculture, Agroforestry, and Envi- ronment Education (ANAFE), which has been supported by Sida, and the Regional University Forum for Capacity in Agriculture (RUFORUM), which has been supported by the Rockefeller Foundation. ANAFE is a network of 117 educational institutions in Africa whose objective is to strengthen the teaching of multidisciplinary approaches to land management. It is the largest working network of educational institutions in Africa. Universities outside Africa may apply for associate membership. ANAFE's policy is to promote the institutionalization of integrated education pro- grams in agriculture and natural resources at African colleges and universities. This will enable the member institutions to produce graduates who are capable of devel- oping, disseminating, and implementing agricultural practices suitable for African farmers and to address natural resources management in a holistic manner to conserve the environment. ANAFE achieves these by promoting and supporting integrated approaches to teaching the science and practice of agriculture and natural resources management. ANAFE seeks the following: · To strengthen the content and delivery of tertiary education in agriculture, forestry, and natural resource management, · To establish and nurture linkages among education, research, and extension in order to maintain relevance of their education programs, and · To build a culture of sharing information and expertise among educational and training institutions in Africa. RUFORUM is an initiative by a consortium of 12 universities in East, Central, and Southern Africa to develop and strengthen human resource capacity in interdiscipli- nary problem-solving. It achieves its goal through grants programs to support research and to address rural (agricultural) development issues, especially community and smallholder needs. The consortium serves to accomplish the following: · To strengthen and promote the roles of universities in agricultural development through capacity building and research for development, · To anchor universities within the broader national agricultural innovation systems, and · To help universities to contribute more effectively to attainment of the Millennium Development Goals and Poverty Reduction Strategy Paper strategies of the mem- ber countries. EMERGING POLICY ISSUES | 209 Current constituent universities of RUFORUM include Egerton University, Jomo Kenyatta University of Agriculture and Technology, Kenyatta University, Moi Uni- versity, and University of Nairobi in Kenya; University of Malawi in Malawi; Eduardo Mondlane University in Mozambique; Makerere University in Uganda; Africa University and the University of Zimbabwe in Zimbabwe; Sokoine University of Agriculture in Tanzania; and University of Zambia in Zambia. RUFORUM has its secretariat in Kampala, Uganda. Through a consultative process, RUFORUM has developed a 10-year strategic plan that outlines five strategic goals: 1. Masters and doctoral programs responsive to stakeholder needs and national or regional development goals, 2. Shared research and training facilities and capacities rationalized for enhanced economies of scope and scale, 3. Innovative training, research, and outreach activities supported by adaptive man- agement structures in universities contributing to policy and development practice, 4. Operational capacity and approaches for innovative, quality, and impact-oriented agricultural research for development in mainstream universities, and 5. A dynamic regional platform for policy advocacy, lobbying, coordination, and re- source mobilization for improved training, research, and outreach by universities. Such successful networks, of which there are quite a number (FARA 2006), should be underpinned by enabling policies that will ensure that their initiatives are responsive to demand based on wide consultations. The policies should favor initiatives that are appropriate to the real needs rather than the convenience of the suppliers and that will achieve and maintain high standards across the board, not just in some subjects at some institutions. They should also aim to produce gradu- ates who are equipped to be entrepreneurs and change agents who are job creators rather than job seekers. In summary, Africa requires comprehensive, forward-looking capacity-strengthening policies that will facilitate innovative approaches that take full advantage of modern teaching and learning approaches and encourage the involvement of all actors, African and non-African, that can make complementary value-adding contributions. Policies for Harnessing Bioscience and Biotechnology Safely Opinions vary about the potential and desirability of employing biotechnology to increase yields of Africa's staple crops, but the available empirical evidence favors judicious applications. This was confirmed by African Ministers for Science and Technology who met recently under the auspices of the African Union and the New Partnership for Africa's Development to discuss Africa's Science and Technology Consolidated Plan of Action. The ministers noted that science, technology, and inno- vation are critical to realizing the United Nations' Millennium Development Goals, 210 | MONTY JONES and they plan to mobilize financial resources and increase expenditure on national research and development. In some countries, such as South Africa and Kenya, biotechnology applications are already quite advanced. Ghana and Nigeria are also committed to employing biotechnology to improve their standard of living. This will involve a wide range of biological specializations related to living organisms, including animals, microbes, plants, and trees that are covered by the term biosciences. Of these genomics (the determination of the DNA sequences that make up the "genetic blueprint" of an organism) and bioinformatics (computer-based analyses of the vast genetic informa- tion produced by genomics research) have perhaps the greatest potential for produc- ing technologies for improving African smallholder and pastoral farming systems. The application of novel computer-based techniques is dramatically accelerating the decoding of the genetic composition of organisms. Achievements to date include decoding the simplest forms of viral life and more complex multicellular organisms, including the landmark completion of the mapping of the human genome in 2001. Bioscience also includes techniques for characterizing and manipulating the genes that regulate specific traits, such as pest resistance or stress tolerance, which can be used in selecting improved breeds of livestock and varieties of crops and trees and in developing more accurate diagnostics and improved vaccines. Indeed, a new genera- tion of safer and more affordable vaccines has already been developed for important human diseases such as meningitis and for animal diseases such as rabies. In food and agriculture, applications of biosciences have led to the development of new crop vari- eties with improved tolerance to pests and diseases and better qualities for storage. Africa needs this new science to reduce poverty and create wealth in sustainable and equitable ways with technologies specifically tailored to unique regional, national, and local circumstances. In addition to taking advantage of knowledge and technologies developed out of Africa, there are problems and opportunities that will only be addressed if Africa has its own capacity to acquire knowledge and make new discoveries to fuel endogenous innovation. This will facilitate research on diseases that are killing the crops and live- stock of poor people that are neglected by non-African research institutes. There is a need for better vaccines and diagnostic tests for some of the most devastating diseases of livestock in Africa such as East Coast fever, while other diseases such as the tick- borne diseases are being inadequately controlled by technologies with undesirable side effects, such as environmental pollution. These diseases are neglected by the major pharmaceutical companies because of the African producers' weak purchasing power. The control of such "orphan diseases" requires the development of specific vaccines and diagnostic tests and better use of indigenous breeds that possess genetic resistance to disease. To make progress with African priorities, NEPAD has promoted the establishment of Biosciences eastern and central Africa (BecA) as one in a continent-wide network of four such centers of excellence with the goal of facilitating the generation and use of cutting-edge science and technology by the continent's researchers to enable them to develop products aimed at solving African problems.1 This will enable Eastern and Central African countries to develop and apply bioscience research expertise to EMERGING POLICY ISSUES | 211 produce technologies that help poor farmers to secure their assets, improve their productivity and income, and increase their market opportunities. Strengthening human and institutional capacities is possibly the most important factor requiring enabling policies (Juma and Serageldin 2007). In addition to enabl- ing the provision of appropriate training opportunities, bioscience policies must be designed to enable sharing of facilities with high-end equipment that are properly staffed and have access to the high-security containment facilities, which individual countries cannot afford. These policies must address a wide range of issues including intellectual property rights and the transfer of materials. The establishment agree- ments for NEPAD's BecA facility, hosted by the International Livestock Research Institute in Nairobi, provides a case study in bioscience-enabling policies. As noted, bioscience research at such facilities will be subject to biosafety policies that, to be implemented effectively, must be agreed at the continental level, as envis- aged in the goals of the Bioscience and Biosafety Policy Platform convened by FARA. Policies for employing high-tech methods must incorporate capacity strengthen- ing, without which the facilities will be underused or even misused. An example of how research and teaching institutions can collaborate better, the BecA Web site statements include a commitment to working with universities to train young scien- tists to M.Sc. and Ph.D. levels and to provide opportunities for postdoctoral fellows. BecA is not a degree-awarding institution; rather it enhances the work of university and research institution laboratories. The close association of these facilities with uni- versities will enable academic staff in Africa to further their professional development and careers through fellowships and secondments that allow them to undertake high- priority research for the poor within first-rate facilities and with direct access to inter- national as well as African resources and scientific skills. It is envisaged that BecA will evolve into a service provider of first resort for bio- sciences research for the poor, while becoming an integral component of the academic and research establishments of Africa. The policy of establishing such centers of excellence will motivate and equip African scientists to reach their full potential within Africa and encourage some of Africa's "best and brightest" scientists to work directly on Africa's development in Africa. Too few established regional collaborations have managed to break their depend- ency on donor funding, and this is a serious handicap in long-term endeavors, especially in view of the inherent competition between African and donor-country institutions. In summary, developing the capacities required for exploiting the potential of biotechnology on a continental scale cannot be addressed in isolation. Good labo- ratories and biotechnologists emerge from systems that provide sound overall education as the foundation for specialization in later stages. Likewise, centers of excellence cannot be established by fiat nor can they exist as intellectual islands. Comprehensive policies are required to be developed within the larger ongoing dialog on enabling Africa and African nations to compete in knowledge economies in which biotechnology is part. For example, funding for building capacity in biotechnology cannot be considered in isolation from the other needs of African agricultural capacity building. 212 | MONTY JONES Policies for Enabling Innovation Research for resource-poor farmers will continue to depend on public investment because the market for smallholder innovations is too small to attract private investors. However, due to disappointment with the impact of agricultural research, there has, until recently, been a drastic decline in public funding from both national and international sources. The attempt by some external donors and national governments to use alternatives such as nongovernmental organizations as an excuse to abandon, rather than reshape, the public sector contribution to agricultural devel- opment was ill advised and needs to be redressed. Rather than reduce funding, and thereby compound the problems, such as high ratios of staff to operating costs, there is need for a more systemic approach to analyses of agricultural research and exten- sion with a view to better integrating the contributions of a wide range of actors. The concept of an agricultural innovation system encompasses a network of inter- institutional linkages that are built up as a support system for economic production across the board (Clark 2001: 40). In this sense, economic creativity is actually driven by the quality of "technology linkages" and "knowledge flows" among and between economic agents. Where the interactions are dynamic and progressive, great innova- tive strides are often made. Conversely, where systemic components are compart- mentalized and isolated from one another, the relevant research bodies often are not very productive. The key property of an agricultural innovation system is therefore not so much its component parts, or nodes, but rather its performance as a dynamic whole. Even if the individual elements of such a system are strong, the system as a whole may be weak. The capability to learn and build new competencies will depend on how well the parts fit together and on the strength of these connections. Such structures will not come about by chance. There will have to be deliberate policy initiatives backed up by redirected and new human and financial resources. The existing hierarchical structures of agricultural research and extension organiza- tions will have to be opened up and decentralized. Policies are required to provide incentives for both organizations and individuals to engage in multidisciplinary, multi-institutional, and multistakeholder systems of innovation. Conclusions In the complex African policy environment, policy makers at different levels need to formulate complementary policies that will collectively enable Africa to make the best use of science and seize the opportunities for and remove the constraints on African agriculture's full potential for improving livelihoods and driving develop- ment. The common aim of all agricultural science policies should be the ethical and safe application of science to improve the welfare of African people, with appropri- ate protection for producers, consumers, farm animals, and critical global biodiver- sity and environmental resources. EMERGING POLICY ISSUES | 213 Compliance by policy makers and advisors with agreed criteria and guidelines as set out in FAAP would go a long way toward harmonizing agricultural science policy in favor of the public good. To take full advantage of science in agriculture, Africa requires comprehensive forward-looking capacity-strengthening policies that can facilitate new and innova- tive approaches that take full advantage of modern teaching and learning approaches and encourage the involvement of all actors, African and non-African, that can make complementary value-adding contributions. Developing the capacities required for exploiting the potential of biotechnology on a continental scale cannot be addressed in isolation. Good laboratories and biotechnologists emerge from systems that provide sound overall education as the foundation for specialization in later stages. Likewise, centers of excellence cannot be established by fiat, nor can they exist as intellectual islands. Comprehensive policies need to be developed within the larger ongoing dialog on enabling Africa and African nations to compete in knowledge economies in which biotechnology is part. For example, funding for building capacity in biotechnology should be considered along with the other needs of African agricultural capacity building. The science policies required will not come about by chance. There will have to be deliberate policies for directing new human and financial resources. To promote innovation, the existing hierarchical structures of agricultural research and extension organizations should be opened up and decentralized. Policies are needed to provide incentives for both organizations and individuals to engage in multidisciplinary, multi-institutional, and multistakeholder systems of innovation. Endnote 1. See www.africabiosciences.org. References Clark, Norman. 2001. Innovation Systems, Institutional Change, and the New Knowledge Market: Implications for Third World Agricultural Development. UNU-INTECH Discussion Paper 2001-10. United Nations University, Institute for New Technologies, Maastricht. Commission for Africa. 2005. Our Common Interest: Report for the Commission for Africa. London: Commission for Africa. (www.commissionforafrica.org.) FARA (Forum for Agricultural Research in Africa). 2006. "Proceedings of a Meeting of African Networks and Associations That Build Agricultural Capacity at Universities." ICRAF House, Nairobi, November 23­25, 2005. Haggblade, Steven, Peter Hazell, Ingrid Kirsten, and Richard Mkandawire. 2004. "African Agriculture; Past Performance, Future Imperatives." In Building on Successes in African Agriculture, ed. Steven Haggblade, chap. 1. 2020 Vision Focus 12. Washington, DC: International Food Policy Research Institute. IAC (Inter Academy Council). 2004a. Inventing a Better Future. Amsterdam: IAC, Royal Netherlands Academy of Arts and Sciences. 214 | MONTY JONES ------. 2004b. Realizing the Promise and Potential of African Agriculture. Amsterdam: IAC, Royal Netherlands Academy of Arts and Sciences. Juma, Calestous, and Ismail Serageldin. 2007. Freedom to Innovate: Biotechnology in Africa's Development; Report of the High-Level African Panel on Modern Biotechnology. Addis Ababa: African Union (AU); Pretoria: New Partnership for Africa's Development (NEPAD). NEPAD (New Partnership for Africa's Development). 2003. "Comprehensive Africa Agricul- ture Development Programme." NEPAD, Midrand, South Africa. Ståhl, Michael, and Richard Hall. 2003. "Is There a Missing Generation of Scientists in Africa?" Paper presented at the second FARA General Assembly, Dakar, May 18­19. World Bank. 2007. Cultivating Knowledge and Skills to Grow African Agriculture: A Synthesis of an Institutional, Regional, and International Review. Washington, DC: World Bank. Closing Remarks ALAIN DE JANVRY Let me start by thanking all of you for this extraordinary experience. We look at the World Development Report (WDR) as a collective good. We all know that this is not a document that can be developed by a team somewhere on the third floor of the main complex of the World Bank. It is to be developed through broad consultations with experts like you so that it is as much as possible a consensus document for the profession. So the conjunction of ideas that we have seen here, and that we hope to pursue in subsequent interactions, is part of the process that we would like to follow. We would like to internalize as much as possible the ideas and suggestions that come from a broad group of academics and practitioners. We would like for you to keep on sharing your ideas, comments, and suggestions with us. However, the objective of the WDR is not to be exhaustive of all subjects, because the world is very big and the issues are very diverse. We cannot pretend to provide guidelines on how to use agri- culture for development in every part of the world. The main messages are the key to usefulness of the report. In a sense, in two days of interactions here, the agenda has been growing larger. What we would like to see now is the agenda getting smaller, more concentrated, and more focused on what are to become the main messages. This is quite difficult, and we have not reached that stage yet. We are still adding to the list. In a sense, what we need to do now is to take the list and see what really matters and decide what we want to stress as essential in using agriculture more effectively for development than has been done in the last 25 years. Let me highlight a few lessons that we have learned in this meeting, which I believe are quite important. I will try to be brief, but the subject is big and multidimensional and the messages are complex. We began with a typology, an empirical approach, and then found three types of countries: what we call the "agriculture-based" economies where agriculture is the engine of growth, the "transforming" economies where growth has accelerated and Alain de Janvry is Professor of Agricultural and Resource Economics at the University of California, Berkeley. Berlin Workshop Series 2008 © 2008 The International Bank for Reconstruction and Development/The World Bank 215 216 | ALAIN DE JANVRY which have large income disparities between rural and urban sectors, and the "urbanized" economies where there is an issue of social inclusion of the rural population in the growth model pursued. I was impressed with the discussion that started on the first evening when François Bourguignon presented us with three the- oretical models of agriculture for development and the correspondence that exists between these models and the country typology we are proposing to use. First we have the agriculture-based economies. This is the classical way of looking at agriculture in terms of win-win: agricultural growth spills over to the rest of the economy, creating an important instrument for mass poverty reduction. This sort of agricultural demand­led industrialization applies very well to the agriculture-based countries. These are mainly African countries where agriculture is key, although glob- alization is changing the classical world of agriculture toward a more open economy. This classical world needs to be reconceptualized in the new context where it applies, and agriculture has to be capitalized if it is to become the main engine of economic growth and poverty reduction. Second we have the transforming economies, in which there is an urgent need to transform the economy to reduce rural poverty, intersectoral disparities, and the political tensions these disparities create. Agriculture is no longer the main engine of growth, and, dollar for dollar, you may be better off putting your money into high- productivity sectors, such as the information technology sector in India or the man- ufacturing sector in China. However, for poverty reduction, agriculture is still the key sector. The high political demand to address rising sectoral disparities will eventually require us to trade off points of growth for points of poverty reduction, which is what explains the outcome of the last elections in India. In China today, you see a lot of social action at the local level because of political demands for a greater sharing of benefits with the rural sector and a fiscal adjustment to relieve the rural sector of the heavy tax burden imposed by local authorities. So we may be in a second-best world in which we are willing to trade off points of growth for more poverty reduc- tion, for which agriculture is the most effective instrument. Third we have the urbanized economies, where investing in agriculture is purely a matter of competitive advantage, and competitive advantage does exist in many sec- tors such as biofuels and cereals in the context of higher prices. Certainly soybeans, cotton, and corn have been successfully grown and exported in Latin America. The key for poverty reduction is, in a sense, to reconcile these comparative advantages with the competitiveness of smallholders and to create more and better jobs in the larger farms and in the rural nonfarm economy driven by agricultural growth. It is hence a question of social inclusion in which agricultural growth in sectors with com- parative advantage help to reduce poverty. These three theoretical perspectives, hence, match fairly well our empirically derived typology. François Bourguignon started from theory, we started from data, and in the end we found a coincidence that works relatively well at identifying arche- types that are empirically defined by theoretical arguments. We need to add to this typology the situation of countries that have not performed up to their agricultural potential because the right policies have not been put into place. There is potential, but the potential has not been captured. The typology of the CLOSING REMARKS | 217 three worlds of agriculture for development presented above is based on past per- formance. Countries where past performance does not reflect potential performance form a category apart. This includes countries like Ethiopia and Mali, which have had relatively weak performance in agriculture and yet have the capacity to mobilize agriculture as an engine of growth under the right policy conditions. Fifth are the countries that have had poor performance, but that have very weak resources in agriculture and hence low potential to do better. In a sense, these coun- tries are barely part of what this WDR should be considering. They have weak resources in agriculture, and so we should be looking for other sources of growth. At the same time, there are typically very large subsistence economies, and this raises the issue of poverty reduction through enhanced land and labor productivity in subsis- tence agriculture. We need to be aware of this fifth category. Finally, we need to go beyond cross-country comparisons and look at regional dif- ferences within countries. The better endowed among lagging areas are very impor- tant to our discussion. Why should we invest in these marginal areas? What type of investments should we seek? There may be an opportunity to invest in biofuels. There may be an opportunity to invest in the provision of environmental services. Some of these areas are marginal because they have been left behind in infrastructure invest- ments. Linking them to markets can help them to attract private investment and offer a source of growth. Going beyond countries and areas within countries, we also need a typology of rural households. Many impoverished rural people are highly dependent on the labor market and on rural nonfarm activities. We find more of these households now than the World Development Report did in 1982, which described smallholder farming as being the main pathway out of poverty for rural households. In fact, data show that agriculture is very rarely the only source of income for farming households, and we always have to look at ways of addressing the income status of poor households by taking into account the combination of opportunities available to them. In addition, we need to take into account the larger commercial sector. Certain farmers are more capable and linked to the market, while others are largely dependent on crops for subsistence. Frequently, successful family farms are able to compete in the market- place precisely because they are linked to the larger commercial farmers. And here we make a frequent mistake when we target social welfare interventions as though the commercial sector did not exist. We tend to target for separation, but we should do the opposite. We need to make sure that there is no separation between our target population and the others because often the solution is for them to develop alliances as opposed to remaining separated. Let me bring up a few additional issues that were mentioned briefly, but not dis- cussed in depth. The first one is political economy. We cannot think of policies with- out thinking of political economy. It is pointless to do policy without taking political motivations into account. We should recall that the 1982 WDR on agriculture had many useful recommendations that have not been implemented because they did not fit in terms of political economy. If the 2008 WDR is to have more impact, recom- mendations have to be compatible with political economy forces. We will hold a workshop in order to think about how to address this issue. 218 | ALAIN DE JANVRY The second is the role of the state. Clearly there is a need to redefine the role of the state in agriculture. It is important to understand the ways in which agriculture depends on the state to overcome market failures. There is a need to reconsider the role of the state in agriculture and the importance of, in that sense, the quality of gov- ernance, the potential for public-private partnerships, and the importance of political capital in influencing state decisions. Our sense is that there is a great deficit of gov- ernance for agriculture, starting at the level of ministries of agriculture and extending to local governments. Improving governance is thus key to using agriculture more effectively for development. The third issue concerns models that simulate the impact of policy reforms such as trade liberalization on social welfare. These models, constructed at high levels of social aggregation, may make us misunderstand important effects that are evident at a deeper level. So it is quite important to go beyond the aggregate effects provided by these models and conduct much more detailed studies that capture the incidence of policies at the more disaggregated level of categories of households. Fundamental in that sense is the contrast between net sellers and net buyers among farm households. We tend to overlook the fact that many more poor farm households are net buyers than net sellers. Under this condition, a rise in the price of food will be detrimental to the welfare of a majority of these households. One of the big changes with globalization is the importance of tradability. In the 1982 report, the main mechanism at work in linking agricultural research to poverty reduction was the impact on the local price of food. It is quite possible that the largest benefit for poverty reduction was achieved indirectly through cheaper food. With greater tradability of food, we need to put more emphasis on rural labor markets in agriculture and the rural nonfarm economy. The work that our profession has done on labor markets is very weak because we focus much more on the farm than on the wider context. For most rural households, there is a link between farm-based activi- ties and the labor market. Do rural labor markets truly provide good opportunities for bringing households out of poverty through upward mobility or more work opportunities? It depends on labor skills and location relative to employment oppor- tunities. It also depends on the availability of local employment opportunities, which depends on territorial development, an issue that needs to be addressed. I was quite taken by Johan Swinnen's presentation on high-value crop exports from Senegal. In a sense, he said that perhaps we should be more optimistic about the ability of both the family farm and the labor market to offer benefits from high- value crop exports. We need to look at this quite carefully in order to identify the conditions under which the high-value revolution in agriculture can be strongly poverty reducing. Quite a bit of the work has been done looking at supermarkets and the ability of family farms to participate in those markets via producer organizations. There are success stories, such as in Senegal, where not only smallholder farming but also the agricultural labor market have been able to make inroads into poverty. We need to look back at vertical contracts and the role of organizations and look for ways to support them. The sixth point is the importance of industrial organization. There have been mas- sive changes in the food chains, in the way consumers access food and producers CLOSING REMARKS | 219 access consumers. In the past, we did not look in great detail at the issues of industrial organization, namely, the structure, conduct, and performance of the food sector, at how margins are set, how prices are bargained, how producer organizations negotiate better terms and access to markets. These are important issues. We need to take a better look at contracts. We need to take a better look at the content of contracts and how they reward those who participate, especially on the supply side and especially for producers. There has also been a rapid process of concentration in agro-industry and the food retail sector, and this is bound to have an impact on prices received by farmers. Developing countervailing power through more effective producer organizations is an important instrument for smallholders to gain a larger share of value chain margins. On the issue of natural resource management, we need to look more carefully at how to internalize externalities. Quite a bit of work has been done on win-win situ- ations at the farm level, but we need to elevate these considerations to include broader institutional arrangements whereby externalities can be internalized with the creation of new markets, new sources of revenue for agriculture, and agriculture as a provider of environmental services. We need to look at the multifunctionality of agriculture and understand how poor people relate to the provision of these services. On the linkages between agriculture and other sectors, we posed the question: Are there opportunities to make these linkages more effective, and what kinds of programs and investments are we looking at to achieve this result? We have larger multipliers and larger spillover effects from agriculture. There obviously are links to the issue of territoriality, links to the issue of more integrated local economies with the ability to capitalize on local systems of production that contribute higher value added. We need to think about how to design projects that enhance the value of linkages, that are tar- geted to regional economies, and that create new employment and new investment opportunities. In that sense, much progress has been made at the community level in terms of local economies, local public goods, and production with the participation of local communities. This has been pursued through the community-driven development approach. But we need to go beyond that level to broader territorial approaches that can support new economic activities, for instance through clusters of firms in particular areas of competitive advantage. We have a very big issue with agricultural poverty in Africa. We are not looking for a silver bullet. The key for Africa will be to provide a new, broader context in terms of governance, higher agricultural productivity, and new spaces for regional trade. The mid-1990s were a turning point for Africa in terms of value added per capita. The idea for Africa is to build on this success and do something different, per- haps with more local participation, new technology that is better adapted to local situations, and new institutions for finance, producer organizations, and local gover- nance. These changes are happening, but they need to gather momentum and be scaled up to have an impact at the aggregate level and make a visible dent in poverty. We need to start by explaining the policy errors of the past. We need to explain why the agenda proposed in the 1982 report has not been sustained. Clearly, agri- culture was badly neglected by governments and international development agencies across the board over the last 25 years. We need to explain what went wrong and 220 | ALAIN DE JANVRY learn from the past. We need to achieve this understanding in order to put forward a new agenda that has a chance of being implemented. What we need to do in the WDR is not go into the details of how to use agricul- ture for development for every particular situation. This will require the subsequent development of local agendas, in an informed and participatory fashion. What we can do is look at the broader picture, reconcile new information with the old per- spective, and create new scenarios based on this information. We need to stop talk- ing about business as usual and begin to think creatively. If we do so, the WDR can help to make a difference in elevating again the role that agriculture can play in devel- opment. Thank you so much for your extraordinary contributions. We look forward to continuing to interact with you. Appendix 1: Program International Bank for Reconstruction and Development The World Bank Development Policy Forum InWEnt--Capacity Building International, Germany* Workshop AGRICULTURE AND DEVELOPMENT in preparation of the WORLD DEVELOPMENT REPORT 2008 Grand Hyatt Hotel, Berlin 4­6 September 2006 PROGRAM In cooperation with the German Federal Ministry for Economic Cooperation and Development (BMZ) *InWEnt (Capacity Building International) was established in 2002 through a merger of the German Foundation for International Development and Carl Duisberg Gesellschaft. 224 | APPENDIX 1: PROGRAM Monday, 4 September 2006 06:15 P.M. Opening Address: Gudrun Kochendörfer-Lucius Managing Director, InWEnt--Capacity Building International, Germany Keynote Address: François Bourguignon Chief Economist and Senior Vice President, Development Economics, The World Bank, United States 07:30 P.M. Dinner Tuesday, 5 September 2006 09:00 A.M. Opening Address: Astrid Kühl Director, Development Policy Forum, InWEnt--Capacity Building International, Germany Introductory Statement: Michael Hofmann Director General, Global and Sectoral Tasks; European and Multilat- eral Development Policy; Africa; Middle East, Federal Ministry for Economic Cooperation and Development (BMZ), Germany Introductory Address: Derek Byerlee Director, World Development Report 2008, The World Bank, United States 09:30 A.M. Session I: Agricultural Policy, Subsidies, and Trade Chair and Speaker: Alain de Janvry Director, World Development Report 2008, The World Bank, United States Speakers: Michael Brüntrup Agricultural Economist, Department II Competitiveness and Social Development, German Development Institute (DIE), Germany Richard Moberly Team Leader, Renewable Natural Resources and Agriculture, Department for International Development (DFID), United Kingdom 11:00 A.M. Coffee Break APPENDIX 1: PROGRAM | 225 11:15 A.M. Session II: Markets and Institutions Chair: Eva Terberger Professor, Department of Economics, University of Heidelberg, Germany Speakers: Benoît Daviron Researcher, French Agricultural Research Centre for International Development (CIRAD), France Johan F. M. Swinnen Professor of Development Economics and Director, LICOS Center for Transition Economics, University of Leuven, Belgium 01:00 P.M. Lunch Buffet 02:15 P.M. Keynote Address: Klaus Töpfer Former Executive Director of UNEP, Professor, UNEP-Tongji Institute of Environment and Sustainable Development (IESD), Tongji University, China 02:30 P.M. Session III: Agriculture, Natural Resources Management, and the Environment Chair: Derek Byerlee Director, World Development Report 2008, The World Bank, United States Speakers: Ade Freeman Director, Targeting R&D Opportunities Theme, International Livestock Research Institute, Nairobi Gunnar Köhlin Assistant Professor, Environmental Economics Unit, Department of Economics, Göteborg University, Sweden Laurence Tubiana Professor at Sciences Po Paris and Director, Institute of Sustainable Development and International Relations (IDDRI), France 04:00 P.M. Coffee Break 04:30 P.M. Session IV: Agriculture-Rural Nonfarm Linkages and Rural Poverty Chair: Boris Pleskovic Research Manager, Development Economics, The World Bank, United States Speakers: Raghav Gaiha Professor, Faculty of Management Studies, University of Delhi, India 226 | APPENDIX 1: PROGRAM Richard Moberly Team Leader, Renewable Natural Resources and Agriculture, Department for International Development (DFID), United Kingdom Alexander Schejtman Head Researcher, Latin-American Centre for Rural Development (RIMISP), Santiago de Chile 06:30 P.M. End of session 07:00 P.M. Dinner/Reception during Boat Trip Wednesday, 6 September 2006 08:30 A.M. Session V: Decentralization, Local Governance, and Rural Development Chair: Christoph Kessler Head of Agriculture & Natural Resources Division, Africa KfW Banking Group, Germany Speakers: Jean-Louis Arcand Professor of Economics, Center of International Development Studies and Research (CERDI), University of Auvergne, France John Farrington Research Fellow, Overseas Development Institute (ODI), United Kingdom John A. Okidi Executive Director and Chief Researcher, Economic Policy Research Centre (EPRC), Makerere University, Uganda Martin Petrick Post-doctoral Researcher, Leibniz Institute of Agricultural Development in Central and Eastern Europe (IAMO), Germany 10:30 A.M. Coffee Break 10:45 A.M. Session VI: Emerging Policy Issues Chair: Ingrid-Gabriela Hoven Deputy Director General, Global and Sectoral Policies, Ministry for Economic Cooperation and Development (BMZ), Germany Speakers: Energy Costs Oliver Henniges Research Associate, Institute for Agricultural Business Administration, Faculty of Agricultural Sciences, University of Hohenheim, Germany APPENDIX 1: PROGRAM | 227 Climate Change Wolfram Schlenker Assistant Professor, Department of Economics, School of Interna- tional and Public Affairs, Columbia University, United States MDGs, Nutrition, and Africa Michiel Keyzer Professor of Mathematical Economics, Director, Centre for World Food Studies, Vrije Universiteit, Amsterdam 12.30 P.M. Closing: Derek Byerlee and Alain de Janvry Directors, World Development Report 2008, The World Bank, United States Closing words: Astrid Kühl Director, Development Policy Forum, InWEnt--Capacity Building International, Germany 01:00 P.M. Farewell Lunch Buffet Appendix 2: Participants 230 | APPENDIX 2: PARTICIPANTS Jean-Louis Arcand Benoît Daviron Professor of Economics Economist Center of International Development French Agricultural Research Centre Studies and Research (CERDI) for International Development University of Auvergne (CIRAD) 65, Boulevard François Mitterand TA 60/15, 73 rue J.F. Breton 63000 Clermont Ferrand 34398 Montepellier Cedex 5 France France Tel.: 0033-4-73177 406 Tel.: 0033-4-6745 7832 Fax: 0033-4-73177 428 Fax: 0033-4-6761 4415 E-mail: arcandjl@alum.mit.edu E-mail: benoit.daviron@cirad.fr François J. Bourguignon Chief Economist and Senior John Farrington Vice President Research Fellow Development Economics Overseas Development Institute (ODI) The World Bank 111 Westminster Bridge Rd 1818 H Street, N.W. SE1 7JD London MSN MC4-404 United Kingdom 20433 Washington, D.C. Tel.: 0044-20-7922 0300 USA Fax: 0044-20-7922 0399 Tel.: 001-202-473 3774 E-mail: j.farrington@odi.org.uk Fax: 001-202-5220 906 E-mail: fbourguignon@worldbank.org H. Ade Freeman Director Michael Brüntrup Targeting R & D Opportunities Agricultural Economist Theme Department II Competitiveness and International Livestock Research Social Development Institute (ILRI) German Development Institute (DIE) P.O. Box 30709 Tulpenfeld 4 00100 GPO Nairobi 53113 Bonn Kenya Germany Tel.: 00254-020 4223000 Tel.: 0049-228-94927-164 Fax: 00254-020 4223001 Fax: 0049-228-94927-130 E-mail: a.freeman@cgiar.org E-mail: michael.bruentrup@die-gdi.de Derek Byerlee Raghav Gaiha Director WDR 2008 Professor of Public Policy The World Bank Faculty of Management Studies 1818 H Street NW University of Delhi 20433 Washington, D.C. 110007 Delhi USA India Tel.: 001-202-458 7287 Tel.: 0091-11-2610 8293 Fax: 001-202-522 0308 Fax: 0091-11-2309 3535 E-mail: Dbyerlee@worldbank.org E-mail: rdg@bol.net APPENDIX 2: PARTICIPANTS | 231 Oliver Henniges Agricultural and Resource Economics Research Associate University of California at Berkeley Department of Agricultural 207 Giannini Hall Economics CA 94720-3310 Berkeley Institute for Farm Management USA University of Hohenheim Tel.: 001-510-642 3348 Institute 410b Fax: 001-510-643 8911 Schloß Osthof-Süd E-mail: alain@are.berkeley.edu 70599 Stuttgart Germany Christoph Kessler Tel.: 0049-711-459 2557 Division Chief Fax: 0049-711-459 3709 Agriculture & Natural Resources E-mail: henniges@uni-hohenheim.de Division, Africa Michael Hofmann KfW Banking Group Director General Palmengartenstraße 5-9 Global and Sectoral Tasks, European 60325 Frankfurt/Main and Multilateral Development Germany Policy, Africa, Middle East Tel.: 0049-69-7431 2895 Federal Ministry for Economic Fax: 0049-69-7431 3559 Cooperation and Development E-mail: christoph.kessler@kfw.de (BMZ) Stresemannstrasse 94 Michiel Keyzer 10963 Berlin Professor of Mathematical Economics Germany Director Tel.: 0049-1888-535 28 00 Centre for World Food Studies Fax: 0049-1888-535 48 00 Vrije Universiteit E-mail: De Boelelaan 1105 michael.hofmann@bmz.bund.de 1081 HV Amsterdam The Netherlands Ingrid-Gabriela Hoven Tel.: 0031-20-598 93 21 Deputy Director General Fax: 0031-20-598 93 25 Global and Sectoral Policies E-mail: m.a.keyzer@sow.vu.nl Federal Ministry for Economic Cooperation and Development (BMZ) Gudrun Kochendörfer-Lucius Friedrich-Ebert-Allee 40 Managing Director 53113 Bonn InWEnt--Capacity Building Germany International, Germany Tel.: 0049-228-535 3190 Friedrich-Ebert-Allee 40 Fax: 0049-228-535-3475 53113 Bonn E-mail: hoveni@bmz.bund.de Germany Tel.: 0049-228-4460 1522 Alain de Janvry Fax: 0049-228-4460 1529 Co-Director WDR 2008 E-mail: Professor gudrun.kochendoerfer@inwent.org 232 | APPENDIX 2: PARTICIPANTS Gunnar Köhlin Department for International Assistant Professor Development (DFID) Department of Economics 1 Palace Street Göteborg University SW1E 5HE London Vasagatan 1 United Kingdom Box 640 Tel.: 0044-20-7023 1530 SE 405 30 Göteborg Fax: 0044-20-7023 0719 Sweden E-mail: R-Moberly@dfid.gov.uk Tel.: 0046-31-773 4426 Fax: 0046-31-773 1043 John A. Okidi E-mail: Executive Director gunnar.kohlin@economics.gu.se Economic Policy Research Centre (EPRC) Christoph Kohlmeyer 51 Pool Road Head of Division P.O. Box 7841 Rural Development; Global Kampala Food Security Uganda Federal Ministry for Economic Tel.: 00256-41-541 234 Cooperation and Development Fax: 00256-41-541 022 (BMZ) E-mail: okidi@eprc.or.ug Adenauerallee 139-141 53113 Bonn Temitope Waheed Oshikoya Germany Division Manager Tel.: 0049-1888-535 3735 Networking and Research Fax: 0049-1888-105 353 735 Partnerships Division E-mail: African Development Bank Christoph.Kohlmeyer@bmz.bund.de B.P. 323 1002 Tunis-Belvedère Astrid Kühl Tunisia Director Tel.: 00216-71-10 2123 Development Policy Forum Fax: 00216-71-10 3779 InWEnt--Capacity Building E-mail: w.oshikoya@afdb.org International, Germany Stresemannstr. 92 Martin Petrick 10963 Berlin Post-doctoral Researcher Germany Leibniz Institute of Agricultural Tel.: 0049-30-43 996 311 Development in Central and Fax: 0049-30-43 996 250 Eastern Europe (IAMO) E-mail: astrid.kuehl@inwent.org Theodor-Lieser-Str. 2 6120 Halle (Saale) Richard Moberly Germany Team Leader Tel.: 0049-345-2928 120 Renewable Natural Resources and Fax: 0049-345-2928 199 Agriculture Team E-mail: petrick@iamo.de APPENDIX 2: PARTICIPANTS | 233 Boris Pleskovic Tel.: 0032-16-3268 57 Research Manager Fax: 0032-16-3267 96 Development Economics E-mail: Jo.Swinnen@econ.kuleuven.be The World Bank 1818 H Street NW Eva Terberger 20433 Washington, D.C. Department of Economics USA University of Heidelberg Tel.: 001-202-473 1062 Grabengasse 14 Fax: 001-202-522 0304 69117 Heidelberg E-mail: bpleskovic@worldbank.org Germany Tel.: 0049-6221-543174 Alexander Schejtman Fax: 0049-6221-543113 Head Researcher E-mail: Latin-American Centre for Rural eva.terberger@awi.uni-heidelberg.de Development (RIMISP) Huelén 10, piso 6, Providencia Klaus Töpfer Santiago de Chile Former Executive Director of UNEP Chile Advisor for Environmental Matters Tel.: 0056-2-2364 557 to the City of Shanghai Fax: 0056-2-2364 558 Professor for Sustainable E-mail: aschejtman@rimisp.org Development Wolfram Schlenker UNEP-Tongji Institute of Environment Assistant Professor and Sustainable Development Department of Economics (IESD) School of International and Tongji University Public Affairs 1239 Siping road Columbia University 200092 Shanghai 420 West 118th Street, MC 3323 People's Republic of China 10027 New York, NY Tel.: 0086-21-6598-8406 USA Fax: 0086-21-6598-8406 Tel.: 001-212-854-1806 E-mail: klaus.toepfer@gmail.com Fax: 001-212-854-5765 E-mail: ws2162@columbia.edu Laurence Tubiana Professor at Sciences Po Paris Johan F. M. Swinnen Director Professor of Development Economics Institute of Sustainable Development Director and International Relations LICOS Center for Transition (IDDRI) Economics 6, Rue du Général Clergerie University of Leuven 75116 Paris Hogenheuvelcollege France Naamsestraat 69 Tel.: 0033-1-5370 2210 3000 Leuven Fax: 0033-1-5370 2145 Belgium E-mail: laurence.tubiana@iddri.org ECO-AUDIT Environmental Benefits Statement The World Bank is committed to preserving Saved: endangered forests and natural resources. · 11 trees The Office of the Publisher has chosen to · 528 lb. of solid print Berlin Workshop Series 2008: Agri- waste culture and Development on recycled · 4,108 gal. of paper with 30 percent post-consumer fiber in waste water accordance with the recommended standards · 990 lb. of net for paper usage set by the Green Press Initia- greenhouse gases tive, a nonprofit program supporting publish- · 8 million Btu of ers in using fiber that is not sourced from total energy endangered forests. For more information, visit www.greenpressinitiative.org. T he Berlin Workshop Series 2008 presents selected papers from meetings held September 4­6, 2006, at the ninth annual forum co-hosted by InWEnt and the World Bank in preparation for the Bank's World Development Report. At the 2006 meetings, key researchers and policy makers from Europe, the United States, and developing countries met to identify and brainstorm on agriculture and development challenges and successes that are later examined in-depth in the World Development Report 2008. This volume presents papers from the Berlin Workshop sessions on issues relating to agricul- tural policy, subsidies, and trade; markets and institutions; agriculture, natural resources manage- ment, and the environment; agriculture-rural nonfarm linkages and rural poverty; decentraliza- tion, local governance, and rural development; and emerging policy issues. IN THIS VOLUME: Introduction by Gudrun Kochendörfer-Lucius and Boris Pleskovic; a welcoming address by Gudrun Kochendörfer-Lucius; a keynote address by François Bourguignon; an opening address by Astrid Kühl; introductory remarks by Michael Hofmann; papers by Michael Brüntrup, Benoît Daviron and Isabelle Vagneron, Johan F. M. Swinnen and Miet Maertens, H. Ade Freeman, Gunnar Köhlin, Laurence Tubiana and Sheila Wertz-Kanounnikoff, Raghav Gaiha and Katsushi Imai, Daniel Bradley, Alexander Schejtman, Jean-Louis Arcand, John Farrington, John A. Okidi and Madina Guloba, Martin Petrick and Andreas Gramzow, Oliver Henniges, Wolfram Schlenker and Michael Roberts, and Monty Jones; and closing remarks by Alain de Janvry. ISBN 978-0-8213-7127-5 SKU 17127