36498 Report No. 36498-GLB Stakeholders, Power Relations, and Policy Dialogue: Social Analysis in Agriculture Sector Poverty and Social Impact Analysis November 2006 Social Development Department Sustainable Development Network Social Analysis and Policy Team The World Bank ©2006 The International Bank for Reconstruction and Development/The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Web site: www.worldbank.org E-mail: feedback@worldbank.org All rights reserved. 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 paper do not necessarily reflect the views of the executive directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of the World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA, 01923, USA; (telephone) 978-750-8400; (fax) 978-750-4470; (web site) http://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; (email) pubrights@worldbank.org. Table of Contents Abbreviations and Acronyms........................................................................................... ................ii Acknowledgments .........................................................................................................................iii Executive Summary......................................................................................................................iv I. Introduction .............................................................................................................................1 PSIA and the Contribution of Social Analysis.......................................................................1 Policy Reforms in the Agricultural Sector.............................................................................3 Analytical Framework of This Study.....................................................................................6 Methodology ..........................................................................................................................8 II. Evidence from Social Analysis: Stakeholders, Power Relations, and the Policy Process in Agricultural Reforms .......................................................................................................11 Types and Reasons for Agricultural Reforms......................................................................11 Stakeholder Identification....................................................................................................14 Power Relations and the Reform Process............................................................................18 PSIA Process and Policy Influence .....................................................................................22 III. Conclusions and Recommendations ....................................................................................34 Stakeholder Identification....................................................................................................36 Power Relations and the Reform Process ............................................................................37 PSIA Process........................................................................................................................38 Annexes Annex A: World Bank-Supported PSIA in Agriculture and Land Reforms............................41 Annex B: Summary of Case Studies........................................................................................42 Tables Table 1: Reasons, Types of Reforms, and Distributional Impacts...........................................12 Table 2: Stakeholder Analysis .................................................................................................15 Table 3: Main Political Economy and Institutional Risks Emerging from PSIA ....................19 Table 4: PSIA Process .............................................................................................................24 Table 5: Findings from the Poverty and Social Impact Analysis Case Studies .......................34 Boxes Box 1: The Stakeholders Influencing Land Reform in Zambia...............................................17 Box 2: Reforming Crop Boards in Tanzania............................................................................22 Box 3: Building Consensus around Land Redistribution in Cambodia ...................................27 References....................................................................................................................................28 i Abbreviations and Acronyms AFR - Africa ADMARC - Agricultural Development and Marketing Corporation CAS - Country Assistance Strategy CEM - Country Economic Memorandum CSA - Country Social Analysis CSOs - civil society organizations CSRP - Cotton Sector Reform Project DAC - Development Assistance Committee DGA - Democracy and Governance Assessment DFID - United Kingdom Department for International Development DOC - drivers of change EAP - Eastern Asia and Pacific ECA - Europe and Central Asia ESW - economic and sector work EU - European Union FAO/TCP - Food and Agriculture Organization Technical Cooperation Programme GOVNET - Governance Network (DAC) GTZ - Deutsche Gesellschaft für Technische Zusammenarbeit [German Agency for Technical Cooperation] GUYSUCO - Guyana Sugar Corporation, Inc. IFAD - International Fund for Agricultural Development IFI - international finance institution IMF - International Monetary Fund LASED - Land Allocation and Social Development Project LCR - Latin America and Caribbean LICUS - low-income country under stress M&E - monitoring and evaluation MDG - Millennium Development Goal MLMUPC - Ministry of Land Management, Urban Planning, and Construction NGOs - non-governmental organizations OECD - Organization for Economic Co-operation and Development PREM - Poverty Reduction and Economic Management Network PRSP - Poverty Reduction Strategy Paper PRMPR - Poverty Reduction Group in the PREM anchor PSIA - Poverty and Social Impact Analysis SDV - Social Development Department Sida - Swedish International Development Agency TFESSD - Trust Fund fore Environmentally and Socially Sustainable Development TIPS - tools for institutional, political, and social analysis TTL - task team leader UNDP - United Nations Development Programme USAID - United States Agency for International Development UNESCO - United Nations Educational, Scientific, and Cultural Organization WB - World Bank ii Acknowledgments This report is based on recent World Bank experience with Poverty and Social Impact Analysis in the agricultural sector. Based on the analysis of 23 PSIA studies conducted in that sector, and in-depth review of nine case studies, this report seeks to stimulate the discussion among users and clients of Poverty and Social Impact Analysis on how to carry out such analysis by better integrating socio-political and institutional factors into policy processes. An extensive literature review and structured interviews with task team leaders and PSIA team members were conducted to complete the analysis. We want to thank the following task team leaders and PSIA team members who contributed their generous time during structured interviews and provided insightful information about their PSIA: Kene Ezemenari and Barbara Verardo (Chad), Naima Hasci and Quentin Wodon (Benin), Antonio Nucifora and Reiner Forster (Malawi), Robert Townsend and Sabine Beddies (Tanzania), Wael Zakout and Andreas Groetschel (Cambodia), Vera Songwe (Mongolia), Sarosh Sattar and Saiyed Shabih Ali Mohib (Tajikistan), Emmanuel Skoufias and Nora Dudwick (Guyana); and Steen Lau Jørgensen and Zlatina Loudjeva (Zambia). We would also like to thank the peer reviewers: Anis A. Dani (Adviser, Social Policy, SDV); Paul Francis (IMF); and Steven Oliver (Senior Agricultural Economist, EASRD). We also thank our colleagues who contributed significantly to earlier drafts, including Caroline Kende-Robb (Sector Manager, SDV), Gillette Hall (Senior Social Development Specialist, SDV), Aline Coudouel (Senior Economist, PRMPR) and Steven Schonberger (Lead Operations Officer, EASRD). Finally, we extend our thanks to Sabine Beddies (Social Scientist, SDV) and Andreas Groetschel (Consultant, WB) who helped with the initial conceptualization of this work. We are grateful to the German Poverty and Social Analysis Trust Fund which contributed funding for this work. * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * Vice President Katherine Sierra Sector Director Steen Lau Jørgensen Sector Manager Caroline Kende-Robb Task Team Leader Estanislao Gacitúa-Marió This report was prepared by a team led by Estanislao Gacitúa Marió (Senior Social Scientist, SDV), and consisting of Renate A. Kirsch (Senior Social Scientist, SDV), Jeffrey Alwang (Consultant, SDV), Mileydi Guilarte (Consultant, SDV) and Kathleen Kuehnast (Consultant, SDV). Peer reviewers were Anis A. Dani (Adviser, Social Policy, SDV); Steven Oliver (Senior Agricultural Economist, EASRD) and Paul Francis (IMF). Andrew Norton (Lead Social Development Specialist, SDV) contributed significantly to the drafting of the report and provided valuable advice. iii Executive Summary The primary message of this report is that poverty and social impact analysis is highly relevant to the great majority of policy reforms, but its application in the agricultural sector is particularly important-- because reforms in this sector have profound and direct impacts on poverty. The high incidence of poverty in rural areas and the overall vulnerability of the livelihoods of the rural poor make it essential to ensure that distributional impacts are consistent with poverty-reduction goals. In other words, without careful attention throughout the design and implementation of agricultural reforms, the poor are often excluded, particularly those who live in remote areas, work in under-valued sectors, and are economically, politically and socially powerless. Based upon the analysis of 23 poverty and social impact analysis (PSIA) studies conducted in the agriculture sector and an in-depth review of nine case studies, this report examines the contribution and experience of conducting social analysis as a part of a multidisciplinary poverty and social impact analyses of agricultural reforms. PSIA is a systematic and interdisciplinary approach to evaluating the distributional impacts of policy reforms on income and non-income well-being (or welfare) of stakeholder groups, with a specific focus on the poor and vulnerable (World Bank 2003a). As such, a PSIA can contribute to the development of country ownership of policies by informing the public debate on the most appropriate policy design and the trade-offs between the various policy choices. The report examines the process whereby the poor become excluded in agricultural reforms. Most notably they are not visible to policy makers as a discrete interest group who will be affected by the reforms under consideration; the issues that matter most to them in relation to the agricultural reforms may not be understood or taken into account; and they may be excluded in the process of designing and implementing reforms. The report in turn emphasizes several ways in which such stakeholder groups are best taken into account and included in the design and implementation of agricultural reforms: i. Mapping all of the stakeholder groups that are likely to be affected by proposed reforms, taking social difference into account, such as gender, caste, and ethnicity--which often tend to be invisible to policy makers--and outlining the various ways in which such groups may be affected by policy change; ii. Highlighting the significance of imbalances in power relations that may be caused by reforms, and how they affect poor and vulnerable groups, in order that the design of reforms can take these effects into account; iii. Improving the design of the poverty and social impact analysis process and ensuring that it includes these stakeholders in all policy deliberations that ultimately will affect them. The objective of this study is to improve standards for social analysis in PSIA as it pertains to the agricultural sector. The study is intended to contribute to a broader, long-term goal of enhancing the capacity of development practitioners and policy-makers to carry out or commission poverty and social impact analysis of policy reforms in the agricultural sector. This study focuses on the contribution of social analysis to PSIAs, particularly in regard to the integration of socio-political and institutional factors into the policy processes. Social analysis in a PSIA focuses on the structure of societies and how their attributes affect reform design and implementation. Social analysis stresses that reforms are manifested through institutional mechanisms and political economy issues (that influence how reform implementation plays out in reality). and emphasizes that policy reforms have a range of impacts on different social groups. While other iv disciplines also deal with these areas, social analysis brings into focus the nature of power relations, as well as an understanding of the social construction of hierarchy, authority, and diversity. Applying a social perspective when examining the poverty and social impacts of reforms involves a range of activities that affects both the analysis and process of the exercise. In terms of analysis, it requires an understanding of the role of social structures to grasp the degree of participation that different social groups have in policy making, and how the interests of these groups are affected by or affect the distributional outcomes of reforms. Typically, social analysis plays a central role at every stage of the process in multidisciplinary efforts of conducting poverty and social impact analysis. Activities include specifying analytical focus, reviewing the secondary literature, interviewing key informants, conducting fieldwork with affected populations using a mix of qualitative and quantitative methods, analyzing results (and including economic analysis), and applying the findings to policy processes. Social analysts in PSIA exercises often pay particular attention toward strengthening the process of the work, ensuring that key stakeholders are engaged, which in turn can build consensus around the policy reform program. KEY FINDINGS The nine case studies examine a range of agricultural reforms, including the agricultural marketing system (Malawi); sub-sector specific marketing reforms in cashmere (Mongolia), coffee, cashews, and tea (Tanzania), cotton (Benin, Chad, Tajikistan, Tanzania), sugar (Guyana); input markets (fertilizer in Zambia); and land (Cambodia, Zambia). The case studies revealed that economic efficiency, fiscal concerns, and improved performance of the agricultural sector were the most common reasons for implementing agricultural reforms, although distributional impacts and equity concerns were also mentioned as significant drivers of reforms. Stakeholder Identification Stakeholders' interests were not only examined along income categories but also disaggregated along characteristics that included household size, household head, gender, ethnicity, locality, and occupation. However, stakeholder identification was not always followed by the analysis of the interests or stakes of these groups in the reform process. Stakeholder identification was particularly useful in identifying those populations in marginalized areas or regions that stood to lose from market reforms (Zambia, Malawi). In Guyana, stakeholder analysis revealed that reforms would have variable impacts on different ethnic groups, as well as the policy implications of this. Stakeholder analysis also helped broaden perceptions about who might have interest or a stake in the reforms, especially in Zambia, Guyana, Tajikistan, and Chad. The analysis also indicated considerable variation in how analysis of social diversity and gender informed the stakeholder analysis in particular, and the PSIA more broadly. In a number of cases, little explicit attention was paid to impacts on women, children, landless laborers, or other specific social groups who tend to have little voice or visibility in policy deliberations. An exception to this is the Zambia case, where social and economic analyses were complementary. Rigorous social analysis, which brings a much thorough understanding of social fault lines, can highlight variables (such as the impact on power relations) which lie outside of the normal parameters for economic analysis, and provide a contextual understanding of how policy reforms might impact different social and livelihood groups. Economic analysis plays many complementary roles, including assessing the likely or actual impacts in quantitative terms, and testing the relevance of findings from particular contexts at a more general level. v Power Relations and the Reform Process Social analysis contributed to understanding the central role that power relations have in the implementation of a reform program. In Zambia, for example, it illuminated how traditional authorities interacted with formal agencies, and the significance of this for proposed land reforms. In other cases, the social analysis identified how power relations could allow some powerful groups capture the benefits of the reform(s), as the Tajikistan case showed. Also, in many cases, the analysis gave better insight into the relationship between the political interests of elites and their influence on the reform proposal, showing how political forces could slow down the reform process. In Benin, for example, weak rural institutions and the state's inability to enforce written laws led to power imbalances and allowed elites to control information. Poor smallholders, due to lack of organization and high transactions costs, had weak bargaining power and were subsequently exploited by the existing arrangements. Most of the political economy risks identified during the PSIA were related to maintaining control over rents. This capture of rents is especially difficult to uncover in rural areas where transactions costs are generally high and, as a result, large marketing margins are common. Other political economy risks included conflicts between different government ministries, inadequate engagement of key stakeholders, and lack of confidence by stakeholders in the efficacy of the reform. In most cases, the PSIA process helped manage these risks. For example, in Tanzania, a task force of two ministries, the ministry of agriculture and food security and the ministry of cooperatives, were engaged in planning and implementing the PSIA; and their engagement supported the analysis and policy dialogue on the design and implementation of the crop board reform. In contrast, the Zambia and Malawi PSIAs did not identify any tension between ministries, and the role of the ministry of agriculture in implementing the reforms was practically ignored. As a result, conflicts between the ministries of agriculture and finance (the sponsoring ministry) continued after completion of the PSIA. The PSIA process helped identify institutional weaknesses that could hamper successful implementation of a reform. This risk was clearly demonstrated in Cambodia where unclear jurisdictions and incomplete delineation of state-owned land were identified as significant threats to land reform. A key concern of several PSIAs was the response of the private sector to the declining role of the state, particularly in agricultural marketing. This issue emerged, for example, in Malawi, where remote farmers were much worse off than those close to agriculture market depot. The social analysis showed that private agents did not operate as intensively in these areas as expected. PSIA Process and Policy Influence The PSIA process often led to changes in the design of proposed reforms. These changes were due to (i) provision of information that changed prior perceptions, (ii) stakeholder participation in the identification of the policy options, and (iii) dissemination of results. In the cases of Cambodia, Tanzania, and Zambia where substantial dissemination took place, stakeholder groups were able to discuss and evaluate findings and recommendations, which had more impact on policy. Similarly, early involvement by the stakeholders with most at stake had a significant impact on design and ownership. For example, in cases where line ministries were not involved until very late, as the Malawi and Guyana cases illustrate, ownership of the reforms by the sector authorities was weak. The cases analyzed indicate that in a PSIA process there is an inherent tension between analytical results, that find evidence in favor of or against a reform, and the political reality, in which stakeholder differences become highly contentious. One of the most effective means of addressing stakeholder vi differences is through dialogue and early inclusion of them in the process. In the case of Tanzania, the government recognized the need to reform the crop boards and formed a respective task force, led by the ministry of agriculture, to review crop board performances and their impacts on crop industries. The joint review of the Government Task Force, the EU, and the World Bank uncovered vested interests determined to maintain the status quo, and realized that more direct engagement of key stakeholders at the political level during the study could have reduced some of the delays in the subsequent policy decisions on the actual reform design and implementation. The PSIAs led to four types of impacts on the policy process and envisaged reforms: · Cases where changes in the proposed policy occurred. In Zambia and Malawi, the findings of the social analysis, which examined institutional capabilities and power structures, led to policy changes. · Cases where increased evidence of desirability of reform was provided. In some cases, the PSIAs did not result in a policy change but contributed to visualizing the different stakeholder interests and to building ownership. The Tanzanian PSIA contributed in advancing the domestic agreement that the cotton, coffee, cashew and tea crop boards need to be reformed in order to improve the performance and competitiveness of export crops. In Tajikistan, the PSIA documented exploitation under the existing system, and helped bridge differences between vulnerable stakeholders and government parties who were mainly concerned with efficiency and fiscal exposure. · Cases where more open and progressive debate about the policy was fostered. Several PSIAs enhanced the local policy dialogue by presenting a menu of options for reform and the mechanisms for addressing social impacts (Guyana, Tajikistan, and Tanzania). Options help create policy space for decision makers, where as part of a consultative, participatory process, they can broaden support for the ultimate decision. · Cases where no change occurred. In some cases, no major changes took place. This finding could be explained by the limited time period between PSIA completion and policy change, such as in Benin and Guyana. In other cases, as indicated in the Tajikistan example, prior beliefs about the efficiency impacts of the proposed reforms were confirmed by the study. The full potential of the poverty and social impact analysis has been constrained by several "process" factors: (i) lack of government support or weak institutional linkages in the PSIAs (e.g., agricultural sector reform PSIAs that do not involve ministries of agriculture as primary partners), (ii) limited efforts at dissemination, and (iii) lack of conviction among donor staff about the ultimate utility of the PSIA. In no case was an adequate monitoring and evaluation system established, and this shortcoming limits the longer-term utility of poverty and social impact analysis. CONCLUSIONS AND RECOMMENDATIONS Policy reforms in the agricultural sector are complicated by a number of factors. Social structures within rural areas are often shaped by unequal power relationships, which in turn have an impact on poor people's livelihoods. Rural areas are characterized by low population density and dispersed groups which have limited access to public services and markets. Reforms can also have indirect positive and negative spillovers and tend to be associated with social upheaval due to their impacts on long-standing political and social relations. vii Social and institutional analyses frequently helped corroborate and expand the findings of economic analyses. It deepened the understanding of the underlying causes of a finding identified during an economic analysis (Benin, Zambia), contributed to the examination of the validity of assumptions underlying the economic models, and was used to identify problems in the economic analysis (Tajikistan). The analysis also helped identify institutional factors that framed economic processes and finally identified and investigated how power relations impact policy changes (Zambia). This study suggest three areas in which the impact of social analysis in the PSIA process can inform agricultural reforms in order to provide better information to policy makers and stakeholders, and to contribute to a more open and inclusive reform process. Stakeholders Stakeholder analysis is important to the PSIA process, especially in rural areas, where it highlights diversity issues, in particular problems of access among remote populations, who tend to be very poor. The stakeholder analysis needs to be underpinned by a sound analysis of social diversity (World Bank 2003d). Statistical analysis tends to overlook different categories of stakeholders, who may not be identifiable from available data sets either because relevant differences are not captured by the data (e.g., ethnic or livelihood groups) or because data sets do not disaggregate below the level of the household. Therefore, a sound social analysis is needed to identify key characteristics of stakeholder groups, including their location of residence, ethnicity, gender, religion, caste, etc. Stakeholder identification benefits when social scientists in the client country participate early in the process to ensure that relevant social groups are clearly identified. This identification is also important to give voice to stakeholders who may lack access to information and, therefore, have little opportunity to participate in the reform process. Without access to information and participation, ownership of the reform and the sustainability of the process are difficult to maintain. Rural stakeholders tend to be dispersed and heterogeneous. Furthermore, the rural poor have little power compared to other stakeholders. Getting buy-ins from government and civil society for a thorough analysis and inclusion of all relevant stakeholders is pivotal. Power Relations and the Reform Process The findings of this report suggest that the political economy of reform processes should be a major focus for social analysis methods and practice in the next generation of PSIAs. One of the critical concerns in the reform process is the analysis of risk--determining the political economy risks and institutional weaknesses that potentially could threaten reform success. The analysis shows that early involvement of government in an inclusive PSIA planning process, early identification and involvement of key stakeholders, and an iterative analysis that builds credibility among stakeholders can help minimize risks to policy reforms. PSIA Process The study has shown that choices made in the design of the PSIA process had a significant bearing on the likely impact of the results. In the agriculture sector, achieving an inclusive process tends to be challenging since most rural stakeholders are at a stark disadvantage in terms of information, mobility, and access. The cases clearly illustrate the issue of timing with respect to PSIA implementation and the policy dialogue. A PSIA must be conducted when the policy environment is still fluid and positions have not solidified. In the agriculture sector, this is essential to allow dispersed and powerless stakeholders to viii articulate their interests and have a chance of participating in the process. The analysis shows that early involvement of institutional stakeholders with the most at stake, including ministries of agriculture and land, affects their ultimate success. Involvement of sector agencies is extremely important to ensure appropriate sector implementation arrangements. Without adequate buy-in, the PSIA becomes a one-off process and impacts are limited. Wider involvement of multiple partners strengthens the PSIA process. Given power differentials between different stakeholders in the agriculture sector, bringing civil society partners into the discussion can help to identify issues that otherwise could have been ignored. The Malawi case and Oxfam's involvement illustrate this clearly. However, it should be noted that country ownership is an over-riding factor in achieving policy impact. The study found that many PSIAs were driven more by donor interests than by clients' interests. PSIAs initiated with the full support of client governments (such as Tanzania and Cambodia) are more likely to have stronger impacts on policy. Because of the complex nature of social and economic relations in rural areas, employing mixed methods and integrating economic and social analyses are essential. The gains from this integration are greater when there is an iterative process, where the economic and social analyses have more opportunity to inform each other. A commitment to a sustained approach to the analysis of poverty and social impacts throughout the reform process is recognized as good practice. This is particularly true in the agricultural sector, where long-term production cycles and investment tend to dominate the organization of the sector. With agriculture sector reforms, further effort is needed to embed the practice of PSIA within the policy cycle in partner countries, to ensure that line ministries and sector stakeholders are involved in the long term to ensure the sustainability of reforms. Most of the case studies reviewed showed shortcomings in disseminating their results and integrating them into the policy process--especially understanding the impact of PSIAs on the policy process--and a related loss of sustainability. The PSIA process should begin by discussing with partners reasonable expectations of what the study can achieve. PSIAs are funded at relatively modest levels and have relatively short time horizons. Even a PSIA that produces the best information cannot be expected to have a long-term impact without institutional buy-in by the host-country government and the World Bank country team. Without long-term sustainability, including dissemination, monitoring, and evaluation, any impact is short-lived at best. In the case of agricultural sector reforms, with dispersed stakeholder populations and competing power relations, long-term commitment is essential for sustained impacts. The study also found that documentation of the PSIA process tended to be weak. This was particularly true of the analytical and field research methods used to inform the social analysis. Going forward, it will be important to ensure that the documentation of social analysis is improved to facilitate comparative analysis and learning. Finally, the cases demonstrate the value of a multidisciplinary approach to poverty and social impact analysis. The future agenda for the social analysis component in agricultural sector reforms should focus on strengthening practice in the following areas: integrating PSIA results into policy formulation; monitoring impact of reforms on diverse social groups; including social groups with low "visibility" and voice in reform processes; analyzing power relations and their significance on reforms and the downstream effect on poor and vulnerable groups; and documenting the PSIA process in general, and the social analysis work in particular, to enable better comparative learning. In all of these areas, close collaboration with other disciplinary perspectives is necessary. Although the focus of this report is on the contribution of social analysis, strengthening PSIA practice requires close co-ordination with sector and economic expertise on both content and process. ix I. Introduction The objective of this study is to promote improved standards for social analysis within Poverty and Social Impact Analysis in the agricultural sector. The study is intended to contribute to a broader, long-term goal of enhancing the capacity of development practitioners and policy-makers who carry out or commission PSIAs of policy reforms in the agricultural sector. This target audience includes policy makers in various government agencies in the World Bank's partner countries, covering central policy, budget functions, as well as and line ministry responsibilities, such as agricultural service delivery. It also includes development practitioners in civil society, research, and the donor community. The study provides original analytical input on effective methods and processes for conducting PSIA. It aims to stimulate discussion among users and clients of PSIA on how to conduct such analysis by better integrating socio- political and institutional factors into policy processes.1 The report focuses on the contribution of social analysis, but recognizes that many of the key areas highlighted here--social difference, political economy analysis, and the policy process--are also important for economic analysis, and that PSIA is most effective when a multidisciplinary approach is used. This paper builds on and complements other work that has looked at PSIA in the agricultural sector.2 PSIA AND THE CONTRIBUTION OF SOCIAL ANALYSIS Poverty and Social Impact Analysis3 is a systematic and interdisciplinary approach to evaluating the distributional impacts of policy reforms on income and non-income well-being of stakeholder groups, with a specific focus on the poor and vulnerable (World Bank 2003a). By examining how these impacts are channeled, PSIA can contribute to better-informed decisions about the design, sequencing, timing, and appropriateness of reforms. The PSIA process enhances understanding of the policy making process and contributes to country ownership of policies as it informs policy-makers on the most appropriate policy options and on the trade-offs between policy choices.4 There is growing awareness among international donor agencies, partner governments, and civil society that policy reform has to be well tailored to the social context and needs sufficient public support in order to be successful. PSIA, if well managed, can assess the potential impacts of reforms and inform the policy debate to this end. Depending on the country and reform context, PSIA involves a number of elements that must be considered in its design and implementation. According to the PSIA User's Guide (World Bank 2003a), these elements include asking the right questions, identifying stakeholders, ascertaining transmission channels for impacts of reforms, assessing institutions, analyzing impacts, assessing enhancement and compensation measures, assessing risks, and promoting policy debate. Further, PSIA is based on a multidisciplinary approach, stressing the application of social and economic analysis via the use of qualitative and quantitative methods. While there is a logical sequence to this listing, it does not imply 1 For further discussion of the conceptual framework and technical guidance on the elements that make up institutional, political, and social analysis for PSIA, see the Tools for Institutional, Political, and Social Analysis (TIPS): Sourcebook for PSIA (World Bank 2005). 2 See, in particular, Lundberg (2005) and Deninger (2005). 3 "Poverty and Social Impact Analysis (PSIA) involves the analysis of the distributional impacts of policy reforms on the well- being of different stakeholder groups, with a particular focus on the poor and vulnerable" (World Bank, 2003, vii). For further discussion on the purpose, conceptual framework, and practice of PSIAs, see A User's Guide to Poverty and Social Impact Analysis (World Bank 2003a); and Using Poverty and Social Impact Analysis to Support Development Policy Operations in Good Practice Notes for Development Policy Lending (World Bank 2004). 4 For a detailed discussion of the different potential trade-offs, see the work of Robb (2003), in which she explains that the objective of a PSIA is "to support country ownership of policies by informing...the trade-offs between policy choices." 1 that all steps need to be undertaken in strict order or that each step is feasible in every country context. (World Bank 2003a, 9). An important element of PSIA practice has been the identification of "transmission channels", through which policy changes impact social and economic relations (outlined in the User's Guide to PSIA (World Bank 2003a)). Delineating channels in this way makes transparent the way, in which policy change is expected to deliver its impact. The five channels originally identified were employment, prices, access, assets, and transfers/taxes. Based on the experience of the first PSIA exercises, the Tools for Institutional, Political, and Social Analysis (TIPS) sourcebook (World Bank 2005) added a sixth channel, through which relations of authority (comprising power, structures, and processes) are directly changed through policy reforms. This channel can operate at the macro-level (e.g., public service reform), at the meso-level (e.g., decentralization of administrative authority), and even at the micro-level (e.g., redirecting welfare payments from men to women). This channel encourages analysis of the likely impact on state actors and citizens of changes in decision-making behaviors and interactions that result from new alignments of rights, obligations, incentives, and sanctions. This study focuses on the contribution of social analysis in the PSIA process, and specifically on the contribuiton of social analyis to informing agricultural reforms in four areas: (i) identifying and including stakeholders in the reform/policy process; (ii) analyzing how the institutional arrangements mediate the transmission of policy impacts to different stakeholders; (iii) disaggregating the impacts across different types of social groups; and (iv) assessing the social risks that might threaten the design and implementation of reforms. The study analyzes recent World Bank experiences with PSIA in the agricultural sector.5 Based on 23 PSIA studies conducted in that sector, and in-depth review of nine case studies, it indicates how the PSIA process has addressed these four dimensions in order to provide better information to stakeholders and policy makers and to contribute to a more open and inclusive reform process. In a generic sense, the term "social analysis" encompasses institutional, political, and social analysis (World Bank, 2005). There are three overlapping areas of analysis, derived from different disciplinary backgrounds, that focus on the underlying rules and relationships of reform outcomes: · Institutional analysis looks at the "rules of the game", which people develop to govern group behavior and interactions in the political, economic, and social spheres of life. Institutional analysis is based on an understanding that these rules, whether formally constructed or informally embedded in cultural practice, mediate and distort (sometimes fundamentally) the expected impacts of policy reform. · Political analysis looks at the structure of power relations and entrenched interests of different stakeholders that affect decision making and distributional outcomes. Political analysis is built on the recognition that political interests underpin many areas of policy debate and economic reform, challenging assumptions about the "technical" nature of policy making. · Social analysis looks at the social relationships that govern interaction at different organizational levels, including households, communities, and social groups. Social analysis is built on an understanding of the role of social and cultural norms in governing relationships within and between groups of social actors, which have implications for the degree of inclusion and empowerment of specific groups. Social analysis focuses on the structure of societies and on the ways, in which their attributes impact reform design and implementation. Social analysis stresses that reforms are manifested through institutional mechanisms, that political economy issues influence the way, in which reform 5 A number of agricultural/rural sector PSIAs have been supported by the World Bank, with topics such as reform of agricultural marketing systems, land reforms, and restructuring of credit systems. 2 implementation plays out in reality, and brings attention to the fact that policy reforms have variable impacts for different social groups. Employing a social perspective to examine the poverty and social impacts of reforms involves a range of activities affecting both the analysis and process of the exercise. In terms of analysis, a social perspective contributes to understanding the country and reform context, the policy implementation process, and the impacts of reforms at the micro-level. For each of these analytical levels, there are a range of tools for both gathering and analyzing data and information.6 By engaging key stakeholders in the analytical and policy process, social analysts have also strengthened the PSIA process so that it generates country ownership and builds consensus around the reform process. This has allowed stakeholder groups, who would otherwise not have had voice in designing policy changes, to participate in either the conducting of the analysis (including gathering data and information), in the governance of the PSIA process (for example, through steering committees) or in deliberating the implications of the analysis for policy design. PSIA has been applied by the World Bank and other donor agencies (e.g., DFID, GTZ) and more recently directly by national governments (e.g., Kenya, Malawi) in the last four years. PSIAs in the agricultural sector are important endeavors because reforms in this sector have profound and direct impacts on poverty. Such reforms can also have indirect positive and negative spillovers and tend to be associated with social upheaval due to their impacts on long-standing political and social relations. Among the analyses conducted by the World Bank, the agricultural sector has the highest number of PSIA cases, providing a robust data source for further analysis. POLICY REFORMS IN THE AGRICULTURAL SECTOR Poverty and Social Impact Analysis is relevant to most policy reform processes, but its application in agricultural reforms is particularly vital. The high incidence of poverty in rural areas and the overall vulnerability of the livelihood strategies of the rural poor make it important to ensure that distributional impacts are consistent with poverty-reduction goals. The World Bank's rural strategy (World Bank 2003b) calls for raising the profile of rural development efforts by (i) fostering an enabling environment for broad-based and sustainable rural growth; (ii) enhancing agricultural productivity and competitiveness; (iii) fostering non-farm economic growth; (iv) improving social well-being, managing and mitigating risk, and reducing vulnerability; and (v) enhancing the sustainability of natural resource management. To achieve these objectives, significant policy reforms are needed, including reforming trade policy, removing obstacles to effective market operations, promoting land markets, improving agricultural financing, introducing sound food security policies, and designing safety nets that directly protect poor rural dwellers from shocks. Over the past 20 years, many countries have embarked on significant reforms of the agricultural sector with different objectives including, among others, subsidizing urban consumers, generating revenue, and stabilizing the economy (de Janvry and Sadoulet 2002; Dorward et al. 2004; Kherallah et al. 2002; Mellor 1986). By and large, the reforms have had many favorable results, but have not yet yielded the desired results in terms of poverty reduction and food security, because their impacts have been muted by partial implementation and structural constraints (Deininger 2005; Gardner 1996; Lundberg 2004). Agricultural reforms are an important part of Poverty Reduction Strategy Papers (PRSPs) in most of the countries, for which these strategies have been prepared.7 Agricultural reforms can significantly impact 6 These are outlined in World Bank (2005). 7 See for example, FAO (2003), "Guidelines to prepare country policy profiles on the state of policy and strategy for food security and agricultural development" (Rome: FAO); Franz Heidhues, 2004, "Development Strategies and Food and Nutrition Security in Africa: An Assessment," IFPRI 2020 Discussion Paper, no. 38 (Washington, DC: IFPRI). 3 poverty reduction but to be successful, they must consider the critical socio-economic, political, and institutional characteristics of the rural poor (von Braun 2003). There still is a gap between the existing PRSPs, the agricultural and rural development strategies followed by the countries, and the desired impacts of the proposed sector reforms.8 Policy reforms in the agricultural sector are complicated by a number of factors. Rural areas are characterized by low population density and dispersed groups with limited access to public services and markets. As a result, transactions costs are high and information is costly. High reliance on rain- dependent agriculture, uneven coverage of public services, and high transaction costs substantially increase risks for the poor and vulnerable. Weak rural institutions are associated with imperfect markets and lower applicability of smooth price-transmission and equilibrium-based economic analyses. These factors add to the burden of the social analysis of reforms as well. Rural areas have often been bypassed by central government investments and policies.9 Rural social structures are shaped by unequal power relationships, which in turn impact poor people's livelihoods. For example, hierarchical caste, ethnic, and gender relations predetermine unequal access to resources and influence the distributional impacts of policy interventions. Women often have little control over productive assets, and even when they do exercise de-facto control, their legal rights are rarely secured. Insecure control over resources reduces the ability to protect one's basic rights and limits access to support services. In many countries, for example, women are precluded from holding title to land, which limits their access to credit and other services, such as agricultural extension. A land policy reform could thus have profound social and economic implications and generate considerable controversy. Due to imperfect information, incomplete markets, and weak institutions, rent-seeking in rural areas has contributed to power consolidation among advantaged groups. As a result, the political sensitivity of some reforms, such as changes in land tenure or in marketing institutions and regulations, can trigger social tensions. These tensions may constitute a political threat for national governments and make them less likely to engage in reforms. Weak rural institutions, caused in part by historical underinvestment and outward migration of better educated rural dwellers, also lower the capacity to implement reforms in these areas. The drivers of reform and the types of reforms in the agricultural sector are also important to consider. Agricultural and rural reforms include: i) sector-wide policies in pricing, marketing and research, and extension institutions and policies; ii) specific sub-sector actions to address problems, for example, in coffee pricing and marketing; iii) land tenure reforms; and iv) actions to reform policies and institutions that have direct implications for rural areas. Pressure to undertake these reforms builds in different ways. Institutions, such as land tenure and agricultural marketing systems, are the product of historical factors and conscious policy decisions over time. As local and global conditions change, the performance of these institutions may be affected. For example, traditional land tenure systems may function perfectly well while population densities are low, but may be less effective as they increase (de Janvry and Sadoulet 2002). 8 See, for example, the 2004 Burkina Faso Poverty Reduction Strategy Paper, joint IDA-IMF staff assessment, and annual progress report that highlights the need for upstream poverty impact assessment on proposed agriculture policies of proposed policies in order to support government strategies aimed at improving rural incomes and at reducing vulnerability or the 2004 Madagascar Poverty Reduction Strategy Paper's first annual progress report (World Bank 2004) and joint assessment which indicates that to improve capacity and assure impact on poverty in the short and medium terms, the government will need to prioritize rural development actions with the largest potential impact on the poor. Similarly, the 2003 Tanzania Poverty Reduction Strategy Paper, progress report, and joint IDA-IMF staff assessment indicate that the operationalization of the agricultural strategy would benefit from a review that more deeply assesses the impact of proposed interventions on poverty. 9 Policy biases against rural areas include over-valued exchange rates, direct taxation of exports, under-investment in rural infrastructure and public services, etc. See Lipton (1993) for an overview of these issues and a cogent discussion of their causes and consequences. 4 Alternatively, evidence may emerge that particular institutions are inefficient or inequitable, and political pressure, both internal and external, may build for change (Grindle 1986 and 2004). One example is government-controlled agricultural marketing systems, which may have been established to generate revenues for government or to protect producers from perceived exploitation. Over time, these systems have become loss-making and unresponsive to changes in the market. The first inefficiency--cost-- implies subsidies which divert government budgets from other priority expenditures. The second-- inflexibility-- leads to inefficiency throughout the agriculture sector and represents a major obstacle to agricultural growth. Because these reforms affect long-standing institutions, and because they affect the distribution of rents and power, they tend to be particularly controversial. As a result, it is important that analysis of their distributional effects be detailed and credible, and that a variety of stakeholders be brought into the process. The recent works by Mattias Lundberg (2005) and Klaus Deininger (2005) discuss relevant conceptual and methodological issues related to the analysis of agricultural market and land-reform PSIAs. Both papers outline the economic factors that influence the reform programs in the agricultural sector, the different policy reform alternatives, as well as some of the economic tools that have been used to analyze the impacts of the reforms. In the case of agricultural market reforms, Lundberg indicates that, in many cases, decision makers and stakeholders are unaware of the magnitudes and distribution of potential reform impacts. PSIAs can provide insights into this classic political economy issue by identifying stakeholder concerns, the winners and losers of reforms, uncovering impact transmission channels, and quantifying the risks to the reform process.10 Deininger (2005) analyzes the PSIA process in the context of land tenure reforms. He argues that land tenure reforms are frequently needed for a number of reasons, some related to economic efficiency (such as incentives to invest in productivity enhancements and access to credit), and others related to equity and distribution (e.g., gender concerns,11 food security, and social safety nets).12 Deininger suggests that PSIAs contribute to effective land policy formulation by (i) linking the general (theoretical) justification for land policy reform to the specific steps or interventions, (ii) evaluating impacts on heterogeneous stakeholders, and (iii) enhancing the policy dialogue by creating flexibility to respond dynamically to experiences and information generated as the reform process progresses.13 Given the high stakes associated with land policy reforms, PSIAs should focus on social outcomes, social processes, and power relationships in addition to economic factors and outcomes. Both authors recognize that there are crucial non-income, hard-to-measure factors and processes that affect the viability and impact of reforms. Heterogeneity within sub-groups of stakeholders, increasing returns to scale, the impact of external factors on the production process, and information differentials, among many others, suggest that economic signals do not fully capture the effects of imperfect markets and weak institutions. The impact of institutions and power relations on policy impacts also need to be considered in the analysis of agricultural reforms. 10 See Bates (1981) for a description of the role of agricultural marketing institutions in Africa with a special emphasis on the political economy driving their continued existence. Anderson and Hayami (1986) present a broader treatment of these political economy concerns. 11See Feder (2002) and Feder, Tongroj, and Tejaswi (1986) for evidence on the relationship between tenure security and outcomes, such as increased investments and productivity of land. Issues of enhanced well-being due to increased security for women and other disadvantaged groups are not generally covered in this literature. 12 See Carter and Olinto (2003) for a detailed discussion of how tenure security affects investments in land. 13 Note that these are some of the known and well-recognized strengths of social analysis and other qualitative methods (Rao and Woolcock). 5 ANALYTICAL FRAMEWORK OF THIS STUDY The analytical framework that guides this ESW is based on the following prior assumptions. First, the value of a PSIA lies in its impact on, and improvement of, the design and execution of policy reform. Second, the PSIA process is not simply a technical research exercise but also contributes to political processes of policy discussion and decision making. PSIAs therefore, can influence policy both through strengthening information,14 and through facilitating deliberation on policy choice. Third, the practice of Poverty and Social Impact Analysis is at its strongest when it is carried out in a multidisciplinary way-- incorporating both social analysis, and economic analysis15. Based on these principles and the existing guidance for good Poverty and Social Impact Analysis, the discussion and analysis of the cases examined in this report are structured around three points: (1) stakeholder analysis; (2) power relations, and in particular their impact on institutional and political risks to implementation; and (3) process elements contributing to feedback into policy choice. By focusing on these elements, it is possible to bring together the main factors that contribute to the social dimension of the PSIA process, while acknowledging at the same time that economic analysis also contributes to the policy process. By considering these three factors, it is possible to examine how a PSIA has (i) contributed to identifying different social groups that will be affected by, or will affect, the proposed reform; (ii) contributed to understanding the socio-political and institutional mechanisms that frame the reform process; and (iii) informed policy discussion and choice. Stakeholder Analysis In the PSIA cases under review, the stakeholder groups, identified during the process, are analyzed. Stakeholder identification allows one to systematically pinpoint stakeholder characteristics, interests, and the nature and degree of their influence on existing or future reforms. Stakeholder analysis recognizes that decision-making outcomes are a function of the interests and power of key policy stakeholders. At this level, the analysis of the cases seeks to illustrate if and how the PSIA focused on the income and non- income distributional impacts across different social groups, disaggregated along various characteristics such as household size, household headship, gender, ethnicity, location, occupation, etc. Also, the analysis examines if the stated or unstated interest of these groups regarding a policy were considered in the framing of the PSIA. Generally practitioners have a sound understanding of the stakeholders involved in a reform process. However, if important stakeholders are not identified at the outset, certain social groups and their interests can remain invisible in the policy discussion process. The analysis thus seeks to examine how the PSIA cases applied stakeholder analysis to include the interests of all stakeholder groups in framing the policy choices. Without an in-depth analysis of social processes and social differences, the interests of some 14 The value of the information provided by a PSIA is shown by the impact that its information has on the decisions made and the resulting design of the policy reform. The PSIA process should be designed to maximize the likelihood that there will be a policy impact. The PSIA user's guide (World Bank 2003a) recognizesthis consideration by noting that reforms should be targeted for PSIAs according to the size and direction of subsequent policy impacts. An information-theoretic (Bayesian) approach would add the idea that PSIAs should target those policy reforms characterized by high degrees of uncertainty about expected size and direction of distributional and social impacts. See Pardey and Smith (2004) for a description of this approach; Alston, Norton, and Pardey (1995) and Norton and Alwang (1997) contain more details and specific examples. The conclusion that research has no value, if it does not lead to changes in policy design or to the likelihood that the policy will be adopted, is not without controversy. Krugman (2004, 80) notes that "it seems very unfair to researchers to declare their work worthless, even if the results are correct and offer major gains to society, simply because politicians refuse to accept their implications." 15 World Bank (2003a) contains a detailed description of economic and social tools broken down into the following: (i) identifying stakeholders, (ii) assessing institutions, (iii) analyzing impacts using social tools, (iv) using economic tools to analyze impacts, (v) assessing risks, and (vi) monitoring and evaluating. See Bourguignon and Pereira da Silva (2003) for a description and assessment of economic tools. For detailed discussion of institutional political and social tools, see the World Bank (2005) TIPS source book. For further discussion of social analysis methods, see also Holland and Campbell (2005); Kanbur (2003); Kumar and Chambers (2002); and World Bank (2002). 6 stakeholders may remain unacknowledged by policy makers. Stakeholder analysis can play an important role in making them visible and highlighting social impacts of planned reforms which may otherwise have been unrecognized. An understanding of social diversity and gender (which is fundamental to social analysis) needs to underpin effective stakeholder analysis. Power Relations and Reform Implementation The second element of the framework refers to understanding how power relations affect the process of designing and implementing policy reforms. In particular, this report examines whether the PSIA process pays attention to how proposed reform can change power relations and how those power relations can affect the way in which the potential impacts are channeled. The intention of the PSIA is to make transparent the way in which power relations are incorporated. The analysis focuses on how different reforms often result in changes in decision making and in new formulations of rights, obligations, incentives, and sanctions that in turn influence the behavior of government actors and citizens. Often, the channels of policy impact are well-understood, but in some cases, these paths are subtle or need special attention. For example, impacts of agricultural marketing reforms might be transmitted largely through changes in relative prices. In developing countries, transaction costs, imperfect information, and the influence of market power, particularly in isolated rural areas, may mean that price formation is extremely complex. An effective social analysis, based on sound empirical material from either secondary sources or field research, can provide insights that assist in the development of an economic analysis that takes account of these factors. This, in turn, improves the robustness of the reform design by reducing uncertainty about key policy parameters. Similarly, the impacts of restructuring of land rights may be transmitted through changes in power relations and authority. It is important to know whether groups might seek to undermine the new decision-making structures or sets of rights by limiting their implementation or by using other influences to stop others from making use of their new entitlements. Risk analysis is a meta-assessment of issues that could go wrong and prevent intended social and poverty impacts of reforms. Explicitly discussing potential risks can help stakeholders adjust the design of reforms or design mitigating measures that address particular risks. Risks to the reform program as well as risks emanating from it should be considered. Four types of risks to the effectiveness of reform programs have been identified in the PSIA literature: institutional risks, political economy risks, exogenous risks (e.g., natural disasters, conflict, economic crisis), and country risks (e.g., political instability, social tensions). This review concentrates on the first two because the power and institutional dimensions of risk raise more compelling issues for PSIA design. Political economy risks encompass the unanticipated capture of benefits by non-poor groups, and opposition or distortion of reform processes by powerful stakeholders. Institutional risks derive from incorrect assumptions about institutional performance or the quality of governance of the reform process.16 Effective analysis of institutional and political economy risks requires detailed understanding of group dynamics, power relations, and political processes.17 Institutional complexities and constraints create additional transaction costs that affect reforms.18 For reasons noted above, these institutional responses are especially critical in rural and agricultural reforms. In this regard, the PSIA process should provide information about the feasibility of change in light of the existing power relations within the institutions and of the norms that regulate their behavior and functions. 16 See World Bank (2003a and 2003d) for complementary frameworks on the analysis of risks. 17 Narayan et al. (2000) discuss the ability of participatory methods to reveal non-economic attributes of institutions and how they affect outcomes. 18 Power relationships in rural areas often exacerbate institutional weaknesses and affect the ability to implement reforms and transmit reform-related signals to stakeholders. For further detail, see Narayan et al. (2000) which discusses the institutional barriers to access to markets, public services, etc., at state and local levels. 7 Process Elements Contributing to Policy Impact Finally, the analytical framework focuses on how choices made through the design and implementation of the PSIA process, impacted the effectiveness of PSIAs in influencing the design of proposed reforms. Relevant design elements include the tools used to analyze stakeholders, impacts, and risks; the mechanisms for integrating social and economic analysis in the PSIA; the extent to which the PSIA design sought to give voice to stakeholders in developing policy options; and the ways in which results were disseminated. An important process element is the nature of integration between the different research components (e.g., social and economic enquiry and analysis). Rao and Woolcock (2004) discuss three approaches for integrating research19: parallel, sequential, and iterative. In parallel approaches, the two types of analysis occur separately: two analysis teams work on the different methods, and then compare and integrate results at the end. Sequential and iterative approaches include dialogue between the social and economic analysis at all phases of the PSIA. Sequential integration involves using findings from one type of research to inform the other (e.g., a social analysis of stakeholders and transmission mechanisms would be used to inform the design of the subsequent economic analysis). In an iterative approach, information and experiences from the two types of analysis are fed back to enhance each. It has been argued that PSIA processes contribute to the design of reforms by "providing key stakeholders and opinion makers with evidence to consider during their internal policy debates" (World Bank 2004, 36), and "ensuring stakeholder involvement and transparency" (World Bank 2004, 39). This study examines the role of the PSIA process in involving key stakeholders and giving voice to their perceptions and interests. The following questions were asked of the case studies presented here: Did they help inform the policy debate? Did they include formerly marginalized groups in the analysis? Did they allow these groups any influence over policy process? METHODOLOGY This study's methodology is based on the analysis of a sample of 23 PSIAs, supported by the World Bank, which addressed agricultural sector reforms.20 From this set of 23, nine cases were selected for further examination. Methods include a comprehensive analysis of supporting documents (PSIA documents, Poverty Reduction Strategy Papers, Country Assistance Strategies, Poverty Reduction Support Credits, Country Economic Memoranda, Poverty Assessments,21 scholarly publications, donor's reports, etc.) and structured interviews with World Bank task team leaders (TTLs), who manage and are responsible for implementing PSIAs, as well as members of the PSIA team. The literature review provided basic background information on all PSIA cases. The analysis was guided by a questionnaire that focused on the type of stakeholder and institutional analysis carried out, how power relations were addressed, and what were the process elements that informed in country policy making. This analysis guided the selection of the nine PSIA reforms used for the in-depth case studies. The key elements used for analyzing the cases were described in the previous section on the "Analytical Framework of this Study." A case-study approach was chosen because it enables a holistic exploration of a range of issues through comparative analysis. Case-study methodology is particularly appropriate for explaining process 19 Social and economic analysis can involve qualitative and quantitative dimensions. Rao and Woolcock's (2004) framework is easily broadened to cover integration of social and economic components of a PSIA. 20 See annex A for a description of the entire universe of agricultural sector PSIAs, as of March 2005. 21 The Country Assistance Strategy is the primary strategic planning document for World Bank country programs. The Poverty Rduction Support Credit is a lending operation by the World Bank supporting a country's poverty reduction strategy. Country Economic Memoranda and Poverty Assessments are diagnostic tools used by World Bank country programs to further policy dialogue. 8 questions and context issues. Case studies can accommodate multiple sources of data. Cases must be carefully selected to allow generalizations. In multiple case studies, the logic for selecting cases is stronger when a theoretical replication model is followed.22 That is, some variables remain constant across all cases, while others are changed to test the relevance of the initial set of propositions. In this case, the constant variable is the assessment of the effectiveness of integrating socio-political and institutional factors into PSIA processes in order to influence agricultural policy reforms. While no statistical generalization is intended with case studies, analytic generalizations based on the theoretical relevance of the case are possible. The following criteria were considered for the selection of the cases: · Status of PSIA process. Cases were divided into three categories based on their status: in design, ongoing, and complete. Of the 23 cases, 2 were still in design, 10 were ongoing, and 11 were complete. The nine cases selected for in-depth analysis were chosen from the "complete" category in order to have a complete view of the process retrospectively and due to data availability. · Type of reform. Based on the specific sector and scope of the reforms, 15 types were identified within the 23 PSIA cases: (i) sector-wide reforms in cashmere, cocoa, coffee, cotton, groundnuts, livestock, maize, rice, sugar, tea, and tobacco; (ii) forestry sector reforms; (iii) agricultural input and output market reforms; and (iv) land reforms. The analysis would have been more diverse if each area for sector-wide reform was represented. However, status, region, and the other selection criteria were also considered in the selection.23 · Regional representation. With the exception of the Middle East and North Africa, all regions were represented in the sample of 23 PSIAs. At least one case was selected from each region for further analysis. Thus, Mali's cotton sector reform was not selected for in-depth analysis because the team had identified two others, Chad and Benin, which also focused on the cotton sector and had similar components to Mali's. It is important to note that most PSIAs have been conducted in Africa, hence the regional distribution in the selection of cases: five in Africa, two in East Asia and Pacific, one in Europe and Central Asia, and one in Latin America and Caribbean. · PSIA generation. Finally, the selection of the cases was made in order to include examples from both the initial generation of PSIAs, as well as newer cases. Five of the nine case studies (Cambodia, Malawi, Zambia, Tanzania, and Chad) were in the first group of PSIAs conducted. This first generation of PSIAs pilot-tested some of the current practices. The practice of poverty and social impact analysis within the World Bank has evolved since then, and the second generation PSIAs benefited from lessons learned in the earlier ones. First generation cases are slightly more represented because more of them were completed. Their selection also sought to examine whether there were lessons to be drawn from their different generational status. Following the selection of cases, a questionnaire was compiled and shared with task team leaders prior to interviewing them. Formulating key questions made it easier to understand the underlying problems that reforms were intended to address. The questionnaire was designed with a diagnosis that listed the chain of cause and effect from policy objectives to constraints, to choices, to impacts. The analysis of the cases also defined a counterfactual, i.e. what the social and poverty impact would be without the reform. 22 See Robert K. Yin (1989), "Case Study Research, Design and Methods," Applied Social Research Methods Series, vol. 5 (CITY: Sage Publications). See also W. Tellis (1997), "Application of a Case Study Methodology," The Qualitative Report 3 (3) [on-line serial], http://www.nova.edu/ssss/QR/QR3-3/tellis2.html. 23 For instance, Madagascar, Malawi, Tanzania, and Zambia all had multiple reforms within the agricultural sector and focused on market liberalization. Madagascar was not included among the nine cases, even though it was complete, because Zambia and Tanzania had been selected and had similar characteristics. The Uganda, Sri Lanka, and Vietnam PSIAs focused on land reform, but were not selected for further analysis, because they were not completed at the time of the analysis. 9 Task team members were asked to provide information regarding the PSIA approach and the progress of the research-policy dialogue, and to clarify questions raised by the secondary documents.24 They were asked to describe the genesis of the PSIA, how country context conditions influenced the decisions made and which PSIA approach was taken, how the PSIA affected the process of policy dialogue, and whether results from the PSIA were used in policy formulation. Special attention was given to discussing the process of dialogue with the government, and government's participation. 24 See annex B for a summary of the information registered during and subsequent to the interviews. 10 II. Evidence from Social Analysis: Stakeholders, Power Relations, and the Policy Process in Agricultural Reforms This section reviews the agricultural reforms carried out in the sample cases, their main characteristics, and the rationale for conducting an exercise in Poverty and Social Impact Analysis. It then examines the Poverty and Social Impact Analysis work carried out for each reform, focusing on the social contribution in three main areas: stakeholder analysis, the impact of power relations on reform design and implementation, and the ways in which choices concerning the PSIA process enhanced or facilitated its impact on policy choice. TYPES AND REASONS FOR AGRICULTURAL REFORMS From the cases analyzed, economic efficiency and improved performance of the agricultural sector were the two most common reasons for implementing agricultural reforms, while distributional impacts and improved food security were also important (See table 2 below.) Sector-wide and sub-sector reforms in agriculture are frequently motivated by efficiency losses associated with state-controlled marketing systems. These losses mean that producers often receive a low share of border prices, that too many resources are being consumed in the marketing process, and that distorted signals are being sent to producers, consumers, and marketing agents. With regard to land, reforms are often needed because existing land allocation systems are inflexible or are associated with inefficiency. State-driven land allocations often reduce incentives for investment in lands, slow up the transfer of lands to their best usage, and can constrain access to agricultural credit. All of these efficiency-related concerns retard agricultural output, reduce export earnings, and cause misallocations of resources. Fiscal concerns can also arise when government budgets are used to finance inefficient state-owned enterprises. Employment in the public sector is increased, and public-sector activity displaces private investment. Equity concerns were also mentioned as a driver of the reforms in the cases examined (Cambodia, Tajikistan, Malawi, and Chad). The poorest of the poor tend to be dependent on agriculture as a source of livelihood and food, so that efficiency-lowering distortions can have a heavy negative impact on them. Policies that raise the producers' share of border prices and increase transparency in price formation generally benefit poor producers. However, in some cases, the poor may be especially vulnerable to market reforms, and PSIA can help create interventions to mitigate their impacts. The type of reforms in the case studies included reforms of the agricultural marketing system (Malawi); sub-sector specific marketing reforms in cashmere (Mongolia), coffee, cashews, and tea (Tanzania), cotton (Benin, Chad, Tajikistan, Tanzania), and sugar (Guyana); input markets (fertilizer in Zambia); and land (Cambodia, Zambia). In each of the marketing system cases, the reform in question sought to reduce or replace state involvement in the marketing system which was known to be inefficient and to incur fiscal costs. In the case of land reforms, private ownership or lease-holding were to replace traditional land allocation systems. In the cases of Tajikistan and Tanzania governments explicitly recognized the necessity of reform. In these cases, the government sought to broaden domestic support for reforms due to disagreement among different stakeholders. In Tajikistan, social analysis uncovered how the current market regime affected stakeholders (mainly by asymmetric information and exploitative power on the part of processing and marketing agents). Economic analysis was used to quantify the degree of loss of vulnerable stakeholders. In Tanzania, stakeholder dialogues and participatory institutional analyses contributed to the policy dialogue and broadened support for policy change. A major motivation for governmental support of the 11 PSIA in Tanzania was the concern about crop board performances and their impacts on crop industries including limited competition, unfair trade practices, low levels of value added, low volume of transactions in input supply, and a perceived decline in product quality. The government recognized that the institutional structure for production and marketing was affecting the competitiveness of the export crops, and that crop boards were part of that structure. Table 1. Reasons, Types of Reforms, and Distributional Impacts Country Area of Distributional Impacts Reform Reason for Reform Type of Reform Emerging from the PSIA AFRICA Benin cotton Improve efficiency in cotton rivatization Mainly on incomes of small- marketing; increase cotton production scale farmers and farm and exports laborers--some of the poorest of the poor Chad cotton Increase efficiency in cotton privatization High level of poverty among marketing; improve agricultural small-scale cotton producers productivity; enhance incomes and food security for small-scale producers Malawi agriculture Increase efficiency in agricultural agricultural marketing Loss of access to marketing market marketing services in underserved, remote rural areas; possible improvements in producer prices from more efficient agricultural marketing Tanzania crop boards Enhance performance and reform of crop boards Choice of exact reform competitiveness of export crops options still subject to a national debate; which stakeholder groups will be affected and how is unknown Zambia fertilizer and Increase efficiency in fertilizer land reform and Disadvantaged poor might land distribution and land allocation; removal/reduction of attain increased land access; reduce government involvement in fertilizer subsidy more remote rural poor might the agricultural sector not be served by private fertilizer markets EAST ASIA AND PACIFIC Cambodia land market Redistributive land reform and Government's reform choices distribution still pending; political commitment to release suitable land Mongolia cashmere Increase efficiency in cashmere assessing the impact Gains to producers from more marketing; enhance production and of trade reform in the efficient marketing exports of cashmere; increase cashmere sector on incomes of small-scale producers the income of the herders EUROPE AND CENTRAL ASIA Tajikistan cotton Improve efficiency in cotton privatization Benefits to remote rural poor marketing; increase cotton production and to smallholders who lack and exports access to key inputs LATIN AMERICA AND CARIBBEAN Guyana sugar Enhance efficiency in the sugar restructuring Employment loss by low- sector; increase sugar production productivity and income workers; social and exports; fiscal sustainability international disruption because sector has competitiveness in long provided social services the sugar industry 12 Selection of Reforms for PSIA From the 23 reforms included in the larger sample, four general patterns emerged regarding the reasons for conducting a PSIA: · In some cases (e.g., Malawi, Tajikistan), funding became available (from donors), enabling the Poverty and Social Impact Analysis to tackle areas of established concern. In such cases, PSIA was viewed as a tool of analysis or a source of funding for an analytical process, not as an ongoing and participatory process.25 · Important distribution-sensitive reforms were being considered by the governments and donors with some disagreement about their appropriateness, and the PSIA process was selected as the framework for assessing positive and negative impacts (for example, in Benin, Tanzania, Chad, and Cambodia). This approach represents the "ideal" cases, where the PSIA is client driven. As will be shown below, a demand-driven selection of a reform is neither necessary nor sufficient for high impact of PSIA results on the reform program. · Donors wanted to "test" the PSIA approach and demonstrate its utility for policy, especially when concerns about distributional impact were high. As a result, they looked for proposed or contemplated distribution-sensitive reforms and applied the PSIA to the problem (e.g., Zambia, Guyana). · Finally, some PSIAs were undertaken to shed additional light on an existing policy question or a controversial issue with little ex-ante movement/momentum toward reform, in order to engage in a national dialogue on the subject. In these cases, the focus on distributional analysis was often neglected, and findings resembled sector analysis studies with generally limited emphasis on stakeholder participation and links to policy design (e.g., Mongolia, Tajikistan). However, the analysis of the cases indicates that while most cases were selected because of a priori evidence of large and distribution-sensitive impacts, or perceived political sensitivity, the concern in some (Tajikistan and Mongolia) was not primarily about distributional impacts. In the latter the PSIA was designed instead to provide insights into efficiency gains/losses attributable to the reforms. Some studies focused on sectors representing a relatively large part of the (agricultural) economy and/or agricultural exports (in Mongolia, cashmere; in Chad, cotton; in Tajikistan, cotton; in Tanzania,cotton, coffee, cashews, and tea; and in Guyana, sugar). In a sub-set of these cases, selection was due to distributional concerns; the sector itself was large, but its reform could potentially have major impacts on disadvantaged populations (Guyana and Chad). Even in the cases in which the economic analysis focused on quantifying efficiency losses associated with existing state-driven marketing systems, the analysis uncovered distributional consequences associated with inefficiencies of the current system, therefore building support for the reform. For example, in Tajikistan, stakeholder and institutional analyses of cotton marketing chains demonstrated that disadvantaged producers were being exploited by other participants in the chain (farm managers, cotton gin owners, lenders), and that in addition to aggregate efficiency gains, disadvantaged producers would benefit disproportionately from liberalized cotton marketing. It is fair to conclude that, even in the cases where the reasons for conducting PSIA were narrow, most PSIA were driven by mixed motivations. Controversy of Proposed Reforms All the reforms subjected to PSIA were politically sensitive due to the size of the reform and the perceived large distributional impacts. Controversy around the proposed reforms was of three general types: (i) uncertainty about the significance and magnitude of distributional impacts; (ii) disagreements 25 Impact of these PSIAs was limited by failure to obtain broad government buy in and failure to engage in an iterative research process. 13 about the relative weights placed on distributional outcomes (and tradeoffs between equity and efficiency outcomes) by different actors; and (iii) questions about the overall efficiency gains of the reform. Issues of regional equity can be particularly sensitive politically. This is a form of "horizontal" inequality which is frequently associated with political conflict and can be a major feature in promoting violent conflict.26 The cases indicate that it is often necessary to pay particular attention to the impacts of reforms on residents in remote rural areas. Historically, agricultural policies (e.g., subsidized or government- controlled marketing or pan-territorial pricing) have supported the spread of productive activities to non- economic, often distant, locations. Marketing reforms that reduce the state's role may have an especially powerful impact on these remote producers, and social policy mechanisms, such as safety nets, may be required to soften the impact of change. It is notable that these groups are least likely to have a strong voice in the policy debate. An effective PSIA can provide isolated groups a voice that they may not have had. Remoteness was a concern in the Zambia, Benin, Chad, Malawi, and Mongolia PSIAs. It can be concluded that the motivation for engaging in a Poverty and Social Impact Analysis needs to be well understood by and transparent to stakeholders in order to influence the design, results, and impacts of a proposed reform. There are indications that an inherent interest in equity concerns (rather than economic efficiency or fiscal concerns) provides a better platform for governments to enhance the public support base for the reforms. A PSIA undertaken in that context would have an enhanced likelihood of resulting in open policy deliberation. STAKEHOLDER IDENTIFICATION The analysis of stakeholders in the case examples is summarized in Table 3 below. Interest groups were not only analyzed along income categories, but also disaggregated along characteristics that included household size, household head, gender, ethnicity, locality, and occupation. How Were Stakeholders Identified? Stakeholders were identified either a priori or in the research process as a result of social analysis: PSIA team members indicated that primary stakeholders were identified by the team members themselves, with secondary stakeholder groups sometimes identified in focus-group discussions with local counterparts. However, stakeholder identification was not always followed by an analysis of their interests or "stakes" in the reform process. In several cases, new stakeholders were identified later in the research process and included for further study. The Zambia study was an exception, where groups with a stake in the reform were identified late in the analysis, and the process was altered to investigate their concerns more fully. In other cases (Malawi, Tanzania, Mongolia), stakeholders were identified a priori, based on knowledge of the country and reform context, or through informal consultations with government or study partners. Such a process is appropriate where ethnic diversity is minimal or stakeholders are "obvious." In Cambodia, appropriate stakeholders were identified through consultations, but many donors were unaware of the political sensibilities related to land distribution. 26 See Stewart (2001). 14 Table 2. Stakeholder Analysis Country Reform Stakeholders Examples of Stakes Identified by the PSIA AFRICA Benin cotton Small-scale The impact on the stakeholders is mainly on their income and producers their lack of minimal social safety nets. Farm laborers Women Marketing agents Chad cotton Cotton farmers Cotton employees would be most affected due to huge lay-offs. Farmer Farmers, farmers' organizations, and local traders would gain organizations from their access to inputs, assets, and capabilities (possibly-- would depend on the response of the private sector). Local traders Cotton employees Government employees Malawi agriculture ADMARC ADMARC (Agricultural Development and Marketing Corporation) market employees employees would be affected by the reform through reduced Producers and employment, the Ministry of Finance - through reduced fiscal consumers in burden, and private traders - through increased market share. isolated areas Producers and consumers in isolated areas would be adversely affected. Ministry of Finance Private traders Local politicians Tanzania crop boards Central government All stakeholders in the export crop industries will be affected. The Farmer associations Ministry of Agriculture and Food Security and the Ministry of Cooperatives and Crop Boards have significant influence over the Traders reform and could be expected to oppose reforms, which reduce Processors and their control over private agents and enhance the need for exporters greater accountability on their part in the crop industries. Traders Ministry of support the reforms, but they may not have as much influence Agriculture and over policy as other stakeholders. As the choice of exact reform Food security options is still subject to a national debate, it remains to be seen Ministry of which stakeholder groups will be affected and how. Cooperatives and Crop Boards Zambia fertilizer Women Women will be affected positively by the reform because land and land Fertilizer marketing legislation does not discriminate by gender. However, under most employees customary systems, women lack access to land. The Ministry of Lands would benefit through revenues from titles, taxes, and Ministry of Lands rents. Small-scale farmers who own valuable land and have Small-scale farmers marketing potential might benefit because their land could be Foreign investors used for collateral. Commercial banks Surveyors EASTERN ASIA AND PACIFIC Cambodia land market Government Landless households would benefit the most from the reform institutions because it would stimulate the economy and alleviate poverty in Non-governmental rural areas where landless households do not have the capacity organizations to provide for their families. The military would be affected negatively (because it expropriated land during the war) and Military 15 Country Reform Stakeholders Examples of Stakes Identified by the PSIA Landless could potentially impede any legislation that may strip it of land. households Mongolia cashmere Herder households Herders, whether individually or via cooperatives, associations, or Cashmere traders and community associations would support the sector reform as and exporters the main beneficiaries. Local cashmere traders, including independent traders at both central and provincial levels, would Community support improvements in cashmere trade. If policy distortions are associations rectified incomes of the herders could increase by over 20 Cashmere percent and the export potential of cashmere could increase by processors over a third. Institutional stakeholders, i.e., the line ministries EUROPE AND CENTRAL ASIA Tajikistan cotton Government of Cotton gin owners, farm managers, and local governments were Tajikistan absorbing income that should accrue to farmers. The inability of Local government farmers to make decisions as to acquiring financing, gin, and marketing their produce has resulted in their exploitation by Cotton gin owners intermediaries. Lenders Small-scale farmers LATIN AMERICA AND CARIBBEAN Guyana sugar Ministry of Finance The ministry of finance would gain from increased revenues, and Sugar sector large private landowners would benefit from increased production workers capacity. Sugar-sector workers would be affected negatively because they could potentially lose their jobs as a result of Small-scale farmers closing inefficient sugar mills. Small businesses that depend on Landowners sugar industry workers would suffer from reduced demand. Small Small businesses farmers would not be able to benefit from the reform because Labor unions they cannot compete with large private landowners. Labor unions Government could strike against state closure mills and could influence the reform through opposition. The government could also impede reform by continuing to support inefficient mills through sugar levies. What Information about Stakeholders Did the Analysis Reveal? For agricultural reforms, stakeholder identification assisted particularly in identifying spatially separated stakeholders. In both the Zambia and Malawi cases, remote smallholders were identified as vulnerable to agricultural marketing reforms. In the Zambia land reform case, however, further analysis may have refined the geographic focus regarding areas with potential for land-related conflict. In cases where tribal or ethnic divisions exist, stakeholder analysis informed the research design of the impact analysis of the PSIA. In Guyana, an extensive stakeholder analysis was undertaken because of strong ethnic divisions. Stakeholder analysis also helped broaden perceptions about who might have an interest or a stake in the reforms, especially in the cases of Zambia, Guyana, Tajikistan, and Chad. The review of the cases found considerable variation in the extent to which analysis of social diversity and gender informed the stakeholder analysis in particular, and the PSIA more broadly. In a number of cases, little explicit attention was paid to the impacts on women, children, landless laborers, or other 16 specific social groups. One exception was Zambia, where stakeholder analysis, informed by a sound understanding of social processes, revealed a couple of key issues for the reform process (see Box 1).27 In cases where economic efficiency concerns were the major motivation to design the reform program, the stakeholder analysis was weaker and traced impacts of reform on stakeholder to a lesser extent than in the cases where distributional concerns were highlighted (Tajikistan, Mongolia). Box 1. The Stakeholders Influencing Land Reform In Zambia The PSIA assessed the poverty and social impacts of a controversial land reform proposed by the Ministry of Land. In its draft land policy, the government proposed titling and converting some of the 94 percent of land under customary law into state-owned land. The reasoning behind the proposal was that secure title would allow longer-term investments in productivity-improving measures, better access through collateral to credit markets and smoother transfer of land to more-efficient uses. Other benefits included improved dispute resolution, possibilities of revenue- generation by local governments, and more equitable access to land, with specific measures to reach the poor and women. Several previous studies identified land titling to be a critical step needed to improve investment and efficiency for smallholders and commercial producers. Many donors had noted that through more market-driven processes many of the problems facing the Zambian agricultural sector would be cured. However, past reforms have not been implemented as originally designed due to controversies and resistance. Consensual solutions again have not had the intended poverty-reducing and economic development effects. There was a clear risk that the 2002 reform proposal would follow a similar path of political conflict and resistance. The PSIA team paid resolute attention to an inclusive and consultative process in order to create ownership for the research findings and identify the common ground amongst civil society, policy makers and donors. The research findings revealed surprising results that made a substantial case for re-examining the underlying logic of privatization of land in Zambia. Findings demonstrated that access to land is only a weak determinant of well-being for most households in rural Zambia. Access to other assets, such as livestock, land quality, farm capital, input and output markets, and agricultural services are equally important parameters. Labor and other capital assets were more important to typical smallholder farmers. Given available labor and existing technologies, smallholders were not able to efficiently farm the land they had. While tenure security (proposed under the reform) is recognized as beneficial to investors, a lot of different stakeholders were skeptical of the central government's ability to guarantee such security. Rural Zambians were skeptical of private land markets and many were of the opinion that it would eventually lead to increased landlessness and rural poverty, instead of more agricultural investment and employment. Women expressed concern about rights to hunt and gather on privately held land and were doubtful as to whether a formal legal system would be as strong as traditional systems in protecting their rights. Although title to land might be an effective form of collateral for agricultural loans, stakeholders had evidence of alternatives (such as group liability, contractual arrangements) and questioned whether credit would be widely available even with title. The most compelling arguments against reform of the traditional tenure system were related to power balances in rural areas: stakeholders felt that by removing the power to allocate land from traditional authorities, a power vacuum would be created in rural areas, possibly undermining peace and cohesion in the communities. The stakeholders' reservations about fast-track titling were identified early on in the process and could be substantiated in the subsequent analysis. These findings persuaded policy makers to rethink the necessity of a major revision of existing land policies in Zambia. Marginal policy changes and more attention to institutional development might instead be enough to stimulate the needed growth in the agricultural sector. In the Tanzania PSIA, stakeholder analysis revealed that existing institutions--specifically crop boards-- served as both regulators and participants in markets. This represented a conflict of interests. Additionally, boards were found to be only accountable to the Ministry of Agriculture but not to the crop 27 In Zambia, the impact of removing land allocation responsibility from traditional authorities was deemed too risky by many stakeholders. Traditional authorities represent the primary (often only due to limited coverage of civil authorities) means of resolving disputes. Their authority would be diminished by loss of land allocation powers, and stakeholders were extremely skeptical of the public sector's ability to fill the vacuum. In the case of fertilizer distribution in Zambia, respondents noted that the two publicly supported programs for fertilizer distribution relied on group formation to ensure loan repayment. These groups noted that their responsibilities had expanded and there were subsequent positive social impacts. 17 industries (their constituents), which they were supposed to serve and from whom they drew their funding. As a result, a more appropriate means of financing the crop boards was proposed. Impacts of agricultural reforms, such as marketing system changes, land reform, etc., are complex; they often have indirect effects; and are not well understood by stakeholders. Reforms are often proposed for efficiency reasons-- to achieve enhanced resource allocation in the rural sector or to promote growth in agriculture-- yet impacts on individual stakeholders are by definition distributional. In rural areas, direct stakeholders (generally small-scale producers and workers in affected sectors) are suspicious of reforms and almost universally feel threatened by them. Even when the existing sectoral structure has not been beneficial to them (e.g., small-scale cotton producers in Tajikistan and coffee, cashew, tea, and cotton producers in Tanzania), reform is a tough pill to swallow. Many feel that efficiency gains will not accrue to them. They are also skeptical of private sector responses.28 It is essential that stakeholders are informed throughout the PSIA process to increase their knowledge about impacts and build their confidence that problems will be addressed. PSIAs that presented policy options for reforms and means of mitigating negative distributional impacts (such as Guyana and Tanzania) generated greater stakeholder support. POWER RELATIONS AND THE REFORM PROCESS Recognizing the central role that power relations play is critical for predicting how the implementation of a reform program will play out, particularly when the reform affects existing structures. For example, in Tajikistan, the "middlemen" in the marketing chains--farm managers, lenders, gin owners, and in some cases, local governments--were capturing most of the rents from the sector. In Zambia, reform of traditional land allocation patterns led to a power vacuum in rural areas as the authority of the traditional authorities was undermined. In Chad, farmer's organizations had the most at stake from privatization of the cotton sector and their lobbying could have prevented the reform from occurring. Tanzania's market liberalization in the early 1990s increased growers' share of export prices and attracted private sector investments. However, these policies have not been sustained or been effective in some cases. Concerns about declining quality of exports, non-competitive behavior, lack of information and credit options, and limited input supply and output marketing channels led to a new round of government regulations that enhanced the power of crop boards. The extent of these issues (including the boards' lack of accountability to industries) and the effects of power relations on industry dynamics and performance were uncovered as a result of the analysis. Market power on the part of cashmere marketing agents was identified as a problem in Mongolia. However, in none of these cases was further analysis applied to determine the implications of this market power (and rent-seeking behavior in general) within the political arena. In the Benin and Tajikistan cases attention was given to the relationship between the political interests of elites and their influence on the reform proposal. The analysis uncovered how political forces were slowing movement toward reform. In Tajikistan, elites were capturing rents under the existing marketing framework, and hence represented an obstacle for the reform. The analysis further revealed that smallholder farmers had the most to gain from the reform, and that power relations impeded what could have been a pro-poor policy outcome. In Benin, weak rural institutions and the government's inability to enforce written laws led to power imbalances and the ability of elites to control information and slow the reform. Poor smallholders, due to lack of organization and high transaction costs, had weak bargaining power and were subsequently exploited by the existing arrangements. In Benin, evidence suggested that cotton production was threatened by a breakdown in existing marketing and credit institutions, and that rent-seeking agents would oppose any effort to reform the current system. Rent-seeking opportunities created incentives for political opposition to further privatization, and conflicts threatened to derail the 28 The theme of inadequate private sector responses to liberalization was highlighted in Malawi, Mongolia, and Tajikistan. In Benin, Tajikistan, Zambia, and Tanzania, social analysis was applied to better understand sources of private sector weakness. Lack of institutional guidance and will was prominent (e.g., no fertilizer policy in Zambia, weak contract enforcement in Mongolia and Benin), as was transparency in private-public sector interactions (Tajikistan, Benin, and Mongolia). 18 entire process. Such rent seeking is not unique to rural areas or agricultural sectors, but in these cases, the interests of geographically dispersed stakeholders may be sacrificed to those of entrenched elites. In Cambodia, many were aware of the power of land allocation in supporting the central government's authority; however, only a detailed analysis of social relations highlighted the importance of this dimension. The high degree of sensitivity to land issues may have reduced the ability of the PSIA to "push" the envelope and suggest controversial elements of the reform: thus, political sensitivity affected the scope of recommendations. Institutional and Political Economy Risks Most of the political economy risks identified during the PSIA were related to power relations: elites had been capturing rents under existing arrangements and would oppose efforts that would reduce or eliminate these rents. (See Table 3 below.) As noted, the Benin, Tajikistan, and Mongolia cases were clear examples of rent capture by elites and resulted in political opposition to the reforms. Capture of rents was identified through both the social and economic analysis, with detailed interviews about the structure of existing marketing relations providing the deepest insights. This capture of rents is especially difficult to uncover in rural areas as transactions costs are generally high, and as a result, large marketing margins are common. Distinguishing between a transaction cost-related marketing margin and rent capture is difficult and requires careful investigation along the marketing channel. Table 3. Main Political Economy and Institutional Risks Emerging from PSIA Country Reform Political Economy Risks Institutional Risks AFRICA Benin cotton Rent-seeking elites would object Private sector response to dismantling of to reforms public marketing system would not be Losses by elites would not be adequate outweighed by gains to poor Laws and regulations would not be producers enforced PSIA recommended an in-depth Weak public institutions have potential to political economy analysis to undermine intended impacts of reforms evaluate the impact of politicization on reform process Chad cotton Some conflict of interest between Without accompanying measures to support those cotton producers who can farmers, the industry runs the risk of afford to buy labor and those who disappearing, or at best, over the long-term, cannot there will be a small core of producers in the West who have knowledge and experience, and access to inputs and markets of neigh- boring countries Malawi agriculture Not identified Private sector response would be inadequate market in remote rural areas Tanzania crop boards Conflict between the role of crop Private sector is unable to take over boards as regulators and previous public service function providers of services in commodity Inability to fund marketing boards markets Partial or incomplete reform in certain Proposed reform options that institutional arrangements without overall reduce controls over private change in institutional environments which agents and increase accountability shape the crop industries of boards to their constituents may be rejected 19 Country Reform Political Economy Risks Institutional Risks Zambia fertilizer and Traditional authorities would Private marketers would not provide land undermine reform essential services Conflicts of interest would develop Private services would not deliver inputs in between "beneficiaries" of land a timely fashion reform and those who would bear Power vacuum would develop in rural the costs areas from loss of authority by traditional authorities Lack of adequate government presence presages inability to enforce new rules and regulations EAST ASIA AND PACIFIC Cambodia land market Existing land allocation system Land conflicts are common, yet the court was an important means of system does not adequately adjudicate keeping regime in power land disputes Strong objections to change were No formal delineation of state land exists detected There is no legal basis for identifying and converting public or private state land for other users Rapid expansion of program would over- tax government Mongolia cashmere Rent-seeking elites would object to Private sector response would be inadequate reforms EUROPE AND CENTRAL ASIA Tajikistan cotton Rent-seeking elites would oppose Private sector response would be inadequate reforms Political economy considerations depended on the particular option LATIN AMERICA AND CARIBBEAN Guyana sugar Political conflicts between two main Access to social services would be lost in ethnic groups would influence reform areas where sugar facilities are closed path. Other social risks exist Fear of unemployment and inability of economy to create alternative employment are risks In Cambodia, the political sensitivity to changes in land policy was under-appreciated by many donors prior to the start of the land reform program. This political sensitivity could have led to national crises had the land reform been poorly managed. Other political economy risks included conflicts between different government ministries, inadequate engagement of key stakeholders, and lack of confidence by stakeholders in the efficacy of the reform. Social analysis contributed to managing all of these forms of risk. For example, an institutional analysis at the start of the PSIA could identify conflicts between government ministries that needed to be addressed. In the Tanzania case, two ministries were engaged in planning and implementing the PSIA via a task force comprised of the Ministry of Agriculture and Food Security and the Ministry of Cooperatives and Marketing. This engagement supported access to stakeholders, often provided sensitive data during the analysis, and enhanced the policy dialogue on the reform design and implementation. In contrast, the 20 Zambia and Malawi PSIA did not identify tensions between ministries, and the role of the Ministry of Agriculture in implementation of the reforms was practically ignored. As a result, conflicts between the Ministry of Agriculture and the Ministry of Finance (the sponsoring ministry) continued after completion of the PSIA. When important stakeholders are not engaged in the process, they can create political obstacles to any reform. Social analysis, by identifying stakeholders, helped mitigate this in Benin, Tajikistan and Zambia, where detailed stakeholder analysis identified groups at risk from the reform. The second major form of risk is related to institutional weaknesses that can prevent successful implementation of a reform. This risk was clearly demonstrated in Zambia, where land tenure policy change was expected to upset existing power relations in rural areas, especially the prerogative of the traditional chiefs to resolve conflicts and adjudicate disputes. Stakeholders expressed little confidence in the ability of the state to replace these functions. In Cambodia, unclear jurisdictions and incomplete delineation of state-owned land could have undermined progress toward the land reform. In Tajikistan, the team found considerable institutional problems and concluded that strong actions would be needed to overcome existing political resistance. In Tanzania, improving the competitiveness of export crops was a key concern, and the government, particularly the Ministry of Agriculture, was concerned with the performance of the crop boards and the export commodity markets. A major conflict of interest was discovered during the course of the PSIA: crop boards compete in the industries they regulate and they currently mix their regulatory responsibilities with their service functions. In Guyana, the donor team uncovered substantial problems with implementation of the reform and management of mitigation measures. A key concern of several PSIAs was the response of the private sector to the declining role of the state in the sector. This issue emerged in Zambia, where concern about the private sector's ability to provide fertilizers, especially to remote rural smallholders, was a primary issue. Similar issues arose in Malawi, where remote farmers were found to be much worse off than those close to ADMARC (Agricultural Development and Marketing Corporation) depots, and respondents in the social analysis indicated that private agents did not operate as intensively in these areas. In Malawi, concerns about private alternatives to ADMARC markets as providers of input and output marketing services were uncovered. The private sector response was identified as a potential risk to remote smallholders. In Tanzania, the performance of private actors in the export commodity markets was a major concern of stakeholders, and concerns about private responses were found in other studies as well. Several of these studies identified the necessity of analyzing potential institutional changes to stimulate private sector involvement, but few conducted sufficient analysis to predict or understand institutional responses. In Benin, weak rural institutions and inability to enforce written laws led to power imbalances and the ability of elites to control information and slow reform. Poor smallholders, due to lack of organization and high transaction costs, had weak bargaining power and were subsequently exploited by the existing arrangements; however, the analysis showed that unless specific actions were taken to improve the legal status of small-scale producers, improve information flows, and enforce market regulations, exploitation would likely continue following the reform. Continued power imbalances constitute a real risk for such reforms. Agricultural and rural sector reforms often imply significant risks to stakeholders. These risks (e.g. increased exposure to market fluctuations, increased institutional risks) were identified through stakeholder involvement as detailed in the Zambia, Chad, Tanzania, and Cambodia Cases. Analysis of risk is often complicated by data limitations, such as inadequate panels.29 In the case studies, social analysis was often the primary tool for understanding risk impacts and risk management. In Guyana, for 29 See Siegel and Alwang (1997) for a discussion. 21 example, both economic and social analyses identified risks and their implications. The stakeholder analysis contributed useful details by soliciting information on hypothetical responses to a more risky environment. In Tajikistan and Benin, political risks to reform were identified through social analysis of marketing processes and power relations. Clearly, the analysis of risk can be approached through a variety of disciplinary and methodological approaches. The case studies show the value of using flexible social analysis approaches to identify stakeholder perceptions, political economy and institutional dimensions of risk. Box 2 illustrates these issues in the Tanzania case. Box 2. Reforming Crop Boards in Tanzania? The structural changes after Tanzania's market liberalization in the early 1990s were intended to improve overall competitiveness of agricultural exports in world markets and to boost efficiency and productivity of Tanzanian farmers, including poor smallholder farmers in remote areas. Some institutional gaps after liberalization, however, contributed to distortions in the local agricultural market, including non-competitive behavior or unfair trading practices of some buyers, limited channels for input supply and output marketing--especially in remote areas--and a lack of accountability of boards to the industries they serve and from which they are financed. During 2003­2004, a joint government, EU, and World Bank PSIA reviewed the regulatory, service, and revenue collection functions of the coffee, cotton, cashew, and tea crop boards and proposed reform options to remove these distortions to enhance competitiveness of export crops. A stakeholder and institutional mapping exercise identified the various stakeholder interests and incentives, analyzed the power relations that have influenced regulatory changes, and examined the formal and informal institutional environments in which decisions are made, specifically in regard to (a) board roles, functions, and financing, and (b) industry dynamics regarding input supply and credit, quality, competition, and fair trade. The aim was to develop an understanding of why the crop industries are organized the way they are and perform the way they do, and to develop options that inform the design of crop board reforms while considering interests of various stakeholders that could impede meaningful reforms. Different power relations emerged with regard to stakeholders' influence, interests, and incentives for reforms, especially regarding the flow of funds, information, and decision making. The Ministry of Agriculture and Food Security and the Ministry of Cooperatives and Crop Boards have significant influence over reforms--they can be expected to oppose reforms that reduce controls over private agents and require greater accountability to producers for service delivery. Traders will support the reforms, but they may not have as much influence over policy. Source: Government of Tanzania and World Bank (2004), "Crop Boards and the Future Prospects for Agricultural Exports of Tanzania," a joint report of the government of Tanzania and the Environment and Socially Sustainable Development Department, Africa region, September 15, Washington, DC; Beddies, Correia, Kolavalli, and Townsend (2006), "Tanzania--Reforms of the Coffee and Cotton Crop Boards," in Analyzing the Distributional Impacts of Reforms: Operational Experience in Implementing Poverty and Social Impact Analysis (Washington, DC: World Bank). PSIA PROCESS AND POLICY INFLUENCE This section discusses process elements that enhanced or facilitated the impact of the PSIA on policy choice. The potential for stakeholder participation in certain process steps, the contribution of PSIA in changing public perceptions, and the role of dissemination of PSIA results are addressed. The section concludes by discussing the impacts that the PSIAs had on the policy process, which ranged from changes in the design of the proposed policy, to abandoning the reform proposal, to no changes at all. While the general issue of the process of conducting PSIA goes beyond social analysis, the focus in this section is on the contribution of social analysis approaches to framing the PSIA process - through providing understanding of stakeholder groups and their interests, power relations and institutional processes. Stakeholder participation in PSIAs can be understood as giving voice and influence to a more diverse group of stakeholders than the direct commissioners of the work (most often a donor agency and one government unit). As outlined by Robb (2003) and a discussion note by Schnell and Poulsen (2005), there are several entry points for stakeholder participation in a PSIA process. These include selection of reforms under study (there were no examples in the case studies reviewed here)30; the design stage of the 30 In practice, provisions for citizen participation in the Poverty Reduction Strategy process are likely to be the most significant forum in low-income countries. See Coudouel et al. (2006). 22 PSIA, including, scope, focus, and choice of research teams (e.g., Cambodia); governance of the process via steering committees or decision-making workshops (Cambodia and Chad); the research process--by diversifying composition of the research team or applying research methods which give voice to stakeholder perceptions of the reform which would otherwise not have been known (e.g. Chad, Tajikistan, Malawi); the development of policy options arising from analysis of findings (e.g., Cambodia and Chad); and monitoring of the results of the reform. The credibility of the technical work underpinning the PSIA remains central to its effectiveness in promoting policy change that benefits poor people, and broader stakeholder engagement needs to be organized to reinforce this.31 The PSIA process often led to changes in the design of proposed reforms. These changes were due to provision of information that changed prior perceptions; participation by stakeholders in the definition of the policy options; and dissemination of results in a manner which influenced policy change. The analysis shows that involving the ministries with at the highest stake in the reforms (including ministries of agriculture and land) early in reform processes affects their ultimate success (e.g., Cambodia, Mongolia, Tajikistan, and Tanzania).32 These ministries are often charged with implementing the reforms, and thus their participation in planning and implementation is critical. PSIAs, even rural and agricultural PSIAs, are most frequently initiated through other ministries, especially ministries of finance,e or entities related to the Office of the President. This institutional relationship is created because they tend to be charged with implementing PRSPs and are in charge of the overall fiscal and growth agenda which are often the main motivation for reforms. Ministries of agriculture have often been brought into the process well after the start of research, and thus ownership by a principal implementing agency has thus been diminished. In Malawi and Zambia, for example, the agriculture ministries were not brought in until the study was completed. In Guyana, the Ministry of Agriculture was not a part of the study and had little use for or confidence in its results. Effective use of the PSIA approach involves building broad acceptance of the study's findings. Dissemination is an important part of the process of evidence-based policy making. The user's guide to PSIAs (World Bank 2005, 35) notes that "the process of policy debate, including among stakeholders, can be just as important as the analysis." However, the experience of the case studies reviewed here indicates that wide dissemination of their findings did not occur, which limited the potential impact on public policy debates. In three cases, substantial dissemination did occur (Cambodia, Tanzania, and Zambia). In each of these cases, stakeholder groups had been brought together to discuss and evaluate findings, and recommendations were changed following these workshops. Team members indicated that these efforts had more impact on policy. In Cambodia, strong participation of other donors and non-governmental organizations in the study and subsequent dissemination was associated with far more evidence of sustainability. Team members attributed sustainability to involvement of many stakeholders in discussions and implementation of the PSIA,33 and the linking of the social concession program to ongoing donor and NGO projects. One part of the explanation for limited dissemination is the relatively short time between completion of the PSIA research and this study. Another part of the explanation is the relatively rapid turnover of donor staff. In Malawi and Benin, greater dissemination was planned, but the initiators of the study were no 31 See Appleton and Booth (2005) and Holland and Campbell (2005) for discussions of participatory methods and quality standards in poverty research. Holland (1998), Robb (1999), and Norton (2001) discuss broader issues of participatory approaches to research and their significance for the policy process. Coudouel et al. (2006) highlights the risks of distortion or capture of PSIA processes by powerful political economy interests and the need for rigor in research standards. 32 Success would be measured in terms of the impact on policy reforms and their adoption by host-countries. Ideally, one would measure success in terms of long-term improvements in impacts; these are difficult to measure, however, due to lack of information on the counterfactual and due to only minimal observed monitoring. 33 For a detail description of stakeholder participation in the Cambodia Case see Schnell and Poulsen (2006, 10). 23 longer involved in country-specific PSIA activities. Policy impacts were limited in these two cases, although the Malawi study helped change donor thinking. In Mongolia, Chad, and Tajikistan, dissemination suffered because the PSIA team had moved on to other assignments. Some case examples highlighted the point that dissemination may be hindered by the reward structure and other obstacles within the donor community, particularly within the World Bank (Chad, Malawi). Several interviewees indicated that ongoing post-PSIA activities are not highly valued by management, including a perception that once the PSIA research is completed, the Bank moves PSIA staff too quickly on to other activities. Other interviewees noted that the cumbersome approval process within the World Bank slows the dissemination process and, since PSIAs are not generally linked to ongoing projects, slowing down dissemination, in fact, disables it. Table 4. PSIA Process Country Reform Changed Perceptions Participation in Process How PSIA Influenced Reform AFRICA Benin cotton Prior beliefs about Unknown impact on PSIA produced additional efficiency losses of acceptance of results evidence in favor of reform cotton market Strong institutional barriers sys-tem were to efficient functioning of confirmed private markets were Vulnerable identified stake-holders were Lack of clear regulations shown to be and irregular enforcement exploited under were identified as important existing system constraints Enforcement of marketing rules was identified as a problem Chad cotton Sector has a strong Unknown impact on PSIA produced additional need to promote private acceptance of results evidence in favor of reform market development for all inputs of cotton production 24 Country Reform Changed Perceptions Participation in Process How PSIA Influenced Reform Malawi agriculture ADMARC reform will Ongoing dialogue between Donor staff perceptions market create gaps in donor team, government, about policy impact were access to services in and NGOs assured that changed by PSIA rural areas confidence was built in World Bank position moved Seasonal differences study's findings closer to that held by other in maize markets donors, NGOs, and and seasonal civil-society representatives differences in access ADMARC should still be to producer services restructured, but with efforts to replace its social functions in remote areas Tanzania crop boards Institutional gaps Joint analysis by PSIA informed design of after liberalization government task force, crop board reforms contribute to EU, WB (including PSIA provided input into distortions in local concept note) policy dialogue by agricultural market, Broad local participation producing a menu of reform Private sector serves (local university options remote areas researchers, industry inadequately representatives) Unfair competition Stakeholder dialogue in by some private parallel to analysis and agents via workshops to discuss Low credit and input findings and reform use--emerging design schemes show potential (farmer groups) Zambia fertilizer and Fertilizer markets Additional stakeholder Land reform process was land underserved remote concerns were identified slowed areas through broader Fertilizer options were Risks associated participation revised with land reform did Stakeholders "accepted" not exceed rewards results Private land markets Power relations were were emerging to underscored meet investor needs Consensus was built among stakeholders EAST ASIA AND PACIFIC Cambodia land market World Bank Ownership of results was PSIA produced additional underestimated the increased by participation of evidence in favor of reform, political complexities local consultants in research but uncovered subtleties that of the country when process were addressed as the it initiated the study process continued Greater appreciation of these complexities was gained The role of land in promoting food security and raising a household's status within the community was better appreciated Land is an important safety net in rural areas 25 Country Reform Changed Perceptions Participation in Process How PSIA Influenced Reform Mongolia cashmere Prior beliefs about Unknown whether PSIA produced additional efficiency losses of participation affected evidence in favor of reform cashmere market outcome system were confirmed EUROPE AND CENTRAL ASIA Tajikistan cotton Prior beliefs about Stakeholders began with PSIA produced additional efficiency losses of suspicion about evidence in favor of reform cotton market sys- liberalization and reliance Need for reform was tem were confirmed, on privatized commodity consistent with the World but magnitudes of marketing Bank position losses were not PSIA documented measured Menu of reform options exploitation and bridged helped create policy space Many vulnerable differences between stakeholders were vulnerable stakeholders Option of "doing nothing" shown to be and government parties was evaluated and shown exploited under who were mainly to be a poor choice existing system. concerned with efficiency and fiscal exposure Participatory social analysis was essential to this convergence of opinion Involvement of stake- holders in the research process increased the credibility of findings, particularly among the most vulnerable LATIN AMERICA AND CARIBBEAN Guyana sugar Social tensions Very little government PSIA appears to have would likely be participation in PSIA some impact after a long exacerbated by Lack of participation hurt time elapsed; the EU proposed reforms acceptance of study's offered the government Gender relations and findings by government.. financial support to access to social Most recently, however, implement the structural services might be the government provided adjustments in the sugar adversely affected some input on the PSIA. sector, but requests for an Sugar households analysis of social impact of were identified as any reform program has highly vulnerable been taken up Sugar employment Menu of reform options closely linked to rest helped create policy space of economy and to Safety net program viewed social sectors as essential in reform areas 26 The PSIAs had four types of impact on the policy process and envisaged reforms: (i) changes in the proposed policy (Zambia, Malawi, Cambodia); (ii) increased evidence of desirability of reform, making policy reform more likely (Tanzania, Benin, Chad, Mongolia, Tajikistan); (iii) more open and progressive debate about the policy, often accompanied by creation of a menu of options (Guyana, Tajikistan, and Tanzania); or (iv) no change, either because the results were not believed or they reinforced prior beliefs against the reform (Benin, Guyana).34 The PSIA process frequently faces tension between analytical results that find evidence in favor of or against a reform, and the political reality, in which stakeholder differences become highly contentious. The Malawi PSIA is the most prominent example of this in the study. One of the most effective means of addressing stakeholder differences is through dialogue and early inclusion of government and other stakeholders in the process. In the case of Tanzania, the government recognized that crop boards needed to be reformed to enhance the competitiveness of these export markets, and formed a respective task force, led by the ministry of agriculture and food security. Researchers, donors, the government, and other stakeholders identified that institutional gaps after liberalization contributed to distortions in the local agricultural market. These included non-competitive behavior, unfair trading practices, and a lack of accountability of the crop boards to the industries they served and by whom they are financed. The PSIA discovered that certain interests were vested in maintaining the status quo, and that engaging these key stakeholders in the process at the political level could reduce delays in policy decision and implementation. From the beginning, stakeholder participation was built into the PSIA design, and this design was shared with and endorsed by key stakeholders. While final policy decisions have not yet been made, indications are that the PSIA recommendations have gained traction within the government of Tanzania. In Guyana, similar degrees of uncertainty existed, but despite having a technically well- designed and iteratively integrated PSIA process, government buy-in was never obtained. The complexity of stakeholder interests in many agricultural policy issues means that a PSIA process, if well managed, can provide a forum for generating consensus around a reform agenda. (See box 3 for the example of Cambodia.) Depending on the level of rural isolation, the transaction costs of organizing and lobbying for rural residents can be quite high. Compounded by entrenched rural interests and concentrated power among a limited number of interest groups, reforms can be stalled. The level of suspicion toward both the private sector and the state in markets reinforces the need for consensus and broad dissemination in the PSIA process. 34 These classes of impact on the reform process are not mutually exclusive; in some instances, the proposed policy did not change, but the PSIA created an enhanced dialogue among stakeholders. 27 Box 3 Building Consensus around Land Redistribution in Cambodia The Cambodia PRSP recommends that social concessions be used in order to distribute vacant state lands to poor households. A comprehensive land reform package was proposed with the intention that social land concessions and legal entitlements of land will have a greater impact on alleviating rural poverty and increasing economic productivity by allowing rural households to use idle land productively and as collateral for capital investment. The PSIA occurred at a time when land concession was a high priority of the government, civil society, the public, and donors. Because of this level of public interest, the PSIA aimed at making more information available about the potential impacts of the land reform, so that the public discussions on could be founded on factual evidence. A great deal of attention was given to stakeholder dialogue throughout the entire process of the PSIA. It was built on a series of consultations with a wide range of stakeholders. The process benefited from a strong governing committee, which included the ministry of land management and Agri-Business Institute Cambodia (a local research NGO), and external financial and technical support from GTZ and Oxfam UK; this was perceived as broadening the perspective on social land concessions and contributed to wider acceptance of the research results. The consultations included the following steps: The research approach was discussed in a workshop in Phnom Penh, attended by the ministry of land management, donors, and consultants, where feedback was solicited to fine tune the approach. As a capacity-building measure, field research teams included a large number of government staff from different line ministries who would be involved in a potential land distribution program. Research results from the different "sub" studies were circulated among the research partners for comments and then distributed for wider discussion. In May 2004, a one-day workshop was conducted to share findings and recommendations from the PSIA. The workshop was attended by nearly 100 participants representing different national and decentralized government institutions, donors, and NGOs. The lively debates provided new insights and focus for finalizing the PSIA recommendations. The consultations helped identify common ground for discussion among stakeholders and strengthened mutual trust. This again elicited new information and suggestions for implementing the reform. The continuous involvement of NGOs, paired with research results, led to the proposal that NGOs can play a major role in a larger-scale social concession program whose implementation would certainly overstretch the capacities of government institutions. Engagement of commune councils and informal local groups helped create important operational relationships for the land concession program and lay the foundation for a participatory process in a World Bank project, initiated by the PSIA team. The research further revealed that it was much more difficult than expected to find and release suitable land for distribution, and that this process would take strong political commitment from the respective government institutions to proceed with the reform. The inclusive process of the PSIA helped bring these issues into the public debate and built alliances for moving the land distribution agenda, which enhanced the likelihood of the reform proceeding with public support. Source: Notes Andreas Groetschel, Chanty, Srey (2004), "PSIA and Cambodia's Land Reform Program," joint exercise of IMAP, OXFAM, GTZ, WB, and Agri-Business Institute Cambodia, PowerPoint presentation at the "PSIA and Agricultural Policy in PRSP Countries" conference, Rome, September 2004. Cases Where Changes Occurred in the Proposed Policy Zambia and Malawi represent cases where the PSIA contributed to changes in policy or in recommendations by donor staff or the respective government. In Zambia, strong momentum to privatize traditional lands was slowed; the PSIA provided evidence that the benefits of policy change had been overstated and costs understated. These findings were mostly gleaned through social analysis, which examined institutional capabilities and power structures. The analysis found weak institutions and potential problems associated with power structures. The economic component of the study also indicated that the amount of land accessed by smallholders was not a significant determinant of household welfare; the household model confirmed this finding. Correspondence between the economic and social analyses helped reduce the sense of urgency with respect to land policy reform. Part of the change in recommendations was due to donor staff reconsidering previously held positions, in part due to recognition by the government and donors that tinkering with land allocation mechanisms could have unintended consequences and could create stress in rural areas. Consensus was formed by involving stakeholders in the PSIA process and by the efforts of the team to build consensus, particularly among 28 representatives of civil society. The process was inclusive from the start, during the design of the research, identification of broader groups of stakeholders, and wider involvement in the policy dialogue.35 Policy space was created because the study was conducted before positions about land reform hardened. In the case of the Malawi PSIA, social analysis inputs shifted the donor's perception of the privatization process of ADMARC: that is, that it would reduce access to essential agricultural services, particularly by those in remote areas. Donor policy subsequently changed from strong support for ADMARC privatization to views consistent with other stakeholders. Decision makers accepted that solutions for the potential distributional consequences of ADMARC privatization had to be addressed prior to embarking on ambitious reform. Consensus was made possible in part because the PSIA findings were consistent with those from a recent Oxfam investigation of ADMARC--and donor position moved closer to that held by other donors, NGOs, and civil-society representatives. Dissemination was not essential because the change in perception was internal to the donor organization. Change in the Malawian position was less pronounced, mainly because the findings supported widely-held prior positions. Cases Where Increased Evidence of Desirability of Reform Was Provided Prior to the PSIA, the government of Tanzania admitted the need to improve the performance of the crop boards to achieve its agricultural growth targets. The government recognized that the institutional structure for production and marketing affects the competitiveness of the export crops, and that crop boards are part of the present structure. The study identified institutional gaps after liberalization that contributed to distortions in the local agricultural market, including non-competitive behavior or unfair trading practices of some buyers and a lack of accountability of boards to the industries they served and from which they were financed. The Tanzania PSIA contributed to increased domestic consensus that reforms of the crop boards were necessary. The PSIA provided input to the policy dialogue by producing a menu of options for addressing current inefficiencies in the export crop industries. Broad local involvement in the research design, stakeholder identification, and use of participatory methods all helped contribute to this consensus. Government support for findings was bolstered by early dialogue with government partners, the decision to share the concept note and PSIA planning efforts with the government, and involvement of several donors. Co-financing and intellectual support from the European Union also broadened acceptance of the study. In Tajikistan, stakeholders walked into the process suspicious about liberalization and reliance on privatized commodity marketing. The PSIA documented exploitation under the existing system and helped bridge differences between vulnerable stakeholders and government parties who were mainly concerned with efficiency and fiscal exposure. Need for reform was also consistent with the donor's position. The PSIA supported this position and built a stronger case for reform. Use of participatory methods was essential to coax the necessary convergence of opinion; involvement of stakeholders in the research process increased the credibility of findings, particularly among the most vulnerable.36 Cases That Fostered More Open and Progressive Debate about the Policy Several PSIAs improved the local policy dialogue by presenting a menu of options for reform and mechanisms for addressing social impacts (Guyana, Tajikistan, and Tanzania). Options can create policy space for decision makers, and when they are the result of a consultative, participatory process, they can broaden support for the ultimate decision. In Tajikistan and Tanzania, broad participation in the PSIA 35 The TIPS website (www.worldbank.org/tips) names three objectives for PSIA: evidence-based, pro-poor, and inclusive policy making. Inclusive policy making is defined here as supporting stakeholder participation and ownership in the PSIA process. 36 Benin had a similar outcome, but not enough time has elapsed to equitably judge the policy impact of the PSIA. 29 helped build consensus around this menu; participants realized that each option had positive and negative dimensions, but were willing to allow the political process to work and produce a solution. In Guyana, no consensus developed around the menu of options and it is unclear whether any of the options will be adopted. It is unfortunate that institutional relationships were inadequately developed. Cases Where No Change Occurred In those cases where no change in the proposed reform was observed, several factors are striking: (i) problems with timing, in particular, insufficient lead time for the PSIA findings to influence the reform program; (ii) no change in prior-held positions--the PSIA confirmed prior beliefs; (iii) inadequate dissemination; and (iv) lack of widespread agreement with the PSIA findings. In Tajikistan and Mongolia, prior beliefs about the efficiency impacts of the proposed reforms were confirmed by the study. The PSIA broadened support among other stakeholders about the desirability of reforms. Several studies had only recently been completed and the window for policy change was too small (examples include Benin and Guyana, which were completed in fiscal year 2004).37 Conclusions: Factors Constraining Impacts of PSIAs The full potential of the poverty and social impact analysis carried out in the cases to inform the policy dialogue has been constrained by several "process" factors: lack of government buy-in or weak institutional linkages in the PSIAs (e.g., agricultural sector reform PSIAs that do not involve ministries of agriculture as primary partners); limited effort at dissemination; and skepticism among donor staff about the ultimate utility of the PSIA. In no case was an adequate monitoring and evaluation system established, and this shortcoming limits the impact/utility of poverty and social impact analysis. Lack of dissemination and ongoing monitoring and evaluation has weakened impacts of many PSIAs. PSIAs are, as practiced, a one-off analytical exercise, although they have generally limited dissemination and virtually no sustainability. In no case were findings of the PSIA used to strengthen monitoring systems or was an ex post analysis either conducted or planned. This problem is caused by three factors: (i) donor staff turnover combined with little country-team ownership, (ii) fiscal-year constraints and administrative conditions that make multiple-year PSIAs difficult, and (iii) limited engagement of line ministries in the PSIA (discussed above). The first two problems are the donor's provenance and need to be addressed by upper administration. In many cases, the PSIA team conducted the analysis without buy- in by country team project managers. The latter group, without this ownership, has shown little interest in PSIA findings, and recommendations are often ignored. Subsequent projects might be used to support monitoring and evaluation efforts, but without institutional buy-in, no such support will emerge. Case studies indicate that monitoring reform processes can improve the policy dialogue in several ways. Reform intentions expressed on paper often differ from reality; sometimes these differences are due to political forces, sometimes due to learning by government as reform progresses. This was found in Zambian fertilizer markets during the 1990s. Monitoring can stimulate interventions to mitigate negative impacts of a policy change and thus help build long-term political support for reforms. Monitoring can also improve the PSIA process by providing information about which impacts were most important, which channels of transmissions created the largest distributional change, and which tools of analysis were most effective. However, as noted, there is no evidence of monitoring and evaluation from either the group of cases or the broader universe of agricultural-sector PSIA. The dialogue in a PSIA process begins with discussions between donor staff and government representatives in the host country on a potential reform area or policy program that would benefit from further scrutiny (Malawi, Chad, Zambia, Mongolia, Tanzania). The intensity of these discussions and the 37 This finding is consistent with results from the broader group of PSIA in agriculture; many of the studies have not yet been completed. 30 degree of agreement on the need for a PSIA are important to the ultimate success of the reforms. For example, the Guyana government only gave lukewarm support for the GUYSUCO PSIA and this limited host country ownership ultimately diminished its impact. Other process factors affecting impact and sustainability are involvement and buy-in by "appropriate" host country government ministries; interactions and linkages within the donor and long-term commitment by country teams to see that PSIA recommendations are pursued; partnerships with other donors and NGOs; involvement of host country institutions, such as representatives of civil society and stakeholders; and capacity created during the process. Budget, skills, and time requirements for each PSIA selected varied according to the needs and capacity of each team. These issues are important to consider but should not be a strict guideline for all PSIAs. For the nine PSIAs reviewed in detail, budgets ranged from US $72,000 to almost $300,000. The time allocated to conduct a PSIA for all of the nine cases ranged from one to two fiscal years. Most PSIA teams stressed that a joint government- and donor-produced PSIA takes significant time, effort, and flexibility to accommodate perspectives, interests, and bureaucratic requirements of partners. Throughout the PSIA process, decision making was subject to negotiation, and thus timeframes and resources for delivering outputs had to be stretched to accommodate multiple stakeholder participation. In some cases, coordination with numerous partners delayed fieldwork and data analysis, which often made it challenging to work with the Bank's conventional fiscal-year delivery targets. As in the case of Tanzania, the PSIA team members expressed that, in addition to time requirement, skills, and budget, the real challenge lies in policy implementation. PSIA begins to make its biggest impacts after the report is completed with the implementation of proposed reform options. The team recommended that required time and funding for reform implementation needs to be adequately budgeted, whether through the respective government or development partners. Additionally, implementation of policy is subject to frequently vested and often conflicting interests among stakeholder groups--even within one group, such as the government. Hence, when designing PSIAs, greater attention is needed on planning adequate resources and timeframes for dissemination, policy dialogue, and reform implementation. Dissemination has been a major issue in most of the cases selected, and has been undercut by lack of funding or weak participation by governments. Impact of PSIAs on the World Bank's Operations Perhaps the biggest challenge the PSIAs had after they were completed was making a broader impact on donor operations. Examples here include Guyana, Tajikistan, and Benin where the findings were not applied to policy design after the analysis was concluded. Poor dissemination, weak participation from governments, and sometimes institutional arrangements within the donor institution weakened the possibilities of an effective impact on policy reform, much less on future donor operations. Most PSIAs, however, fed into the PRSP and occasionally provided analytical inputs to PRSC and ongoing reforms in each of the sectors, such as in Chad and Benin, where the PSIA augmented the ongoing 1999 Cotton Sector Reform Project (CSRP). In the case of Guyana, given that the PSIA was just completed at the time of this analysis, the impacts from the PSIA on operations have been difficult to establish. Initially, the PSIA was intended to be linked to other donor operational instruments, such as the PRSC II. Although the PSIA was basically completed with little government buy-in, the fact that the European Union had proposed a special annual fund in March 2006, prompted the Guyanese government to submit its Sugar Action Plan detailing how it planned to deal with the new EU sugar price regime. The plan was immediately criticized because it did not consider the social consequences of the new sugar price regime. Given these developments, new attention was given to the PSIA and its potential operational implications. Naturally, under the circumstances, the PSIA report is now of considerable interest and relevance to the Guyanese government and could potentially be used as a baseline for future donor operations. 31 Cambodia serves as a model of PSIAs that aim to have a bigger impact on donor operations. Cambodia's Land Allocation for Social and Social Development (LASED), a new lending operation, drew on the results of its PSIA. The composition of the Bank team and the design of the analysis ensured long-term sustainability of the PSIA process in Cambodia. Engagement of community councils and informal local groups was found to be a source of success for the land concession program and lay the foundation for a participatory process in the LASED project. By the time the PSIA was completed, the process of implementing the Social Land Concession Program had made a decisive step forward. The Royal Government of Cambodia began preparation for the LASED project. In November 2005, a translation of the State Land Management sub-decree, approved by the Council of Ministers, was shared with the donor team--one of the prior requirements for LASED and a Consultative Group Joint Monitoring Indicator. Building Confidence in the PSIA Process: Contributions of Social Analysis within a Multidisciplinary Approach The findings of this study confirm the value of a multidisciplinary approach to PSIA. The relationship between social and economic analysis in the case studies varies, and it is important to acknowledge that there are no blueprints in either field, and that there are many overlapping areas of concern. In general, however, social analysis focused on understanding causal relationships (particularly with reference to power relationships), testing and expanding conclusions from economic analysis, and exploring stakeholder perceptions. Economic analysis provided essential elements of quantifying impacts, generalizing findings, and testing posited causal relationships through quantitative analysis. It would be fair to regard the contrast between economic and social analysis as a continuum--with more contextual and qualitative approaches at one end, and more generalized and quantitative approaches at the other. Many economists, however, utilize political, social, and institutional analysis as a core part of their work, and many social analysts work extensively with quantitative economic data.38 High quality PSIAs require effective integration of economic and social analysis. In the case studies reviewed here, social and institutional analysis frequently corroborated or developed the findings of economic analysis. Examples of this include Guyana, where risks and risk management mechanisms were identified using survey techniques, confirmed, and later explored using focus group discussions. In Benin, quantitative measures of poverty were created, and social analysis was used to confirm the quantitative findings39; in Zambia, a limited relationship between holding size and household well-being was found using the household model and later confirmed using ethnographic and focus group techniques. The use of mixed analytic approaches to examine a specific relationship tends to build confidence in results among both the analysts and stakeholders. Social analysis also deepens understanding of the underlying causes of a relationship identified during an economic analysis. In Tajikistan, economic analysis showed that farmers received only a small percentage of the value of the cotton they produced. Focus-group analysis along the marketing chain led to an understanding of how rents were being captured by old-guard farm managers, by money lenders, and by gin owners. These groups captured rents by withholding information and by utilizing exploitative power relations and connections with local governments. In Benin, the quantitative survey showed that many farmers had quit producing cotton because of high indebtedness, low prices, and poor yields. Focus groups indicated that the indebtedness was due to delays in payment and high seasonal cash needs, that the low prices received by farmers were related to imperfect information and probable collusion among marketing agents, and that lack of access to extension workers and other crop-related information contributed to the low yields. 38 The TIPS sourcebook (World Bank 2005) was prepared to guide development professionals from all backgrounds in using institutional, social, and political analysis tools for PSIA. 39 Patterns of findings based on quantitative and qualitative measures were rather different in the Benin case, but little analysis of the source of these differences was conducted. 32 The analysis contributed to examining the assumptions underlying elements of the economic analysis. One example includes Zambia, where the analysis of livelihoods showed that food security is a key objective of most smallholder households: the household model was adjusted accordingly and household behavior was modeled while meeting a food security constraint. In Benin, Tajikistan, Mongolia, and Chad, assumptions about smooth price transmission in liberalized markets were shown to have only limited validity. Similarly, all the studies indicated the presence of high transactions costs and unequal power relations in rural areas. Maximum use of these social analysis findings requires an integrated research process. Without sequential or iterative integration, social and economic analysis cannot fully complement each other. In several cases, institutional analysis assisted in identifying relationships which economic analysis was then able to quantify and test. In Zambia, the rural participatory analysis indicated that the lack of timely access to fertilizers was an important constraint to increased maize production. This issue was identified only after detailed discussions with groups, undertaken during an institutional analysis of marketing chains. Impacts of timing of access were examined in detail using the household economic model. The household model quantified impacts of a finding identified using social analysis. In Benin, focus groups generated hypotheses which were later tested with an econometric model of the relationships between cotton prices, access to credit, and acreage in and productivity of cotton production. Finally, it is worth noting that elaborating the impact of reforms on power relations was an area where the social analysis component often comprised the bulk of the substantive analysis in the PSIA. An example of this was the social analysis in the Zambia PSIA, which identified a potential power vacuum in rural areas if the authority of traditional leaders was weakened by a change in land tenure. This can be seen as an area where social analysis can strengthen its contribution by highlighting variables which are outside the scope of traditional economic analysis. For example, in a general social transition from a smallholder peasantry to an agricultural labor force, the impacts on the people concerned are about more than just income. They may also lose security and autonomy. Social analysis can make a significant contribution to understanding the impacts of policy choice by outlining the significance of non-economic variables, such as empowerment and agency.40 40 Christopher Delgado, personal communication. 33 III. Conclusions and Recommendations The study finds that social analysis is an essential part of agricultural PSIAs at several levels. First, it contributes fundamentally to the analysis of country and reform contexts, complementing standard descriptions of the economic and policy environment. Second, meso-level stakeholder and institutional analyses yield data on the range of opportunities and obstacles for policy implementation. This input is especially important in rural areas, where stakeholders tend to be geographically isolated and diverse, governing institutions tend to be weak, and power structures tend to be unequal and asymmetric. Heavy dependence on agriculture and widespread, deep poverty raise the importance of sector reforms. Third, social analysis contributes to understanding the distributional impacts of policy reform. Fourth, social analysis adds to risk analysis at the macro-, meso-, and micro-levels, and provides guidance on how policy can be designed and modified to address risks. With regards to process, social analysts have worked with colleagues from other disciplines to strengthen the PSIA process leading to enhanced country ownership and consensus building around the reform process. This has allowed stakeholder groups who would otherwise not have had voice in the process of designing policy change in either the conducting of the analysis (including gathering data and information), the governance of the PSIA process (for example, through steering committees) or deliberations on the implications of the analysis for policy design. To fully realize its potential, social analysis needs to be applied as part of a multidisciplinary approach, where other skill sets (such as economic and statistical analysis) bring complementary strengths, including a greater emphasis on the quantification of impacts, and the ability to test the applicability of findings of intensive qualitative studies beyond the specific social context in which they were conducted.41 Table 5 provides a summary of the main policy-related and process-related findings of the different PSIA case studies reviewed here. The rest of this chapter outlines key conclusions under the three thematic areas outlined earlier, namely, stakeholder identification, power relationships and reform processes, and the PSIA process and policy influence. Table 5. Findings from the Poverty and Social Impact Analysis Case Studies Policy Related Process Related 1. Zambia The land reform proposed by the ministry of The World Bank is open to dialogue. land could hurt poor people's access to, and Local perception of the Bank's position on land control over, land. reform and fertilizer subsidies shifted from "fixed" Land titling in an environment where land is to "willing to discuss." not a scarce resource provides few advantages to farmers and instead increases their risks, since the government's capacity to enforce land allocation was identified as very weak By disallowing traditional authorities to allocate land, the resulting power vacuum could undermine peace and stability 41 Please note that the conclusions and recommendations outlined in this section might not apply equally to all agricultural sectors, given the dynamics of political economy issues in each sector (e.g., marketing reforms and land reforms). 34 Policy Related Process Related 2. Tanzania Crop boards require clearly delineated Policy dialogue during design and analysis functions and revised accountability was essential for ownership and access to structures. stakeholders and sensitive data. Agricultural crop boards are vulnerable to A multidisciplinary design, analysis, and team conflict of interest because they fulfill dual (economic, social, governance) was utilized; a roles as both regulators and market process partnership formed as a joint study participants between the government, EU, WB, and two Because smallholders are not represented local universities on crop boards, the boards fail to represent the poor 3. Benin Positive long-term reform impacts must be Coalition building is possible, even around fostered by short-term protective measures contentious reform. to ensure that the reform objective can be A broad coalition was formed between achieved. government, NGOs, and donors that Unless addressed upfront, the weak court outlived the PSIA system, lack of formal delineation of state This coalition assessed the potential and lands, and resulting conflicts will impede pitfalls of the proposed land distribution redistribution of land to the poor program via participatory consultations with Parallel investments in support services affected families, and facilitated policy and basic infrastructure (education, water, discussion and decision-making electricity) for new land recipients are Involving critical stakeholders in the PSIA critical to the program's success or failure and subsequent reform design fostered higher public support for the reform program 4. Cambodia Missing elements in the legislative frame- "Coalition building is possible, even around work will undermine land distribution. contentious reform." Unless addressed, the weak court system, A broad coalition was formed between lack of formal delineation of state lands, government, NGOs, and donors which and resulting conflicts will impede outlived the PSIA redistribution of land to the poor This coalition assessed the potential and Parallel investments in support services pitfalls of the proposed land distribution and basic infrastructure (education, water, program via participatory consultations with electricity) for new land recipients are affected families, and facilitated policy critical to the program's success or failure discussion, and decision-making Involving critical stakeholders in the PSIA and subsequent reform design fostered higher public support for the reform program 5. Guyana Workers and their families made Government participation is fundamental to unemployed by sugar industry reforms achieve impact. require safety nets. Lack of participation from government has Closure of sugar mills will result in about generally weakened the impact of the study. To 8,000 unemployed workers, with greater date, the government has not reviewed report, ripple effects on their families and on and no safety net program has been economic activity throughout affected implemented communities A social safety net program will be needed to offset negative coping by affected households (e.g., withdrawing children from school) 35 Policy Related Process Related 6. Tajikistan Farmland has been privatized, but Civil society provides access to key distortions in cotton production and stakeholders, in this case, poor, dispersed marketing chains impede rural poverty cotton farmers. reduction. Consultations with civil society were essential State farm managers, loan brokers, and gin to accessing local conditions among cotton owners currently extract rents from cotton farmers, who are dispersed and without a production at the expense of poor farmers unified bargaining platform Liberalizing production and marketing chains in addition to land markets would result in significant poverty reduction among cotton farmers 7. Mongolia While cashmere trade reform is necessary, Not available weak legal and administrative systems limit the benefits for poor herders, particularly females. Inadequate marketing and distribution systems caused the capture of most of the profits of the cashmere trade by elites, to the exclusion of producers (herders), the poorest of whom are often female Legal and administrative systems control- ling grazing and water access need to be strengthened; the optimal solution is secure land rights for herders 8. Chad The fast pace of privatization in cotton Political interests often interfere with markets might prevent farmers from intended reforms, weakening results. participating in and benefiting from reforms. Economic agents, often with backing from Cotton farmers, particularly in the favored political interests, may decide to ignore traditionally excluded south, lack transport, new rules (delaying or failing to make infrastructure, and organization payments to cotton farmers), undermining the Fast-paced privatization without ancillary reform process support would likely exclude southern farmers from the new marketing chain altogether 9. Malawi The crop marketing board actually has a PSIA findings can be difficult to leverage positive impact on household welfare in without public dissemination and information poor, remote areas. campaigns. While the crop marketing boards have been With no prior commitment from government to weakened over time, in fact, they play a net disseminate results, public information was positive role in very poor, rural limited. The process led to public controversy communities, raising per capita and decision-making without the full benefit of consumption by up to 20 percent the findings Crop boards should be privatized as planned in less remote areas where markets are well-established, but maintained in remote areas, and where the government should instead focus on investing in road and market infrastructure STAKEHOLDER IDENTIFICATION Stakeholder analysis is an important element of the PSIA process, especially in rural areas, as it highlights diversity issues--in particular, problems of access among remote populations, which tend to be very poor. This identification is also important in the process of giving voice to stakeholders who may lack access to information and, therefore, have little opportunity to participate in the reform process. Without such opportunity to access information and to participate, ownership of the reform and the sustainability of the process is difficult to maintain. In order to achieve this, stakeholder analysis 36 needs to be underpinned by a sound analysis of social diversity (World Bank 2003d). Statistical analysis tends to overlook different categories of stakeholders, therefore, social characteristics of the stakeholders must be considered, including their location of residence, ethnicity, gender, religion, caste, etc. Stakeholder identification requires early participation of social scientists in the client country. Receiving support from government and civil society for a thorough analysis and inclusion of stakeholders is pivotal. This buy-in helps make the entire PSIA an inclusive process. Rural-sector stakeholders tend to be dispersed and heterogeneous, and building support for reforms is a long-term endeavor. Without solid institutional support from rural-specific institutions, this long-term commitment will not emerge. Chad, Tanzania, and Cambodia are examples where buy-in by appropriate ministries and engagement of important NGO and rural-sector actors contributed to sustainability and wider support of the reforms among the most vulnerable stakeholders. A key aspect of the reform process is the balance between the participatory process and the in-country established policy decision making process, since the rural stakeholders are at a stark disadvantage in terms of information, mobility and access. Therefore, logistic and cost related consequences must be considered as part of the planning of a PSIA process. Under the planning rubric, one of the major challenges is the choosing the most cost- and time-effective analytical methods that will capture the range of the heterogeneous stakeholders. Most of the proposed reforms were expected by the donor to have positive long-term efficiency impacts but with short-term and some distributional costs. Non-governmental stakeholders were generally more concerned than the government with distributional impacts, particularly how the reform would affect their well-being. Many stakeholders hold a general perception that privatization and liberalization will have negative consequences. The possibility of exploitation by marketing agents and increased exposure to risks are common concerns. Marketing reforms were expected to have a positive impact on stakeholders through improved price transmission to producers and better market efficiency, which would increase the producers' share of final price. These factors imply that producers are more exposed to market-related risks, a factor noted by several PSIAs but not analyzed in detail by any of them. Government stakeholders were generally more supportive of marketing system reforms and were favorably disposed to information about efficiency dimensions, but were concerned about loss of public-sector employment and loss of government ability to regulate a privatized marketing system. POWER RELATIONS AND THE REFORM PROCESS An analysis of the ways in which differentials in power among key stakeholder groups are likely to affect the outcomes in practice of reform processes was a major contribution of the social analysis component of the PSIAs in many of the cases reviewed here. The findings of this report suggest that the general area of the political economy of reform processes should be a major focus for the methods and practice in the next generation of PSIAs, and that social analysis has a key role to play in this area. One of the critical concerns in the reform process is the analysis of risk--what are the political economy risks and institutional weaknesses that potentially could threaten reform success? The analysis shows that early involvement of government in an inclusive PSIA planning process, early identification and involvement of key stakeholders, and an iterative analysis that builds credibility among stakeholders can help minimize risks to policy reforms. A common issue found in the PSIAs reviewed here was an emphasis on powerful groups which would likely oppose reform efforts. In some cases, this was because of perceived loss of rent-seeking opportunities (Tajikistan, Benin), in other cases, because significant authority structures would see elements of their position eroded (Zambia, Cambodia). The cases reviewed here provide a strong case for a robust analysis focused on the issue of the political economy risks to reform implementation. 37 THE PSIA PROCESS The study concludes that the process of carrying out the PSIA42 has a significant bearing on the likely impact of the results. Sound analysis, underpinned by effective social and economic analysis, is obviously fundamental. Beyond this a range of other factors also have a significant bearing on the outcome. The cases illustrate clearly the issue of timing with respect to PSIA implementation and the policy dialogue. A PSIA must be conducted at a time when the policy environment is still fluid and positions have not solidified. The case studies indicate that the policy environment can be most influenced when the PSIA produces a menu of options and trade-off that has worked to build consensus about the elements of the reform before positions had thoroughly hardened. The analysis shows that early involvement of the ministries with most at stake, including ministries of agriculture and land affects their ultimate success. While there may be reasons why central actors in planning and finance take the lead in defining reform agendas, the cases analyzed here show that involvement of the ministries charged with implementing the reforms in the details of the PSIA is important for impact. Identification of appropriate host government partners, therefore, is critical to impact and sustainability. In agricultural reforms, research quality and sustainability depends on involvement of agriculture and other rural-related ministries. These ministries contribute institutional knowledge and are usually directly affected by the reforms. Ongoing monitoring and evaluation of reform implementation and impacts should at least be coordinated with these ministries. Capacity building is essential, but often the time and financial resources for effective capacity building is not available. Another key relationship is between donor staff, other donors, and NGO representatives. Wider involvement of external partners strengthens the PSIA process. Key examples are found in Malawi and Tanzania. In Malawi, consensus was made possible in part because the PSIA findings were consistent with those from a recent Oxfam investigation of ADMARC--that is, the donor position moved closer to that held by other donors, NGOs, and civil-society representatives. In Tanzania, the partnership with the EU broadened support for subsequent findings. Because of the complex nature of social and economic interactions, well-designed PSIAs employ mixed methods. An important process element is the nature of integration between the economic and social analysis. Of the three forms of integration, parallel integration was the most common. In parallel approaches, social and economic analyses occur separately: two analysis teams work separately and integrate results at the end. Guidelines (World Bank, 2003) imply a linear PSIA process that suggests parallel integration. But, the case studies and conceptual framework indicate real gains from sequential or iterative integration, whereby the economic and social analyses have more opportunity to inform each other, as widely documented in the literature (Rao and Woolcock 2003; Holland and Campell 2005). In several PSIA cases, non-parallel integration led to synergies. Iterative integration does, however, have its costs. A fully iterative process requires constant feedback and adjustment; time and budget constraints may preclude such a process, but the PSIA guidelines should reflect the need for feedback between the different forms of analysis. Often time concerns dictate that one component of the analysis will need to be completed prior to others; in such cases full integration is not be possible. A weakness of the case studies reviewed was a general lack of documentation on analytical and field research methods (especially for the social analysis elements). As a result, it has not been possible to examine critically what methodological approaches showed the best results. Going forward, it will be 42 While some of these recommendations could also apply to other sectors, they are essential for the agricultural sector because of the complex nature of social and economic relations in rural areas, where long term production cycles and investment tend to dominate the organization of the sector. 38 important to ensure that the documentation of social analysis is improved to facilitate comparative analysis and learning. A commitment to a sustained process of analysis of poverty and social impacts throughout the reform process is recognized good practice. In the studies reviewed here, the PSIAs were one-off exercises with little sustainability and weak integration into monitoring and evaluation processes. Further effort is needed to embed the practice of PSIA within the policy cycle in partner countries. The study found that many PSIAs were driven more by donor interests than by clients. The experimental nature of early-generation PSIAs (such as Zambia, Malawi, and Cambodia), as well as the donor mechanisms that made resources for PSIAs available, also add to the donor-driven emphasis in the PSIAs. PSIAs initiated with the full support of client governments (examples include Tanzania and Cambodia) are more likely to have stronger impacts on policy. PSIAs with little or no buy-in from client governments have the least impact. In cases where governments are unwilling to take on the need to carry out poverty and social impact analysis, it may be necessary for donor agencies to do their own PSIA work to be sure that policies under discussion adequately take account of distributional issues and the interests of poor people. The cases, demonstrate, however, that this kind of approach should be a last resort. In several case studies, PSIAs suffered from lack of engagement with government or lack of government demand for the final product, notably Guyana and Malawi, but also, to a limited extent, Zambia. In Guyana, the government "approved" the study, but because it was not involved in the process, it did not fully buy into findings. As a result, it is unknown if the PSIA will have any impact on policy, illustrating the need for a "demand-driven" PSIA process, with early involvement and buy-in from appropriate government partners. The findings indicate that practice in terms of the dissemination of results and their integration into the policy process tends to be weak in many cases. The study reveals a major shortcoming in the knowledge about the impacts of PSIAs on the policy process, lack of adequate dissemination, and a subsequent loss of sustainability. The PSIA process should begin with a discussion with partners about a reasonable expectation with respect to impacts. PSIAs are funded at relatively modest levels and with relatively short time horizons. Even a PSIA that produces the best information cannot be expected to have a long-term impact without institutional buy-in on the part of the host-country government and the World Bank country team. Without long-term sustainability, including dissemination and monitoring and evaluation, any impact is short-lived at best. In the case of agricultural sector reforms, with dispersed stakeholder populations and competing power relations, long-term commitment is essential for sustained impacts. The three cases where substantial dissemination did occur were Cambodia, Tanzania, and Zambia. In each case, stakeholder groups were brought together to discuss and evaluate findings, and following these workshops, recommendations were revised. In other cases, rapid staff turnover, inadequate budget, and a relatively short project time line hampered dissemination. Based on the findings of the cases examined, government-driven PSIAs have the most impact. Many arrangements between donor staff, client governments, and other external partners are possible. For example, the PSIA might be conducted with direct involvement of the donor team or it might be primarily sponsored by the Poverty Reduction and Economic Management Department. This internal arrangement may be related to funding sources, interest in affecting the Country Economic Memorandum or Country Assistance Strategy of the World Bank, or inter-personal relationships and experiences. Decisions about internal arrangements relations involve trade-offs between having an immediate impact on policy dialogue versus having longer-term impacts on government policy and subsequent operations. Many PSIAs have not been fully integrated into country team operations; agricultural reforms clearly need inputs/support of country teams and may require adjustments to ongoing operations. 39 Without adequate commitment (and participation of appropriate ministries), the PSIA becomes a one-off process and its impacts are limited. Therefore, social analysis has a role to play in identifying cases where client government demands for PSIA are high. The reviewed case studies suggest that social analysis, effectively carried out as part of a multidisciplinary effort, helps engage decision makers and stakeholders. Such was the case of Zambia where the PSIA was a donor-driven process that later gained traction among the government due to the high degree of participation of stakeholders in the process. Stakeholder engagement also helped in Tanzania and Cambodia, where strong interactions at the country level increased uptake of findings. In the case of agricultural reforms, the presence of multiple actors in the decision process and multiple impacts over diverse population groups increase the importance of this process element. Ultimately, the analysis of these PSIA cases indicated that in order to increase the impact of PSIAs on the policy process it is essential that the results of the PSIA are fed into a well-informed, deliberative process in which relevant stakeholders can voice their interests. At the same time, it is crucial to have clear monitoring and evaluation mechanisms that will allow all interested parties to follow-up changes and measure progress. Only through a commitment to a sustained process of analysis of Poverty and Social Impact Analysis throughout the reform process is it possible to sustain the needed reforms in the agricultural sector. 40 Annexes Annex A: WB-Supported PSIA in Agriculture and Land Reforms (as of March 2005) Country Reform Sector Main Feature Status AFRICA Benin cotton agriculture Privatize state-run SONAPRA complete Botswana livestock agriculture Sector viability ongoing Burundi coffee agriculture Liberalize coffee market ongoing Chad cotton agriculture Privatize/liberalize cotton market complete Congo, DR Assess distributional implications of forestry agriculture forestry reform in design Côte d'Ivoire cocoa and coffee trade Trade and tariffs of cocoa and coffee ongoing Kenya maize market agriculture Liberalize the maize market ongoing agriculture reform Madagascar (rice, fertilizer, land, infrastructure, trans- agriculture action, extension) 5 Reforms within the agricultural sector complete Malawi tobacco agriculture Liberalize tobacco market ongoing Malawi agriculture market agriculture Assessment of agriculture market closures complete Mali cotton agriculture Liberalize the cotton sector ongoing Rwanda tea agriculture Tea sector privatization ongoing Senegal groundnut agriculture Groundnut sector liberalization ongoing Tanzania crop boards agriculture Reform of crop boards complete Land policy reforms in northern Uganda Uganda land and implications for resettlement and land recovery ongoing Zambia fertilizer and land agriculture Removal/reduction of fertilizer subsidy and land reform complete EAST ASIA AND PACIFIC Cambodia land market land Impacts after the de-collectivization. complete Indonesia rice trade Rice tariffs/trade complete Assess the impact of trade reform in the Mongolia cashmere trade cashmere sector on the income of the herders complete Vietnam land land Assessment of land policy in design EUROPE AND CENTRAL ASIA Tajikistan cotton agriculture Privatization of cotton farmland complete SOUTH ASIA Sri Lanka land land Assessment of land policy ongoing LATIN AMERICA AND CARIBBEAN Restructuring productivity and Guyana sugar agriculture international competitiveness in the sugar industry complete 41 Annex B: Summary of Case Studies Table B.1. Description of Case Studies Country Reform Theme Why Chosen Team Distributional Counterparts Stakeholders Composition impacts Benin Cotton sector; Heavy state involvement in WB staff, Mainly on incomes of Ministry of agriculture, Small-scale State- the cotton sector that FAO/Rome, small-scale farmers ministry of state, producers, hired dominated provides employment and NRI (UK), local and farm laborers-- permanent secretariat farm laborers marketing incomes for the poor consultants and some of the poorest of of the PRSP sector Cotton accounts for academics the poor approximately 70% of export earnings Proposed reforms (privatization and liberalization) could have major impacts on the poor Cambodia Land market Land is most important WB Staff, IMF Impacts on poor rural Ministry of land Land owners, safety net available in rural country team, households management, urban farmers, day areas Oxfam GB, GTZ planning, and laborers, government Land concession is a high PSIA team construction, Agri institutions, NGOs, government priority Business Institute donors, Cambodia, local NGO Chad Cotton Importance of cotton to WB Staff, local High poverty among Ministry of finance Cotton farmers, marketing producer families participation small-scale cotton farmer organizations, Accounts for 65% of exports producers local traders, Cotton Major employer Chad employees, government officials Guyana Sugar sector sugar sector represents 16 WB Staff, local Employment loss by Office of president Ministry of finance, % of GDP consultants low-income workers; sugar sector high inefficiency and state social disruption workers, small-scale involvement; ongoing reform because sector has farmers long provided social (No stake-holder services analysis conducted) 42 Malawi Agricultural Concern about impact of WB staff; two 65% of population Ministry of finance; ADMARC, National marketing-- privatization on remote teams: one for below poverty line; National Economic Association of Small- privatization of smallholders quantitative, 90% in rural regions, Planning and holder Farmers, parastatal one for dependent on Development; National Economic qualitative; local agriculture Agricultural Council, Grain and academics Production and Milling Limited, contributed Marketing Board Smallholder Farmer (ADMARC) Fertilizer Revolving Fund, World Food Programme, Employees, isolated small-holders Mongolia Cashmere Concern about social impact of WB staff Large number of poor Ministry of agriculture Representatives of sector trade reform in the cashmere (ESSD, PREM, herders dependon the and livestock central and local sector and on the incomes of the EAP); local market for their livelihood, management; government, cash- herders; Second largest producer consultants which is a highlyrisky ministry of water mere traders, of raw cashmere, but poor public business (disease, management, and households, line and private institutional capacity climate, draught, harsh ministry of trade and ministries to guide industrial policy winters); especiallyimpact industry development leaves it extremely on female herders was vulnerable to market changes. considered. Tajikistan Cotton farmland Potential for strong social Multidisciplinary Remote rural poor; Ministry of Government of privatization; impacts team; local smallholders who lack Agriculture, National Tajikistan, local cotton National debate surrounding academic access to key inputs Social Investment governments, gin marketing welfare impacts counterparts Fund owners, lenders, sector reforms small-scale farmers Need to quantify degree of inefficiency in system Tanzania Coffee, cotton, Liberalization of agricultural Multidisciplinary Upstream analysis with Ministry of agriculture Central and local cashew and tea markets in 1990s; institutional team; local reform options-- and food security government, crop crop board gaps after liberalization academic distributional impacts boards, producers, reforms contributed to distortions in the counterparts can be assessed once farmer associations, local agricultural market government has traders, processors chosen actual reform and exporters design Zambia Land and Land access is key to rural WB Staff Disadvantaged poor Ministry of finance; Isolated rural poor, agricultural welfare might attain increased ministry of agriculture fertilizer marketing marketing Current allocation system land access; more and lands employees thought to be inefficient remote rural poor might not be served by Fertilizer policy creates private fertilizer distortions and hinders markets growth 43 Table B.2. Components of Case Study Analyses Country Transmission Institutional Analysis Risk Analysis Process Sustainability Channels Benin Price, employment, Detailed analysis of cotton Analysis of risk-adjusted Government involved in May be compromised power relations and marketing institutions and welfare change concept; some follow-up of by government's access to goods and their performance under PSIA findings unwillingness to release services different models of documents; some liberalization conflict between WB Washington team and country team Cambodia Employment, prices, Three different studies were PRA, FGD, and HHS were Government was consulted Problems of access to goods and conducted, each included used as a part of risk in design and synchronization of services, assets, and some aspects of institutional analysis, and took place in implementation program transfers and taxes. analysis all four agro-ecological implementation, zones especially given the large number of priorities Chad Employment, producer Detailed analysis of cotton Social risk analysis; implicit Government supportive; lots None: team never prices, access to marketing institutions in many of the qualitative of dialogue prior to start; completed report markets, assets, social analyses minimal participation of process NGOs and civil society Guyana All: Employment, prices None conducted Social risk management Government not fully None: No ownership or and access to services (SRM) used as conceptual engaged; Bank team participation by were the most framework; report contains changed mid-PSIA; government; little important; transmission detailed analysis of risk originators of PSIA are no follow-up from WB staff; channels determined by management strategies longer engaged WB staff turnover theory Malawi Employment, prices, Minimal None Ministry of agriculture not Limited: Very little access to services engaged; analysis ownership in conducted prior to reforms; government; some dissemination not a high credibility among NGOs priority and donors 44 Mongolia Employment, access to Some institutional analysis Risk analysis was integrated Ministry of Agriculture and Limited do to goods and services, was carried out through throughout the PSIA process Livestock Management, the inadequate legal and assets and power cashmere supply chain Ministry of Trade and institutional markets, relations analysis, data on trade, Industry, and poor herders problems with draught, census, and annual were a part of the process waste, access, and statistics, budget and fiscal rehabilitation of wells data, employment data, etc. Tajikistan Prices (of outputs and Analysis of cotton input and Availability of inputs for Government not fully Limited: Weak civil inputs), access to goods output market structure cotton producers engaged; key informant society in country; very and services, transfers interviews used to identify little dissemination; WB and taxes; power stakeholders; analysis Staff left; constrained relations were conducted during ongoing timeframe due to WB considered during reforms fiscal year analysis Tanzania Price, access to goods Financial review of all 8 crop None Joint study of government, Integration into WB and services, assets, boards; institutional mapping WB and donors (EU), local lending cycle, in- power relations ,and of coffee, cotton, cashew, researchers (2 universities); country policy dialogue, taxes and transfers tea crop industries; concept note shared with institutional Household survey with government and EU, broad arrangements indicate coffee and cotton producers local participation of industry sustainability is possible stakeholders, multidisciplinary design, analysis, team (economic, social, governance), 2 stakeholder workshops (coffee, cotton) Zambia Primarily employment Availability of alternative None Ministry of agriculture not Limited: Very little (in fertilizer distribution), institutions for dispute included as partner analysis ownership in prices, access to resolution conducted prior to reforms government; some services; political power credibility among NGOs was later identified as a and donors; some transmission channel indigenous capacity created 45 Table B.3. Analysis Types and Their Integration Quantitative Analysis Qualitative Analysis Country Methods Findings Methods Findings Integration Complementarity Household survey (IFPRI Cotton production Field visits, Power relations in rural Iteratively High degree of and QUIBB datasets)-- positively related to HH individual and areas may increase integrated: complementarity: analysis of impacts of well-being; strong link focus rent capture by favored quantitative multidisciplinary policy change on poor between cotton prices group/key groups; lack of analysis informed techniques helped and well-being; indirect informant transparency in rules; and was informed generate hypotheses that effects--employment and interviews lack of extension and by the qualitative were tested using Benin expenditure effects--are other services may analysis econometric models; also important have adversely qualitative analysis affected cotton yields identified mechanisms behind quantitative findings Household survey Study showed that 80% Participatory Relevant concepts of Iteratively Highly inclusive process, of the population resides rural appraisal social land concession integrated; but also slow and tedious in rural areas and (PRA), Focus program were quantitative depends on agriculture, group explained to the teams analysis informed especially rice production. discussion by government official; and was informed Survey can play a major (FGD) at dialogue proved by the qualitative Cambodia role in a large scale social community, significant to analysis; the concession program that district and stakeholders, allowing process itself was can overstretch the provincial the strengthening of transformative capacities of government levels with key mutual trust. institutions. officials, at village level with key informants Naïve estimates of Not presented/unknown Stakeholder Low bargaining power Unclear; most Unclear; likely none impact, comparison to analysis, of cotton producers, likely parallel established baseline, ownership lack of private sector theory-based evaluation assessment, alternatives social impact Chad assessment, and institutional analysis 46 Household survey; Household welfare better Structured Low access to public Sequential: Qualitative analysis summary statistics on off in sugar communities; interviews with schools; high demand qualitative deepened understanding sugar and non-sugar sugar households are not household for education; conducted last; of how differences in households; econometric worse off compared to members GUYSUCO important sequencing access to public services model estimating others; education is an provider of social allowed qualitative explain statistical findings; Guyana determinants of well important determinant of services; detailed analysis to address qualitative analysis being (consumption well-being analysis of risk specific issues allowed use of SRM expenditures per capita) management framework strategies Household survey, Proximity to ADMARC a Focus-group Private traders are Parallel: No None summary statistics and statistically significant discussions, replacing ADMARC as integration well-being regression on determinant of well-being key informant buyers and sellers; Malawi determinants of interviews little private sector household well-being involvement in input markets Household Income Two categories of herder Focus group Recognition of the lack Parallel: Little Qualitative analysis Expenditure Survey; households identified: discussions of risk management integration deepened the awareness cashmere supply chain professional herders, with groups at practices, especially in about the impact that risk analysis, data on trade, primary income is that of household a post-command management practices census, and annual herding; and households levels, economy era; lack of could potentially have on Mongolia statistics, budget and with livestock, not primary community, knowledge about the cashmere industry; fiscal data, employment source of income. institutional general risk quantitative analysis data, etc. and line minimization practices. demonstrated the extent ministries to which the formal and informal economy depends on cashmere industry Partial equilibrium Data availability limited Stakeholder Individual tenure Parallel: No Qualitative analysis led to analysis of farmland the possible quantitative analysis opposed by current integration deeper understanding of privatization; quantitative analyses; private tenure farm managers, loan how cotton sector analysis of distortions in shown to be better for brokers and gin structure contributed to Tajikistan input and output markets individual farmers; large owners; central high poverty; qualitative quantitative losses from government would analysis allowed analysis current system; poverty benefit, except it would of distribution will grow without reforms lose control over farms 47 Financial review of crop Producer prices offset Stakeholder Lack of board Iterative: findings Qualitative analysis boards; household with input supply and analysis, accountability to from different informed quantitative survey (incidence transport costs; Pricing Institutional industries and mixed components were analysis, i.e., institutional analysis ) to analyze mechanisms and quality mapping public/ private board used to inform mapping generated producer prices and rewards not recognized activities; low credit other components household survey Tanzania production support; and by many smallholders and input use, but hypotheses analysis of auction data emerging schemes show potential (e.g., farmer groups), uneven private sector participation Statistical analysis; Land holding not strongly Institutional Lack of confidence in Iterative: Highly complementary programming household related to well-being; mapping, state to supply input household model model labor is key constraint; participatory services, enforce and qualitative fertilizer not profitable stakeholder disputes; importance of methods were Zambia given current conditions interviews, group dynamics in rural used to check ethnographic areas; emergence of results studies endogenous solutions to land problem 48 Table B.4. Impacts of PSIAs on Policy Maker Perceptions Country Key Parameters Prior Perceptions Up-dated Perceptions (due to PSIA Information) Benin Sensitivity of rural producers to policy Low linkages to export markets; Strong linkages between cotton prices, change Markets will be efficient production and poverty; government may have to intervene to protect poor producers; access to finance critical determinant of ability to produce cotton Cambodia Land concession program for poor Underestimated the political complexities of the Importance of involving all levels of families without land, poor families losing country: state land management was a major tool stakeholders in the reform process, which their land, and land without people. to keep the political regime in power. took place at a time when land concession was of high importance to the government, civil society, policy makers and other stakeholders Chad Cotton sector reform through liberalization Political interests interfere with the reform process. Better understanding of the importance of and privatization of CotonChad Weak institutions; inability to enforce laws understanding their rights and institutional entitlements; pace of privatization must give farmers the chance to build enough institutional capacity to be an active part of the process Guyana Welfare among sugar-producing Job security is a serious problem for poor sugar- Sugar households' head of household are households; access to services in sugar- producing households less educated than non-sugar household producing areas head. Yet, non-sugar households are more likely to be unemployed or out of the labor force Malawi Impact of ADMARC on disadvantaged ADMARC highly inefficient with little positive impact ADMARC inefficient, but provides needed producers on rural poor services, especially to remote producers Mongolia All stakeholders in the cashmere sector, Because of command economy era, farmers have Livestock production became a household especially herders and their families, little understanding of risk management; transition coping mechanism for non-professional processing workers, and processing economy brought many more into the cashmere herders; pastureland utilization is closely agencies will be affected, but will be market; lack of legal and institutional markets. linked to the availability of water wells; affected differently by the reforms. risk management strategies must be integrated into markets; in spite of the risks, Mongolia is well suited for goat production Tajikistan Impacts of land tenure on farmer welfare Privatization preferred for improving farmer welfare; Study confirmed prior perceptions Rents controlled by different actors in Rents gathered by farm managers and actors in cotton supply system supply chain 49 Tanzania Performance of crop boards and impact Market liberalization will lead to increased Need to tailor reform options due to on industries' competitiveness and efficiency, private sector will provide all previous differences in production and marketing in agricultural growth; private sector marketing board services; each crop industry and geographical participation Levies on farmers are the most efficient means of locations; limited competition, unfair Regulatory, service, and revenue finance practices, low value adding, breakdown of collection functions of boards; institutional input supply system (particularly in remote arrangements/environments, and areas); boards financed according to incentives in production and marketing of selected reform and changed roles (e.g,. tea coffee, cotton, cashew, tea via industry, coffee and cashews via government budget, and cotton jointly) Zambia Efficiency of land tenure system; Traditional tenure inefficient; Current land tenure system not a major existence of privatized lands; land access Few private landholdings exist; land access is an constraint to poverty reduction; endogenous and well-being; poverty impacts of important determinant of smallholder well-being; private landholding institutions have fertilizer programs Private sector suppliers of fertilizer will outperform emerged; labor and access to inputs are state; more important; lots of holes in private supply system; remote residents are Fertilizer supply programs reduce poverty and underserved; problem with access to increase access fertilizer is acute in rural areas 50 References Alston, Julian, George W. Norton, and Phillip G. Pardey. 1995. Science under scarcity: Principles and practice for agricultural research evaluation and priority setting. Ithaca, NY: Cornell University Press. Anderson, Kim, and Yujiro Hayami . 1986., The political economy of agricultural protection. Sydney, Australia: Allen and Unwin. Appleton, Simon, and David Booth. 2005. "Strong fences make good neighbors: survey and participatory appraisal methods in poverty assessment and poverty reduction strategy monitoring." In Methods in development research: Combining qualitative and quantitative approaches, ed. Jeremy Holland and and John Campbell. London: ITDG. Bates, Robert M. 1981. Markets and states in Africa: The political basis of agricultural policies. Berkeley, CA: University of California Press. Beddies, Sabine, Maria Correia, Sashi Kolavalli, and Robert Townsend. 2006."Tanzania: Reforms of the coffee and cotton crop boards." In Analyzing the distributional impact of reforms: Operational experience in implementing poverty and social impact analysis, ed. A. Coudouel, Anis Dani, and Stefano Paternostro. Washington, DC: World Bank. Bourguignon, Francois, and Luiz A. Pereira da Silva, eds. 2003. The impact of economic policies on poverty and income distribution: Evaluation techniques and tools. Washington, DC: Oxford University Press and the World Bank. Carter, Michael R., and Pedro Olinto. 2003. "Getting institutions right for whom? Credit constraints and the impact of property rights on the quantity and composition of investments." American Journal of Agricultural Economics 85(1):173­86. Coudouel, A., and S. Paternostro, eds. 2004., Analyzing the distributional impact of selected reforms. Vol.1. Washington, DC: World Bank. Coudouel, A, Anis Dani, Stefano Paternostro, eds. 2006. "Lessons from the implementation of poverty and social impact analysis of reforms." In Analyzing the distributional impact of reforms: Operational experience in implementing poverty and social impact analysis. Washington, DC: World Bank. ------. 2006. Poverty and social impact analysis of reforms: Lessons and examples from implementation. Washington, DC: World Bank. Deininger, Klaus. 2005. "Land policy reforms." In Analyzing the distributional impact of reforms: A practitioners' guide, ed. Stefano Paternostro and A. Couduel, 213­259. Washington, DC: World Bank. 51 de Janvry, Alain, and Elisabeth Sadoulet. 2002. "Land reforms in Latin America: Ten lessons toward a contemporary agenda." Paper prepared for the World Bank's Latin American Land Policy Workshop, Washington, DC. Dorward, Andrew, et al. 2004.,"Institutions and economic policies for pro-poor agricultural growth." DSGD Discussion Paper, no. 15. Washington, DC: International Food Policy Research Institute. Feder, Gershon. 2002. "The intricacies of land markets: Why the World Bank succeeds in economic reform through land registration and tenure security." Paper presented at the International Federation of Surveyors, Washington, DC. Feder, Gershon, O. Tongroj, and R. Tehaswi. 1986. "Land ownership, security, and access to credit in rural Thailand." ARD Discussion Paper, ARU 53, World Bank, Washington, DC. Gardner, Bruce L. 1997. "Policy reform in agriculture: An assessment of the results in eight countries." Department of Agriculture and Resource Economics, University of Maryland, Baltimore, MD. ------. 2004. "Returns to policy-related social science research in agriculture." Chapter 7. In What's economics worth? Valuing policy research, ed. Philip G. Pardey and Vincent H. Smith. Washington, DC: IFPRI and Johns Hopkins University Press. Government of Tanzania and the World Bank. 2004. "Crop Boards and the future prospects for agricultural exports of Tanzania." Environmentally and Socially Sustainable Development Department, Africa Region, World Bank, Washington, DC. Grindle, Merilee S. 1986., State and countryside: Development policy and agrarian politics in Latin America. Baltimore, MD: The John Hopkins University Press. ------. 2004. "Good enough governance: Poverty reduction and reform in developing countries." Governance 17(4):525­48. GTZ, IFAD, and the World Bank. 2004. "Poverty and social impact analysis and agricultural policy in developing countries." Workshop proceedings, Rome, September 15­16, 2004. Hirshleifer, J., and J. G. Riley. 1992. The analytics of uncertainty and information. New York: Cambridge University Press. Holland, Jeremy, and James Blackburn. 1998. Whose voice? Participatory research and policy change. London: ITDG Publishing. Holland, Jeremy, and John Campbell, eds. 2005. Methods in development research: Combining quantitative and qualitative approaches. London: ITDG Publishing. Holtzmann, Robert, and Steen Lau Jorgensen. 1999. "Social protection as social risk management: Conceptual underpinnings for social protection sector strategy paper." SP Discussion Series 9904, Human Development Network, Social Protection Unit. Washington, DC: World Bank. 52 Jorgensen, Steen Lau, and Zlatina Loudjeva. 2005. "A poverty and social impact analysis of three reforms in Zambia: Land, fertilizer and infrastructure." Social Development Papers, Social Analysis Paper, no. 85. Washington, DC: The World Bank. Kanbur, Ravi, ed. 2003. Q-Squared: Qualitative and quantitative methods of poverty appraisal. New Delhi, India: Permanent Black Publishers. Kherallah, Mylène, et al. 2002. Reforming agricultural markets in Africa.. Baltimore, MD: IFPRI and Johns Hopkins Press. Krugman, Paul. 2004. "Assessing the benefits of economics research: What are the problems?" Chapter 3. In What's economics worth? Valuing policy research,. Ed. Philip G. Pardey, and Vincent H. Smith. Washington, DC: IFPRI and Johns Hopkins University Press. Kumar, Somesh, and Robert Chambers. 2002., Methods for community participation. London: Intermediate Technology Publications. Linder, Robert. 1987. "Toward a framework for evaluating agricultural economics research." Australian Journal of Agricultural Economics, 31(2):95­111. Lipton, Michael. 1993. "Urban bias: Of consequences, classes, and causality." Journal of Development Studies 29(4):229­58. Lundberg, Mattias. 2005. "Agricultural market reforms." Pp. 145-212 In Analyzing the distributional impact of reforms: A practitioners' guide: Trade, monetary and exchange rate policy, utility provision, agricultural markets, land, and education ed. A. Couduel and Stefano Paternostro. Washington, DC: World Bank. Mellor, John W. 1986. "Agriculture on the road to industrialization." In Development Strategies Reconsidered, ed. J.P. Lewis and V. Kallab, 67­89. New Brunswick, NJ: Transaction Books. Moyo, Sibusiso, George W. Norton, and Jeffrey Alwang. 2004. The economic impact of peanut research on poverty reduction: Resistance strategies to control peanut viruses in Uganda. Denver, CO: American Agricultural Economics Association Annual Meetings. Narayan, Deepa, Raj Patel, Kai Schafft, Anne Rademacher, and Sarah Koch-Schulte. 2000. Voices of the Poor: Can Anyone Hear Us? Washington, DC: Oxford University Press and the World Bank. Norton, Andrew et al. 2001. "A rough guide to PPAs." In Participatory poverty assessmen: An introduction to theory and practice. London: Centre for Aid and Public Expenditure, Overseas Development Institute. Norton, George W., and Jeffrey Alwang. 1997. "Measuring the benefits of policy research." American Journal of Agricultural Economics 79(5):1534­38. Pardey, Philip G., and Vincent H. Smith. 2004. What's economics worth? Valuing policy research. Washington, DC: IFPRI and Johns Hopkins University Press. 53 Rao, Vijayendra. 2002. "Experiments in participatory econometrics: Improving the connection between economic analysis and the real world." Economic and Political Weekly 22(20):1887­91. Rao, Vijayendra, and Michael Woolcock. 2003. "Integrating qualitative and quantitative approaches in program evaluation." Chapter 8. In The impact of economic policies on poverty and income distribution: Evaluation techniques and tools, ed. Francois Bourguignon and Luiz A. Pereira da Silva. Washington, DC: Oxford University Press and World Bank. Robb, Caroline M. 1999. Can the poor influence policy? Participatory poverty assessments in the developing world. Washington, DC: The World Bank. ------. 2003. "Poverty and social impact analysis--Linking macro-economic policies to poverty outcomes: Summary of early experiences." IMF Working Paper WP/03/43. Washington, DC: IMF. Schnell, Sabina, Peter Poulsen, Ann Condy, Mari Tertsunen, and Jeremy Holland. 2005. "Principles for PSIA process in policy cycles and stakeholder participation." Discussion note prepared for the German Agency for Technical Cooperation (GTZ) and the UK Department for International Development (DFID). http://www2.gtz.de/dokumente/bib/06-1183.pdf. Accessed November 11, 2006. Siegel, Paul B., and Jeffrey Alwang. 1999. "An asset based approach to social risk management." SP Discussion Series 9926, Human Development Network, Social Protection Unit. Washington, DC: The World Bank. Stewart, Frances. 2001. "Horizontal inequalities: A neglected dimension of development." Working Paper, No. 1. Queen Elizabeth House. Oxford: University of Oxford. Tellis, W. 1997. "Application of a case study methodology." The Qualitative Report 3(3). On-line serial. http://www.nova.edu/ssss/QR/QR3-3/tellis2.html. von Braun, Joachim, with María Soledad Bos, Mary Ashby Brown, Sarah A. Cline, Marc J. Cohen, Rajul Pandya-Lorch, and Mark W. Rosegrant. 2003. "Overview of the world food situation food security: New risks and new opportunities." Brief prepared for the annual general meeting of the Consultative Group on International Agricultural Research, Nairobi, Kenya, October 29, 2003. World Bank. 2003a. A user's guide to poverty and social impact analysis. Washington, DC: World Bank. ------. 2003b. Reaching the rural poor: A renewed strategy for rural development. Washington, DC: World Bank. ------. 2003c. "Reforming the Malawi agricultural development and marketing corporation: Synthesis report of the poverty and social impact analysis." Washington, DC: World Bank. ------. 2003d. Social analysis sourcebook: Incorporating social dimensions into Bank-supported projects. Washington, DC: World Bank. 54 ------. 2004. "Using poverty and social impact analysis to support development policy operations." Good Practice Note 2, Operations Policy and Country Services Department. Washington, DC: The World Bank, 33­51. http://siteresources.worldbank.org/INTPSIA/Resources/490023- 1120840449856/GPNonPSIA9204.pdf .2004. Madagascar Poverty Reduction Strategy Paper (PRSP) ­ First Annual Progress Report. http://imagebank.worldbank.org/servlet/WDSContentServer/IW3P/IB/2004/09/29/000012009_20 040929111856/Rendered/PDF/30036.pdf ------. 2005. "Tools for institutional, political and social analysis (TIPS): Sourcebook for PSIA." Washington, DC: World Bank. http://web.worldbank.org/WBSITE/EXTERNAL/TOPICS/EXTSOCIALDEVELOPMENT/EXT TOPPSISOU/0,,menuPK:1424015~pagePK:64168427~piPK:64168435~theSitePK:1424003,00.h tml. Yin, Robert K. 1989. "Case study research, design and methods." Applied Social Research Methods Series. Vol 5. New York: Sage Publications. 55