WORLD BANK TECHNICAL PAPER NO. 407 \A/TP(40"P Work in progress W vzk I: for public discussion M a Designing Rules for Demand-Driven Rural Investment Funds iliw ,ij x l II n /1' tollfl\p t/( RECENT WORLD BANK TECHNICAL PAPERS No. 333 Jha, Ranson, and Bobadilla, Measuring the Burden of Disease and the Cost-Effectiveness of Health Interventions: A Case Study in Guinea No. 334 Mosse and Sontheimer, Performance Monitoring Indicators Handbook No. 335 Kirmani and Le Moigne, Fostering Riparian Cooperation in International River Basins: The World Bank at Its Best in Development Diplomacy No. 336 Francis, with Akinwumi, Ngwu, Nkom, Odihi, Olomajeye, Okunmadewa, and Shehu, State, Community, and Local Development in Nigeria No. 337 Kerf and Smith, Privatizing Africa's Infrastructure: Promise and Change No. 338 Young, Measuring Economic Benefitsfor Water Investments and Policies No. 339 Andrews and Rashid, The Financing of Pension Systems in Central and Eastern Europe: An Overview of Major Trends and Their Determinants, 1990-1993 No. 340 Rutkowski, Changes in the Wage Structure during Economic Transition in Central and Eastern Europe No. 341 Goldstein, Preker, Adeyi, and Chellaraj, Trends in Health Status, Services, and Finance: The Transition in Central and Eastern Europe, Volume I No. 342 Webster and Fidler, editors, Le secteur informel et les institutions de microfinancement en Afrique de l'Ouest No. 343 Kottelat and Whitten, Freshwater Biodiversity in Asia, with Special Reference to Fish No. 344 Klugman and Schieber with Heleniak and Hon, A Survey of Health Reform in Central Asia No. 345 Industry and Mining Division, Industry and Energy Department, A Mining Strategyfor Latin America and the Caribbean No. 346 Psacharopoulos and Nguyen, The Role of Government and the Private Sector in Fighting Poverty No. 347 Stock and de Veen, Expanding Labor-based Methods for Road Works in Africa No. 348 Goldstein, Preker, Adeyi, and Chellaraj, Trends in Health Status, Services, and Finance: The Transition in Central and Eastern Europe, Volume II, Statistical Annex No. 349 Cummings, Dinar, and Olson, New Evaluation Procedures for a New Generation of Water-Related Projects No. 350 Buscaglia and Dakolias, Judicial Reform in Latin American Courts: The Experience in Argentina and Ecuador No. 351 Psacharopoulos, Morley, Fiszbein, Lee, and Wood, Poverty and Income Distribution in Latin America: The Story of the 1980s No. 352 Allison and Ringold, Labor Markets in Transition in Central and Eastern Europe, 1989-1995 No. 353 Ingco, Mitchell, and McCalla, Global Food Supply Prospects, A Background Paper Preparedfor the World Food Summit, Rome, November 1996 No. 354 Subramanian, Jagannathan, and Meinzen-Dick, User Organizationsfor Sustainable Water Services No. 355 Lambert, Srivastava, and Vietmeyer, Medicinal Plants: Rescuing a Global Heritage No. 356 Aryeetey, Hettige, Nissanke, and Steel, Financial Market Fragmentation and Reforms in Sub-Saharan Africa No. 357 Adamolekun, de Lusignan, and Atomate, editors, Civil Service Reform in Francophone Africa: Proceedings of a Workshop Abidjan, January 23-26, 1996 No. 358 Ayres, Busia, Dinar, Hirji, Lintner, McCalla, and Robelus, Integrated Lake and Reservoir Management: World Bank Approach and Experience No. 360 Salman, The Legal Frameworkfor Water Users' Associations: A Comparative Study No. 361 Laporte and Ringold, Trends in Education Access and Financing during the Transition in Central and Eastern Europe. No. 362 Foley, Floor, Madon, Lawali, Montagne, and Tounao, The Niger Household Energy Project: Promoting Rural Fuelwood Markets and Village Management of Natural Woodlands No. 364 Josling, Agricultural Trade Policies in the Andean Group: Issues and Options No. 365 Pratt, Le Gall, and de Haan, Investing in Pastoralism: Sustainable Natural Resource Use in Arid Africa and the Middle East No. 366 Carvalho and White, Combining the Quantitative and Qualitative Approaches to Poverty Measurement and Analysis: The Practice and the Potential (List continues on the inside back cover) WORLD BANK TECHNICAL PAPER NO. 407 Designing Rules for Demand-Driven Rural Investment Funds The Latin American Experience Thomas Wiens Maurizio Guadagni The World Bank Washington, D.C. Copyright © 1998 The Intemational Bank for Reconstruction and Development/THE WORLD BANK 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America First printing May 1998 Technical Papers are published to communicate the results of the Bank's work to the development community with the least possible delay. The typescript of this paper therefore has not been prepared in accordance with the proce- dures appropriate to formal printed texts, and the World Bank accepts no responsibility for errors. Some sources cited in this paper may be informal documents that are not readily available. The findings, interpretations, and condusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. The World-Bank does not guarantee the accuracy of the data in- cluded in this publication and accepts no responsibility for any consequence of their use. The boundaries, colors, de- nominations, and other information shown on any map in this volume do not imply on the part of the World Bank Group any judgment on the legal status of any territory or the endorsement or acceptance of such boundaries. The material in this publication is copyrighted. Requests for permission to reproduce portions of it should be sent to the Office of the Publisher at the address shown in the copyright notice above. The World Bank encourages dissem- ination of its work and will normally give permission promptly and, when the reproduction is for noncommercial purposes, without asking a fee. Permission to copy portions for dassroom use is granted through the Copyright Clearance Center, Inc., Suite 910, 222 Rosewood Drive, Danvers, Massachusetts 01923, U.S.A. ISSN: 0253-7494 Thomas Wiens is a principal agricultural economist and Maurizio Guadagni is an agriculturalist in the Environ- mentally and Socially Sustainable Development Department, Latin America and the Caribbean Region. Library of Congress Cataloging-in-Publication Data Wiens, Thomas B. Designing rules for demand-driven rural investment funds: the Latin American experience / Thomas Wiens, Maurizio Guadagni. p. cm. - (World Bank technical paper; no. 407) Includes bibliographical references. ISBN 0-8213-4229-0 1. Rural development projects-Latin America-Finance. I. Guadagni, Maurizio. R. Title. III. Series. HC130.E44W54 1998 307.1'412'098-dc2l 98-21015 CIP Contents F OREWORD ............................................................... v ABSTRACT .............................................................. vi ABBREVLATIONS and ACRONYMS .............................................................. vii GLOSSARY ........................................................................ viii ACKNOWLEDGMENTS .............................................................. i2x EXECUTIVE SUMMARY .............................................................. xi INTRODUCTION ...............................................................1 Study Objectives ...............................................................1 Basic Concepts ...............................................................2 FRAMEWORK OF OBJECTIVES ...............................................................3 Policy Objectives ...............................................................3 Other Objectives ...............................................................4 THE ACTORS AND THEIR OBJECTIVES ...............................................................7 World Bank ...............................................................7 Central Government ...............................................................9 Project Management Units .............................................................. 10 Intermediate Government ...............................................................11 Local Government .............................................................. 11 Traditional Local Elites .............................................................. 12 THE CONTEXT .............................................................. 13 Brazil ..... . 13 Colombia ..... 15 Mexico ..... 17 Poverty Targeting versus Political Imperatives ........................................................... 19 Demand-Driven versus Feasibility and Sustainability ........................................................... 22 Coronelismo versus participation at the municipal level ......................................................... 30 External Priorities and Eligibility Criteria versus Demand-Driven Project Selection ............. 33 Efficiency and Accountability versus Poverty Targeting ........................................................ 35 iii IMPLEMENTATION AND ENFORCEMENT MECHANISMS .......................................... 41 Dissemination ................................................ 41 Monitoring and Evaluation ................................................ 42 Enforcement and Sanctions ........................................... 42 CONCLUSION ........................................... 44 Findings ........................................... 44 Next Steps .......................................... 47 ANNEXES 1 Targeting ........................................... 49 2 Subproject Eligibility Criteria .......................................... 57 3 Procurement and Disbursement .......................................... 64 BIBLIOGRAPHY ........................................... 68 BOXES 1 The participatory planning process in Bolivia .......................................... 25 FIGURES 1 Actors in participatory microplanning ........................................... 24 2 Flow of funds .......................................... 39 3 Evaluation indicators .......................................... 43 TABLES 1 Objectives and rules comparison matrix .......................................... 20 ANNEX FIGURES A1.1 Investment per capita versus poverty .......................................... 55 A1.2 Investment per capita versus poverty: Mexican states .......................................... 56 ANNEX TABLES A2.1 PAC/FUMAC cofinancing matrix .......................................... 58 A2.2 Fondo DRI Cofmancing matrix .......................................... 60 A2.3 Fondo de Desarrollo Campesino cofmancing matrix .......................................... 62 A3.1 Procurement methods .......................................... 64 A3.2 World Bank prior review of procurement decisions .......................................... 65 A3.3 Documentation of expenditures ................................65............................. 65 A3.4 Withdrawal of the proceeds of the loans ...... 66 iv FOREWORD Demand-driven rural investment funds (DRIFs) offer a major new means for improving the design, implementation and sustainability of rural development programs. DRIFs are mechanisms through which central government transfers funds to local governments and communities to use to address their own priorities. This approach to rural development represents a sharp break with the past, when central governments selected and implemented rural development activities, often with little input from communities - and little impact on development. However, governments and donors are concerned that local governments are too weak to make decisions about which projects to undertake and lack capacity to implement all but the simplest projects. They worry that funds will be spent improperly or appropriated by local elites. To counteract these problems, central governments have often constrained the choices of communities by limiting the types of projects eligible for financing, and requiring specific procedures for procurement and disbursement. They monitor compliance and retain veto power over community choices of subprojects. This study explores the extent to which well-designed DRIF rules and incentive structures can substitute for central control, and truly empower communities to take charge of their own development. It identifies the many objectives that DRIFs are meant to satisfy, and shows which are being met and which are not. And it offers practical guidance about how to design DRIFs so that they effectively and sustainably promote rural development. The analysis was based primarily on the evaluation of World Bank-financed projects in Mexico, Colombia and Brazil which represent early examples of DRIFs. The research was carried out as part of the World Bank's Decentralization, Fiscal Systems and Rural Development Research Program, supported by the Swiss Agency for Development Cooperation, the Royal Ministry of Foreign Affairs of the Government of Norway, and the World Bank. Alex F. McCalla, Director and Chair, Sector Board Rural Development Environmentally and Socially Sustainable Development Network v ABSTRACT The demand-driven rural investment fund (DRIFs) is a new mechanism for decentralizing decisionmaking authority and financial resources to local governments and communities to use for investments of their choice. They offer great promise for improving the design, implementation and sustainability of rural development programs. However, to counteract weak capacity of local governments to choose and implement projects well, central govermments have often constrained the choices of communities by limiting the types of projects eligible for financing and requiring specific procedures for procurement and disbursement. They also monitor compliance and often retain veto power over community choices of subprojects. This study explores the extent to which well-designed DRIF rules and incentive structures can substitute for central control. It looks at the different, and often conflicting, motivations of donors, central governments and communities and explores how rules can be devised to allow all actors to achieve their objectives. It identifies the many objectives that DRIFs are meant to satisfy, and shows which are being met and which are not. And it offers practical guidance about how to design DRIFs so that they effectively and sustainably promote rural development. vi ABBREVIATIONS and ACRONYMS DRD Decentralization and Regional Development (Mexico) DRIF demand-driven rural investment fund FDC Fondo de Desarrollo Campesino (Small-farmer Development Fund, Bolivia) Fondo DRI Fondo Desarrollo Rural Integrado (Integrated Rural Development Fund, Colombia) FPM Fundo de Participacion Municipal (Municipal Participation Fund, Brazil) FUMAC Fundo Municipal de Apoio Comunitario (Municipal Community Fund, Northeast Brazil) IRR internal rate of return MPC municipal participatory committee MSF Municipal Solidarity Fund (Fondo de Solidaridad Municipal, Mexico) NRDP Northeast Rural Development Program (Brazil, phase 1, 1986-93, and phase 2, reformulated NRDP, 1993-95) NRPAP Northeast Rural Poverty Alleviation Program (Brazil, phase 3, 1995-96) PAC - Programa de Apoyo Comunitario (Community Support Program, Northeast Brazil) PDIC Programa de Desarrollo Integrado Campesino (within Fondo DRI) (Integrated Small-Farmer Development Program, Colombia) PIDC Programa Integrado Desarrollo Campesino (within FDC) (Integrated Small Farmer Development Program, Bolivia) SEDESOL Secretaria Desarrollo Social (Social Development Secretariat, Mexico) vii GLOSSARY Caudillismo, coronelismo, caciquismo. The caudillo historically was the military strongman who could maintain order, though often with iron-fisted brutality. At present what remains of the caudillo system is defined as "an informal system of power, practiced by individuals or groups in strategic positions in the economic and political structure. It implies asymmetric relationships among people of different social classes with reciprocal but uneven obligations, characterized by latent threatening." A controversial parallel of the caudillo in the indigenous social structure is the cacique. Historically the cacique was in charge of organizing and supervising work groups which - according to the Inca and Aztec fiscal legislation - were obliged to give some days of free labor for public works. (This form of tax was called mita.) The controversy derives from the extremely variable role of the cacique, who did not based his power on his own army and had a smaller group of subordinates than the caudillo. Throughout this report, these terms are used to refer to a single politician who retains arbitrary decisionmaking authority and discretionary control over resources. Clientelismo. Use of resources to reward one's patrons or party. Favoritism based on personal relationships rather than efficiency criteria. Municipal Assembly. Meeting of the municipal council which is open to all citizens. Municipal Council. Elected body representing the local legislative power. Municipal Government. The smallest administrative unit which is composed of the mayor and his assistants. The municipal council is not formally part of the executive organs of the municipal government. However, the term "municipal government" often refers to all the public institutions and organs of municipal powers. In this case, it also comprises the municipal council. Municipal participatory committee (MPC). An informal body created to administer the resources of demand-driven rural investment funds with variable levels of responsibility. In some cases it may be independent of the municipal government. Municipality. Generic term for the geographical delimitation of the administrative unit and/or its level of decisionmaking. It is broader than the Municipal Government, as it does not only represent the public, institutional, municipal powers. This term is used though this report with the Spanish connotation, which is not limited to city and town. This report is strictly concerned with rural municipalities. viii ACKNOWLEDGMENTS The authors are grateful to the following people for their comments, support, and suggestions: Fernando Aguilar, Wendy Ayres, Joao Barbosa, Tulio Barbosa, Hans Binswanger, Mark Cackler, Raimundo Caminha, Luis Coirolo, Angela Esteve, Elsie Garfield, Thereza Lobo, Andrew Parker, Suzanne Pirou-Sall, Mary Rieth, Jean Claude Sallier, and Andrea Silverman. The authors also wish to thank all government staff and community members and representative met during the field visits in Brazil, Colombia, and Mexico. Wendy Ayres of the World Bank edited the paper. ix EXECUTIVE SUMMARY Traditionally, central governments have been in charge of designing rural infrastructure projects, often with more or less input from local communities. To ensure efficient execution and minimal costs, project rules often required the use of centralized contracting and procurement. Now, however, many countries are giving communities more control over project choice and execution through the use of "demand-driven investment funds" (DRIFs). Broadly, DRIFs are pools of funds available to communities to invest in their preferred projects. Communities must submit applications for the funds describing in some detail how the funds will be used. However, communities are not free to use the funds in whatever way they want. Central government constrains local decisionmaking with eligibility rules - defining acceptable project types and providing incentives to choose projects of particular sorts - and constrains local implementation power with rules of procurement and disbursement. Generally central authorities monitor compliance with the rules, and also often retain the power to endorse or reject communities' choices of subprojects. If local governments all functioned effectively, and projects were small and localized, then there would be no need for the rules. Communities could be given access to untied resources from revenue sharing or general purpose block grants, and invest in the projects they prefer, constrained only by availability of resources. However, local governments often lack the capacity to choose and implement projects well: staff with essential skills are lacking, institutions for decisionmaking are weak, and mechanisms of transparency and accountability are absent. In Latin America, as elsewhere in the developing world, coronelismo and clientelismo are pervasive in rural areas. These shortcomings provide central government with a rationale for overseeing local decisions. This case study explores the extent to which a carefully-devised rule and incentive structures can replace central control and centrally-imposed constraints, while assuring that coronelismo and clientelismo do not undermine the objectives of "demand-driven rural investment funds" (DRIFs). It focuses on decentralized projects which have been in operation long enough to have a track record, particularly Colombia's Fondo DRI, Brazil's PAC/FUMAC, and Mexico's Solidaridad projects. Assessing the rules and incentives governing these projects is a difficult task since they are changing from year to year (partly in recognition of imperfections noted below). Therefore not all of the "facts" reported will still be valid when this paper is read. Moreover, we make no attempt to evaluate the performance or design of any of these projects as a whole, relative to each other or relative to previous projects. We rather assume that DRIFs as a class are at the cutting edge of the best approaches to rural poverty, and focus on ways to further improve this instrument. DRIFs are distinguished from the social fund-type projects of Africa and Latin America by targeting rural areas and by supporting mainly productive infrastructure and natural resources xi management activities. DRIFs borrow from social funds the goal of poverty alleviation, a thrust toward participatory identification and implementation of many small subprojects, and often a focus on strengthening institutions at the community and municipality level. The similarities between DRIFs and social funds are greater than the differences, and most of the conclusions of this study are equally applicable to social funds. Goals of DRIFs DRIFs are intended to meet many policy objectives, which are often in conflict. Various institutions involved in developing, financing, and implementing projects often also have conflicting objectives. The primary objectives are: * Target rural poverty (goal 1). Grant-funded projects should primarily benefit the poor. DRIF rules should be established so that most of the benefits go to the poor, or at least so that the benefits are distributed more equitably than normally. - Give communities control over project choice and implementation (goal 2). When community members choose their projects, and help pay for and implement them, they are much more likely to value and sustain them. Full community involvement usually requires decentralized decisionmaking. * Build municipal capacity (goal 3). DRIFs should help local governments develop permanent capacity to plan and execute small projects, where this is lacking. Where local governments already possess adequate capacity, DRIFs can help transfer responsibility for selecting and implementing small projects to local governments. Either way, DRIFs may be designed to foster decentralized decisionmaking and implementation. Other objectives are common to international development projects: * Choose projects which are economically viable (goal 4). Curiously, this most universal of goals conflicts with the goal of involving local people in choosing and implementing projects (goal 2). Local communities may choose projects which cannot pass the test of viability (generally defined as projects with acceptable economic rates of return). Requiring viability gives outsiders the power to veto community choice. This goal may also conflict with the goal of targeting poverty (goal 1). The poor often live in resource-poor and isolated regions, in conditions that make it particularly difficult for projects to succeed. * Implement projects efficiently (goal 5). This sensible goal may conflict with the goal of involving local people in choosing and implementing projects (goal 2) and perhaps that of targeting poverty (goal 1). Generally, implementing projects efficiently requires hiring experts with skills and experience from outside the xii community. Poor communities are less likely than others to have members with the necessary expertise and experience to implement projects efficiently. * Manage funds efficiently and honestly (goal 6). Donor and governments have developecd complex regulations to control procurement, accounting, and auditing to ensure the efficient and honest use of funds. However, these rules, which authorities designed in response to experiences with centralized projects, frequently conflict with goals I and 2. Targeting poverty means working with poor communities that typically lack the technical capacity to satisfy the regulations. * Effectively manage project operation and maintenance, including recurrent financing (goal 7). This goal suggests that local people should participate in the daily management of the project, at least of small subprojects for which professional management is not feasible. * Respond to priorities of people outside the community (goal 8). Central governments or external donors may give priority to certain types of subprojects, and not others. There are many reasons why. For example, national constitutions may require countries to provide specific goods or services to its citizens, such as access to education or health care, which demand priority. Countries may require that projects or programs be clearly separate in content, so new projects do not duplicate older ones. Whatever the reasons for the preferences, this goal conflicts with that of giving communities control over project choice (goal 2). Finally, some important but largely unstated goals of individuals or institutions may influence design and implementation: * Conform to institutional imperatives (goal 9). Institutional or bureaucratic rules do not always promote - and may even be at cross-purposes with - project goals. Yet, policymakers may require projects to follow institutional imperatives anyway, even though this has a negative impact on project design. * Administer projects cost-effectively (goal 10). The goal is to minimize the time and effort required to develop, implement, and supervise the project. This is often a high priority and other goals are often sacrificed in pursuit of cost-effectiveness. * Get reelected (goal 11). This is often the most important goal for national and local political leaders. Indeed, politicians are likely to promote other goals only to the extent that they facilitate getting reelected. Of course many of the other goals do not particularly fturther this one. * Reward one's patrons or political supporters (goal 12). This is a crucial goal of politicians, which is even more important where localities restrict the number of terms xiii politicians can serve. This is the definition of clientelismo. Usually politicians reward patrons by giving them contracts outside the competitive bidding process. Hence clientelismo conflicts with the goals of choosing economically viable projects and executing and managing them efficiently (goals 4-7). Participation as a means. We do not take "participation" as a general goal, even though it may be central to this kind of project. This is because participation at different stages of the project cycle can exist and serve quite different goals, at the possible expense of other goals. Project Background The Fundo Municipal de Apoio Comunitario (FUMAC) and Programa de Apoio Comunitario (PAC) in Northeast Brazil. The Brazilian Northeast Rural Poverty Alleviation Program (NRPAP) represents one decade of efforts to alleviate rural poverty in northeast Brazil. Development of the program has passed through three different phases, each incorporating lessons of experience from the previous one: 1) The original Northeast Rural Development Program (NRDP) lasted from 1986 to 1993. The program-had a small demand-driven component which distinguished itself from the rest of the program through its capacity to respond to the needs of small producers and its high disbursement rate. 2) The Northeast Rural Development Program was reformulated in 1993, by expanding and improving upon its demand-driven component. The beneficiaries choose from among eligible types of subprojects investments meeting their most critical needs. The reformulated Northeast Rural Development Program worked through two different delivery mechanisms: Programa de Apoio Comunitario (Community Support Program, PAC) in which rural communities submit their proposal directly to the State. The State screens, approves, and releases funds for subprojects interacting directly with the beneficiaries. Fundo Municipal de Apoio Comunitario (Municipal Fund for Community Support, FUMAC), in which a municipal participatory committee is created to screen, prioritize, and submit proposals to the State. The municipal participatory committee is not a formal municipal structure, though the municipal government is represented, together with representatives of communities and civil society. 3) The Northeast Rural Poverty Alleviation Program (NRPAP) started in 1995-96. This phase expanded the use of FUMAC and introduced a third delivery mechanism (which has barely begun to function). Fundo Municipal de Apoio Comunitario-Piloto (Pilot Municipal Fund for Community Support, FUMAC-P), under which the municipal participatory committee assumes xiv additional implementation responsibilities (besides those of FUMAC), receives a firmn annual budget, and manages the resources. The new mechanism is intended to transfer as much responsibility to the municipal level as possible. Fondo DRI Like the Northeast Brazil Program, the Colombian Fondo DRI has been operating for two decades to reduce rural poverty, and has evolved from a centralized rural development program. By contrast to the program in Brazil, its conversion to a community based program has been piecemeal, and is still not complete. Colombian municipalities, which have been efficiently soliciting financing and implementing most subprojects for years, are responsible for its success. In 1990 with the support of a Bank loan, the program began to finance subprojects proposed and managed by the municipalities, reinforcing the mayor's authority. Even now, however, the resources are rigidly allocated among specific types of investments: technical assistance, organization and training, watershed management and fisheries, rural women, marketing, rural roads, water supply and sewerage, electrification, and flood control. For example, the resources available for roads cannot be used for technical assistance, and vice-versa. Mexico Municipal Solidarity Fund (MSF). The National Solidarity Program, now renamed the Superaci6n de la Pobreza, was the Mexican Government's primary tool to reduce poverty and provide services to underserved communities. It was'initiated in 1988 and included 30 subprograms to target health, education, nutrition, employment, infrastructure, agriculture, institutional developrnent, and income-generating subprojects for the poor. Therefore it is not a single program, but a general approach to reducing poverty, and an umbrella for many different components. The program was built on four principles common to the whole program: (a) community participation; (b) poverty targeting; (c) decentralization; and (d) responsibility and cost sharing among federal government, states, municipalities, and beneficiary communities. While the umbrella program also finances large subprojects, the Municipal Solidarity Fund finances only small subprojects, which are completely managed at municipal and community levels. Subprojects were identified through community assemblies, and were selected by a specially-formed municipal participatory committees called Solidarity Councils (one per municipality). The Solidarity Committees (one per subproject) were in charge of implementing the subprojects. The Solidarity Councils decided the final priorities and identified the investments, given simple eligibility criteria and minimal procedural rules. Findings How well have the three DRIFs achieved the eight goals most important to the World Bank? Which rules (or types of rules) have proven most useful, and least useful, in achieving these goals? * Target rural poverty (goal 1). It appears that the DRIFs have not been particularly effective in targeting poor localities, although we have no way of comparing how well xv DRIFs target poverty compared with other programs. This is because formulas have not generally been used to allocate funds; when used, formulas have included several criterion unrelated to poverty; and additional eligibility conditions which would exclude the nonpoor from the benefits of the subprojects have not been applied or enforced. Colombia's Fondo DRI is the most successful of the three projects in targeting poor localities. Project managers used a cofinancing matrix which heavily favored poorer municipalities to allocate funds. * Give communities control over project choice and implementation (goal 2). In all three DRIFs, communities are responsible for identifying projects and submitting proposals for financing. It is unclear, however, to what extent the process is really participatory, or is instead dominated by outside influences (the mayor, contractors, or nongovernmental organizations). The requirement that beneficiary communities provide some proportion of subproject funds should ensure that beneficiaries at least desire the subprojects even if they do not regard them as their highest priorities. After all, all the DRIFs constrain freedom of choice by imposing eligibility, cofinancing, and feasibility criteria. Colombia imposes the greatest constraints on communities: Fondo DRI, until recently, rigidly allocated funds according to subproject type. In Mexico, where guidelines emphasize participation, experience has been mixed and generally fallen far short of the guidelines. There are few rules governing allocation of resources from the Municipal Solidarity Fund, and enforcement is weak. Therefore outcomes depend largely on the attitudes of local political and administrative authorities. Bolivia, with its requirement that communities adopt a participatory process of planning and priority setting, has developed the most elaborate participatory scheme. * Build municipal capacity (goal 3). In all but Brazil's PAC, municipal level authorities set priorities and choose subprojects among those proposed by communities. Municipalities have received considerable technical assistance to help strengthen their investment management capacities. The extent to which the mayor, the existing municipal council or a parallel municipal participatory council has dominated subproject selection has varied among countries and localities. On the whole, Colombia's Fondo DRI has tended to reinforce the mayor's authority: the municipal participatory committees have not functioned as envisioned. The Brazil and Mexico DRIFs made more of an effort to include beneficiaries and other members of civil society in decisionmaking (reflecting central government distrust of local government). This effort has been especially successful in Brazil, perhaps because civil society at the grassroots level (in the municipalities eligible for FUMAC) may be stronger than in Mexico, and because Brazil encouraged communities to develop mechanisms for indirect, rather than direct representation, of beneficiaries. xvi * Choose projects which are economically viable (goal 4). The Mexico DRIF has relied entirely on beneficiary participation to achieve the goal of viability and sustainability. Most of the subprojects financed by the Municipal Solidarity Fund appear to have addressed basic social needs at substantially lower unit cost than other government projects. There have been too few productive-type projects under the Mexican program to judge whether participation would also have been effective for these projects. However experience from a limited number of subprojects in Chiapas suggests that productive subprojects may be less sustainable than subprojects addressing basic needs. There is no evidence that subprojects in Mexico were of higher quality than in the other countries, even though Mexico screened the subproject proposals for feasibility. However, there are too little data available on project performance following completion to allow us to draw firm conclusions at this time. It is clear that requiring full economic evaluation (plus environmental impact and gender analysis) of every subproject is a costly strategy, and probably an unrealistic goal given technical capabilities in local areas in these countries. Unfortunately little effort has been devoted to developing and applying rules of thumb which could be used to screen out most unviable or unsustainable subprojects quickly and easily. * Implement projects efficiently (goal 5). Is it better to implement subprojects through contractors or participatory self-management? There is too little information to be able to say for sure. We do know that in Mexico, which allows communities to retain saved funds to expand subprojects, communities have completed projects at lower unit costs than in the other countries. Although routine monitoring data provide information only on whether or not projects, as originally proposed, were completed, physical audits show that communities were able to expand projects by an average of 10 percent using the original budgets. M Manage funds efficiently and honestly (goal 6). The three programs have all experienced difficulties with disbursements, which were late or difficult to access because central governments were unable to advance funds. The World Bank's disbursemnent procedures are not designed to deal with this situation. But authorities have increasingly been able to find ways to manage disbursements to meet the needs of dispersed and decentralized communities, yet conform to World Bank rules. A weakness has been the failure of governments to enforce subproject rules through sanctions: this would demonstrate that government takes seriously the need for honest funds management. And, where the World Bank and implementing agencies (at central or local levels) differ significantly in their commitment to various project goals, the World Bank should disburse funds directly to these agencies to give them the incentive to observe agreed rules. This is the lesson from the Mexican experience. * Effectivelly manage project operation and maintenance, including recurrent financing (goal 7). None of the programs have performed particularly well in xvii achieving this goal, according to the little information available. Indeed, information from local or partial surveys suggests that a high proportion of subprojects may not be sustainable. With infrastructure subprojects, communities need help to strengthen their capacities to better manage operations and maintenance. Sustainability is not guaranteed simply because communities formally agree to take responsibility for operating and maintaining subprojects. With productive subprojects, sustainability can be improved with careful analysis of project proposals for financial viability and sustainability before they are started. Respond to priorities of people outside the community (goal 8). Through the use of incentives, all three projects encourage communities to choose projects which conform to external priorities - limiting true community identification (goal 2). Brazil limits the cost of subprojects by type. Colombia, until recently, rigidly earmarked funds by subproject type. And Mexico gives priority to subprojects meeting basic needs, requiring communities to pay a larger proportion of costs for amenity infrastructure subprojects than other types. Because countries classify projects differently, it is impossible to see how this has affected the distribution of projects by type, so we cannot be sure how extensively the imposition of external priorities has distorted local preferences. This study has identified the conflicts among objectives, and the advantages and disadvantages of various rules which mediate between them. Unfortunately, we have not been able to identify the optimal DRIF design. This is in part because data available on the three projects are insufficient. In addition, differences in multiple variables among DRIFs make it difficult to identify with certainty relationships between subproject rules and outcomes (there are too many variables, and the sample contains only three projects). Still, some important conclusions can be drawn: * Design DRIF rules to reflect local conditions. There is no one optimal design suitable for DRIFs in all places at all times. The particular set of rules agreed for each DRIF, at each stage of its evolution, must result from negotiations among actors with different preferences and objectives - in different political, social, and economic contexts. * Consider offering communities with different capacities, different financing packages. Within the same country communities differ in their level of development, organization, and method of decisionmaking. Brazil has dealt with this by offering different programs to different communities depending on their capacities to implement subprojects. Communities with greater capacities take greater responsibility for implementing an annual development program, which FUMAC supports. Those with more limited capacities receive financing to complete a single subproject, with financing under PAC. xviii * Define objectives, and the tradeoffs among them clearly. Specifying objectives and recognizing the tradeoffs among them can help all actors choose and implement projects more successfully. It is also important that DRIF designers understand which tools (or rules) to apply in particular contexts. Next Steps The first priority is to collect more data on subproject sustainability. Simultaneously, much more needs to be learned about how specific tools or rules actually function. Finally, analysts should conduct small studies to find answers to the following two questions: * How much responsibility for planning and priority setting should be assigned to communities, municipal governments, and municipal participatory committees? * What can be done to better ensure quality at entry and long-run sustainability of subprojects, short of conducting full appraisals? xix INTRODUCTION Traditionally, central governments have been in charge of designing rural infrastructure projects, often with little input from local communities. To ensure efficient execution and minimal costs, project rules often required the use of centralized contracting and procurement. Now, however, many countries are giving communities more control over project choice and execution through the use of "demand-driven investment funds" (DRIFs). Broadly, DRIFs are pools of funds available to communities to invest in their preferred projects. Communities must submit applications for the funds describing in some detail how the funds will be used. However, communities are not free to use the funds in whatever way they want. Central government constrains local decisionmaking with eligibility rules - defining acceptable project types and providing incentives to choose projects of particular sorts - and constrains local implementation power with rules of procurement and disbursement. Generally central authorities monitor compliance with the rules, and also often retain the power to choose (or at least approve) subprojects. If local governments all functioned effectively, and projects were small and localized, then there would be no need for the rules. Communities could be given access to untied resources from revenue sharing or general purpose block grants, and invest in the projects they prefer unconstrained except by available resources. However, local governments often lack the capacity to choose and implement projects well: staff with essential skills are lacking, institutions for decisionmaking are weak, and mechanisms of accountability are absent. In Latin America, as elsewhere in the developing world, coronelismo and clientelismo characterize many local governments, and provide a rationale for central government oversight of local decisions. DRIFs have much in common with social funds common in Africa and Latin America: they are both designed to alleviate poverty, they operate with a high degree of participation by community members in the identification and implementation of many small subprojects, and they serve to strengthen community and municipal institutions. DRIFs differ from social funds mainly by targeting rural populations and supporting primarily productive infrastructure investments and natural resource management activities. DRIFs also more frequently give a strong role to local government institutions. Otherwise DRIFs and social funds are more alike than different, and most of the conclusions of this study apply equally to both types of funds. Study Objectives Can carefully-devised rules and incentive structures for DRIFs replace central control and allow communities greater power over project choice, while assuring that coronelismo and clientelismo do not corrupt the process and lead to poor project outcomes? Which rules and incentive structures offer most promise of achieving this objective? This study examines in detail the rules and incentives structures of several World Bank- supported DRIFs in Latin America under implementation long enough to generate useful information. The projects are Colombia's Fondo Desarollo Rural Integrado (Fondo DRI), Brazil's Programa de Apoyo Comunitario (Community Support Program) and Fundo Municipal de Apoio Comunitario (Municipal Community Fund) (hereafter referred to as PAC/FUMAC), and Mexico's Solidaridad projects. The operating rules and incentive structures of these projects are changing from year to year as communities and central governments gain experience with them, and modify them to make them more effective. Therefore some of the information reported here may no longer be valid. We do not assess the performance or design of the projects, or try to rank them relative to each other or to other types of projects. However, we do believe that DRIFs offer considerable promise as tools to combat rural poverty. The purpose of this paper is to identify the design features which can make them most effective. To provide an analytical framework for the empirical analysis and a primer for people designing DRIFs, we begin by defining some key concepts and with a proposition: the main function of DRIF rules is to reconcile conflicts among multiple goals shared in varying degrees by different actors (such as the World Bank, central and local governments). To help in understanding the differences among the three DRIFs on which we focus, we discuss relevant aspects of the political and administrative frameworks of the countries and projects. We then identify the structures of rules governing these and similar projects (as described in the project documents and as implemented in practice), their rationale in relation to core objectives and country circumstances, their advantages and disadvantages, and some of their likely consequences. There are of course tradeoffs among goals, which makes it impossible to define an optimal set of rules for DRIFs appropriate in all places and at all times. Nevertheless this analysis does provide information on the way particular rules affect behavior, useful for people designing new DRIFs, and identifies issues to explore through further experimentation. Basic Concepts Rules, like laws, are prescriptions which people must follow to gain rewards or avoid penalties. Rules are enforced through formal sanctions or social pressure. This is in contrast to guidelines, which are not enforceable, and to incentives, which motivate people to behave in desirable ways through rewards or penalties which generally are directly proportional to the degree to which people follow the prescriptions. We are concerned primarily with rules intended to meet objectives or establish priorities among conflicting objectives. Project rules may be few and simple, or many and complex. Monitoring and enforcement are easier when project rules are few and simple, which can speed project implementation. However, when rules are too few and simple, they may not adequately serve to encourage behavior which serves the project objectives. Establishing rules is not the only, or necessarily the best, way to encourage desired behaviors. Where behavior is the result of habit, tradition, fashion, or ignorance, and not due to conflicting objectives, education (training or propaganda) may more effectively change behavior than the imposition of rules. 2 The effectiveness of rules depends on both the probability they will be enforced and the magnitude of the reward or penalty. If the payoff from a crime outweighs the penalty if caught, many people would commit the crime (for example, polluters). Put another way, a given penalty will be more effective in stopping crimes with small payoffs than ones with large rewards. FRAMEWORK OF OBJECTIVES In analyzing the framework of rules and incentives for a DRIF, we examine the framework of objectives or goals at each level of the decisionmaking hierarchy: the World Bank and task managers, central government (elected officials, political appointees, and civil servants), intermediate government, such as the state, provincial or departmental governments, local or municipal government (mayors and governing councils), community leaders and community members. What are the policy objectives and their rationales? Do different actors have different goals, creating confliict among actors? Are goals or objectives which are in conflict fimdamental and enduring (for example, reflecting deeply held values and beliefs), or are they malleable? Policy Objectives DRIFs are designed to meet a number of different policy objectives. They are to: * Target rural poverty (goal 1). Grant-funded projects should primarily benefit the poor. DRIF rules should be established so that most of the benefits go to the poor, or at least soc that the benefits are distributed more equitably than normally. Nonpoor communities can either benefit from existing programs, or can afford to arrange to pay for their projects through taxation or public borrowing, and therefore should not be eligible for grant funding.' * Give communities control over project choice and implementation (goal 2). When community members choose their projects, and help pay for and implement them, they are much more likely to value and sustain them. Full community involvement usually requires decentralized decisionmaking. * Build municipal capacity (goal 3). DRIFs should help local governments develop permanent capacity to plan and execute small projects, where this is lacking. Where local governments already possess adequate capacity, DRIFs can help transfer responsibility for selecting and implementing small projects to local governments. The criteria for targeting can be expressed more rigorously in several ways: as ensuring that the largest proportion of the poor population receive benefits, as ensuring that most of the nonpoor do not receive benefits, and as ensuring that the level of benefits received per capita is proportional to severity of poverty. Borrowing from statistical theory, the first can be rephrased as minimizing errors of exclusion (Type I errors) and the second as minimizing errors of inclusion (Type II errors). Strict targeting should minimize both types of errors. The third criterion makes benefits progressive. As Annex 2 shows, in theory a single formula can encompass all three criteria. In practice, however, targeting must be based on imperfect indicators of poverty and rural residence at the level of some administrative unit. 3 Either way, DRIFs may be designed to foster decentralized decisionmaking and implementation. To ensure that new capacity is permanent, existing local institutions - not transitory project implementing units - should execute the project. The goal of building local capacity may conflict with the goals of reducing poverty or involving local people in choosing and implementing projects. Local governments do not necessarily respond to the needs of the poor (indeed the central government is often more concerned with the interests of the poor), and leaders may prefer to build projects which will strengthen their political power. Other Objectives Other objectives are common to international development projects: * Choose projects which are economically viable (goal 4). Curiously, this most universal of goals conflicts with the goal of involving local people in choosing and implementing projects (goal 2). Local communities may choose projects which cannot pass the test of viability (generally defined as projects with acceptable economic rates of return). Requiring viability gives outsiders the power to veto community choice. This goal may also conflict with the goal of targeting poverty (goal 1). The poor often live in resource-poor and isolated regions, in conditions that make it particularly difficult for projects to succeed. Furthermore, carrying out competent feasibility studies of the many small subprojects which communities desire is not possible. * Implement projects efficiently (goal 5). This sensible goal may conflict with the goal of involving local people in choosing and implementing projects (goal 2) and perhaps that of targeting poverty (goal 1). Generally, implementing projects efficiently requires hiring experts with skills and experience from outside the community. Poor communities are less likely than others to have members with the necessary expertise and experience to implement projects efficiently. However, if project activities are fairly simple and the community fairly well-organized, local people may well be able to manage the project more efficiently than outside experts or central government employees using force-account approaches. * Manage funds efficiently and honestly (goal 6). Donor and governments have developed complex regulations to control procurement, accounting, and auditing to ensure the efficient and honest use of funds. However, these rules, which authorities designed in response to experiences with centralized projects, frequently conflict with goals 1 and 2. Targeting poverty means working with poor communities - that typically lack the technical capacity to satisfy the regulations. Involving local people in implementing projects means putting untrained and often uneducated people in charge of affairs requiring skills and training. Furthermore, local governments often lack the capacity or integrity to manage accounting, procurement, and expenditures. 4 * Effectively manage project operation and maintenance, including recurrent financing (goal 7). This goal suggests that local people should participate in the daily management of the project, at least of small subprojects for which professional management is not feasible. External financing and government agency assistance is often not available after the initial investments, so if works are to be maintained and used efficiently, communities must often finance and manage small subprojects through their own financial contributions and efforts. * Respond to priorities of people outside the community (goal 8). Central governments or external donors may give priority to certain types of subprojects, and not others. There are many reasons why. For example, national constitutions may require countries to provide specific goods or services to its citizens, such as access to education or health care, which demand priority. Countries may require that projects or programs be clearly separate in content, so new projects do not duplicate older ones. They may restrict the use of public funds or require public ownership of facilities constructed with public funds, or they may support projects favored by politically important constituents. Whatever the reasons for the preferences, this goal conflicts with that of giving communities control over project choice (goal 2). Finally, some important but largely unstated goals of individuals or institutions may influence design and implementation: * Conform to institutional imperatives (goal 9). Institutional or bureaucratic rules do not always promote - and may even may be at cross-purposes with - project goals. Yet, policymakers may require projects to follow institutional imperatives anyway, even though this has a negative impact on project design. * Administer projects cost-effectively (goal 10). The goal is to minimize the time and effort required to develop, implement, and supervise the project. This is often a high priority of institutions and individuals involved in project preparation and implemenatation. Indeed other goals are often sacrificed in pursuit of cost- effectiveness. * Get reelected (goal 11). This is often the most important goal for national and local political leaders. Indeed, politicians are likely to promote other goals only to the extent they do not facilitate getting reelected. Of course many of the other goals do not particularly further this one, as discussed below. Local government leaders in Latin America are less concerned with this imperative than those in other regions, because often they cannot serve for successive terms in the same office (they can serve in other offices). Rather, rewarding one's patrons (goal 12) is the most important goal for local government leaders in Latin America. * Reward one's patrons or political supporters (goal 12). This is a crucial goal of politicians, which is even more important where localities restrict the number of terms 5 politicians can serve. This is the definition of clientelismo. Usually politicians reward patrons by giving them contracts outside the competitive bidding process. Hence clientelismo conflicts with the goals of choosing economically viable projects and executing and managing them efficiently (goals 4-7). Sometimes the need to reward patrons will conflict with the goals of targeting poverty and giving communities control over project choice and implementation (goals 1 and 2). Participation as a means. We do not take participation as a general goal, even though giving communities control over project design and implementation is often central to DRIFs. This is because participation at different stages of the project cycle can serve quite different goals. * Participation in project selection. Community members should be involved in choosing subprojects (goal 2), though this goal may conflict with that of building municipal capacity (goal 3). The public funds available for communities are likely to be limited, and insufficient to finance subprojects of reasonable scale each year in every community. Thus participation in choice usually means community members identify and nominate subprojects for implementation at the community level, and perhaps, influence final allocative decisions at the municipal level. * Participation in project finance (community members provide labor, cash, or materials). This helps ensure that the community members truly desire the projects (goal 2) and helps to satisfy the World Bank's country counterpart requirements. But, because poor communities can rarely contribute much cash or many materials, in practice community members contribute mostly their labor (except when family members working elsewhere send cash to their own families to substitute for the labor requirement, which may be difficult to meet when family members at home are primarily women and children.) This goal can conflict with that of efficient implementation (goal 5) if, for example, local laborers are less skilled than outsiders, yet contractors prefer to use skilled labor. * Participation in implementation. With proper incentives (most practically, the right to retain and utilize a proportion of funds budgeted but not spent due to efficient implementation), community management of construction work, rather than management by contractors or government officials, can promote efficient implementation, funds management, and operation and maintenance (goals 5-7), as well as community involvement (goal 2).2 * Participation in operations and maintenance. While this can promote effective operations and maintenance and cost recovery (goal 7), it is often very difficult to implement, requiring a long-term organizational effort which is difficult for many 2 This incentive is often prohibited for government projects. Instead savings are supposed to be returned to the treasury. 6 communities to sustain. Indeed, experience shows that community members are most likely to actively participate in the project during the initial stage of identification, and much less thereafter. Although local government may perceive all forms of participation as conflicting with its prerogatives, in fact participation contributes to the formation of a strong civil society. Indeed the existence of a strong civil society is essential to building municipal capacity (goal 3), since it is needed to offset any tendencies toward coronelismo and clientelismo. Discretionary control over project resources. Retaining discretionary control over resource allocation or other aspects of projects is not a fundamental goal, but rather a means of achieving the goals of getting reelected and rewarding one's patrons or political party (goals 10- 11). In its most extreme forn, coronelismo may be seen as the complete control of resources by politicians to use to get reelected or reward those who appoint them to office. Requiring the use of a strict formula to determine the distribution of resources reduces the power of politicians by restricting their control over funds. Strictly applying a formula also rigidly fixes the weighting of goals underlying it. By contrast, politicians may respond rapidly to changing external circumstances or shifts in voter priorities. Thus it seems surprising that politicians would be willing to surrender their control over resources, allowing the use of a formula instead. However, they may well be willing to require others to use formulas (for example, national politicians may require local politicians to use a formula, or the legislative branch of government may require the executive branch to do so). They may also accept the use of formulas if external donors or foreign governments require it. It may be easier for members of political bodies, such as legislatures, to agree on formulas to allocate resources rather than on the specific allocations. Or civil society groups may demand the use of formulas to defeat coronelismo and clientelismo. THE ACTORS AND THEIR OBJECTIVES The key actors - the World Bank, central, intermediate, and local governments - may have conflicting goals. The following describes the characteristics and motivations of the actors. World Bank The World Bank is generally reasonably clear about its objectives (which typically include all non-political goals, 1-10). However, transparently resolving conflicts, making tradeoffs, and establishing priorities among objectives or between the World Bank and the government is not typically part of the World Bank corporate style. Project task managers are often aware of, and perhaps troubled by, conflicts or tradeoffs, particularly the potential conflict between the goals of targeting rural poverty (goal 1) and those seeking efficiency and quality (goals 4-7): often the poorest communities are least able to execute projects efficiently. The World Bank can manage this conflict (and a similar one between community management and efficiency) in various ways: 7 * Impose subproject eligibility rules which limit project type to those which are relatively simple, cost less than a defined amount, or can be completed within a short period of time. * Provide external technical assistance to help communities develop capacity to manage projects. * Allow less-poor communities to participate in the project along with the poorest communities. Efficiency and timeliness of project execution, on average, is likely to improve. * Target poor communities for priority access to funds, but, if they fail to meet other requirements, reallocate unutilized funds to less-poor communities capable of satisfying the requirements (this effectively allows only communities with adequate implementation capacity to undertake projects). Bank staff. The people who prepare projects sometimes see a conflict between coimmunity participation (goal 2) and municipal capacity building (goal 3). In the long-run, however, participation builds and strengthens municipal capacity by strengthening civil society. The conflict is instead between the objectives of the World Bank and those of local politicians. To help resolve program designers can: * Delineate in detail the responsibilities and implementation mechanisms of community groups and local government at each stage of the subproject cycle. Specify the allocations to each for technical assistance and training. * Give a third party (for example, a nongovernmental organization) the role of arbitrating between levels by having them disburse project funds. * Disburse project funds directly to municipal governments or community organizations. Very often the World Bank favors certain types of subprojects over others (goal 8). However these are not always the same type of projects which communities favor (goal 2), so these desires are sometimes in conflict. Generally the World Bank gives priority to supporting the projects it favors at the expense of community control over project choice (goal 2). The World Bank's favored project types often reflect donor preferences (for example, environmental or social services projects). The preferences also arise from the need to conform to institutional imperatives (goal 9), for example, avoiding financing activities eligible for support under another project. They also reflect the desire to avoid using public funds for "frivolous purposes" (basketball courts) which have vote appeal (goal 11) or which reflect coronelismo. Sometimes they reflect mistrust of the mechanisms put in place to ensure that projects are truly demand- driven. Of course they may also reflect the preferences of World Bank staff who believe they 8 know what is really in the best interests of their clients, in contrast to the client's expressed desires. World Bank preferences also reflect goals of World Bank staff other than task managers: procurement and disbursement staff, lawyers, and others who review and influence project design. The World Bank should first be very clear on its relative priorities, and simplify its objectives as much as possible, before contending with conflicting national goals. Central Government In designing DRIFs, World Bank staff would often prefer to minimize the role of the central government, instead giving local governments or beneficiary communities maximum control over project resources. However, the three levels of government - central, intermediate, and local - must at least consent to (if not enthusiastically endorse), project concepts and the implementation rules. There is likely to be considerable conflict between World Bank goals and goals of government at all levels. There is also likely to be conflict between different levels of government and among actors at the same level (ministers, civil servants, political parties). Project rules are shaped in response to the need to reconcile multiple goals. World Bank staff are likely to be particularly concerned with the extent to which the central government supports the targeting of rural poverty (goal 1), community control over subproject choice (goal 2), and municipal capacity building (goal 3) - and empowers lower levels of government to achieve these. Central government actors are likely to take positions that further their desires ito get reelected or reward their supporters (goals 11 and 12), which may conflict with the goals of the World Bank or local government. Most central governments in Latin America and the Caribbean do not give high priority to rural poverty alleviation. The exceptions are countries where the party in power has a strong rural base or is competing with parties or forces (including guerrilla forces) that do. The region is among the most urbanized in the world (71 percent of the population live in cities), so rural people have little political voice. However, ministries of agriculture within central government do respond to rural concerns, and often support projects intended to alleviate rural poverty. If they did not, these projects would not be implemented in the region. Government (at all levels) may also support projects which further its various ideological or strategic goals. For example, government may be more interested in maximizing employment over its term in office than in investing in education and health care that will raise incomes over the long-run. Agencies sponsoring projects will usually emphasize their own, often limited, mandate in choosing subprojects. This limits community control over subproject choice (goal 2) to a set of project types which sponsoring agencies or World Bank divisions are willing to support. The third major conflict is likely to arise when governments are not fully committed to municipal capacity building (goal 3). Each country differs in the extent to which central 9 government defers to intennediate or local authorities. Central government's attitude towards lower levels of government may be affected by: * Mistrust of local or intermediate authorities (especially if the political party or factions in control of central government differs from the ones controlling local or intermediate government); * Desire to strengthen political support of local or intermediate authorities (especially if the same political party controls the local or intermediate governments); or * Desire to obtain voter support even if this comes at the expense of local or intermediate authority (especially if the party controlling central government differ from the ones controlling lower levels). The executive and legislative branches of government and the civil service are likely to have very different views, and may even be unable to reach a consensus on the issues. There are a variety of ways of resolving major differences between the World Bank and central government. If the World Bank is financing part of a wider program, and identifies conflicts with government's objectives, the World Bank can limit its financing to the project activities (or geographic areas) that fit within its objectives, for example, by creating special rules. Money is fungible, however, so the World Bank's claim that, by restricting its support, it does not compromise its objectives is frequently self-deception. Project Management Units The civil service is intended to carry out government policies, not create them. But every bureaucracy has its own independent objectives -primarily to maintain itself and maximize the resources under its control. These goals may heavily influence project design and implementation. This is because political leaders are rarely attentive to design and implementation details, and because detailed rules are usually developed after the start of implementation. The preferences or objectives of the people staffing project management units may depend on how secure they feel in their jobs. Staff of small units which are relatively understaffed and not under threat, may prefer to minimize their workload by decentralizing and sharing power and minimally enforcing rules. People of larger or overstaffed units, if insecure in their jobs, may seek to justify their positions by centralizing and monopolizing power and strongly enforcing rules. Managers desiring increased power or resources may seek to centralize power and increase the enforcement responsibilities of their staff. In any case, unit staff would likely desire to decentralize the least attractive or most burdensome tasks (field supervision), and retain the better tasks for themselves (annual budgeting). And staff of most central units would seek to maximize resources available to them (vehicles and office equipment). 10 Intermediate Government Local government officials will be primary players in managing DRIFs, so defining their roles and powers are among the most crucial issues in designing DRIF rules and incentives. But, the interests of intermediate levels of government - the province, state or department, which may or may not share the attitudes or objectives of local and central government - must also be taken into account. Indeed intermediate governments generally have some role in managing DRIFs, although this role is not always explicit in the project rules, or well-monitored by the World Bank. Still, it is essential that DRIF designers give sufficient thought to the 'role of intermediate levels of government when designing rules and incentives. Local Government Local government's primary concern is likely to be vote appeal and patronage (goals 11 and 12). If civil society is weak, there will be no checks on local officials pursuing these objectives. Even if civil society is reasonably strong, it may function primarily in the urban areas (cabeceras). If this is true then "rational" local politicians will focus on satisfying urban voters and patrons, and neglect the concerns of rural people. In allocating resources, politicians with strong urban bases are likely to favor: * Densely populated areas, since more people will benefit for the money invested * Communities with a tradition of high voter tumout * Communities the members of which are sources of financial contributions, taxes, or counterpart funding * Communities which are home to people who can influence others (to vote in a particular way or donate money) * Projects with high visibility or which narrowly target key supporters. Communities of poor rural dwellers do not usually meet these criteria, because they: * are situated in sparsely-populated areas, where households are widely dispersed * have a history of low voter turnout * cannot afford to contribute much to political campaigns, or more than labor for projects in their communities are unlikely to have strong organizations to articulate their demands, and * have minimal technical capacity to undertake projects on their own. 11 At the same time it is not unusual for local politicians to want to pay back the people or groups which provide them with financial support or arrange for kickbacks - the quid pro quo of clientelismo. Municipal clientelismo does not necessarily conflict with community control over subproject choice. The mayor may care little which projects are undertaken as long as there is money to be made from them. Clientelismo does imply that mayors, rather than community organizations, will be in charge of selecting contractors to execute projects. To appear powerful and win votes, mayors (and other elected leaders) want credit for key decisions. Transferring real decisionmaking power to communities is not usually part of the mayor's agenda - increasing his own discretionary power is. Under what circumstances would mayors be willing to transfer real power? There are several possibilities: * A quid pro quo. For example, communities may be willing to pay taxes if they have a say about how the money is spent (this is likely to favor wealthier areas). * A change in voter attitudes. Voters may stop rewarding politicians for pork-barrel delivery of resources, and judge performance on other criteria. * Growth in community organizations which track and influence politicians' behavior. Civil society may become stronger and pressure the mayor to be responsive and honest. * The existence of rules which constrain mayoral powers. Thus, unless rural civil society is reasonably strong, elected officials will likely discriminate against poor rural areas, retain the power to make decisions, and reward their financial backers with municipal contracts. The goal of strengthening municipal government is thus best achieved by taking action to strengthen civil society - not simply by providing technical assistance and training for the mayor and his staff, as often occurs today. Thus, there is no real conflict of objectives between community-based decisionmaking and implementation on the one hand, and the strengthening of municipal capacity on the other. The joint objective is to strengthen civil society so that it can oversee - and force changes in - politicians' behavior. Where civil society is well developed, decentralized and participatory approaches generally work well. By contrast where civil society is weak, local elites or caciques may capture or subvert the process. Where civil society is sufficiently strong, there is little justification for DRIF designers to bypass municipal governments and impose extra-legal authority on ad hoc community organizations. Traditional Local Elites Where civil society is weak, traditional community elites control political structures. Elites are not necessarily only the wealthy or educated members of the community; elites may be from particular racial, cultural or occupational groups. Traditional community leaders may 12 maintain their status through affiliation with a political party or officials of higher levels of government. However, many community elites will not support the participatory goals or transparency requirements of externally-funded projects, either because they prefer to make the key decisions themselves or they wish to reward their supporters or cronies (vote appeal and patronage, goals 11 and 12). It is naive to assume that decisionmaking at the community level necessarily reflects the desires of the majority of community members. Thus people designing DRIFs must also take into account the functioning of government at the community level. Where elites are likely to control project outcomes but not work in the best interests of most community members, then DRIFs should contain rules to constrain or bypass traditional community leaders. Rules requiring "real" community participation may threaten the prerogatives of the traditional elite. Hopefully, new leaders will emerge, who are better able to guide the community through change. Sorne traditional leaders will adapt to the changing political climate. Others may successfully collEude with the mayor or others and claim that participatory rules have been followed, when they have not. THE CONTEXT Brazil Political framework. Brazil is a federation of 27 states with 4,974 municipalities. Excluding the eight large urban agglomerates, populations of municipalities average 28,600. Brazil's citizens choose their representatives at all levels of government through popular elections. These include the president (who serves a nonrenewable four-year term) and the congress (senate: 81 mnembers serving eight-year terms; chamber: 503 members serving four-year terms) at the national level, governors and assembly council members at the state level, and mayors (prefeitos) and municipal council members (camara de vereadores) at the local level. Municipal resources. Municipalities receive financial resources from taxes, income- generating activities, and transfers from higher government levels. For small municipalities, neither taxes nor income-generating activities generate much revenue, so transfers from higher levels of government often account for more than 90 percent of total resources. The most important source of funds from higher levels of government is the Municipal Participation Fund (Fundo de Participacion Municipal). Funds originating from specific federal taxes flow automatically to municipalities based on a variety of indicators, the most important being population. However, there is scope for negotiation and political manipulation regarding final allocations from this fund. This can cause severe problems for small municipalities who depend on the Fundo de Participacion Municipal for a large proportion of its funds, but who have little negotiating power. Often these communities receive barely enough funds to cover operating costs, much less new investments. Municipal responsibilities. Brazil's 1988 constitution transformed municipalities from administrative entities to relatively autonomous political institutions. Municipalities took on 13 new responsibilities, including municipal land administration and the provision of local public services. Municipalities also assumed the major responsibilities for water supply and sewage, sanitation, primary and secondary education, health, local and regional development, the construction and maintenance of basic infrastructure and local roads, and a variety of other services. Today, to provide these many services, rural municipalities often employ large numbers of people. Indeed it is not uncommon for municipalities to provide low-income employment to one-quarter of the economically active population within their jurisdictions. They thus serve as a kind of public employment program. Participation. During Brazil's years of dictatorship (1964-1988) the heads of government at all levels had great power, and popular participation was very limited. For example, mayors had much greater power than the municipal councils (camara de vereadores). These councils, comprising 9-22 elected members, could issue municipal laws, approve specified general items of the municipal budget, and impeach the mayor. However, the mayor was often revered as the symbol of the municipality and its benefactor, granting resources received from the central government, and councils rarely exercised their power to impeach him. It is interesting to note that, during the period of dictatorship, elections were regularly held only at the municipal level. The -Fundo Municipal de Apoio Comunitario and the Programa de Apoio Comunitario in northeast Brazil. The Brazilian Northeast Rural Poverty Alleviation Program has been underway for ten years working to alleviate rural poverty in northeast Brazil. The program has passed through three distinct phases, each incorporating lessons of experience from the previous one: * Phase 1. The original Northeast Rural Development Program (NRDP, also known by its Portuguese acronym - PAPP (Programa de Apoio ao Pequeno Produtor)) lasted from 1986 to 1993. The program was divided into 10 separate loans, one for each state of northeast Brazil - Sergipe, Rio Grande do Norte, Pernambuco, Bahia, Piaui, Ceara, Paraiba, Maranhao, Alagoas - and part of Minas Gerais. These are the poorest areas in the country. The program had a small demand-driven component called the Apoio as Pequena Comunidades Rurais (support to small rural communities) which operated differently than other components through its ability to respond to the needs of rural communities. * Phase 2. After a field visit to the Solidarity Program in Mexico, Brazil's state governors decided to reformulate the Northeast Rural Development Program in 1993, by expanding and improving upon its Programa de Apoio Comunitario component. The reformulated program provides matching grants to rural community associations to use for eligible small-scale subprojects of their choice. Of the eligible subprojects, the beneficiaries choose those meeting their most critical needs. The reformulated Northeast Rural Development Program works through two different delivery mechanisms: Programa de Apoio Comunitario (Community Support Program, PAC), under which rural communities submit their proposals directly to state governments. State 14 government officials screen, approve, and release funds for subprojects, working directly with beneficiaries throughout the process. This is to reduce the risk that resources are lost to intermediate government levels. Evaluation of the first phase of the program showed that more than 50 percent of targeted resources failed to reach the beneficiaries. Fundo Municipal de Apoio Comunitario (Municipal Fund for Community Support, FUMAC), under which a municipal participatory committee (MPC) is created to screen, prioritize, and submit proposals to state government. The committees are not part of a formal municipal structure. They do however contain representatives of municipal government together with representatives of communities and civil society. The committees of FUMAC municipalities allocate project resources, set priorities for subprojects, promote local-level consensus-building, foster transparency, and monitor subproject inmplementation. * Phase 3. Nortiheast Rural Poverty Alleviation Program (NRPAP) started in 1995-96. This program, which builds on the successes of the reformulated Northeast Rural Development Program, was recently launched in the three states of Bahia, Ceara, and Sergipe, and is being prepared in three others. This phase expands the use of FUMAC and introduced a third delivery mechanism: Fundo Municipal de Apoio Comunitario-Piloto (Pilot Municipal Fund for Community Support, FUMAC-P), under which municipal participatory committees, in addition to their responsibilities ifor administering FUMAC funds, assume implementation responsibilities, receive firm annual budgets and manage the resources. This new mechanism is intended to transfer as much responsibility for administering FUMAC funds as possible. A municipality can participate in only one of the programs, PAC, FUMAC, or FUMAC-P. Thus communities within FUMAC municipalities cannot apply for funds under PAC. Rather they must operate through municipal participatory committees and cannot deal directly with state governments. In fact the newer FUMAC and FUMAC-P systems distribute resources more equitably than the PAC system by setting priorities across communities rather than leaving it up to individual communities to present proposals to the state. In the 18 months immediately following the reformulation, the program provided US$85 million for subprojects: US$74 million under the PAC and US$11 million under the FUMAC. FUMAC-P, which new projects are using, has not yet begun to disburse. Project management units, which are part of the planning secretariat, implemented the second and third phases of the Northeast Rural Development Program. Colombia Political framework. Colombia is a republic with 32 departments and 1,050 municipalities. Excluding the five largest towns, population of municipalities averages about 23,000. During 1985-90 Colombia began decentralizing government functions and resources after a long period of central control. It passed several laws, issued a presidential decree, and 15 adopted a new constitution giving more resources, responsibilities, and decisionmaking authority to municipalities. Since 1988 citizens have chosen their mayors through popular elections to serve for three-year non-sequentially-renewable terms. The 1991 constitution established the right of the people to elect departmental governors. Currently citizens elect the heads of all branches government: the president and national congress (senate: 102 members; and chamber: 163 members serving four-year terms); governor and departmental assembly (11-31 members); and mayor and municipal council (7-21 members). Municipal resources. Municipalities have three sources of funds: taxes, income- generating activities and transfers from higher level of governments. Transfers are the most important source of funds for small municipalities. Transfers from central to municipal governments have been rising in recent years, increasing from 2.6 percent of GDP in 1980 to 5.5 percent of GDP in 1994 (World Bank 1995). They should reach a minimum of 22 percent of total national revenues by 2002 (Ferreira, and Valenzuela 1993). Colombia uses a cofinancing system with a matching-grant mechanism to foster municipal investments, almost 12 percent of public investments in the country. For this reason, municipalities use central government transfers primarily for more urgent expenditures, often recurrent costs. Sector institutions administer the cofinancing system with specific rules and mechanisms. One of these institutions is Fondo DRI, analyzed below. Municipal responsibilities. Municipalities are supposed to use their resources to administer schools up to the secondary level, hiring teachers and providing and maintaining infrastructure; administer local hospitals and clinics, providing and maintaining necessary infrastructure; provide basic water and sanitation, energy and telephone services; administer low- income housing programs; build and maintain urban and rural roads; provide agricultural extension and environmental protection services; provide cultural, sport, and recreational services; and, in some cases, provide security or police services. Participation. Government has created numerous ways for communities to participate in selecting and implementing projects and programs - individually or collectively, through formal and informal channels. However, the degree to which community members actually influence outcomes still varies considerably. In many communities, mayors - who usually have poor relationships with municipal councils - often make key community decisions alone (World Bank 1995). But many communities in southern Colombia, where the Indian custom of donating labor for tasks which benefit the entire community (minga) endures, community participation is quite high. Indeed the practice of participation is so strongly embedded in some communities that even senior officials shovel debris and pour cement in a minga. Fondo Desarrollo Rural Integrado (Fondo DRI). Like the northeast Brazil Rural Poverty Alleviation Program, the Colombian Fondo DRI has been operating for two decades to reduce rural poverty, and has evolved from a centralized rural development program. By contrast with the program in Brazil, its conversion to a community-based program has been piecemeal, and is still not complete. Colombian municipalities, which have been efficiently soliciting financing for and implementing subprojects for years, are responsible for its success. 16 Fondo DRI has evolved in three phases: Phase 1 (1976-82, with total investments of around US$200 million) covered 213 municipalities; Phase 2 (1982-88, with total investments of around US$330 million) covered 337 municipalities; and Phase 3 (1988-96, known as Programa de Desarrollo Integral Campesino (PDIC) with a total investment of US$250 million) covered all 1,050 municipalities in the country. Phase 4 is currently under preparation. Prior to Phase 3, Fondo DRI was a traditional integrated rural development institution, whose managers, with the support of technical staff, selected and implemented all activities. Beneficiary participation was limited to consultation. During Phase 3, which started in 1990 with the support of a World Bank loan, the program began financing subprojects which municipalities proposed and managed. Even now, however, resources are rigidly allocated among a limited set of activities: technical assistance, organization and training, fisheries and watershed management, rural women's programs, marketing, rural road construction, water supply and sewerage construction, electrification, and flood control. There is no flexibility in how resources are allocated: funds available for rural roads cannot be used for technical assistance. Under Phase 3, Fondo DRI funds in theory can go directly to nongovernmental organizations and organized communities, but in practice nearly all the resources are channeled through municipal governments. Mexico Political framework. Mexico is a federation of 32 states divided into about 3,000 municipalities. Excluding the eight largest cities, municipalities average about 23,000 people. Citizens choose their representatives at all levels through popular election: president and national congress representatives (senate: 128 members for six-year terms; chamber: 500 members for three-year terms); governors and state congress representatives; and mayors and municipal council members. Although the 1983 constitutional reform increased the authority and responsibilities of the municipal government, municipal autonomy is still limited. Municipal resources. The central government transfers 22 percent of total federal revenues to states. The states are obliged to transfer a minimum of 20 percent of these resources to municipalities (4.4 percent of the federal revenues). States allocate the funds according to specific criteria - which vary from state to state - such as total municipal population, total number of commurnities within a municipality, or municipal "marginality index number" (a poverty indicator based on distance from sources of services). Municipalities also directly levy and retain property taxes: these are significant only for large, urban municipalities, and contribute about 10 percent of resources (including transfers) available to rural municipalities. Most rural municipalities have very limited resources, which are barely sufficient to cover operating costs, let alone new investments. Hence, programs such as Solidarity were created with the purpose of financing local investments. Responsibilities. The sweeping constitutional reforms of Article 115 in 1983 gave municipalities the responsibility for water supply and sewage, roads, public area lighting, sanitation, and slaughterhouses. The same law gave municipalities the right to collect and retain property taxes and user fees to fund the services they provided. In practice, since they must 17 receive state approval for their budgets, tax rates and tariffs, and other local laws, municipalities remain highly dependent on state governments. Participation. Decentralization is quite recent in Mexico. Debates regarding participation occur more at the state and national levels than at the municipal level. The potential for participation should be analyzed in light of the one-party rule that has characterized Mexico's political system for the past 66 years. People of remote rural municipalities in most states participate little, if at all, in municipal affairs. However, Mexico does have some traditional participatory organizations at the community level through which people donate labor (faenas) for tasks which benefit the entire community. The links between traditional participatory organizations and local governments are generally weak. Mexican Municipal Solidarity Fund (MSF). The National Solidarity Program, now renamed the Superacion de la Pobreza, was the Mexican government's primary tool to reduce poverty and provide social services to communities that were underserved. It started in 1988 and included 30 subprograms targeting health, education, nutrition, employment, infrastructure, agriculture, institutional development, and income-generating subprojects for the poor. Thus it was not a single program but a general approach to reducing poverty, under which there were different components. The program was built on the principles of community participation, poverty targeting, decentralization, and responsibility and cost sharing among federal government, states, municipalities, and beneficiary communities. Within the Solidarity Program, World Bank loans (the First and Second Decentralization and Regional Development Projects, Loans 3310 and 3790) supported the Municipal Solidarity Fund (now absorbed into the Fund for Social Municipal Development). The first project covered four states, Chiapas, Guerrero, Hidalgo, and Oaxaca, and involved investments of nearly US$1.4 billion, of which the World Bank financed only US$350 million. The second project, currently under implementation, added four states, Michoacan, Puebla, Veracruz, and Zacatecas, and involved investments of US$1 billion, of which the World Bank financed US$500 million. Within the program, the Municipal Solidarity Fund was established to finance small subprojects, which are completely managed at the municipal and community levels. Community assemblies identified potential subprojects, and specially-formed municipal participatory committees (one per municipality), called "solidarity councils," made the final selections in accordance with simple eligibility criteria and minimal procedural rules. Table 1 presents the major features of and rules governing the Brazilian, Colombian, and Mexican programs, structured around the eight principal policy objectives or goals. The table also includes the Bolivian Fondo para el Desarrollo Campesino, encompassing the Programa Integrado Desarrollo Campesino (PIDC), a more recent project with some interesting features, discussed later. 18 Poverty Targeting versus Political Imperatives The most important instrument for attaining the major goals of DRIFs is the budgetary allocation among basic administrative units (commonly municipalities). In the three programs analyzed here, the major compromises among conflicting goals are made through the budgetary allocations. This need not be the case. DRIF rules could merely define eligibility criteria for projects and programs, and not a budget constraint for municipalities. Municipalities would apply for funding to a higher-level body, which rations the available funds (a number of social funds in the Latin America and Caribbean Region operate this way). However, imposing a budget constraint on municipalities helps to build municipal capacity to manage projects (goal 3) by making municipalities responsible for decisions regarding small projects. It also helps to target rural poverty (goal 1) by making it more difficult for the more technically-competent or resource-rich areas to obtain a disproportionate share of funds. Requiring municipalities to provide counterpart funding would also create an implicit budget constraint. But this approach would not serve the goal of poverty targeting (goal 1), since the better-off communities are the ones best able to provide counterpart funding. Moreover, it is usually not practical in projects of national scope to impose a budget constraint at the community level, rather than the municipal level, as communities (pueblo or ejido in Mexico) usually lack legal status, administrative apparatuses, or detailed information on demographic and economic conditions. Within municipalities, funds are allocated to either communities or, more often, subprojects. To avoid excessive fragmentation of funds, municipalities do not generally give all communities an equal share of funds. There are many possible criteria for allocating budgetary funds among municipalities, including municipal need, first-come - first-served, demonstration of satisfactory past performance, development potential, or existence of adequate capacity for subproject execution. Funds may also be allocated through political negotiation. To avoid giving political actors excessive discretionary control over allocations, DRIF designers must define the allocative criteria explicitly in terms of objectives and monitorable indicators. Unfortunately, there are few useful indicators that can be applied at the municipal level, making it difficult to use some criteria such as the development potential or adequacy of executing capability. Generally some indicators of municipal need exist, including total population, rural population (or ratio between the last two), percentage of people living in poverty (or the percentage of people with unsatisfied basic needs), degree of isolation, number of communities or localities, and the extent of existing budgetary transfers. DRIF designers can use a mathematical formula, assigning weights to the indicators in relation to different objectives, concepts of equity, and contexts. For instance, the decisionmakers in Chiapas, Mexico considered the proportion of indigenous people living in a municipality as important, and gave this indicator extra weight in their allocative formula. To satisfy political imperatives (goals 11 and 12), some DRIFs do not entirely exclude wealthier regions from eligibility for funds. They use formulas to allocate funds among certain administrative levels but not others, or they leave a proportion of resources for political actors to 19 Table 1 Objectives and rules comparison matrix Brazil Colombia Mexico Bolivia PAC, FUMAC Fondo DRI Municipal Solidarity Fund Fondo de Desarrollo Campesino Rural Poverty Within a specified poor Budget allocation formula Priority to poor states (inter- Priority poor areas. Funding Targeting region, villages < 7,500 among departments; state allocation formula strictly for rural productive people. First-come, first- villages < 2,500 people; recently introduced). groups, who have limited served (in theory). positive list excluding most Within-states, budget access to resources, and who Political influence may be urban projects. Cofinancing allocation formula, varying are using traditional significant. matrix varying with by state; specified shares of technologies. municipal poverty level. funds for rural areas. Community Community groups propose Rigid fund allocation by Community groups propose Legal entity representing Identification projects, subject to (low) type of subproject. High projects, subject to beneficiary community beneficiary cofinancing (but varying) cofinancing (moderate) cofinancing, proposes projects, subject to varying by general type. requirements specified by discouraging amenity (moderate) cofinancing Nongovemmental detailed positive list, with infrastructure and varying by type. Heavy organization intermediation no distinction between encouraging technical nongovemmental in some cases. municipal and final assistance, with "basic organization intermediation. beneficiary cofinance. needs" orientation. Strong participatory planning process. Municipal Within FUMAC, municipal Municipal govermment Municipal participatory Bolivia had its first local Capacity participatory committees selects, cofinances, committees, with elections in 1995, and the Building prioritize with civil society implements. Municipal representatives of existing municipal role is still (FUMAC-P implements). participatory committees beneficiary groups, select limited. project PAC bypasses municipality. were created to fulfill projects (but sometimes management unit selects State screens and approves. central government merely rubber stamps them). subprojects, which are requirements. Either groups or municipal subject to full appraisal. government implement Community legal entity projects. manages projects. Economic Subject to technical Subject to technical Unit cost < regional average Full technical and financial Viability and feasibility, environmental feasibility, environmental (?). Assured budget for study: IRR > 12 percent. Sustainability sustainability. sustainability. staffing. No underutilization Gender and environmental of existing infrastructure. impact analysis. Beneficiary self-management. Efficient < US$50,000; I year. No maximum cost per < US$45,000; 1 year, no < US$250,000; I year. Implementation Disbursement against subproject. One year staged projects. Disburses budgeted costs. Savings duration (in practice). against budgeted costs; locally retained (?). savings locally retained. Efficient Funds Advance disbursement (one Disbursement of World Advance disbursement (two Limited local bidding for Management tranche) to municipality Bank's share with contract tranches) from govemment procurement. from World Bank via between municipality and (but often delayed). Mainly definition as grant contractor. Mainly local direct contracting. Disbursement for entire shopping and direct annual program for contracting. qualified municipalities. Mainly local shopping and direct contracting. Effective Beneficiary association is Operations and maintenance Budget for staff of schools In principle, beneficiaries Operations and owner of subprojects, and costs included in proposal. and clinics must be assured. are responsible for Maintenance and responsible for its Municipality is responsible. operations and maintenance, Cost Recovery operations and although they may maintenance. The subcontract its execution. association can transfer some types of subprojects to local govemment or agency. External Negative list being piloted. Rigid fund allocation Cofinancing penalty for Strong attention to gender, Priorities according to type of amenity infrastructure. environment. Cofinancing subproject. Cofinancing Productive projects only percentages modestly reflect percentages strongly reflect being piloted. Some extemal priorities. extemal priorities. excluded activities are I financed by other projects. Source: Authors' elaborations 20 use at their discretion (for example, the Mexico Municipal Solidarity Fund allocated 15 percent of total funds to the state governors of Oaxaca and Guerrero for this purpose). The Colombia and Mexico DRIFs both allocate funds according to a formula. In fact, the Mexico Municipal Solidarity Fund used eight different formulas to allocate resources among municipalities in the eight states involved in the program (while the national government required states to use formulas involving some combination of objective indicators, it left the specific choice of indicators and weightings to the governors). The Mexican DRIF has not in the past used a formula to allocate resources among states, but its newest incarnation does include formulas for doing so. The fund amounts have been publicized, so transparency is high. Colombia's Fondo DRI uses a formula to allocate resources among the 32 departments of the country. Departments use a different approach to allocate funds among subprojects (described below as a subproject approach). The Brazil DRIF does not currently use formulas to allocate funds. However t:he phase 3 will use them to allocate resources among the FUMAC-P municipalities (see Annex 1 for the formulas). Extent of pcverty targeting in practice. A statistical analysis of resource allocation among municipalities was carried out for Brazil's PAC/FUMAC, Colombia's Fondo DRI, and Mexico's Municipal Solidarity Fund to measure the degree to which these DRIFs target poverty. Specifically, correlations between municipal poverty levels and investments (per capita allocation, DRIF financing, and beneficiaries' contributions) at the municipal level were looked at. These measures provide a rough indication of whether subsidies are actually allocated according to strict formulas, encompassing all three poverty targeting criteria (ensuring that the largest proportion of poor people receive benefits; that nonpoor do not; and that per capita, the poor receive benefits in proportion to the severity of their poverty) (see Annex 2). The statistical analysis was complicated by certain problems of data incompatibility, so only transfers related to the DRIFs were measured, and other budgetary transfers or municipal revenues were ignored (these are likely to be strongly inversely correlated with poverty levels, but also mostly used to pay municipal operations costs).3 In addition it is impossible to compare national programs with subnational prograrns without data on excluded areas: subnational areas were chosen to participate because they were poorer than average, so they do not represent the national averages. With these major caveats, the following conclusions can be drawn: * Brazil's PAC/FUMAC (1993-95). The correlation between municipal poverty levels and investments was 7 percent within the generally-poor region where the projects were implemented. 3 The poverty measures are significantly different among countries, as well as the definitions of the target population. To reduce the impact of these incompatibilities on the analysis, relative rather than absolute measures of poverty level were used; and investments per capita, considering the total population, were calculated, although this is not the target o:f DRIF (estimates of the rural or poor population were problematic because of varying definitions). Nevertheless, the usual poverty measures - the percentage of population with unsatisfied basic needs (in areas such as education, health services, water supply, sewerage, and housing) are imperfectly correlated with income. 21 The Colombian Fondo DRI (1994). The correlation was 18 percent among all the eligible municipalities of the country. The Mexican Municipal Solidarity Fund (1994). The correlation was strong within the states (20 percent in Chiapas and 26 percent in Hidalgo) where allocation formulas were properly applied. Correlations were negative in other states (-2 percent in Guerrero, -8 percent in Oaxaca). However, correlations across all participating states were slightly negative (-4 percent) because no formulas were used to allocate resources among states. These correlations at best are only moderately positive. Still, it is likely that these funds reach more poor people than other funds available to municipalities. There are several reasons why DRIFs may have had limited success in targeting poverty: countries may be using allocation formulas that give weight to factors like the number of municipalities or total population size, and which are not positively correlated with poverty but are politically relevant; and countries may not be systematically applying formulas to allocate funds. Targeting rural areas. In addition to using budget allocation formulas (which may or may not include measures of regional rural populations), some funds have used subproject eligibility criteria to assure that rural areas within administrative units have access to fund resources. For instance, Colombia's Fondo DRI targets rural areas: only villages with fewer than 2,500 inhabitants are eligible for fund resources. Brazil's PAC/FUMAC provides resources to villages with 7,500 inhabitants or less. Mexico's Municipal Solidarity Fund requires that 50 to 75 percent of resources be invested in areas outside the municipal urban core (cabecera municipal).4 In addition, Colombia's Fondo DRI, unlike most other funds, earmark specific amounts of financial resources for specific activities (such as technical assistance, rural roads, marketing, fisheries, water supply, rural women, training, environmental protection, and municipal strengthening), violating the principle that communities should choose projects, but reducing the otherwise substantial discretionary power of the mayor and the tendency for the cabecera to appropriate a large proportion of the resources. Earmarking has recently been eliminated. Demand-Driven versus Feasibility and Sustainability Subproject approach. Allocating municipal funds (or funds of other small administrative units) equally among communities would lead to excessive fragmentation of resources, leaving each with insufficient funds to implement subprojects of reasonable size. Therefore municipalities generally need to ration the funds among communities. Typically, they allocate 4 Minimal share of funds outside the municipal urban core (cabecera): * 50 percent in "A" municipalities, with less than 5,000 inhabitants in the cabecera; * 75 percent in "B 1" municipalities, with more than 5,000 inhabitants in the cabecera, but less than one-third of the total municipal population; and * 60 percent in "B2" municipalities, with more than 5,000 inhabitants in the cabecera, and more than one-third of the total municipal population. This presumes there is a geographic and statistical base sufficient to identify the cabecera. 22 funds for the highest-quality subprojects among those proposed by communities. Other rationing schemes are possible, for example municipalities may offer communities the opportunity to apply for funds on a rotating basis, or they may ration funds on the basis of community needs or political imperatives. The Mexico Municipal Solidarity Fund gives priority to communities lacking basic needs, and little emphasis to subproject quality. In some cases funds may be allocated according to a formula at one level, but on the basis of subproject-quality at a different administrative level, such as the state, municipal subunit, or community.5 Aside from the fragmentation issue, there is a tradeoff between having lower-level or higher-level administrative units handling the funds. Lower-level units are closer and sometimes more responsive to the clients, but higher-level units have the skills to implement larger subprojects. In Nicaragua it is planned to leave options open for larger projects, such as intermunicipal roads or large watershed environmental projects, if all localities involved accept them and provide the necessary counterpart funding. In Colombia funds are allocated according to a formula at the departmental level, but on the basis of subprojects within each department. Participatory microplanning. At the municipal level and below, ensuring that community members truly select projects requires that mechanisms for participation are in place. Participation means that beneficiaries (with the other actors, depicted in Figure 1) identify priority needs and propose investments to meet them in accordance with a defined development strategy. The major challenge is ensuring that DRIFs do not fund projects which beneficiaries select but which are not actually feasible and sustainable. This is often accomplished by making higher administrative levels responsible for final approval. But DRIFs can also include mechanisms to help communities avoid generating unrealistic wish-lists. DRIF designers need to consider how much structure to impose on the participatory processes through a combination of rules and outside assistance. There are three general approaches. In the first, there is no formal process, but, to be considered, project proposals must come from a minimum number of beneficiaries or a nongovernmental organization speaking on their behalf. Higher levels are responsible for reviewing and approving the proposals. In the second, beneficiaries choose from a positive list of eligible project types and conditions for each, reflecting the judgments of outsiders regarding project feasibility, sustainability, and importance. In the third, there is a formal process to set priorities among proposals offered by beneficiaries, and to roughly evaluate their feasibility and sustainability. Nongovernmental organizations or local leaders assist with the process. Regardless of approach, the participatory process works best when beneficiaries know DRIF rules, project eligibility criteria, and fund availability. This requires active dissemination of information. S For example, veredas in Colombia, regimientos in Mexico. 23 Figure 1 Actors in participatory microplanning National nternationalAi Government Agencies oll' ~~~~~~~(World Bank) Local Technical Experts Government * I *.~-------~ Traditional _- _ _ Participatory v U Municipal Participatory ' 'o m m ittee , Civil Society > C X (NGOs, trade unions, Beneficiary Groups -private sector, - - - - - - - - - -_- - - - - - or Communities religious groups (vulnerable groups) Source: Authors' elaborations Choice of approach may depend on whether the emphasis is on generating large numbers of subproject proposals quickly, or on strengthening community participation processes. Each of the approaches has disadvantages. The first, because it is relatively unstructured, can lead to divisiveness and abuse by individuals (including contractors) promoting particular subprojects, and an increased burden on higher-level decisionmakers charged with approving project proposals. The second - if the positive list is too narrow - can stifle true community development planning. The third, the most structured of the three approaches, requires considerable time, expensive training, and the assistance of nongovernmental organizations. Full-scale participatory development planning without known budget constraints (usually set at the municipal level) can easily degenerate into wish-list creation. In practice, municipalities and communities often do not know how much money will be available to them until late in the planning process, which makes it impossible for them to know which projects are likely to be accepted by higher-level authorities. In addition, not all municipalities can meet eligibility criteria and deadlines in order to qualify for funds, so in the first round of planning, budgeted resources are likely to exceed disbursements for eligible projects. The problem could be resolved by overallocating budgetary resources (if government regulations permit), or by producing two-stage development plans, the first reflecting a known budget constraint, and the 24 second to be implemented if additional resources become available. In Mexico communities do just this: every municipality knows its allocation for the following fiscal year, and also that it may be receive up to 20 percent more resources in a second-round. Indeed, planners could use a two-stage procedure to address multiple objectives. For example, in the first stage planners could target rural poverty by allocating funds through a formula, and in the second stage reallocate surplus funds to projects on the basis of quality or other criteria. (See Box 1 for an example of the participatory planning process). Box 1 The participatory planning process in Bolivia The Bolivia Rural Communities Development Project supports a participatory planning process designed to help rural communities and municipalities (the last created only in 1995) to be effective players in decisionmaking, and stimulate grassroots participation in local planning and project identification. The participatory planning process involves a series of municipal workshops in which representatives from government entities, communities, development organizations, and iinterested individuals - the civil society - participate. Between workshops, community delegates return home to inform community members of what they have learned and listen to feedback. Since many communities in Bolivia make decisions by consensus, this process can be lengthy and tedious. The participatory planning process is divided in five stages. The first is the preparatory phase, during which objectives and methodologies are introduced. The second is the self- diagnostic phase, during which communities reach a better understanding of their economic and social strengths and weaknesses to help with formulating achievable socioeconomic goals. Process leaders assist communities to identify constraints, their causes, and solutions, through the use of participatory rural appraisal techniques - such as participatory mapping and seasonal calendars. The third is the plan phase, during which communities prepare municipal development plans, and a list of subprojects ranked in order of priority. Subprojects are screened according to how well they further the development strategy, meet community requirements, match funds availability, are technically feasible, and suitable for DRIF financing. The fourth is the plan adjustment phase. The fifth is subproject preparation phase during which communities turn subproject ideas into concrete implementation proposals. Beneficiary contribution. To ensure that beneficiaries truly desire subprojects, many DRIFs require government bodies at different levels and beneficiaries to contribute financial resources to the project. DRIFs have different rules for this: some require beneficiaries to contribute fixed amounts, others vary the requirements depending on project type or degree and extent of poverty in the community. Some DRIFs require both municipalities and beneficiaries to contribute, others do not. In the most complex cases, a cofinancing matrix is used to vary the required contribution depending on the type of subproject and the poverty level of the administrative unit or municipality (see examples in Annex 1). Cofmancing matrixes can be useful in a number of ways. Central government can use them to encourage communities to follow its investment priorities (at the expense of their own), without using positive or negative lists. Decisionmakers can use them to ensure that DRIF resources go primarily to the poorest areas, even if these communities contribute less counterpart funding. The Colombia Fondo DRI 25 targets poorer localities by tailoring its cofinancing requirements according to the level of poverty in a municipality. Municipalities generally contribute financial resources, while beneficiaries contribute either cash or unskilled labor, local material, land or other noncash resources.6 Municipalities with a large proportion of poor people may have limited financial resources, so requiring them to contribute can discriminate against the poor. On the other hand, valuing beneficiaries' contributions of labor or materials at market rates often works to the advantage of the poor. There are dangers of trying to meet too many objectives (such as external subproject preferences) through the cofinancing matrix when beneficiary contributions are limited to labor and local materials. Although even the poorest beneficiaries have some resources to contribute, their ability to contribute labor is limited by other demands on time, and may vary widely with the season. This means that labor contributions will represent variable proportions of project costs. In addition, some subprojects may be too technologically sophisticated to productively employ much local labor. To require beneficiaries to contribute a higher proportion of project costs than they can reasonably afford has risks: it can lead communities to choose projects that are not in their best interests, or encourage them to cheat through accounting maneuvers (such as valuing contributed gravel at high shadow prices).7 Subproject cost limits. DRIFs finance small subprojects, whose planning and operations and maintenance can be managed by rural communities and governments of poor municipalities (goals 3, 4 and 5); and which are smaller in scale and lower in quality than those sought by urban areas and richer municipalities (goal 1). Corruption is also less likely with small projects (goal 6). Therefore many DRIFs contain rules limiting the cost of subprojects. Mexico's Municipal Solidarity Fund limits subproject cost to US$45,000, and Brazil's PAC/FUMAC limits it to US$40,000 (later increased to US$50,000). Colombia's Fondo DRI does not have a limit. Consequently, Fondo DRI finances costlier projects than the Mexico and Brazil DRIFs, averaging around US$38,000 compared with US$10,000 for the Mexico Municipal Solidarity Fund and US$20,000 for Brazil PAC/FUMAC. A few large subprojects comprise a large share of Colombia Fondo DRI financing: in 1994 the fund financed 148 subprojects costing over US$100,000 each out of 2,983 subprojects. This was 25 percent of total costs, but 5 percent of the total number of projects.8 To avoid cost limits, communities sometimes divide subprojects into stages. To enforce the limits, it may be necessary to prohibit this practice. To encourage communities to choose projects which are simple and discourage them from starting but not finishing projects, the 6 The valuation of some goods (such as land) or services (labor) can vary greatly. Operational manuals usually set some percentage limits for contributions in kind, varying them by types of subproject. 7 In many cases, it is extraordinary how precisely communities meet exactly the minimum counterpart requirements! 8 In reality, the large subprojects financed by Fondo DRJ are part of a different program (asignaci6n especifica) which has different objectives. The problem is therefore that the same institution manages different programs with different objectives. 26 Mexico Municipal Solidarity Fund requires communities to complete subprojects within a year of starting. To adhere to these rules, communities tend to choose simple public infrastructure projects, not complex subprojects even when these would be justified. In addition, when communities face subproject cost limits rather than budget constraints (and can meet counterpart funding requirements without too much difficulty), they often select the largest subproject allowed under the DRIF rules. Subproject feasibility, economic viability, and sustainability. Subprojects are evaluated according to how well they achieve the generic goals of international development projects (goals 4-7). To better assure project quality, technical committees or units should follow three procedures. First they should screen subprojects ex ante by evaluating how well they meet the eligibility criteria. Second they should evaluate them ex post, and reward or punish the people responsible for the results depending on performance. Third, they should ensure that beneficiaries particilpate in all phases of subproject development (selection, design, implementation, moriitoring, operations and maintenance, and evaluation) to ensure that projects truly reflect community preferences. Each procedure has its advantages and drawbacks. Evaluating many thousands of small subprojects rigorously ex ante in accordance with World Bank internal procedures is likely to be very time-consuming and costly, and a major bottleneck to subproject implementation. For instance, the Boliviain Fondo de Desarrollo Campesino (FDC) requires a minimum internal rate of return (IRR) of 12 percent for every subproject, forcing communities to undertake cumbersome preparation and appraisal processes. Instead, some simple rules-of-thumb could be defined (for example, thresholds could be established for subproject unit costs or costs per beneficiary; requirerments that specific investigations be carried out such as market analysis could be imposed; or negative lists of project types could be developed). Altematively, DRIF managers could identify some typical subprojects, and subject them to economic and sensitivity analyses. Evaluating subprojects only ex post allows communities to start implementing subprojects faster. But failing to screen subprojects before communities start to implement them may also lead to the financing of many poor quality subprojects. Denying communities financing for future subprojects - the only practical punishment - is not generally appropriate except in cases of fiaud or deliberate breech of rules (these are common hazards, however). Strong participation by community members, though difficult to organize and enforce, is the best way to encourage economy, honesty, efficiency, and sustainability. However, without training, beneficiaries often lack the managerial, technical, or evaluative skills needed to manage subproject execution. Beneficiaries may also fail to take adequate care to minimize environmental damage. Most DRIFs require an ex ante analysis to demonstrate, at a minimum, the technical feasibility and environmental sustainability of every subproject. It is important that subprojects are technically feasible, since a common monitoring indicator used to judge their success is the number of subprojects completed: a subproject cannot be completed if it is not feasible. The 27 requirement of ex ante evaluation of environmental impacts derives from donor and nongovernmental organization sensitivity to this issue. DRIF managers often either fail to analyze ex ante the financial sustainability of subprojects, or expect beneficiaries to assure this. This no doubt contributes to observed weaknesses in operations and maintenance for infrastructure projects and a relatively high failure rate for productive projects. Several DRIFs contain special mechanisms to foster female participation. For instance, Colombia's Fondo DRI has a rural women's component which is financing technical assistance and small productive subprojects of organized women groups. In 1994, this component financed US$1.5 million (2 percent of the annual portfolio) to help rural women, primarily for small subprojects. The Bolivian Fondo de Desarrollo Campesino requires analysts to conduct gender analyses for each subproject prior to approval, which imposes an extra burden on those evaluating subprojects. Generally the ex post evaluation of subprojects has been given little emphasis. None of the studied DRIFs systematically collected information on the operation of subprojects once construction was complete. Authorities carry out ex post evaluations routinely in two situations. The first is when subprojects of a particular type develop problems (rural market subprojects, financed by Colombia's Fondo DRI, developed generic problems). The second is when donors require ex post evaluations before making new loans (as for the multiple descendants of PAC/FUMAC in Brazil). For the Brazil case, no more than summary results of the ex post evaluations are available. Nonetheless, informal feedback from field visits during supervision and from small-scale evaluations has been significant, and decisionmakers have used the informnation to modify many specific rules. Of the three DRIFs studied here, only the Mexico Municipal Solidarity Fund based its quality control strategy almost exclusively on beneficiary participation. Until 1994 this DRIF applied only two efficiency criteria ex ante: unit costs were not to exceed regional averages, and existing infrastructure of the same type should not be underutilized. Neither requirement did more than eliminate extreme abuses. Feasibility analyses of subprojects were not required, in part because the DRIF funded primarily basic social infrastructure subprojects, and few productive subprojects. In 1994, a new rule was established requiring that state governments and the Secretaria Desarrollo Social (Social Development Secretariat, Mexico) had to approve technical documents (expediente t&cnico) for each subproject. However, municipal-level governments could approve the technical documents for small subprojects, costing less than 40,000 pesos (in the future, this requirement may also be waived). In any case, because funds can be disbursed in advance against budgeted amounts, and because there are no penalties if communities "expand" subproject specifications, communities have a strong incentive to "stretch" the funds to expand project scope. Simple completion statistics do not generally reveal these practices.9 Subprojects financed by the Mexico DRIF appear to be as high quality as those financed by the other programs, although no formal evaluation has been carried out. It is unclear 9 Actual expenditures are documented by receipts, but need not match itemized budgeted requirements. 28 whether this demonstrates the effectiveness of beneficiary participation or the ineffectiveness of the use of ex ante criteria. One way to control costs ex ante is to require that projects conform to specified design standards. This has been common practice in the Brazil PAC\FUMAC. In Mexico some DRIFs require projects to conform to design standards. For example, the Decentralization and Regional Development Project II requires that some types of projects at the national level (for example, classrooms), and other types of projects at the state level conform to particular design standards. The design standards should represent cost-effective and technically-feasible options for the most numerous and simple types of subprojects. Obviously the standards should be flexible and easy for communities to iimplement (they should not require over-design, as is common in programs other than DRIFs). UJse of such designs should not be compulsory. Instead they should be used to technically pre-qualify proposals, eliminating the need for contracted designs and reviews, saving on costs of technical assistance. The danger is that communities may choose projects with approved designs, rather than ones which are more complicated but perhaps more urgent. Operations and maintenance. Managing operations and maintenance effectively (goal 7), and raising the funds from users to make this possible, has been a problematic area for DRIFs. This is in part because various levels of government and the beneficiaries share the costs and the responsibilities of various parties are often not clearly defined. Unfortunately little information is available on this important issue. Only a few localized surveys have been carried out, and there has been minimal ex post evaluation of how well operations and maintenance is being handled in subprojects. DRIF managers, during the participatory planning process, should make sure that beneficiaries are committed to participate in operations and maintenance, and make cash payments if appropriate. To be eligible for subprojects, communities should be required to provide estimates of operations and maintenance costs and the user fees needed to cover them - and propose plans for sharing the costs with government. Government may take responsibility for some operational costs, such as staff costs for schools or clinics. In such cases, DRIF managers may require that communities specify how these will be met before approving subprojects. For example, the Mexican Municipal Solidarity Fund will not finance subprojects until the community provides assurances that resources for recurrent costs, such as salaries for operational staff, will be available from other budgetary resources - not that such assurances can be entirely trusted. Colombia's Fondo DRI requires communities to provide estimates of operations and maintenance costs as part of their subproject proposals. This gives beneficiaries (the municipality), who are responsible for these costs, a clear understanding of the costs of their commitment. Beneficiaries' agreement to pay all operations and maintenance costs should also be clearly stated in the legal agreements - a requirement of Brazil's FUMAC/PAC. However, it is unrealistic to assume that just because formal commitments have been made, operations and maintenance will be satisfactory. It is necessary to have either a competent local government agency with clear responsibilities - and funding - for maintenance, or a strong beneficiary organization capable of collecting fees from its members and excluding free riders. Also critical is that the project implementing unit take seriously its responsibilities for operation and maintenance. This is unlikely when there is little or no ex post monitoring of subproject performance. 29 Coronelismo versus participation at the municipal level Municipal participatory committees. Community proposals need to be evaluated against alternatives and in light of budgetary constraints and the region's development strategy. From a juridical point of view, municipal councils are best equipped to undertake these evaluations and make decisions on levels of investment. While members of municipal councils are elected, in most of Latin America they have the reputation of being either bodies with little true political power that rubber-stamp mayoral decisions, or as excessively-politicized bodies. In addition, municipal councils often represent just the cabecera, or urban core, of the municipality. Consequently, DRIF designers often do not want to give municipal councils too much power within the participatory planning process, while recognizing that in the long-run it is necessary to integrate the participatory planning process into the legal structure of the municipality. DRIF designers have several options: enlarge the existing council, making sure it has rural representation; create a parallel institution such as a municipal participatory committee which selects projects (or plans investments in general); or create appeals mechanisms. For the Mexico Municipal Solidarity Fund, project designers have permitted either of the first two approaches. Whether they choose to enlarge existing municipal councils or create parallel institutions, DRIF designers must make sure that beneficiary communities are represented by community leaders or by organizations (such as nongovernmental organizations). Also, the enlarged or new council must be given legal authority to exercise its functions, either by local government or by higher authorities.'° Municipal participatory committees have been important actors in all the DRIFs studied here, with the exception of PAC in Brazil (PAC communities work directly with state level authorities). In Colombia mayors create municipal participatory committees under the authority of municipal councils. Representatives of municipal government have most of the voting power on municipal participatory committees (in Mexico, the mayor alone has voting power in the committees). The functioning of municipal participatory committees varies: some are dominated by mayors, who turn them into rubber-stamp bodies or institutions that build consensus for his or her plans. Thus, in Mexico, although government rules emphasized participation, actual beneficiary participation has varied considerably from state to state and municipality to municipality. In Colombia, only 450 municipal participatory committees have been created, although the country has 1,050 municipalities. Those that have been created often do not perform the functions they were expected to. One reason is that Fondo DRI from the beginning focused on strengthening the capacity of municipal governments, rather than increasing participation or eliminating caudillismo of the mayor. Strengthening municipal government is regarded as the priority, since Colombia has had elected municipal governments for only a few years. In Brazil, on the other hand, the municipal participatory committees seem stronger and more independent. This is because of the strength of civic organizations (unions and other 10 The commonly used terminology is ambiguous. Municipal councils are the elected legislative organs of municipalities, while municipal participatory committees merely allocate DRIF funds. In Spanish, the municipal participatory committees are called consejo municipal (consejo = advice) while the municipal council is called concejo municipal or ayuntamiento. In Brazil, the municipal councils are called camaras de vereadores, and the municipal participatory committees are called conselho municipal. 30 associations) whose leaders constitute the majority in municipal participatory committees, and because FUMAC rmunicipalities - which are the only ones with municipal participatory committees - are selected by the state based on municipal implementation capacity." To function well, municipal participatory committees should be independent and represent a broad cross-section of the community. Indeed, municipal participatory committees play an important role in developing a dialogue between public institutions and citizens and articulating the interests of groups which have lacked voice in the government. Therefore higher levels of govermment or DRIFs usually impose rules on municipal participatory committees. For example, DRIFs may require that they: Have a minirmum number of members (larger than that of typical municipal councils). Include representatives of both public institutions (mayor, municipal director of public works, representatives of the municipal council and of the state/department) and civil society (labor unions, nongovernmental organizations, churches, nonprofit organizations, subproject committees, rural community representatives). All three DRIFs analyzed require municipal participatory committees to have representatives of civil society: in Brazil they mnust be the majority. Have a minimum number of representatives of weaker social groups such as the poor, women, and indigenous people. Make choicc-s through one-person-one-vote elections (however the president of the committee, usually, but not always the mayor, often has the tie-breaker vote, also called the quality or minerva vote). The rules do not specify how or by whom individuals are selected or elected- potentially the most difficult and controversial issue in project design. If a formal participatory planning process is part of the project, this can form the basis of an electoral system of representation (as in Bolivia). If on the other hand, civil organizations are strong and coronelismo insignificant, it may be more practical to allow the mayor to appoint members, as long as there exists an appeals mechanism to handle abuses. In Mexico, in some cases, the top leaders of subproject beneficiary groups (comites pro obra) automatically become members (leading to large and unwieldy municipal participatory committees as the number of subprojects proliferates). In Chiapas, Mexico, specific municipal participatory committees are created to represent regions with a high proportion of indigenous people. " This does not represent a legal change in municipal powers, since municipal participatory committees have authority only to spend ]DRIF funds. These are managed by the state government and are not part of the official municipal budget. 31 Generally, municipal participatory committees are in charge of managing the participatory process, including such activities as: * Participatory planning * Advertising the project and building consensus for it at local level * Providing - directly or through contracts with others - training and technical assistance in issues of subproject design and operations and maintenance * Screening and reviewing subproject proposals * Approving subprojects - once it is verified that they meet eligibility criteria, and that funds are available (only in FUMAC and FUMAC-P) * Monitoring and supprvising implementation * Managing funds (in FUMAC-P). DRIFs do not generally define the rules of procedure for the municipal participatory committees. Brazil's PAC/FUMAC does require that municipal participatory committees approve internal by-laws through which their operating rules are defined. These must be consistent with DRIF objectives and allow sound participatory planning. Some examples of by- laws require that: * Terms of municipal participatory committees are limited (usually to two to three years). * Individuals have the right to participate, speak, and vote. * No more than a specified period of time passes between meetings. * Any member who fails to participate in more than a specified number of meetings without justification is excluded from the committees (some municipalities allow three unjustified absences). * Particular procedures be followed in calling for meetings. 0 Persons participating in meetings are remunerated (particularly participants without public salaries). * The executing secretariat and internal auditing department be organized in a particular way. 32 Issues. Many municipal participatory committees have problems attracting beneficiaries to participate in the planning process. In the past many rmunicipal participatory committees were formed merely to comply with DRIF requirements, and did not perform the functions that were envisioned for them. There are several reasons why. One is that municipal participatory committees often lack real power. (This is not the case in Brazil, where municipal participatory committees do have real power. These committees also appear to have been more successful in involving beneficiaries). A second reason may be that participants - except for salaried public employees and professional representatives of civil society - are not being remunerated for their time. Of course paying beneficiaries to participate in meetings to make decisions about their own investments is peculiar. However, rural dwellers may have to walk several days to participate in meetings, taking time from subsistence and income-earning activities. Providing a minimal per diem allowance would enable people who would not otherwise be able to afford the time to participate. There are other issues which DRIF designers might consider. For example, should the private sector be represented on municipal participatory committees? Although the private sector is an important part of civil society, there would be a conflict of interest if potential contractors have influence over public investment choices. How should the terms of municipal participatory comnmittees relate to the normal political cycle: should the terms overlap to ensure continuity, or not? A severe problem affecting DRIFs is the lack of continuity in municipal government. In many countries mayors appoint nearly all municipal staff, yet they themselves cannot serve for more than a single term. Therefore institutional strengthening efforts must be repeated after each election. External Priorities and Eligibility Criteria versus Demand-Driven Project Selection Collective benefit. A key eligibility criteria (which has not been spelled out, but is understood) is that DRIF activities may not directly finance activities of individuals but only those of groups (allhough DRIFs do not usually specify how large the group must be, it can clearly be smaller than an entire community). This is in part due to the existence of legal restrictions on the transfer of public property to private hands, and the desire to minimize the potential for corruption, limit money spent per beneficiary, or encourage group responsibility. The requirement that DRIFs fund groups, not individuals, forces communities to choose collective activities which, in the case of productive subprojects, may not be appropriate. However, in some cases, group revolving funds may finance individual activities. A simple indicator of whether the benefits of matching grants are going to only a small number of people is the "cost per beneficiary" measure. This indicator is not always easy to estimate: often it is not easy to quantify the number of beneficiaries, and the distinction between direct and indirect beneficiaries can be unclear. Project implementation documents should (but often do not) give clear operational definitions of "beneficiary" for each type of subproject. None of the studied cases enforces a maximum cost per beneficiary, even though data on numbers of beneficiaries are recorded. 33 Restrictions on subproject type. The Brazil and Colombia DRIFs finance only specific kinds of investment expenditures, listed as eligible subprojects in operational manuals (Annex 2 provides some examples). These DRIFs also encourage beneficiaries to select projects reflecting external priorities (goal 8) by varying the required counterpart contributions according to project type. By contrast, the Mexico DRIF has a uniform counterpart requirement, except in the case of a few types of projects. The Brazilian FUMAC and FUMAC-P use a negative list approach. Subprojects can be classified in various ways: * Investments in public goods, or social infrastructure. These include schools, health centers, roads (constructing, rehabilitating, or improving), water supply and sewerage systems, market structures, electrification, or storage facilities. * Investments in activities generating private income, or investments in productive infrastructure. These include workshops, tractors, and the creation of revolving funds for the purchase of agricultural inputs, agroindustrial facilities, or irrigation. a Expenditures on services, such as technical assistance and training, including for project preparation. * Expenditures to generate short-term employment (workfare), which is exempt from beneficiary counterpart requirements. * Investments to protect the environment, and improve natural resources management. * Amenity investments, such as for playgrounds, parks, social halls, and others. The first category of subprojects are similar to those financed by various emergency social funds in many Latin American countries. It is important to consider whether some of these subprojects are better provided by the public or private sectors (electrification and storage facilities, for instance). Another is whether DRIFs are unnecessarily competing with other programs in providing such investments. Permitting DRIF resources to be used for productive infrastructure - rare in Mexico and Colombia, common in Brazil - can be controversial (largely in the World Bank). This is because the practice can encourage communities to invest in collective organizations rather than allow the private sector to fulfill the need, and to invest in larger rather than smaller collective enterprises. It can also generate resentment if some community members benefit from subsidized loans or grants, while others must rely on private loan finance. Generally, formal financial institutions have little interest or ability to make the many small loans that the poor and rural dwellers require. Informal financial institutions primarily provide short-term working capital - so the competition is more hypothetical than real. However, the relatively high cost of productive projects compared with social infrastructure on a per beneficiary basis also discourages communities from choosing them if funds are limited and the process of selection is democratic. 34 The remaining categories also involve special considerations. Technical assistance and training may be treated either as part of subproject preparation, or as a separate subproject, or both. It is usual to budget a-reasonable proportion of project costs to finance subproject preparation and to exempt this from counterpart financing requirements to encourage beneficiaries to seek such assistance - of course these incentives are not always sufficient. Mexico's Solidarity Program has not been particularly successful in persuading communities to use such funds for their intended purpose. Technical assistance and training (including extension services), if the community demands them, are appropriate stand-alone subprojects, although single-year financing; may unnecessarily limit their ability to provide benefits. Employment generation schemes are popular with politicians, and are legitimate safety- net programs, but obviously are inconsistent with the requirement of community provide counterpart funding in the form of labor. They make sense primarily in places where community members are unable to volunteer labor because they are too busy meeting their needs for survival, or where subprojects, such as roads, mainly benefit people outside the community. In Mexico, components of the Solidarity Program (other than the Municipal Fund Program) have disbursed funds for such purposes in some remote areas. Investments to protect the environment, because they generate positive externalities, sometimes justify the use of subsidies. Finally, local politicians have long favored amenity investments (most such investments being located in the cabecera and highly visible). Mexico's Municipal Solidarity Fund requires high counterpart contributions to discourage communities from choosing these types of subprojects. Efficiency and Accountability versus Poverty Targeting Focusing on poor rural communities and requiring participation, however essential, can make project preparation less efficient. The low managerial and technical capacity of the communities can make it difficult for them to fulfill national and World Bank requirements on procurement and disbursement. It can also lead to low technical quality of subprojects.'3 Community capacity. Two obstacles discourage communities from participating in projects. First, a country's legal system may require beneficiary communities to organize and obtain legal status through registration, which may involve cumbersome procedures. Second, communities must have the basic capacity to manage finances in order to play an active role in managing subprojects. When regulations are complicated (for example, in Mexico it is difficult for solidarity committees to achieve legal status, making them dependent on the legal authority of the municipal government), and extensive training is not feasible, then communities must rely 12 For example, tree-planting subprojects may not be high in the priorities of local communities, but may be very popular as make-work schemes. Because labor is the majority (not just 10-30 percent) of costs, workers get paid for tree-planting.. 13 World Bank Operational Directives on the issue: When the project includes social objectives, the procuremnent procedures and contract packaging should be adapted as appropriate, within the framework of the procurement guidelines (OD 4.15). Contracts with nongovernmental organizations may need to reflect the fact that these organizations differ from commercial contractors. The contract might therefore stipulate, that nongovernmental organizations should involve the community in planning and implementation (OD 14.70). Similarly, special measures may be required to ensure that nongovernmental organizations have sufficient liquidity to fulfill contracts. 35 on intermediaries such as nongovernmental organizations or municipal governments to act on their behalf. The use of intermediaries tends to increase project costs (especially when the intermediaries need training) and sometimes reduces the sense of ownership on the part of beneficiaries as they resent the intermediary's. There are several ways to solve these problems. Government could, and should, change the laws to simplify registration procedures. The World Bank and government could simplify procurement rules. Project designers could use standardized designs to limit the need for intermediaries. Finally, project managers could use intermediaries to prepare for full self-management (including obtaining legal status). At the same time, DRIFs can require that intermediaries meet specific eligibility criteria. For example, they may demand that they have been active with the sector, region, or ethnic group for a minimum period of time; have available specialized personnel and equipment and access to technology; submit their plans to beneficiaries for approval, and agree to a beneficiary-based performance review. In addition, DRIFs could include incentive schemes or risk-sharing arrangements for intermediaries. This should be done with caution, however, since it would discourage nongovernmental organizations from working with communities where risks of failure are higher. Procurement methods. The traditional method of competitive bidding, designed to ensure that subproject implementation and funds management are efficient (goals 5-6), is often not useful in cases where most subprojects are small (averaging US$20,000 in Brazil, US$38,000 in Colombia, and US$10,000 in Mexico), implemented in dispersed and remote areas, and can be managed by rural communities themselves'. Creating procurement packages to increase their size could in theory help to facilitate competitive procurement. However, because distribution costs and the risks of mistiming are high, this is generally not feasible or advisable. For these reasons, the World Bank and governments have permitted communities to use local shopping and direct contracting to procure goods and services when the normal practice would have been to use competitive bidding, and these are the predominant procurement methods for DRIF-financed subprojects (see Annex 3 for thresholds). The Mexican and Brazilian DRIFs limit total subproject size to US$50,000, which is within the World Bank threshold for direct contracting. Other types of procurement are possible. In Asia, for instance, force account is often used in situations when government officials (for example, local project management units) hire and supervise community labor. Of course, allowing procurement to be dispersed, while easing one aspect of implementation, can create problems for another - disbursement. And, ex ante review of contracts by the World Bank or the local implementing agency becomes impractical. Using local shopping and direct contracting extensively does raise concerns that available resources will not be used efficiently and that materials and civil works may be poor technical quality. These potential problems are least likely in subprojects based on standardized designs and that use materials for which the unit prices are well known. Otherwise the project should include incentives to encourage people responsible for procurement to conserve funds. The Mexican Municipal Solidarity Funds Program, by disbursing against budgeted costs rather than against purchases or contracts, allows the community to use savings to expand subprojects, creating a strong incentive for efficiency. As a result, communities avoid overdesign, and implement subprojects costing 30-50 percent less than those built with normal municipal 36 contracting processes. The Brazilian PAC\FUMAC uses similar procedures. Fondo DRI uses the more traditional practice, disbursing funds against contracts between municipalities and contractors. Both approaches, however, encourage communities to overbudget to minimize the chances of cost overruns. Disbursements. Without disbursements, projects cannot be implemented. If some actors have discretionary control over disbursements - from the World Bank to the community - they can achieve their goals at the expense of others. People with discretionary control can make disbursements contingent on the completion of particular actions, providing strong incentives to others. However, these incentives are lost if there is no link between the actor dispersing the funds and those benefiting from them. For example, the Secretaria Desarrollo Social in Mexico receives funds under a World Bank-financed project, but only indirectly (see figure 2 for details of the flow of funds under the three projects studied here). The World Bank disburses funds to the central bank, which holds them in accounts which the agency cannot access directly. Instead the agency gets its funds from congress through the annual budgeting process and other sources. So the agency is not linked to the World Bank in any way, and has no reason to act in accordance with the World Bank's preferences nor to observe rules legally binding under the World Bank's loan. This is also a risk with decentralization of finance, whereby central agencies lose control of the disbursement process. Rules might then become simply guidelines. Indeed this is what is happening in Mexico under the latest incarnation of the Solidarity Program - municipalities have little reason to observe the rules and preferences of the Secretaria Desarrollo Social. If beneficiary communities or municipalities handle procurement, they must have funds disbursed to them. This poses two problems. First, grassroots implementing institutions (whether nongovermnental organizations or community groups) and small contractors generally lack liquidity. Unlike large contractors or suppliers, small contractors or implementing agents have no assets against which to secure World Bank financing of working capital. Second, there are generally huge numbers of small expenditures and receipts, posing an administrative problem. Neither the World Bank nor most govermments are equipped to provide advance payments, although the Mexican government has been able to do so. Therefore, governments must reform their procedures - a major effort that disbursement officers favor - or the World Bank must find a way to make advance payments without endangering the principle of accountability. By normal World Bank regulations, an amount sufficient for less than 30-days' expenditures may be withdrawn in advance from a Special Account; documentation of past expenditures is required for subsequent withdrawals. The 30-days of grace is insufficient for DRIFs to allow expenditures to be completed and documentation to be submitted, regardless of whether this comprises receipts or certification of use. The transfer of funds alone can take 15 days, especially when rural banks deliberately delay informing recipients that funds have arrived. World Bank guidelines are now allowing "Second Generation Special Accounts" under exceptional circumstances": when "(a) there is a clear need for such an account (such as logistical problems involving the transfer of funds from the central special account to distant rural areas where project entities are located); (b) the borrower is unable to provide project entities with adequate working capital to prefinance the World Bank's share of expenditures; and 37 (c) the borrower has the necessary authority to enforce accounting and auditing controls over the Second Generation Special Accounts and the administrative capacity to manage the financial aspects of the project.""4 Of course, (b) and (c) rarely go together. Under all circumstances, to monitor and control many small Special Accounts is extremely difficult, and therefore managers in the World Bank discourage this practice."5 An alternative, the "Simplified Disbursement Procedure," developed in response to similar problems in Africa, but not yet tried in the Latin America and Caribbean Region, extends the grace period to 90 days for advancing funds from the Special Account before expenditures have been certified."6 The Brazilian Northeast Poverty Alleviation Program uses an innovative solution: it disburses funds against contracts signed between DRIF management units (at the state level, with access to funds from the Special Account) and the agencies which will implement subprojects (municipal participatory committees or communities). There are two types of contracts in use. In PAC and FUMAC, project management units and beneficiary communities sign contracts involving specific subprojects. In FUMAC-P, DRIF management units and municipal participatory committees sign contracts to implement a set of subprojects over the course of a year, or a detailed annual work plan. Under the first type of contract, DRIFs disburse 100 percent of subproject financing at the time the parties sign the contract.'7 This is not considered an advance from a legal standpoint, since the contract between two legal entities constitutes acceptable supporting documentation. In the second case, DRIFs disburse funds in tranches in amounts equal to the planned expenditures of a single month. To disburse against contracts in this way is acceptable to the World Bank because the legal documents avoid distinguishing between goods and services: the only disbursement category for subprojects is termed "grants" (see Table 1). With this approach municipal participatory committees or communities need not inform the World Bank of how they are spending the money (local regulations are generally stricter). A statement of satisfaction or completion is sufficient to certify subprojects. Through this approach funds are disbursed not against inputs (which normally requires close monitoring of contracts, invoices, and other documentation), but against outputs (completed schools, water supply "' Funds from the primary Special Account are transferred into Second Generation Special Accounts as an advance, rather than a reimbursement. (World Bank, Operational Policy 12.20, Annex B, Subaccounts and Second- Generation Special Account, May 1994. See Annex 3) 1 For a short period of time, Fondo DRI used a mechanism similar to the Second Generation Special Account, which was managed by an independent institution, the fiduciaria. The results were not satisfactory, as thefiduciaria slowed the flow of funds in order to earn interest, and the mechanism was abandoned. 16 "The special 90-day advance procedure is allowed in social sector investment project when (a) the project covers an entire sector or subsector program and a wide geographical area or the entire country; (b) project implementation authority is decentralized to numerous regional, provincial, district, and other administrative levels; (c) NGOs and the beneficiary community at large participate in implementation; and (d) multiple donors provide funding for these projects." The last requirement seems irrelevant, since the problems occur with only a single donor. (Memorandum from the Director of the Loan Department on "Implementation of Simplified Disbursement Procedures under Investment Operations," December 14, 1994). 17 This procedure requires accurate cost estimates. When actual costs exceed estimates, project management units decide whether to finance the difference. When actual costs are below estimates, the savings are usually - but not automatically - left to the beneficiaries. 38 systems, or other investments), obviating the need to collect hundreds of receipts from every subproject (between 10,000-16,000 per year for the program), and greatly simplifying disbursement procedures.. The "grant" has become a disbursement category similar to "works," in which disbursements are made against outputs defined in a legally binding contract. The Northeast Poverty Alleviation Program has also created a system giving municipalities graduated responsibility for their programs. Municipalities which have shown they can implement subprojects successfully, are given resources to implement an entire annual work plan (FUMAC-P system). Those with more modest capacity are given resources to implement only one specific subproject at a time. Use of graduated approaches which vary with the capacity of the actors at a given point in time is an elegant approach. Figure 2 Flow of funds Brazil, Northeast Rural Poverty Alleviation Program: The Special Account disburses against a contract between project management units and municipal participatory committees (FUMAC) or communities (PAC) Loan Account D Special Account Municipal Account Contractor and/or (Washington) (State capital) (FUMAC-P only) Supplier State Counterpart Community Account (old project, also national) (PAC/FUMAC) (m unicipal ity and/or ) Is The Colombian project was designed before 1990, much earlier than the other two projects. Subsequently, the World Bank placed more emphasis on resolving procurement and disbursement issues. 39 Figure 3 Flow of funds (cont.) Colombia, Fondo DRI: The Special Account advances in tranches Loan Account pec_a ccount Municipal Acc contractor and/or (W ashington) (B ogota) _S upplier National Other Implem enting Counterpart Institutions j * (Departments, NGOs) Benefcary as Counterpart (municipality and/or community) Mexico, Municipal Solidarity Funds: The Special Account reimburses the general budget of the nation (not the budget of the line agency) Loan Account Special Account General National (Washington) (Mexico City) Budget Line Agency State Account Municipal Account . Community Account (SED ESOL) } ~~~~~~~Im plem enting / \ b g ~~~~~~~~Institution | State Account 9 ounterpart JI Contractor and/or Beneficiary Cash Supplier Counterpart || rn(unicipality and/or J community) Source: Authors' elaborations 40 IMPLEMENTATION AND ENFORCEMENT MECHANISMS Dissemination Good dissemination is crucial if DRIF objectives are to be achieved and coronelismo and clientelismo reduced. The first step in encouraging rural people to actively participate in subproject planning and implementation is to inform them of the principles and regulations governing DRIF activities. This is far from simple, and disseminating this information often constitutes a obstacle to subproject realization. Among the major issues are the following: * Communications can be extremely difficult between project managers and beneficiary communities, who are dispersed and often situated in remote localities with poor transportation and communication infrastructure. Many communities can be reached only after several hours, or even days, of walking. * DRIF regulations are complex. Even educated readers can find operational manuals difficult to understand. * Intermediaries can distort the flow of information. For example, contractors often disseminate only information which can benefit them. Intermediaries can obtain information more easily than rural dwellers, and they often offer rural communities ready-made subprojects. Communities need do nothing but approve the subprojects, and the contractor prepares all the paper work. This hinders many important objectives of the DRIFs, including building community capacity for project management, and implementing only subprojects which communities truly desire. • Dissemination generates expectations amnong the beneficiaries, which can be difficult to manage politically. Dissemination strategies must be designed to reach dispersed and remote rural communities. Projects should devote sufficient resources so that different methods can be used simultaneously. Radio campaigns - in different local languages, if necessary - have proven effective in reaching remote areas. However, DRIF managers often prefer to use television rather than radio, perhaps because television is more prestigious or to reach political decisionmakers. This approach makes little sense since rural households often lack television. Simple booklets and pamphlets, supported by graphics and drawings (comic books, for example), may be distributed, perhaps supplemented by workshops explaining their content to semi-literate rural people. Posting billboards or posters in places where subprojects are underway, which many government agencies favor, can make implementation more transparent and assist community members monitor progress by providing details on subprojects (cost, financial contributors, contractors, and others). 41 Monitoring and Evaluation It is problematic to monitor projects comprising thousands of subprojects being implemented by a similar number of local government units. It is easy to falsify numbers and signatures. Furthermore, quantitative indicators of implementation cannot capture subtle but important differences - such as whether committees merely rubber-stamp decisions made by the mayor or others, or truly make decisions. DRIFs can employ different strategies to better monitor projects: entrusting implementation to people in the local area who understand the objectives of the project and who are not beholden to local authorities, such as DRIF staff or employees from other central government agencies; creating management information systems (MIS) based on quantitative indicators as monitoring tools; and creating a beneficiary-based monitoring system. No single one of these strategies is sufficient on its own, so projects usually adopt all three simultaneously. Evaluating results is also complex. Indicators must be clearly defined to accurately and realistically measure whether subproject goals are being achieved. Achieving goals means more than simply completing subprojects (usually the main indicator). Creating community capacity through participation, and strengthening civil society and local government may be important goals, so it is important to evaluate the process through which communities identified, planned, and completed subprojects. As individual subprojects may be small, replicability is important. On-going evaluations following project completion can indicate whether the subprojects are truly sustainable. Figure 3 provides some examples of indicators which are useful for evaluating subprojects. If projects are intended to further goals not included here, then different indicators may be required. Enforcement and Sanctions Because most of the actors involved in project implementation have incentives to hide irregularities, it is essential to develop a methodology for enforcing subproject rules. It is necessary to distinguish violations which are due to fraud and those which are due to irregularities, and to select sanctions appropriate to the violation. Unlike fraud, irregularities do not interfere with achieving the overall project objectives."9 However, it is not always clear whether fraud or irregularities have been committed. When the situation is ambiguous, applying sanctions can deter others from committing irregularities. None of the three DRIFs studied here define clearly what constitutes irregularities versus fraud. Rather it is left up to the monitoring and legal departments of DRIF implementing units to make these judgments. 19 The authors observed an irregularity during a field visit: a community was using funds for two separate water supply subprojects to implement a different, single subproject. The funds were diverted to cover costs of the implemented subproject which exceeded the limit. However, the implemented subproject was very important for the beneficiaries, and all stakeholders agreed on the change. Therefore, although the community violated a rule, it achieved the project objectives. 42 Figure 3 Evaluation indicators Sustainability Replicablity > Effective Use c S N ostainab ility \ Human capacity / Patterns of diya g n~Bulding,, Soure A s oepalic aboility dSustainability Replicability Effective Use s Subproject qualitt and o Optimal use t Ability to expand services maintenance p Number of users S Transferability a Human capacity l Patte ins of daily and development seasonal use Cost recovery Source: Authors' elabo:rations The DIuFs studied here currently use three types of sucti Recovuspend disbursements; recover disbursed funds; and prosecute liable presentatives through the legal system. Legal prosecutions often take considerable time and expense, so are normally used only for cases of fraud. Community leaders or mayors are often in office for a lirnited period of time, so they rarely face penalties, if ever, while in office. Suspending disbursements and delaying approval of new subprojects are milder sanctions, which unfairly penalize the entire communitd or municipality. Most of the actors involved in project implementation do not favor this approach because everyone is interested in quickly completing the subproject. Recovering disbursed finds (or the goods on which they were spent) is more common, and is used under Brazil's PAC/FUMAC. Fondo DRI uses an interesting mechanism: when a municipality commits a noncritica'l irregularity (which does not merit suspension of disbursement), it must provide insurance -a sort of performnance bond or guarantee -to be eligible for new funds. 43 Informing communities of sanctions. This is essential if sanctions are to achieve their objectives - deterring others from committing irregularities. In particular, it is important that all relevant actors understand why they have been penalized, and what they need to do to correct the problem and end the sanctions. It is also important that people other than just the community representatives or mayors are subject to the sanctions. None of the three DRIFs studied here disseminate news of sanctions. CONCLUSION Findings How successfully have the three DRIFs achieved the eight goals most important to the World Bank? Which rules (or types of rules) have proven most useful, and least useful, in achieving these goals? * Target rural poverty (goal 1). It appears that the DRIFs have not been particularly effective in targeting poor localities.20 There are several reasons why. First, generally formulas have not been used to allocate funds. Second, even when used, formulas have included several criterion unrelated to poverty. Third, additional eligibility conditions which would exclude the nonpoor from the benefits of the subprojects have not been applied or enforced. Colombia's Fondo DRI is the most successful of the three projects in targeting poor localities. Project managers there used a cofinancing matrix which heavily favored poorer municipalities to allocate funds. * Give communities control over project choice and implementation (goal 3). In all three DRIFs, communities are responsible for identifying projects and submitting proposals for financing. It is unclear, however, to what extent the process is really participatory, or is instead dominated by outside influences (the mayor, contractors, or nongovernmental organizations). The requirement that beneficiary communities provide some proportion of subproject funds should ensure that beneficiaries at least desire the subprojects even if they do not regard them as their highest priorities. After all, all the DRIFs constrain freedom of choice by imposing eligibility, cofinancing, and feasibility criteria. Colombia imposes the greatest constraints on communities: Fondo DRI, until recently, rigidly allocated funds according to subproject type. In Mexico, where guidelines emphasize participation, experience has been mixed and generally fallen far short of the guidelines. There are few rules governing allocation of resources from the Municipal Solidarity Fund, and enforcement is weak. Therefore outcomes depend largely on the attitudes of local political and administrative authorities. Bolivia, with its requirement that communities adopt a participatory process of planning and priority setting, has developed the most elaborate participatory scheme. 20 We have not compared the performance of DRIFs in targeting poverty with that of other programs. 44 * Build municipal capacity (goal 3). In all but Brazil's PAC, municipal level authorities set priorities and choose subprojects among those proposed by communities. Municipalities have received considerable technical assistance to help strengthen their investment management capacities. The extent to which the mayor, the existing municipal council or a parallel municipal participatory council has dominated subproject selection has varied among countries and localities. On the whole, Colombia's Fondo DRI has tended to reinforce the mayor's authority: the municipal participatory comnmittees have not functioned as envisioned. The Brazil and Mexico DRIFs made more of an effort to include beneficiaries and other mernbers of civil society in decisionmaking (reflecting central government distrust of local government). This effort has been especially successful in Brazil, perhaps because civil society at the grassroots level (in the municipalities eligible for FUMAC) may be stronger than in Mexico, and because Brazil encouraged communities to develop mechanisms for indirect, rather than direct representation, of beneficiaries. * Choose projects which are economically viable (goal 4). The Mexico DRIF has relied entirely on beneficiary participation to achieve the goal of viability and sustainability. Most of the subprojects financed by the Municipal Solidarity Fund appear to have addressed basic social needs at substantially lower unit cost than other government projects. There have been too few productive-type projects under the Mexican program to judge whether participation would also have been effective for these projects. However experience from a limited number of subprojects in Chiapas suggests that productive subprojects may be less sustainable than subprojects addressing basic needs. There is no evidence that subprojects in Mexico were of higher quality than in the other countries, even though Mexico screened the subproject proposals for feasibility. However, there are too little data available on project performance following completion to allow us to draw firm conclusions at this time. It is clear that requiring full economic evaluation (plus environmental impact and gender analysis) of every subproject is a costly strategy, and probably an unrealistic goal given technical capabilities in local areas in these countries. Unfortunately little effort has been devoted to developing and applying rules of thumb which could be used to screen out most unviable or unsustainable subprojects quickly and easily. * Implement projects efficiently (goal 5). Is it better to implement subprojects through contractors or participatory self-management? There is too little information to be able to say for sure. We do know that in Mexico, which allows communities to retain saved funds to expand subprojects, communities have completed projects at lower unit costs than in the other countries. Although routine monitoring data provide information only on whether or not projects, as originally proposed, were completed, physical audits show that communities were able to expand projects by an average of 10 percent using the'original budgets. 45 * Manage funds efficiently and honestly (goal 6). The three programs have all experienced difficulties with disbursements, which were late or difficult to access because central governments were unable to advance funds. The World Bank's disbursement procedures are not designed to deal with this situation. But authorities have increasingly been able to find ways to manage disbursements to meet the needs of dispersed and decentralized communities, yet conform to World Bank rules. A weakness has been the failure of governments to enforce subproject rules through sanctions: this would demonstrate that government takes seriously the need for honest funds management. And, where the World Bank and implementing agencies (at central or local levels) differ significantly in their commitment to various project goals, the World Bank should disburse funds directly to these agencies to give them the incentive to observe agreed rules. This is the lesson from the Mexican experience. * Effectively manage project operation and maintenance, including recurrent financing (goal 7). None of the programs have performed particularly well in achieving this goal, according to the little information available. Indeed, information from local or partial surveys suggests that a high proportion of subprojects may not be sustainable. With infrastructure subprojects, communities need help to strengthen their capacities to better manage operations and maintenance. Sustainability is not guaranteed simply because communities formally agree to take responsibility for operating and maintaining subprojects. With productive subprojects, sustainability can be improved with careful analysis of project proposals for financial viability and sustainability before they are started. * Respond to priorities of people outside the community (goal 8). Through the use of incentives, all three projects encourage communities to choose projects which conform to external priorities - limiting true community identification (goal 2). Brazil limits the cost of subprojects by type. Colombia, until recently, rigidly earmarked funds by subproject type. And Mexico gives priority to subprojects meeting basic needs by requiring communities to pay a larger proportion of costs for subprojects providing amenities (for example, sports facilities) than other types. Because countries classify projects differently, it is impossible to know how the use of these incentives has affected the distribution of projects by type. Thus, we cannot be sure how extensively the imposition of external priorities has distorted local preferences. This study has identified the conflicts among objectives, and the advantages and disadvantages of various rules which mediate between them. Unfortunately, we have not been able to identify the optimal DRIF design. This is in part because data available on the three projects are insufficient: so far only limited efforts have been made to monitor and evaluate the subprojects developed under DRIFs, much less to monitor compliance with the rules and the 46 functioning of the institutions responsible for implementation.2' In addition, differences in multiple variables among DRIFs make it difficult to identify with certainty relationships between subproject rules and outcomes (there are too many variables, and the sample contains only three projects). Still, some irnportant conclusions can be drawn: * Design DRIF rules to reflect local conditions. There is no one optimal design suitable for DRIFs in all places at all times. The particular set of rules agreed for each DRIF, at each stage of its evolution, must result from negotiations among actors with different preferences and objectives - in different political, social, and economic contexts. * Consider offering communities with different capacities, different fmancing packages. Within the same country communities differ in their level of development, organization, and method of decisionmaking. Brazil has dealt with this by offering different programs to different communities depending on their capacities to implement subprojects. Communities with greater capacities take greater responsibility for implementing an annual development program, which FUMAC supports. Those with more limited capacities receive financing to complete a single subproject, with financing under PAC. * Define objectives, and the tradeoffs among them clearly. Specifying objectives and recognizing the tradeoffs among them can help all actors choose and implement projects more successfully. It is also important that DRIF designers understand which tools (or rules) to apply in particular contexts. Next Steps The first priority is to collect more data on subproject sustainability. Simultaneously, much more needs to be learned about how specific tools or rules actually function. Finally, analysts should condLuct. small studies to find answers to the following two questions: * How much responsibility for planning and priority setting should be assigned to communities, municipal governments, and municipal participatory committees? Do the social benefits generated by communities planning and executing subprojects outweigh the losses from reduced efficiency in project implementation and the costs that communities bear? Does involvement in project planning increase the likelihood that communities successfully fulfill responsibilities for operations and maintenance? Why are some municipal participatory committees effective while others are not? Under what circumstances should municipal participatory committees be formed and 21 There are notable exceptions, such as the studies by Angela Estebe and Fernando Aguillar respectively of the software and hardware aspects of project implementation under the Mexico Municipal Solidarity Fund. 47 given the power to allocate resources or implement projects, instead of municipal authorities? (Analysts would use selected case studies as data sources.) * What can be done to better ensure quality at entry and long-run sustainability of subprojects, short of conducting full appraisals? Are there simple eligibility criteria (or rules of thumb) which can used to assess subprojects to maximize the probability that subprojects will be viable and sustainable, especially productive projects? Can the use of standardized designs reduce costs, improve quality, and compensate for weak technical capacities of poor, rural communities, without excessively biasing choice? Can designs be created that work well in a variety of geographical settings? Can subprojects be designed to efficiently utilize local labor and materials? What are the advantages and disadvantages in terms of project feasibility, economic viability, and sustainability, of community's implementing projects themselves versus implementing them through contractors? What are the conditions under which self-management can or cannot succeed? (Analysts would collect data from the following sources: designs currently in use; ex post quantitative evaluation data for productive and other projects; subproject completion reports representing a variety of management arrangements; and site interviews, leading to guidelines for operational manuals). 48 ANNEX 1 TARGETING This annex presents the formulas used to allocate funds, and the detailed results of the statistical analysis done for the Northeast Rural Alleviation Program in Brazil (after the 1993 reformulation): Fondo DRI in Colombia, and the Municipal Solidarity Fund, active in four states of Mexico. The primary objective of the analysis was to assess how successful countries have been in targeting poverty. To make this assessment, we measured the correlation between investment per person and poverty levels in all municipalities where the DRIFs were active. Formulas Mexico: Municipal Solidarity Fund In Mexico, authorities used a formula to allocate resources to the municipalities within each state: no clear criteria were used to allocate resources among states. Each state used different formulas, developed through simulations and extensive negotiations. States informed each municipality about fund availability at the beginning of the year. The formulas used in the four states of the First Decentralization and Regional Development Project (Loan 3310) follow. Chiapas. The formula used in Chiapas weights "marginality" heavily, an indicator of poverty based on distance from sources of services. It contains four elements, each weighted differently: 70 percent to marginality, 10 percent to total population, 10 percent to the number of localities within the municipality, and 10 percent to indigenous population (measured as municipal proportion of total state indigenous population). The number of localities is correlated with the number of communities, and is an indicator of municipal dispersion. The result is the following equity formula: FE= TFx CFMgM x PM x07+PMx 01+ LM x0.1+PIM x0-1) PPMgM PT TL PIT Where: FE = yearly municipal allocation TF = yearly financial resources available in the state FMgM = municipal marginality factor PM = municipal population PPMgM = weighted population PT = total state population LM = number of localities in the municipality TL = total number of localities in the state 49 PIM = indigenous population in the municipality, and PIT = total indigenous population in the state. Guerrero. The formula used in Guerrero is based on a municipal classification derived from the urban-rural population ratio. "A" municipalities have fewer than 5,000 inhabitants in the municipal urban core (cabecera); "B1" municipalities have more than 5,000 inhabitants in the cabecera, but less than one-third of the total municipal population; and "B2" municipalities have more than 5,000 inhabitants in the cabecera, and more than one-third of the total municipal population. It is interesting to note that 15 percent of funds have been left for "state priority," or allocation according to the governor's discretion. FE-T~FrF Pm PM PM1 FE = TF x 0(x 0.45 + -x 0.25 + -x 0.1 5) + PE x 0.l S5 PA + PB1 PA PBI j Where: FE = yearly municipal allocation TF = yearly financial resources available in the state PM = municipal population PA = population of A municipalities P"B I" Population of B 1 municipalities, and PE = state priority (15 percent of total funds). Hidalgo. The formula used in Hidalgo is based on marginality. An iterative process is used to rank localities and municipalities. The result is a marginality factor which is classified in eight ranks. F=3 xLl xPl +2xL2xP2+L3 xP3 Where: F = marginality factor L1, L2, L3 = number of localities with very high, high, and medium marginality Pl, P2, P3 = population of localities with very high, high, and medium marginality. (The limits between very high, high, and medium are not given.) 1. Oaxaca. The formula used in Oaxaca weights most of the criteria. It is interesting to note that, in this unique case, 30 percent of funds have been allocated equally among municipalities. (1I PM LM PEM FE = TFx 0-x.30+ x 0.25+-x 0.l0+ x 0.35) x 0.80 + PE x 0.20] L(NM PT TL TPEM Where: FE = yearly municipal allocation 50 TF = yearly financial resources available in the state NM = number of municipalities in the state PM = municipal population PT = total state population LM = number of localities in the municipality TL = total number of localities in the state PEM = marginality weight (very high = 5, high =4, medium 3, low =2, very low = 1); TPEM = state marginality weight, and PE = state priority (expressed as percentage). Colombia: Fondo DIU In Colombia, a mathematical formula was used to allocate resources among the country's 32 departments, using the following criteria: rural population: 30 percent; poverty level (percentage of population with unsatisfied basic needs): 50 percent; and per capita budget transfers from central government: 20 percent (the higher the transfer, the lower the allocation). Every department was assigned a ranking from I to 10: the authorities compared this number to the countrywide total to calculate the percentage each department would receive. FD = TF x KP x 0.30 +NBID x 0-10 + B x 0.35 PT ~ NBJT PBTDxO3 PBTT Where: FD = yearly departmental allocation TF = yearly financial resources available in the country PD = departmental population PT = total country population NBID = percentage of departmental population with unsatisfied basic needs NBIT = percentage of national population with unsatisfied basic needs PBTD = per capita departmental budgetary transfers, and PBTT = national per capita budgetary transfers. Brazil: PACIFUMA4C Formulas were not used to allocate funds under PAC, which directly targeted communities. However the pilot FUMAC component of the new program (phase 3) will use allocation criteria. The formula for the states of Bahia, Ceara and Sergipe will comprise the following elements: municipal size (both population and area), ratio of rural to total population, poverty rate, number and degree of dispersion of poor rural communities, availability of public resources per capita, number of approved subprojects (small discriminatory factor), performance 51 of the municipality over the previous year, and willingness of the mayor to participate in the program. Statistical Analysis The basic mathematics of targeting are as follows: Let lij = investment subsidy of ith individual in jr municipality Yij = income of ith individual p = poverty line (defined in terms of income). Then strict targeting of the poor would require that the government provided a subsidy to each individual in accordance with the following equation: (1) Iij = cc + ,8 (p - Yij) if Yij < p, and Ijj = O if Yij 2 p where cc 0 and B Ž 0 but a = ,B =0 is impossible. Note that: If a > 0 and ,B 0, subsidies of equal amounts go to all people below the poverty line, regardless of the degree of poverty; * If a = 0 and , > 0, subsidies go in strict proportion to the degree of poverty; * If a = 0 and 3 = 1, subsidies equal the poverty gap; and , > 1 is possible, indeed necessary, if the subsidy is invested to bring the incomes of the poor sustainably to the poverty line. The formulas can be aggregated over the entire population of a municipality and manipulated to yield an equivalent single formula at the municipal level. Letting np = the number of poor people in municipality j, and n = the total population of the municipality, then: (2) Zj,ij/n = [a + p (p - YnPYij/np)] (np/n) This says that investment subsidies per person for municipalities equals the product of (a) an expression which inter alia compares the average income of the poor in a particular municipality with the poverty line (this drops out if f3 = 0 (the subsidy level is intended to be the same for all of the poor), and (b) the proportion of poor in a municipality. Although data on the per capita investment subsidy and the proportion of the poor may be available for municipalities, the average per capita income of the poor in a locality is not. It may be hypothesized that the poor living in regions with a high proportion of poor people are likely to be poorer than those in other regions. If this relationship holds reliably, then it is likely that the two are strongly negatively correlated, and that the proportion of people living in poverty could be used as a proxy for local incomes of the poor. 52 The loosest test for assessing whether investment subsidies are well-targeted is the value of the correlation coefficient between per capita subsidy level and the proportion of the poor. One could further estimate a quadratic regression (with no intercept tenn) and examine the sign and significance of the coefficients. If the target subsidy level is the same for all of the poor, only the first coefficient should be significant. In practice, targeting efforts may qualify use of such a formula with the argument that, in areas where there are few poor, extending subsidies might lead to Type II errors (the nonpoor receive an unintended subsidies). Disqualifying municipalities where the percentage of poor falls below a defined cutoff level from the program, or limiting the program to areas where the percentage of the poor is high (Brazil uses both practices) are legitimate ways of trading off the risks that Type I errors occur as opposed to Type II errors, but they also reduce the fit of the equation to observed data (it becomes a step rather than a monotonic function). The analysis is complicated by problems in comparing poverty indicators and target populations, which vary greatly across countries. To reduce the impact of incompatibilities on the analysis, relative rather than absolute measures of poverty level were used, and investments per person in the total population were calculated (estimating the number of rural or poor people was problematic because countries use varying definitions of "rural"). Brazil uses the percentage of "indigent-families" living in the municipality as its poverty indicator, provided by the Istituto Pesquiza Economica Aplicada, from the Mapa da Fome. Colombia uses the percentage of population of the municipality with "unsatisfied basic needs" (such as education, health services, water supply, sewerage, and housing) - which is imperfectly correlated with income - as its measure of poverty.. Mexico uses the "marginality index" as an indicator of the level of marginality of the municipality, ranking municipalities on a scale of -2 to 3, (-2 means minimal marginality, and +3 means maximum marginality). Included in the analysis were 1,089 of 1,500 municipalities in Brazil (data on Bahia were not available), 900 of 1,050 municipalities in Colombia (the 150 larger cities were not eligible for DRI financing in 1994), and all 840 municipalities of the four Mexican states (Chiapas, Guerrero, Hidalgo, and Oaxaca). The investments were allocated (although not necessarily disbursed) during 1994 for Colombia's Fondo DRI and Mexico's Municipal Solidarity Fund, and during 1993-95 (from the reformulation to the October 1995) for Brazil's Northeast Rural Poverty Alleviation Program. Investments per capita, in United States dollars, were calculated on the basis of total population. The correlations resulting from the analysis are: +6.8 percent for Northeast Rural Poverty Alleviation Program in Brazil, +18.4 percent for Fondo DRI in Colombia, and -4.5 percent for the Municipal Solidarity Fund in Mexico. Note that the Northeast Rural Poverty Alleviation Program did not use a formula to allocate resources among municipalities; Fondo DRI did use a formula to allocate resources among 32 departments; and Mexico used different formulas to allocate resources within the four states, but did not use any formula to allocate resources among states. These results are shown in a scatter graphs in Figure Al.1 for the country programs, and in Figure 5 for the four Mexican states individually. 53 Similar data have sometimes been presented in graphic form, with the average investment per capita among all municipalities for a given poverty range represented by bars. However, presenting the data contained in the scatter diagrams below in bar graphs would convey a totally different impression. The bar graphs are deceptive, because they hide the variance of subsidies (which represent Type I or Type II errors) within poverty ranges, and give equal weight to subgroups regardless of size. In addition, analysts can easily manipulate the impressions the data provide by changing the grouping ranges. 54 Figure A1.1 Investment per capita versus poverty Brazil: Northeast Poverty Alleviation Program PAC/FUMAC investments in eight states (except Bahia), 1993-95 6 0 _------ + ---------- +, + + . J + + + + O- 11111 + _ 20 A ~~44"+ 4 + 15 > 15 <25 < 59 Productive > 20 0 <20 < 59 Social >10 > 10 <30 < 59 Leisure >15 > 15 < 25 < 59 Source: Project operational manual Mexico: Municipal Solidarity Fund a) There should be one Solidarity Committee per subproject and one Municipal Development Council per Municipality. (Chicken-egg problem: before or after the subproject application has been submitted?) b) The total cost of the subproject must be less than US$45,000. (The purpose of this criterion is to keep the subprojects small in scale, so they can be managed locally.) c) Beneficiaries contribute a minimum 20 percent of project costs. (In order to assure beneficiary participation). They contribute 30 percent for road paving project costs, 40 percent of costs of public gardens, parks, civic squares, sports facilities, and social centers (casas ejidales), and nothing for technical assistance. d) There should be no land tenure problems. (In order to avoid future ownership problems.) 58 e) The implementation period is not to exceed one year. (Similar to the criterion (b), this is to assure that implementation is simple. This criterion is also a consequence of the temporary status of the solidarity fund, which may not exist infuture years.) f) Subprojects are not to be divided in stages. (In order to avoid sneaking around rules (b) and (e).) g) Operational staff for classrooms and health centers must be assured. (In order to assure good operation) h) There should be no subutilization of the existing infrastructure. (In order to assure utilization of the subproject.) i) The subproject must have significant social impact. (It is not clear how this would be measured.) j) Unitary costs must not be higher than the regional average. (This criterion is defined incorrectly: the cost cannot be always inferior to the average, because of the average concept. Instead the criterion should be defined as average plus a certain percentage. However, in reality, is quite difficult to measure the average. In practice to limit costs, beneficiaries are allowed to use the savings from the planned budget.) Rules for specific subprojects: k) Road paving: permitted only where water supply and sewerage already exist. Project cost must not exceed N$300,000 (US$89,000) per municipality, and 25 percent of total municipal allocation. Colombia: Fondo DRI The eligibility criteria of the PDIC (Programa de Desarrollo Integral Campesino) of the Fondo DRI (Rural Integrated Development Fund) in Colombia are as follows: a) Minimum beneficiary contribution. A cofinancing matrix defines the share of costs among Fondo DRI, departmental, and municipal government. Cost shares vary depending type of subproject and development level of the beneficiary municipality. Direct beneficiaries do not participate in subproject costs. b) Community participation. All subprojects must originate in the beneficiary community, and their designs should encourage community participation in all subproject phases. 59 Table A2.2 Fondo DRI cofinancing matrix Municipal Categorya C1 C2 C3 C4 C5 C6 C7 C8 (Maximum Fondo DRI cofinancing percentageb) Basic extension 70 70 70 60 60 60 50 50 Specific extension 90 90 85 80 80 70 50 50 Agricultural mechanization 60 60 50 45 45 35 15 15 Irrigationc 90 90 85 80 80 75 40 40 Fisheries technical assistance 90 90 85 80 80 70 50 50 Fisheries infrastructure 70 70 65 60 60 55 30 30 Rural marketing infrastructure 70 70 65 60 60 55 30 30 Rural marketing technical assistance 90 90 85 80 80 70 50 50 Rural roads 60 60 50 45 45 35 15 15 Bridges 60 50 45 35 15 15 15 10 Water supply, sewerage 60 60 50 45 45 35 15 15 Rural electrification 65 65 60 55 55 50 35 35 Watershed management technical assistance 90 90 85 80 80 70 50 50 Watershed management infrastructure 70 70 65 60 60 55 30 30 Environmental impact mitigation 90 90 85 80 80 70 45 45 Flood control 60 60 60 60 55 55 45 45 Institutional development 90 90 80 70 70 60 45 45 Rural women development 90 90 80 70 70 60 40 40 a/ Categories of municipalities: Cl are less developed municipalities, and C8 more developed ones (established by decree 2132 of 1992.) b/ Municipal governments or the final beneficiaries finance the remaining project costs. c/ Irrigation is regulated by law 41 of 1993. Source: Project operational manual c) Type of subproject. The subproject should pertain to one of the following 11 types: (a) institutional development, (b) agricultural extension, (c) watershed management, 60 (d) fisheries, (e) rural marketing, (f) rural roads, (g) sewerage and water supply, (h) rural electrification, (i) irrigation, (j) flood control, (k) agricultural mechanization. d) Technicalfeasibility, which is specified for every kind of subproject. e) Economic andfinancialfeasibility and sustainability. f) Environmentalfeasibility. An enviromnental analysis is necessary for all subprojects. When a subproject presents a negative environmental impact, mitigation measures should be guaranteed. g) Gender analysis. This analysis is necessary in all subprojects, which should foster women participation. h) Existence of an executing agency, with satisfactory levels of technical, administrative, legal, financial, environmental, gender, and participation performances/records. i) Ineligible expenditures: debt cancellation, land and shares purchase, duties and taxes. Bolivia The eligibility criteria of the Investment Program for Campesino Development (PIDC, within Campesino Development Fund, FDC) in Bolivia are: a) Target groups. The subproject should benefit specific target groups, defined as rural inhabitants associated in productive groups with limited access to resources and services, adopting traditional technologies for agricultural activities. b) Integration of women. The integration of women has to be demonstrated through a gender analysis, which is necessary for subproject approval. c) Priority regions. The subproject has to be located in a priority region of the country. The priority regions are defined on the basis of poverty level, availability of natural resources, and institutional development. d) Manageability and implementation accessible to beneficiaries. Subproject implementation and operation are the responsibilities of the beneficiaries, who may subcontract some or all subproject implementation, but who will remain uniquely responsible for managing the subproject. e) Technical feasibility should be demonstrated for subproject implementation, operation, and maintenance. 61 f) Economic feasibility. All subprojects should demonstrate a minimal internal rate of return of 12 percent. g) Sustainability and multiplier effect. Sustainablity is a consequence of economic feasibility. The multiplier effect refers to the possibility of the activity to be replicated. h) Environmental feasibility. An environmental analysis is necessary for all subprojects. i) Existence of an executing agency, which is legally responsible for the subproject. Ideally, the same beneficiary community should be the executing agency, but rural communities rarely have legal status. Table A2.3 Fondo Desarrollo Campesino cofinancing matrix Type of subproject Percent direct beneficiaries Percent municipal governmenta Irrigation 20 5 Rural road - 5 15 Bridge 5 15 Pedestrian bridge 10 10 Rural market 5 20 Storing facility 15 5 Workshop 15 5 Erosion control 15 5 Atajados 20 5 Agriculture extension 10 0 Training, extension 5 0 Revolving fund 30 0 Watershed management. 10 0 Agroforestry 20 0 Institutional strengthening 10 0 Pre-investment 0 10 a/ The central government finances the remaining project costs through a credit of the International Development Association of the World Bank Group. Source: Project operational manual j) Implementation period not exceeding one year. (This criterion is meant to assure simple implementation. It is also a consequence of the temporary status of the 62 Fondo Desarrollo Campesino, which may not exist in future years.) This rule has exceptions for special cases (extension). k) Subproject type. The projects supports investment in: (i) basic infrastructure (irrigation, rural roads, bridges, rural markets, storing facilities, workshops, defensivos, atajados); (ii) production support (agricultural extension, training of extension agents, revolving fund, watershed management, agroforestry); (iii) institutional strengthening (to executing agencies, to beneficiary communities, preinvestment). 1) Minimum beneficiary contribution. This varies depending the subproject type, as detailed in Table A2.3. 63 ANNEX 3 PROCUREMENT AND DISBURSEMENT The following tables provide information on procurement and disbursement thresholds to allow easy comparison across programs of the thresholds. These tables provide information only on the project components which finance subprojects (for Mexico, only the Municipal Solidarity Funds provide such financing), not components financing institutional assistance to project management units . Since no subproject can cost more than US$50,000 in Mexico and Brazil, the only procurement methods used are direct contracting in Mexico and local shopping in Brazil (Table A3.1). The situation in Colombia is more complex, since subprojects are not subject to cost limits. These considerations apply to the prior review by the World Bank of procurement decisions (Table A3.2) and disbursement documentation (Table A3.3), whose thresholds are always greater than the cost of a subproject. Therefore, as prior review is not , the World Bank will review procurement decisions only ex post; the only documentation of expenditure will be the statement of expenditure. Table A3.1 Procurement methods (US$) Direct contracting Shopping Noncompetitive bidding Brazil PAC/FUMAC Works/goods for grants (by associations) (by public entities) (by public entities) <40,000 (phase two) < 100,000 > 100,00 < 50,000 (phase three) Colombia Fondo DRI Civil works < 50,000 50,000-1,000,000 Goods < 25,000 25,000-200,000 Mexico Municipal Solidarity Fund Civil works < 50,000 n.a. n.a. Goods < 100,000 100,000-350,000 n.a. means not available. Source: Project documents 64 Table A3.2 World Bank prior review of procurement decisions (US$) Brazil PA(C/FUMAC Goods and works World Bank guidelines Consulting services from firms > 100,00 Consulting services from individuals > 50,000 Colombia Fondo DRI Civil works > 500,00 Goods > 200,00 Consulting services > 0 Mexico Municipal Solidarity Fund Goods and works > 200,00 Consulting services from firms > 100,00 Consulting services from individuals > 50,000 Source: Project documents Table A3.3 Documentation of expenditures (US$) Full documentation Statement of expenditures Brazil PAC/FUMAC Goods and works > 100,00 < 100,00 Consulting services from firms > 100,00 < 100,00 Consulting services from individuals > 50,000 < 50,000 Colombia Fondo DRI Equipment > 200,00 < 200,00 Civil works > 500,00 < 500,00 Mexico Municipal Solidarity Fund Goods and works > 350,00 < 350,00 Consulting services from firms > 100,00 < 100,00 Consulting services from individuals > 50,000 < 50,000 Source: Project documents The Brazil project is the only one using the denomination "grants" as a disbursement category (Table A3.4). By doing so, the possibility of differentiating the World Bank percentage of cofinancing of different categories has been lost, but the almost impossible estimation of works, goods, and technical assistance has been avoided. 65 Table A3.4 Withdrawal of the proceeds of the loans Disbursement category Percentage of expenditures to be financed Brazil PAC/FUMAC Grants for subprojects 59 Consulting services for institutional development and project administration 100 Incremental administrative costs 20 Colombia Fondo DRI Works, goods, and consulting services for subprojects 85 Works, goods, consulting services, and 70 percent of local expenditures, operating costs for institutional strengthening 100 percent of foreign expenditures Mexico Municipal Solidarity Fund Works and construction materials for MSF subprojects 50 Goods (other than construction materials) for MSF subprojects 90 Technical assistance and training for MSF subprojects 100 Goods for institutional strengtiening 90 Long-term consulting services for institutional strengthening 50 Technical assistance and training for institutional strengthening 100 Source: Project documents 66 OP 12.20-Annex B May 1994 Subaccounts and Second-Generation Special Accounts 1. Under a special account controlled by a central agency, subaccounts may be established within the same depository World Bank and operated by agencies that are implementing the project. Although funds in the special account are not transferred from the special account, internal accounting records monitor disbursements against amounts allocated to each of the subaccounts. This system may facilitate access to special account funds by project implementation agencies. 2. In exceptional circumstances, when arrangements such as separate subaccounts are not feasible, the World Bank may authorize the borrower to withdraw funds from an subaccounts and use them as an advance into a second-generation special account. The World Bank must be satisfied that projects requiring a second-generation special account meet the following criteria: (a) there is a clear need for such an account - for example, there may be logistical problems involving the transfer of funds from the central subaccounts to distant rural areas where project entities are located.22 (b) the borrower is unable to provide project entities with adequate working capital to prefmance the World Bank's share of expenditures. (c) the borrower has the necessary authority to enforce accounting and auditing controls over the second-generation special account and the administrative capacity to manage the financial aspects of the project. 3. In addition, when conversion of subaccount funds into local currency is anticipated, adequate arrangements must be in place to ensure that the value of the subaccount is maintained. 22 However, a government's unwillingness to change procedures that deny prompt access to Special Accounts by a project entity does not justify the use of Second Generation Special Accounts. 67 BIBLIOGRAPHY Alcide do Nascimento. 1994. Sistema Municipal de Governo. Projecto Aridas. Instituto Internacional para la Cooperaci6n en Agricultura. Teresina, Brazil Banks, Arthur (ed). 1995. Political Handbook of the World 1994-1995. CSA Publications, State University of New York, Binghamton, New York. Ball, Gwen. 1994. "Institutional and Organizational Development in Decentralized Projects." Available from Maurizio Guadagni, Latin America and Caribbean Region, Environmentally and Socially Sustainable Development (LCSES), World Bank, Washington, D.C. Bernal, Fernado (ed.). 1994. "Programa Nacional de Rehabilitaci6n (PNR): 11 Anios de Camino. La Sociedad Civil en Marcha." Presidencia de la Republica. Santafe de Bogota, D.C., Colombia. Besley, Timothy, and Ravi Kanbur. 1990. "The Principles of Targeting." Working Paper 385. World Bank, Washington, D.C. Cernea, Michael. 1983. "A Social Methodology for Community Participation in Local Investments. The Experience of Mexico's PIDER Program." Working Paper 598. World Bank, Washington, D.C. 1992. The Building Blocks of Participation: Testing Bottom-up Planning. World Bank Discussion Paper 166. Washington, D.C. Cornelius, Wayne, Ann Craig and Jonathan Fox (ed.). 1994. Transforming State-Society Relations in Mexico: The National Solidarity Strategy. San Diego: University of California. Croich, Colin and David Marquand. 1995. Reinventing Collective Actions. From the Global to the Local. Cambridge, Mass.: Blackwell Publishers. Ferreira, Ana Maria, and Luis Carlos Valenzuela. 1993. Descentralizacion Fiscal: El Caso Colombiano. Naciones Unidas, Comisi6n Econ6mica para America Latina y el Caribe. Santiago de Chile. Fox, Jonathan, and Josefina Aranda. 1994. "Community Participation in Mexico's Municipal Funds Program: The Case of Oaxaca." Paper presented at the Workshop on Participatory Development, May 17-20, World Bank, Washington, D.C. Fox, Jonathan. 1995. "Governance and Rural Development in Mexico: State Intervention and Public Accountability." Journal of Development Studies 32: 1-30. 68 Fuhr, Harald. 1994. "Municipal Institutional Strengthening and Donor Coordination: The Case of Ecuador." Public Administration and Development 14: 169-86. Gelbach, Jonah, and Lant Pritchett. 1995. "Does More for the Poor Mean Less for the Poor?: The Politics of Tagging. " Policy Research Working Paper 1523. World Bank, Washington, D.C. Gopal, Gita. 1995 Procurement and Disbursement Manual for Projects with Community Participation. World Bank Discussion Paper 312. Washington, D.C. Healy, Kevin. 1984. Caciques y Patrones: Una Experiencia de Desarrollo Rural en el Sud de Bolivia. Cochabamba, Bolivia: Ceres. Kottad, Conrad, Al-berto Costa, and Rosane Prado. 1994. "A study of Popular Participation in Brazil: Northeast Rural Development Program." Paper presented at the Workshop on Participatory Development, May 17-20, World Bank, Washington, D.C. Marc, Alexandre, aind Gita Gopal. 1994. "Procurement and Disbursement." Paper presented at the Workshop on Participatory Development, May 17-20, World Bank, Washington, D.C. MacDonald, Charles, and Peter Sollis. 1994. "Mexico: Fondos Municipales de Solidaridad." Available from Maurizio Guadagni, Latin America and Caribbean Region, Environmentally and Socially Sustainable Development (LCSES), World Bank, Washington, D.C. Parker, Andrew. 1995. "Decentralization. The Way Forward for Rural Development?" Policy and Research Working Paper 1475. World Bank, Washington, D.C. Shah, Anwar. 1994. The Reform of Intergovernmental Fiscal Relations in Developing and Emerging Market Economies. Policy and Research Series 23. World Bank, Washington, D.C. Tendler, Judith. 1993. New Lessons from Old Projects: the Workings of Rural Development in Northeast Brazil. Operations Evaluation Department, World Bank, Washington, D.C. Tomic, Blas. 1984. Examen Critico de Tres Intentos Oficiales de Planificaci6n Descentralizada y Participativa. World Labor Organization. Santiago, Chile. World Bank. 1995. "Colombia - Local Government Capacity: Beyond Technical Assistance." Report 14085-CO. Latin America and Caribbean Region, Country Department 3, Country Operations Division 1, World Bank, Washington, D.C. van Zyl, Johan, Tulio Barbosa, Andrew Parker, and Loretta Sonn. 1995. "Decentralized Rural Development and Enhanced Community Participation. A Case Study from Northeast Brazil." Policy and Research Working Paper 1498. World Bank, Washington, D.C. 69 Distributors of COLOMBIA GERMANY IRELAND MEXICO PHILIPPINES Mundi-Prensa Barcelona lnfoenlace Ltda. UNO-VeYag Govemment Supplies Agency INFOTEC Intemational Booksource Cenler Inc. Consell de Cent, 391 W orld Bank Carrera 6 No. 51-21 Poppelsdorler Allee 55 Oig an tSoldthair Av. San Femando No.37 1 127-A Anlipolo SI, Barangay, Venezuela 08009 Barcelona Apartado Aereo 34270 53115 Bonn 4-5 Harcourt Road Col. Toiello Guerra Makali Cily Tel: (34 3) 488-3492 Publications Santahi de Bogotd, D.C. Tel: (49 228) 949020 Dublin 2 ~~~~~14050 Mexico, D.F. Tel: (63 2)896 6501; 6505; 6507 Fax: (34 3) 487-7659 Prices and credit terrns varyi om Tel: (571) 285-2798 Fax: (49 228) 217492 Tel: (353t1)45-30 TF F e 2896 1 E mundiprensa.es cutcountr outy oal u Fan: (57 1) 285-2798 UL: illp:lowA.unovo-nrlag.dlo Fan: (353 1) 475-2670 Fan: (52 5) 624-2822 PLN R AK,TEMLIE loa ditribountory. placing anrEml noea~oo E-mail: infoolc@rln.noe.mxsOADSILNK,TEMLIE local distrbutor before placmg an COTE D'IVOIRE ISRAEL URL: httpt/rln.nel.mx Intemational Publishing Service Lake House Bookshop order. CenterdEdilion el deillousion Alricaines GHANA YozmBot LMerature LId. Ul. Piek0na 31/37 100, Sir Champalam Gardiner Mawatha (CEDA) Epp Banks Services P.0. Ban 56055 Mundl-Preona Mexico S.A. de C.V OD-677 Warzawa Colombo 2 ARGENTINA 04 B.P 541 PO. Bax 44 3 Yohanan Hasandlar SIreet cRio Panco. 141-Colonia Cuauhtemnoc Tel: (482) 628-089 Tel: (94 1)32105 Oiicina del Libro Intemacional Abidjan 04 TUC Tel Aviv 61560 06500 Mexico, D.E Fax: (48 2) 621-7255 Fax: (94 1) 432104 Av. Cordoba 1877 Tel: (225)246510;246511 Accra Tel: (9723) 5285-397 Tel: (525) 533-5658 E-nail: books%ips@ih,.atmcor.pI E-mail: LHLOsfi.lanka.net 1120 Buenos Aires Fax: (225)250567 Fax: (9723)5285-397 Fax: (525)514-6799 URL: hIpJ/Awm.ipscg.wawptipstexport Tel: (541) 815-8354 GREECE PORTUGAL SWEDEN Fax: (541)8154156 CYPRUS Papasotiriou S.A. R.O.Y lnrematlonal NEPAL Livrada Podugal Wennergren-Williams AB E-mail: oliltb@sallink.com Cenler for Applied Research 35, Sloumara Str. PO Box 13056 Evermst Media Inlemalinaal Services (P) Ltd. Apartado 2681, Rua Do Casno 70-74 P 0. Box 1305 Cyprns College 106 82 Athens Tel Aviv61130 GPO Box5443 12M Lisbon S-171 25 Solna AUSTRALIA, FIJI, PAPUA NEW GUINEA, 6, Diogenes Street, Engomi Tel: (30 1)364-1826 Tel: (9723)0461423 Kathmandu Tel: (1)3474982 Tel: (468)709-97-50 SOLOMON ISLANDS, VANUATU, AND P.6. Box 2006 Fax: (301) 304-8254 Fax: (9723)5461442 Tel: (9771)472152 Fax: (1)347-0264 Fax: (46 8)27-00-71 WESTERN SAMOA Nicosia E-mail: rnyil@nelvision.net.il Fax: (977 1) 224 431 E-mail: mail@wwi.se D.A. Infomiation Services Tel: (357 2) 44-1730 HAM ROMANIA 648 Whitehorse Road Fax: (357 2) 46-2051 Cukure Diffusion Palestinian Authority/Middle East NETHERLANDS Compani De Libradi Bucuresti S.A. SWITZERLAND Mitcham 3132 5, Rue Capos Iodex Information Services De LindebonmtlnOr-Publ0m5tes Str. Lipscani no. 26, sector 3 Libraire Payot Service lnstliulionnel Victoria CZECH REPUBLIC C.P 257 PO.B.19502 Jeusalem P.O. Box 202,7480 AE Haaksbergen Bucharest Codes-de-Montbenon 30 Tel: (61) 3 9210 7777 Nalional InolomatiOn Cenler Pod-au-Prince Tel: (972 2) 6271219 Tel: (31 53) 574-0004 Tel: (401) 613 9645 1002 Lausanne Fax: (61)392107785 prodejna, Konviklaska 5 Tel: (509)239260 Fax: (9722)6271634 Fax: (31 53) 572-9296 Fax: (401)312 4000 Tel: (41 21)341-3229 E-mail: service@dadirecl.com.au CS-11357 Pmague 1 Fax: (509)234858 E-mail: llndeboo@woddonline.nl Fax: (4121) 341-3235 URL: htpl/www ,dadirecl.com.au Tel: (42 2) 2422-9433 ITALY URL: haptiwwww .woddonline.nil-lindeboo RUSSIAN FEDERATION Fax: (42 2) 2422-1484 HONG KONG, MACAO Licosa Commissionaria Sansoni SPA Isdatelstvo ADECO Van Diexnen EddionsTechniques AUSTRIA URL: hffp://www .nis.czt Asia 2000 LId. Via Duca Di Calabda, 1/1 NEW ZEALAND 9a, Kolpachniy Pereuok Ch. de Lacuez 41 Gerold and Co. Sales & Circulation Deparment Casella Poslale 552 EBSCO NZ LId. Moscow 101831 CH1807 Blonay Weihburggasse26 DENMARK SeabirdnHouse, unit1101-02 50125 Finnze Privale Mail Bag 99914 Tel: (7095) 917 87 49 Tel: (41 21)943 2673 A-1011 W0en SamfundsLMteratur 22-28 Wyndham Stmeet, Conral Tel: (55) 645415 New Markel Fax: (7 095) 917 92 59 Fax: (41 21) 943 3605 Tel: (43 1) 512-47-314- Rosenoems Alls 11 Hong Kong Fax: (55) 641-257 Auckland Fax: (431) 512-47-31-29 DK-1970 Frederiksberg C Tel: (852) 2530-1409 E-mail: licosa@flbcc.il Tel: (649)524-8119 SINGAPORE, TAfWAN, THAILAND URL: hltp://www.garvId.co/al.onlinx Tel: (45 31) 351942 Fax: (852) 2526-1107 URL: hOp/wwwwfebhcc.llicosa Fax: (649)524-8067 MYANMAR,BRUNEI Cental Baoks Distnbution Fax: (45 31) 357822 E-mail: sales@asia2000.com.hk Ashgate Publishing Asia Paciic Pte. Ltd. 306 Silom Road BANGLADESH URL: htpi/wwwasl.cbs.dk URL: htipi/waww .asia2000.com.hk JAMAICA NIGERIA 41 KalLang Pudding Road #04-03 Bangkok 10500 Micro Industries Development Ian Randle Publishers LId. University Pness LUmfed Golden Wheel Building Tel: (66 2) 235-5400 Assislance Sociely (MIDAS) ECUADOR HUNGARY 206 Old Hope Road, Kingslon 6 Three Crowns Building Jericho Singapore 349316 Fax: (66 2) 237-6321 House 5, Road 16 Libd Mundi Euro nfo Service Tel:876-927-2085 Privale Mail Bag5095 Tel: (65) 741-5166 Dhanniondi RtArea Libreda Inlemacional Margilszgeli Europa Haz Fax: 876-977-0243 lbadan Fax: (65) 742-9356 TRINIDAD & TOBAGO Dhaka 1209 PO. Box 17-01-3029 H-1 138 Budapesl E-mail: irpl@colis.com Tel: (23422)41-1356 E-mail: ashgale@asiauxonnect.com AND THE CARRIBBEAN Tel: (860 2) 326427 Juan Leon Mora 851 Tal: (361) 350 8024, 350 80 25 Fax: (234 22) 41-2056 Systematics Studies LId. Fax: (850 2) 811188 Quho Fax: (361) 350 90 32 JAPAN SLOVENIA St. Augusline Shopping Cenoer Tel: (5932)521-606; (5932)544-185 E-mail: euroinfo@mail,malavhu Easlem Book Senvice NORWAY Gospodarski Vestnik Publishing Group Eastem Main Road, St. Augusline BELGIUM Fax: (593 2) 504-209 3-13 Hongo 3-chome, Bunkyo-ku NIC lnfo AS Dunajska ceslaS 5 Trinidad & Tobago, Wosl Indies JeanDe Lannoy E-mail: libfmuol@librimundi.onm.ec INDIA Tokyo 113 BookDepadmenl, Postboks 6512 Eoerlad 1000 ubljana Tel: (868)645-8466 Av. du Roi 202 E-mail: libnrmu2@librimundi.com.ec Allied Publishers Lid. Tel: (81 3)3818-0861 N-0606 Oslo Tel: (386 61)1338347;1321230 Fax: (868) 845-8467 1060 Brnussels 751 Mounl Road Fax: (81 3)3018-0864 Tel: (4722)97-4500 Fax: (38661)1338030 E-mail: lobe@Itdnidad.nel Tel: (32 2) 538-5169 EGYPT, ARAB REPUBLIC OF Madras - 600 002 E-mail: onrders@svt-ebs.co.jp Fax: (47 22) 974545 E-mail: repansekj@gvestnik.si Fax: (32 2) 538-0841 Al Ahram Distibutlin Agency Tel: (91 44) 852-3930 URL: hfp//wwwa bekkoame.orjp/-svi-ebs UGANDA Al Galaa Streel Fax: (91 44) 852-0649 PAKISTAN SOUTH AFRICA, BOTSWANA Gustm Ltd. BRAZIL Caiw KENYA Mirza Book Agency Forsingle tliles: PO Bax 9997, Madhvani Building Publicacoes Tocnicas Intemacionais Ltda. Tel: (202)578-6083 INDONESIA Aflca Baok Service (E.A.) Lid. 65, Shahrah-e-Quaid-e-Azann Oxford Univensy Press Soulhem Africa Plot 164 Jinja Rd. Rua Peixool Gomide, 209 Faxn (20 2) 578-6833 Pt. Indira Limded Quaran House, Mfangano Street Lahore 54000 Vasco Boulevard, Goodwood Kampala 01409 Sao Paulo, SP JalanBorobudur20 PO. Box 45245 Tel: (9242) 735 3601 P.O. Box 12119, Ni Cfty7463 Tel: (256 41) 251 467 Tel: (55 11) 259-6644 The Middle East Observer PO. Bax 181 Nairobi Fax: (92 42) 576 3714 Cape Town Fax: (256 41) 251 468 Fax: (5511) 258-6990 41, Sherd Streel Jakara 10320 Tel: (254 2) 223 641 Tel: (27 21) 595 4400 E-mail: gus@swmituganda.com E-mail: postmasler@pli.uol.br Cairo Tel: (62 21) 390-4290 Fax: (254 2) 330272 Oxford University Press Fax: (27 21) 595 4430 URL: hbpl/www .uol.br Tel: (20 2) 393-9732 Fax: (62 21) 3904289 5 Bangalre Town E-mail: oxford@oup.co.za UNITED KINGDOM Fax: (20 2) 393-9732 KOREA, REPUBLIC OF Sharae Faisal Microino Lid. CANADA IRAN Daepn Trading Co. LId. PO Bax 13033 Ebrsubscvnplion orders: PO. Bnx 3, Ahon, Hampshire GU34 2PG Renouf Publishing Co. LId. FINLAND Kelab Sara Co. Publishers PO. Box 34, Youida, 706 Seoun Bldg Katachi-75350 Inlemalional Subscription Service England 5369 Canolek Road Akateominen Kirjakauppa Khaled Eslamboli Ave., 6th Streel 44-6 Youido-Dong, Yeongchungpo-Ku Tel: (9221)446307 PO. Box 41095 Tel: (44 1420) 56848 Odawa, Ontaoio KiJ 9J3 PO. Box 128 Delafroz Alley No. 8 Seoul Fax: (92 21) 4547640 Craighall Fax: (441420) 89889 Tel: (613) 745-2665 FIN-00101 Helsinki P.O. Bax 15745-733 Tel: (82 2) 785-1631/4 E-mail: ouppak@ThoOffice.nel Johannesburg 2024 E-mail: wbank@ukminfo.demon.co.uk Fax: (613) 745-7660 Tel: (3580) 121 4418 Tehran 15117 Fax: (822) 784-0315 Tel: (2711)880-1446 URL: htopwAww.microinfo.co.uk E-mail: order.dept@anoufbooks.cnm Fax: (3580)121-4435 Tel (98 21) 8717819,8716104 Pak Book Corporalion Fux: (27 11) 880-6248 URL: htp:.// www.renoufbooks.com E-mail: akalilaus@slockmann,fi Fax: (98 21) 8712479 MALAYSIA Azki Chambers 21, Queen's Road E-mail: issis.co.za VENEZUELA CHINA ~ ~ ~ ~~ ~~~UR:lfflwwaaemnn.on -al:ktbsr~nd,o.rUnivemsity of Malaya Cooperalive Lahore Tocol-Cionciu Libros, S.A. CHINA URL: htipit .akaleeminen.comt E-mail: ketab-sara@neda.nel.ir Bookshop, LimpLed Tel: (92 42) 636 3222:636 0885 SPAIN Centlw Cuidad Comercial Tamanco China Financial & Eoonomic FRANCE Kowkab Publishers PO. Box 1127 Fax: (92 42) 636 2328 Mundi-Prensa Libros, S.A. Nivel C2, Caracas Publishing House Wodd Bank Publications PO. Bax 19575-511 Jalan Panlai Baru E-mail: pbcCbrain.nef.pk Castello 37 Tel: (58 2) 959 5547; 5035; 016 8, Da Fo Si Dong Jie 66, avenue dI1sna Tehran 59700 Kuala Lumpur 28001 Madrid Fax: (58 2) 959 5636 Bejing 75116 Pads Tel: (98 21) 258-3723 Tel: (60 3) 756-5000 PERU Tel: (341) 431-3399 Tel: (86 10) 6333-8257 Tel: (33 1) 40-69-30-56/57 Fax: (98 21) 258-3723 Fax: (603) 7554424 Ediorial Desarmllo SA Fax: (34 1)575-3998 ZAMBIA Fax: (86 10) 6401-7365 Fax: (331) 40-69-30-68 E-mail: urmkoop@fnm.nol.my Aparlado 3024, Uma 1 E-mail: libmeda@mundipmensa.es University Bookshop, Univershy of Zambia Tel: (51 14) 285300 URL: hfap:llwww .mundiprensa.es/ Great East Road Campus China Book Import Cenlre Fax: (51 14) 286628 PO. Box 32379 RO. Bux 2825 Lusaka BeOijing Tel: (260 1) 252 576 Fax: (2601) 253 952 RECENT WORLD BANK; TECHNICAL PAPERS (continued) No. 367 Colletta and Reinhold, Review of Early Childhood Policy and Programs in Sub-Saharan Africa No. 368 Pohl, Anderson, Claessens, and Djankov, Privatization and Restructuring in Central and Eastern Europe: Evidence and Policy Options No. 369 Costa-Pierce, From Farmers to Fishers: Developing Reservoir Aquaculturefor People Displaced by Dams No. 370 Dejene, Shishira, Yanda, and Johnsen, Land Degradation in Tanzania: Perceptionfrom the Village No. 371 Essama-Nssah, Analyse d'une repartition du niveau de vie No. 372 Cleaver and Schreiber, Inverser la spriale: Les interactions entre la population, l'agriculture et l'environnement en Afrique subsaharienne No. 373 Onursal and Gautam, Vehicular Air Pollution: Experiencesfrom Seven Latin American Urban Centers No. 374 Jones, Sector Investment Programs in Africa: Issues and Experiences No. 375 Francis, Milimo, Njobvo, and Tembo, Listening to Farmers: Participatory Assessment of Policy Reform in Zambia's Agriculture Sector No. 376 Tsunokawa and Hoban, Roads and the Environment: A Handbook No. 377 Walsh and Shah, Clean Fuels for Asia: Technical Options for Moving toward Unleaded Gasoline and Low-Sulfur Diesel No. 378 Shah and Nagpal, eds., Urban Air Quality Management Strategy in Asia: Kathmandu Valley Report No. 379 Shah and Nagpal, eds., Urban Air Quality Management Strategy in Asia: Jakarta Report No. 380 Shah and Nagpal, eds., Urban Air Quality Management Strategy in Asia: Metro Manila Report No. 381 Shah and Nagpal, eds., Urban Air Quality Management Strategy in Asia: Greater Mumbai Report No. 382 Barker, Tenenbaum, and Woolf, Governance and Regulation of Power Pools and System Operators: An International Comparison No. 383 Goldman, Ergas, Ralph, and Felker, Technology Institutions and Policies: Their Role in Developing Technological Capability in Industry No. 384 Kojima and Okada, Catching Up to Leadership: The Role of Technology Support Institutions in Japan's Casting Sector No. 385 Rowat, Lubrano, and Porrata, Competition Policy and MERCOSUR No. 386 Dinar and Subramanian, Water Pricing Experiences: An International Perspective No. 387 Oskarsson, Berglund, Seling, Snellman, StenbAck, and Fritz, A Planner's Guidefor Selecting Clean-Coal Technologies for Power Plants No. 388 Sanjayan, Shen, ancl Jansen, Experiences with Integrated-Conservation Development Projects in Asia No. 390 Foster, Lawrence, arnd Morris, Groundwater in Urban Development: Assessing Management Needs and Formulating Policy Strategies No. 392 Felker, Chaudhuri, Gyorgy, and Goldman, The Pharmaceutical Industry in India and Hungary: Policies, Institutions, and Technological Development No. 393 Mohan, ed., Bibliography of Publications: Africa Region, 1990-97 No. 394 Hill and Shields, Incentives for Joint Forest Management in India: Analytical Methods and Case Studies No. 395 Saleth and Dinar, Satisfying Urban Thirst: Water Supply Augmentation and Pricing Policy in Hyderabad City, India No. 396 Kikeri, Privatization and Labor: What Happens to Workers When Governments Divest? No. 397 Lovei, Phasing Out Leadfrom Gasoline: Worldwide Experience and Policy Implications No. 398 Ayres, Anderson, and Hanrahan, Setting Prioritiesfor Environmental MAnagement: An Application to the Mining Sector in Bolivia No. 399 Kerf, Gray, Irwin, Levesque, Taylor, and Klein, Concessionsfor Infrastructure: A Guide to Their Design and Award No. 401 Benson and Clay, The Impact of Drought on Sub-Saharan African Economies: A Preliminary Examination No. 402 Dinar, Mendelsohn, Evenson, Parikh, Sanghi, Kumar, McKinsey, and Lonergan, Measuring the Impact of Climate Change on Indian Agriculture No. 403 Welch and Fremond, The Case-by-Case Approach to Privatization: Techniques and Examples No. 404 Stephenson, Donnay, Frolova, Melnick, and Worzala, Improving Women's Health Services in the Russian Federation: Results of a Pilot Project No. 405 Onorato, Fox, and Strongman. World Bank Group Assistancefor Minerals Sector Development and Reform in Member Countries No. 406 Milazzo, Subsidies in World Fisheries: A Reexamination 14\ THE WORLD BANK ISIS I I St rc.t. N.\\. \\ Tshini tou I1). . 211433 1 S \ I1Ktcphoui I: 2112-477-123.4 I ls,,l:2112-47,-y0(l.' 'I'Cc\: \I(:I 04145 \\UI{I.I)11 \\Nk \1( . 248423n \\ O10.1)1.) \N1\ \\WI LI \\I it:c \\ch: I : tt 1): \\ .\\ ,,I-t .il II'I,,ll : 1ho,d,1X I \\,,1.1(1;1,,1 wl' I 11111!13111 14229lil" 9 78021 0342299 ISBN 0-8213-4229-0