social funds - world banklnweb90.worldbank.org/.../$file/social_funds.pdf · 44 social funds’...

Social FundsAssessing Effectiveness

2002THE WORLD BANK

Washington, D.C.

W O R L D B A N K O P E R A T I O N S E V A L U A T I O N D E P A R T M E N T

http://www.worldbank.org/oed

ENHANCING DEVELOPMENT EFFECTIVENESS THROUGH EXCELLENCE AND INDEPENDENCE IN EVALUATION

The Operations Evaluation Department (OED) is an independent unit within the World Bank; it reports directlyto the Bank’s Board of Executive Directors. OED assesses what works, and what does not; how a borrower plansto run and maintain a project; and the lasting contribution of the Bank to a country’s overall development. Thegoals of evaluation are to learn from experience, to provide an objective basis for assessing the results of theBank’s work, and to provide accountability in the achievement of its objectives. It also improves Bank work byidentifying and disseminating the lessons learned from experience and by framing recommendations drawn fromevaluation findings.

OPERATIONS EVALUATION DEPARTMENT

© 2002 The International Bank for Reconstruction and Development / The World Bank

1818 H Street, NW

Washington, DC 20433

All rights reserved.

Manufactured in the United States of America

First Printing May 2002

1 2 3 4 03 02 1

The findings, interpretations, and conclusions expressed here are those of the author(s) and do not necessarily reflect the

views of the Board of Executive Directors of the World Bank or the governments they represent.

The World Bank cannot guarantee the accuracy of the data included in this work. The boundaries, colors, denomina-

tions, and other information shown on any map in this work do not imply on the part of the World Bank any judgment

of the legal status of any territory or the endorsement or acceptance of such boundaries.

Rights and Permissions

The material in this work is copyrighted. No part of this work may be reproduced or transmitted in any form or by any

means, electronic or mechanical, including photocopying, recording, or inclusion in any information storage and retrieval

system, without the prior written permission of the World Bank. The World Bank encourages dissemination of its work

and will normally grant permission promptly.

For permission to photocopy or reprint, please send a request with complete information to the Copyright Clearance

Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, www.copyright.com.

All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher,

World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail [email protected].

Cover photo: Courtesy of Curt Carnemark.

ISBN 0-8213-5141-9

Library of Congress Cataloging-in-Publication Data has been applied for.

Printed on Recycled Paper

i i i

Contents

vii Acknowledgments

xi Foreword, Prólogo, Avant-Propos

xv Executive Summary, Resumen, Résumé Analytique

xxiii Abbreviations and Acronyms

xxv Overview

1 1. Introduction1 Social Fund Projects Have Diverse Features . . .2 . . . And Their Objectives Have Evolved2 Therefore, the Evaluative Framework Must Be Broad

5 2. Outcome5 What Have Social Fund Projects Aimed to Achieve?6 To What Extent Have Social Fund Projects Been Relevant, Efficacious,

and Efficient?19 What is the Record of Bank Performance in Social Funds?

21 3. Subproject Sustainability21 Subproject Sustainability Can Be Assessed by Direct Observation . . .24 . . . Or by Examining Conditions for Sustainability

31 4. Institutional Development Impact32 Most Social Fund Agencies Are Independent and Competent . . .34 . . . And Their Wider Institutional Effects Have Been Limited44 Social Funds’ Institutional Role Continues to Change

47 5. Future Bank Support to Social Funds: Issues to Consider

53 6. Recommendations

Annexes57 Annex A. Social Funds Portfolio61 Annex B. Review Instruments65 Annex C. Regional Distribution of Social Funds67 Annex D. Social Fund Financing by Activity69 Annex E. Social Funds 2000 Results83 Annex F. Promotion, Outreach, and Targeting91 Annex G. Social Fund Winners and Losers93 Annex H. Cost Data97 Annex I. Bank Performance99 Annex J. Stakeholder Survey Results

101 Annex K. Highlights of Beneficiary Assessment Findings on Participation103 Annex L. Community-Based Contracting105 Annex M. OED Household Survey Results133 Annex N. Best Practice Examples/Elements of Social Fund Projects137 Annex O. Community Participation and Social Capital149 Annex P. Statement of the External Advisory Panel151 Annex Q. Management Response161 Annex R. CODE Report

165 Endnotes

179 Bibliography

Boxes3 Box 1.1. How Social Fund Agencies Expect to Achieve Development

Effectiveness4 Box 1.2. Social Funds Face Tradeoffs To Be Effective

11 Box 2.1. Quantity, Access, and Utilization of Physical Infrastructure HasImproved

12 Box 2.2. Outcomes and Welfare Impacts Are Varied16 Box 2.3. Lessons From Experience: Allocation Rules for Reaching the

Poor18 Box 2.4. Solving Transport Problems Is Not Just a Matter of Building

Roads23 Box 3.1. Zambia: Tackling Tradeoffs Between Community Self-Help and

Construction Quality27 Box 3.2. Learning Curve: Trends in Sustainability Mechanisms29 Box 3.3. When Should Social Funds Support Communities “Acting

Alone” to Provide Goods and Services, and When ShouldGovernment Be Involved?

34 Box 4.1. Institutional Development Effects Have Many Channels42 Box 4.2. Social Capital Impacts in Four Surveyed Countries Have Been

Mixed50 Box 5.1. What Key Factors Influence the Suitability and Design of Social

Funds?

S o c i a l F u n d s : A s s e s s i n g E f f e c t i v e n e s s

i v

Tables6 Table 2.1. Most Social Funds Have Many Objectives

23 Table 3.1. Household Perception of Operations and MaintenancePerformance

32 Table 4.1. Institutional Development Objectives Have Become MoreLocal

38 Table 4.2. Social Fund Budgetary Arrangements

Figures13 Figure 2.1. Social Funds Have Mildly Progressive Geographic Targeting14 Figure 2.2. Social Funds Have Neutral or Very Mildly Progressive

Household Targeting

C o n t e n t s

v

v i i

Acknowledgments

This Operations Evaluation Department(OED) review responds to the interestof the World Bank’s Board of Executive

Directors in the development effectiveness of theBank’s rapidly expanding social funds port-folio, reiterated most recently at the Board dis-cussion of the Social Protection Sector StrategyPaper in September 2000. Work on this reviewcoincided with, and benefited from, a parallelself-evaluation of social fund projects in sixcountries managed by the Human Develop-ments Network’s Social Protection Unit and thePoverty Reduction and Economic ManagementNerwork’s Poverty Analysis Unit, Social Funds2000 Impact Evaluation. In addition to pre-senting an independent perspective on issuescovered in the self-evaluation, including theissues of targeting and subproject sustainability,this OED review provides fresh insights intoother areas, notably the participatory process insocial fund projects and their institutional devel-opment impacts in the government, non-government, and community arenas.

The social fund portfolio is relatively young,with only 23 closed projects in the OED data-base as of June 30, 2000. This review is basedon a portfolio assessment of all 66 social fundprojects approved by the Bank as of end-fiscal1999, a literature review, task manager inter-views, stakeholder surveys, household surveysand community-level key informant interviewsand focus groups in 4 countries, and institu-tional analysis based on field research in 7countries. A number of tools were developedor applied in the context of this review: theory-

based methodology; a social fund evaluationtoolkit (institutional analysis at the govern-ment and social fund agency levels); householdsurveys comprising quantitative and qualitativequestions and complemented by focus groupsand key informant interviews (community-levelanalysis); and an Internet-based “Activity Room”that served as a discussion forum. The findingsand recommendations of this review are pre-sented in a spirit of stock-taking, learning, andknowledge-sharing related to the developmenteffectiveness of social fund projects.

The task manager of the review was SoniyaCarvalho. This report was written by SoniyaCarvalho, Gillian Perkins, and Howard White,with contributions from Caroline Bahnson,Anju Gupta Kapoor, and Svenja Weber-Venghaus. It was an activity of the UrbanCluster, coordinated by Roy Gilbert and com-prising Anna Amato, Soniya Carvalho, andRonald Parker. William Hurlbut edited thereport. In addition to the immediate team,background analysis and/or data collectionwere undertaken by Charles Abelmann, RodrigoCisneros, Geske Dijkstra, Jean Paul Fauget,Jennifer Green, Ana Maria Ibanez, ThomasKuby, Sohail Malik, Julian May, AzedineOuerghi, Vijayendra Rao, Susan Razzaz, DavidSahn, Rodrigo Serrano, Jan Kees van Donge,and Stephen Younger. Contributions from theDevelopment Economics, EASHD, PRMGE, andthe World Bank Institute are much appreciated.Andres Jarrin and Stanley Khaila managed thehousehold survey work and qualitative researchon behalf of Urbana Consultores, Ecuador, and

the Center for Social Reseach, Malawi. GeorgeCampbell and Ruel Cooke, Environmental Solu-tions, Jamaica, conducted additional fieldwork.Can Adamoglu, Alexander Arenas, Arif SyedHusain, Enver Kamal, Stig Kjeldsen, KalpanaMehra, Hakon Nordang, Raylynn Oliver, andSamir Stewart contributed to various aspects ofthe database. Jacintha Wijesinghe advised onthe budget. Romayne Pereira provided admin-istrative support. Maria Mar, Alex McKenzie, andKevin Warr set up the Internet-based ActivityRoom. Simon Appleton, Robert Chase, BerkOzler, Menno Pradhan, Mark Schacter, andRoger Slade provided technical advice.

Social fund agency managers and staff aregratefully acknowledged for their contribu-tions: Clare Barkworth, Scarlette Gillings, SamKakhobwe, Carlos Lacayo, Wedex Illunga,Edward Male, Cosmas Mambo, and CharlesMandala. OED also gratefully acknowledges theinput of numerous community members andother stakeholders across various countries,who patiently participated in interviews andsurveys and provided valuable perspectives.

The External Advisory Committee for thereview consisted of Samuel Morley (Interna-tional Food Policy Research Institute), JoanNelson (Woodrow Wilson Center of the Smith-sonian and School of International Science atAmerican University), Elinor Ostrom (IndianaUniversity), Judith Tendler (Massachusetts Insti-tute of Technology); Christina Paxson (Prince-ton University) and Carol H. Weiss (HarvardUniversity) advised on methodological aspects.

This review would not have been possiblewithout the generous support of several donors.Substantial technical and financial collaborationwith the Swiss Agency for Development Coop-eration, in particular with Carmen Alvarado,Jurg Benz, Peter Bischof, Walter Hofer, RuthHuber, Alexandre Widmer, and HanspeterWyss, as well as contributions from Belgium,Denmark, GTZ, the Netherlands, and Norwayare gratefully acknowledged.

Many Bank staff reviewed drafts and pro-vided comments at various stages. Extensiveadvice was provided by Christine Kessides,Jennie Litvack, Christina Malmberg-Calvo,Alexandre Marc, Laura Rawlings, Lynne

Sherburne-Benz, David Steel, Susan Stout, andJulie Van Domelen, and review by LiviaBenavides, Judy Baker, Anush Bezhanyan,Christopher Chamberlin, Samantha De Silva,Shantayanan Devarajan, William Dillinger,Philippe Dongier, David Ellerman, LauraFrigenti, Maurizio Guadagni, Navin Girishankar,Robert Holzman, Kathryn Johnston, SteenJorgensen, Qaiser Khan, Elizabeth King, NorbertMugwagwa, Nginya Mungai Lenneiye, TimothyMarchant, John Newman, Deepa Narayan,Jennifer Sara, Marguerite Salah, Norbert Schady,Michael Stevens, Maurizia Tovo, AndreaVermehren, David Warren, Borge Wietzke, andMichael Woolcock. The review also benefitedfrom useful comments from other Bank staff,in particular, Joseph Baah-Dwomoh, JaimeBiderman, Michael Bamberger, HansBinswanger, Caroline Cederlof, LaurenceClarke, Anis Dani, Cornelis de Haan, WilliamDorotinsky, Samir El-Daher, Yasser El-Gammal,Peter Ellis, Talib Esmail, Kene Ezemenari, Clau-dia Fumo, Alan Gelb, Eroll Graham, JefferyGutman, James Hicks, Ulrich Lachler,Dominique Lallement, Frannie Leautier, RobertLiebenthal, Marlaine Lockheed, Brian Levy,Nicholas Manning, Caroline Mascarell,Chistopher Mausolff, Prasad C. Mohan, HidekiMori, Veronica Nyhan, Andrew Parker, AnthonyPellegrini, Patti Petesch, Derek Poate, AnaMaria Sandi, Kinnon Scott, Parmesh Shah,Najma Siddiqi, Graham Smith, Willem Struben,Juliana Weissman, Dana Weist, MarkWoodward, and Thomas Zearley.

The following OED staff provided advice andcomments, which are much appreciated: RemaBalasundaram, Shamit Chakravarti, Linda Dove,Victoria Elliott, John Eriksson, Alison Evans,Osvaldo Feinstein, Madhur Gautam, ChristopherGerrard, Gita Gopal, Patrick Grasso, PoonamGupta, Catherine Gwin, Nagy Hanna, JohnHeath, Wendy Jarvie, Timothy Johnston, Pierre-Joseph Kingbo, Laura Kullenberg, Nalini Kumar,Ruben Lamdany, Patricia Laverly, SimoneLawaetz, Hernan Levy, Andres Liebenthal,Fernando Manibog, Ridley Nelson, Keith Pitman, Diana Qualls, Binyam Reja, KlasRingskog, Anwar Shah, Masaharu Shimizu, andWarren van Wicklin.


v i i i

This study was published in the Partner-ships and Knowledge Group (OEDPK) by theOutreach and Dissemination Unit. The taskteam includes Elizabeth Campbell-Pagé (taskteam leader), Caroline McEuen (editor), andJuicy Qureishi-Huq (administrative assistant).

Director-General, Operations Evaluation: Robert Picciotto

Director, Operations Evaluation Department:

Gregory Ingram

Manager, Sector and Thematic Evaluation: Alain Barbu

Task Manager: Soniya Carvalho

A c k n o w l e d g m e n t s

i x

x i

FOREWORD

This study is the first inde-pendent evaluation of social

fund projects by OED. It reviews thedevelopment effectiveness of socialfund projects and draws implica-tions for future Bank support tothem. It responds to the interest ofthe World Bank’s Board of ExecutiveDirectors for an independent evalua-tion of this fast-expanding portfolio.The review is based on an examina-tion of existing data and literature,and brings to bear new informationrelating to the participatory processin social fund projects and theirinstitutional development impacts atthe government, nongovernment,and community levels.

The report finds that socialfund projects have been highlyeffective in delivering small-scaleinfrastructure, but much less so inachieving consistently positiveand significant improvements inoutcomes and welfare impacts.While social fund projects havedelivered slightly more than pro-portional benefits to the poor andthe poorest, there have also beena significant number of non-poorbeneficiaries. Most social fundbeneficiaries are satisfied with thefinanced subprojects, but thegreatest community problemshave not necessarily beenaddressed and there is no assur-ance that the selected subprojectsensure the highest net benefits tothe community. While social fundfacilities are generally operatingand equipped, they have notbeen immune to staffing andequipment shortages. Social fund

PRÓLOGO

Este estudio constituye laprimera evaluación inde-

pendiente de proyectos relaciona-dos con fondos sociales que llevaa cabo el DEO; en él se examina laeficacia de dichos proyectos parapromover el desarrollo y seextraen conclusiones que incidi-rán en el respaldo que el Bancoles ha de prestar en el futuro. Eltrabajo se realizó a raíz del interésdel Directorio Ejecutivo del BancoMundial por contar con una eva-luación independiente de esta car-tera en rápida expansión. Se basaen el examen de datos y publica-ciones existentes, e incorporanueva información acerca del pro-ceso de participación en proyectosde fondos sociales y sus conse-cuencias para el desarrollo institu-cional en los ámbitos estatal, nogubernamental y comunitario.

En el informe se llega a laconclusión de que los proyectosvinculados a fondos sociales hansido sumamente eficaces parasuministrar infraestructura enpequeña escala, pero no lo hansido tanto para lograr mejorasapreciables y duraderas en losresultados y en el bienestar dela población. Si bien este tipo deproyecto ha traído aparejadosbeneficios apenas más que pro-porcionales para los pobres eindigentes, también ha habidoun número considerable debeneficiarios que no son pobres.La mayor parte de los beneficia-rios de los fondos sociales estánsatisfechos con los subproyectosfinanciados, pero ello no signi-

AVANT-PROPOS

Cette étude est la premièreévaluation indépendante des

projets de fonds sociaux réaliséepar l’OED. Elle examine l’efficacitéde ces projets sur le plan du déve-loppement et leurs implications surl’appui futur de la Banque. Le pré-sent examen a été réalisé à la de-mande des administrateurs de laBanque mondiale qui souhaitaientobtenir une évaluation indépendan-te de ce portefeuille en rapide ex-pansion. Il repose sur une analysedes données et des études existan-tes et met à profit les nouvelles in-formations collectées surl’application d’un processus parti-cipatif dans les projets de fonds so-ciaux et sur l’impact de cesderniers sur le développement ins-titutionnel au niveau gouvernemen-tal, non gouvernemental etcommunautaire.

Le rapport établit que les pro-jets de fonds social ont permis demettre en place de manière trèsefficace des infrastructures detaille modeste mais n’ont eu derésultats ni d’impact sur le bien-être de la population systémati-quement positifs et notables.Même si ces projets ont profitéproportionnellement un peu plusaux groupes pauvres et extrême-ment pauvres, ils ont aussibénéficié à un nombre non négli-geable de non pauvres. La plu-part des bénéficiaires sontsatisfaits des sous-projets finan-cés, mais ceux-ci n’ont pas né-cessairement réglé les problèmesles plus importants, pas plusqu’ils n’ont permis de garantir

EN

GL

ISH

ES

PA

NO

L

FR

AN

ÇA

IS

agencies have developedcapacity as effective andinnovative organizations,but wider impacts on exist-ing institutions have beenmuch more limited.

Improving social fundperformance warrants more

attention to the following issues:the appropriateness and effective-ness of the social funds’ demand-driven mechanism; whether orwhen social funds can beexpected to serve long-termdevelopment objectives; how theydepend on, and affect, other pub-lic institutions; what scale andsectors of operation are appropri-ate; and what transformation orexit strategies are indicated.

Depending on the country con-text, addressing these issues maybe a matter of modifying socialfund institutional designs, bettercoordination with complementaryinterventions outside the socialfund, adopting an alternativeinstrument, and the like. The cre-ation of new social funds—andincreases in the scale of supportto existing social funds—requirescaution and careful analysis of thetradeoffs between specific objec-tives in a particular country con-text. The report recommendsgreater transparency and selectiv-ity in the use of this instrument.


x i i

EN

GL

ISH

fica, necesariamente, quese hayan abordado losproblemas más graves dela comunidad, ni que lossubproyectos escogidosaseguren los máximosbeneficios netos para ella.Aunque, por lo general,

las instalaciones de los fondossociales se encuentran en fun-cionamiento y están equipadas,han sufrido los efectos de laescasez de personal y equipos.Los organismos de los fondossociales se han convertido enorganizaciones eficaces e inno-vadoras, pero los efectos en lasinstituciones existentes han sidomucho más limitados.

Para que el desempeño de losfondos sociales mejore, se debeprestar mayor atención a lascuestiones siguientes: qué utili-dad y eficacia tiene el meca-nismo, impulsado por lademanda, de los fondos socia-les; cuándo cabe esperar que losfondos sociales contribuyan alos objetivos de desarrollo alargo plazo; cómo se relacionancon otras instituciones públicasy cómo las afectan; qué escala ysectores de operación son apro-piados, y qué estrategias detransformación o de solución serecomiendan. Según la situacióndel país, al ocuparse de estascuestiones habrá que modificarlos diseños institucionales de losfondos sociales, mejorar la coor-dinación con intervencionescomplementarias ajenas al fondosocial, adoptar un instrumentoalternativo, o tomar las medidassimilares necesarias en cadacaso. La creación de nuevos fon-dos sociales —así como la inten-sificación del respaldo a losexistentes— exige un análisis

que les sous-projets rete-nus sont les plus avanta-geux, sur une base nette,pour les communautés. Siles installations financéespar les fonds sociaux sontdans l’ensemble opération-nelles et disposent des

matériels et fournitures nécessai-res, elles ne sont pas à l’abri depénuries au niveau des ressour-ces humaines et physiques. Lesfonds sociaux ont renforcé leurscapacités et sont devenus desorganismes efficaces et nova-teurs, mais leur impact au niveaudes institutions existantes estresté beaucoup plus limité.

Pour améliorer la performancedes fonds sociaux, il faudra por-ter une plus grande attention à :la pertinence et l’efficacité d’unmécanisme impulsé par lademande ; la définition desobjectifs (un fonds social doit-ilservir à poursuivre des objectifsde développement à long termeet, le cas échéant, dans quellesconditions ?) ; les interdépendan-ces entre les fonds sociaux et lesautres institutions publiques ;l’envergure des activités et lessecteurs visés ; et les stratégiesindiquées à des fins de transfor-mation ou de désengagement.Pour régler ces problèmes, ilpeut être nécessaire, par exem-ple, selon le contexte de chaquepays, de modifier le modèle insti-tutionnel du fonds social, d’amé-liorer la coordination desopérations financées par le fondsavec d’autres interventions oud’adopter un instrument différentdans certaines conditions. Avantde créer de nouveaux fonds soci-aux — et d’accroître l’ampleur del’appui aux fonds sociaux exis-tants — il faudra faire preuve de

FR

AN

ÇA

IS

ES

PA

NO

L

cauto y detenido de laelección entre objetivosespecíficos en la situaciónparticular de cada país. Enel informe se recomiendauna mayor transparencia yselectividad en el uso deeste instrumento.

prudence et soigneuse-ment analyser les compro-mis nécessaires entredifférents objectifs dans lecontexte particulier d’unpays. Le rapport recom-mande d’employer cetinstrument de manière

plus transparente et plussélective.

F o r e w o r d

x i i i

FR

AN

ÇA

IS

ES

PA

NO

L

Robert PicciottoDirector-General, Operations Evaluation

x v

EXECUTIVESUMMARY

The World Bank’s social fundportfolio, although relatively

young, is growing and is set to con-tinue expanding at a rapid pace. Asof end-fiscal 1999, about $2.4 billionhad been approved for 66 projects in42 countries. By May 2001, total Bankinvestment in social funds stood at$3.5 billion for more than 98 projectsin 58 countries. This reviewresponds to the interest of the WorldBank’s Board of Executive Directorsfor an independent evaluation of thisportfolio.

Social fund agencies channelresources to small-scale subpro-jects proposed by stakeholdersand screened using eligibility cri-teria. Unlike a typical projectimplementation unit, social fundagencies usually have a highdegree of independence from lineministries and sectoral budgetsand make decisions on allocationof resources among alternativeinvestments—both across andwithin sectors and Regions.

How Have Social FundProjects Performed?

OutcomeThe operational focus of theseprojects has been broadly consis-tent with Bank and governmentpoverty-reduction objectives.However, the discussion of socialfunds in the Bank’s Country Assis-tance Strategies (CASs) has tendedto focus on the activities that thesocial fund projects aim tofinance, with little attention to theappropriate role of the social fund

RESUMEN

Aunque relativamente nueva,la cartera de proyectos del

Banco Mundial relacionados confondos sociales está creciendo, ytodo indica que continuará amplián-dose rápidamente. A fin del ejerci-cio de 1999, se habían aprobadounos US$2.400 millones para 66 pro-yectos de ese tipo en 42 países. Paramayo de 2001, el total de inversionesdel Banco en fondos socialesascendía a US$3.500 millones, distri-buidos entre más de 98 proyectos en58 países. El presente examen sur-gió del interés del Directorio Ejecu-tivo del Banco Mundial en contarcon una evaluación independientede esa cartera.

Los organismos de los fondossociales encauzan los recursoshacia pequeños subproyectos pro-puestos por los interesados yseleccionados entre los que cum-plen determinadas condiciones. Adiferencia de las unidades corrien-tes de ejecución de proyectos, losorganismos de los fondos socialessuelen tener un alto grado deindependencia de los ministeriosoperativos y los presupuestos sec-toriales, y adoptan decisionessobre la asignación de recursoseligiendo entre distintas inversio-nes, tanto entre diferentes sectoresy regiones como dentro de ellos.

¿Qué desempeño han tenidolos proyectos relativos afondos sociales?

ResultadosEl tema central de estos proyectosha sido, en líneas generales,

RÉSUMÉ ANALYTIQUE

Quoique relativement récent,le portefeuille de fonds soci-

aux de la Banque mondiale s’accroîtrapidement et semble devoir conti-nuer dans cette voie. À la fin del’exercice 99, quelque 2,4 milliardsde dollars avaient été approuvés autitre de 66 projets dans 42 pays. Enmai 2001, le montant total investi parla Banque dans ces fonds s’élevait à3,5 milliards de dollars au titre de98 projets dans 58 pays. Le présentexamen a été réalisé à la demandedes administrateurs de la Banquemondiale qui souhaitaient obtenirune évaluation indépendante de ceportefeuille.

Les organismes des fonds soci-aux aiguillent les ressources versde petits sous-projets proposéspar les parties prenantes et sélec-tionnés suivant des critères d’éligi-bilité. Contrairement aux cellulesd’exécution de projets tradition-nelles, les fonds sociaux jouissentd’habitude d’une grande indépen-dance vis-à-vis des ministères detutelle et des budgets sectoriels ;ce sont eux qui décident de l’allo-cation des ressources entre diffé-rents investissements, à un niveauaussi bien intra ou inter sectorielet régional.

Quels sont les résultats obtenus par les projets defonds social ?

RésultatsSur le plan opérationnel, ces pro-jets cadrent avec les objectifs deréduction de la pauvreté de laBanque et du gouvernement. On

EN

GL

ISH

ES

PA

NO

L

FR

AN

ÇA

IS

within the country’s institu-tional framework.

Social fund projectshave been highly effectivein delivering small-scaleinfrastructure, and thisadditional infrastructurehas increased facility

access and utilization. But theoutcomes and welfare impacts—for example, enrollment rates;academic achievement; inci-dence of diarrhea, wasting, andinfant mortality—have varied.While such projects have deliv-ered slightly more than propor-tional benefits to the poor andthe poorest, there have been asignificant number of non-poorbeneficiaries.

Most beneficiaries were satis-fied with the infrastructure, butthe social funds’ subprojectchoices and selection processeshave not guaranteed that thebiggest community problems areaddressed, nor have they pro-vided assurance that the selectedsubprojects delivered the highestnet benefits to the community.Several factors were found toinfluence the “demand-driven”process, including the role of“prime movers” (that is, localleaders) who were critical in themobilization of support andpreparation of a successful sub-project proposal, and whose inter-ests were driven by the nature oftheir position (such as headmas-ters mobilizing support forschools).

Sustainability of BenefitsSurveys of social fund facilitieshave found that staffing andequipment levels were at least asgood as those in comparatorfacilities. But both kinds of facili-


x v i

EN

GL

ISH

coherente con los objetivosde reducción de la pobrezaque persiguen el Banco ylos gobiernos. No obstante,en las estrategias de asis-tencia a los países diseña-das por el Banco, elexamen de los fondos

sociales se ha centrado, por logeneral, en las actividades que losproyectos relacionados con ellosvan a financiar, y se ha prestadoescasa atención a la función quedebe tener el fondo social en elmarco institucional del país.

Los proyectos de fondos socia-les han sido sumamente eficacespara proveer de pequeñas obrasde infraestructura, que han facili-tado el acceso a instalaciones y suutilización. Pero sus resultados ylas consecuencias para el bienes-tar de la población han sido dis-pares, como se refleja, porejemplo, en las tasas de matricula-ción, los logros académicos, y laincidencia de la diarrea, la con-sunción y la mortalidad infantil. Sibien dichos proyectos han arro-jado beneficios apenas más queproporcionales para los pobres ylos indigentes, también existe unnúmero considerable de benefi-ciarios que no son pobres.

La mayoría de los beneficiariosse han mostrado satisfechos conla infraestructura, pero las opcio-nes y los procesos de selecciónde los subproyectos de los fon-dos sociales no han garantizadoque se abordaran los problemasmás graves de la comunidad, nitampoco que los subproyectosescogidos produjeran los máxi-mos beneficios netos para ella.Se llegó a la conclusión de queen el proceso “impulsado por lademanda” influyen diversos fac-tores, entre ellos la función de

notera toutefois que, dansle cadre des Stratégiesd’assistance de la Banqueaux pays (SAP), l’attentiona, dans l’ensemble, surtoutporté sur les activités queles projets de fonds socialvisent à financer au détri-

ment du rôle qu’un fonds socialdoit assumer dans le cadre institu-tionnel du pays considéré.

Les projets de fonds social ontpermis de mettre en place demanière très efficace des infras-tructures de taille modeste qui ontfacilité l’accès aux installations etleur utilisation. Toutefois leursrésultats et leur impact sur lebien-être de la population — parexemple les taux de scolarisation ;les niveaux d’instruction ; l’inci-dence de la diarrhée, de la dénu-trition et de la mortalité infantile— sont variables. Même si cesprojets ont profité proportionnel-lement un peu plus aux groupespauvres et extrêmement pauvres,ils ont aussi bénéficié à un nom-bre non négligeable de nonpauvres.

La plupart des bénéficiaires sedisent satisfaits de l’infrastructure,mais le choix et le mode de sélec-tion des sous-projets n’ont pastoujours permis de s’attaquer auxproblèmes les plus importants,pas plus qu’ils n’ont permis degarantir que les sous-projets rete-nus sont les plus avantageux, surune base nette, pour les commu-nautés. Plusieurs facteurs ont étéidentifiés, qui influencent le pro-cessus de « réponse à lademande », et notamment les« principaux agents du change-ment » (les responsables locaux)qui ont contribué dans unemesure cruciale à mobiliserl’appui et les ressources nécessai-

ES

PA

NO

L

FR

AN

ÇA

IS

ties suffered from short-ages. The sustainability ofeducation and healthinvestments has generallybeen better than invest-ments in other sectors,such as water or roads.Overall, insufficient atten-

tion has been given to ensuringfrom the outset that the scopeand scale of activities undertakenare guided by an assessment ofrelevant capacity in each of theproposed sectors of social fundintervention.

Institutional Development ImpactsSocial fund agencies have beeneffective and competent organiza-tions. They have developed inno-vative procedures for projectmanagement that have beenadopted in other Bank projects.By fostering partnerships amonggovernment agencies, the privatesector, and other stakeholders,social funds have mobilized com-munity resources and stimulatedprivate contracting capacity.Impacts at the central governmentlevel have been limited. Giventhat social fund agencies makedecisions on allocation ofresources, coordination with lineministries that also perform thisfunction is crucial, especially forlarge-scale social funds.

Coordination has, however,proved difficult, particularlyregarding sectoral planning andthe capital-recurrent expenditurebalance. Impacts at the local gov-ernment level have been morepositive but have varied with thenature of social fund engagement,the degree of responsibility dele-gated to local governments orcommunities, and the alignmentof social fund operations with the

E x e c u t i v e S u m m a r y

x v i i

EN

GL

ISH

los “principales motores”(es decir, los dirigenteslocales), quienes eran deimportancia crucial paramovilizar el respaldo y lapreparación de una pro-puesta de subproyectosatisfactoria, y cuyos inte-

reses se guiaban por la natura-leza de su posición (losdirectores, por ejemplo, reuníanapoyo para las escuelas).

Continuidad de los beneficiosEn los reconocimientos de las ins-talaciones de los fondos socialesse determinó que la dotación depersonal y el nivel de equipa-miento eran, como mínimo, tansatisfactorios como en las que seutilizaron como parámetro decomparación. Pero en los dostipos de instalaciones se observa-ron carencias. Las inversiones eneducación y salud han resultadomás duraderas que aquellas enotros sectores, como los serviciosde agua o los caminos. En térmi-nos generales, no se ha dedicadoatención suficiente a asegurar,desde un principio, que elalcance y la magnitud de las acti-vidades emprendidas se guiaranpor una evaluación de la capaci-dad pertinente en cada uno de lossectores en los que se propone laintervención de los fondossociales.

Consecuencias para el desarrollo institucionalLos organismos de los fondossociales han sido organizacioneseficaces y competentes. Han ide-ado, para la administración deproyectos, procedimientos inno-vadores que se han aplicado enotros proyectos del Banco. Alfomentar la colaboración entre los

res à la préparation de pro-positions de sous-projetsvalables, et dont les intérêtssont dictés par la naturemême de leurs fonctions(comme les directeursd’école, qui mobilisent unappui en faveur des équi-

pements scolaires).

Pérennité des avantagesLes études consacrées aux fondssociaux montrent que les dota-tions en effectif et les ressourcesphysiques des installations qu’ilsfinancent sont au moins aussisatisfaisantes que celles des instal-lations de référence. Il est vraique des pénuries ont été obser-vées dans les deux cas. Les inves-tissements se sont généralementrévélés plus durables dans lesdomaines de l’éducation et de lasanté que dans d’autres secteurs,comme l’eau et les routes. Enrègle générale, on ne s’est passuffisamment préoccupé des’assurer, dès le départ, que laportée et l’ampleur des activitésentreprises étaient fondées surune évaluation des capacités réel-lement disponibles dans chacundes secteurs proposés pour uneopération de fonds social.

Impacts sur le développementinstitutionnelLes organismes de fonds socialsont efficaces et compétents. Ilsont mis au point des procéduresnovatrices de gestion des projetsqui ont été adoptées pour d’autresprojets de la Banque. En encoura-geant la constitution de partena-riats entre les organismes publics,le secteur privé et les autres par-ties prenantes, les fonds sociauxont mobilisé des ressources auniveau des communautés et sti-

ES

PA

NO

L

FR

AN

ÇA

IS

decentralization frame-work. At the communitylevel, the evidence indi-cates mixed social fundimpacts on capacity build-ing and social capital incomparison withnon–social fund communi-

ties. Overall, social funds haveoperated as users rather than pro-ducers of social capital.

What Are the Implications?Initially set up as emergencyresponse mechanisms, the focusof the funds has shifted to longer-term development impact andinstitutional development objec-tives, a transition that is some-times difficult to implement. Thenew focus requires significantchanges in an agency’s perform-ance incentives, staffing, andskills mix. Improving performancewarrants more attention to:• The appropriateness and effec-

tiveness of the social funds’subproject menus and selectionmechanism

• Whether or when social fundscan be expected to serve long-term development objectives

• How they depend on, andaffect, other public institutionsand public expendituremanagement

• What scale and sectors ofoperation are appropriate

• What transformation or exitstrategies are indicated.

Addressing these issues may bea matter of modifying social fundinstitutional designs, better coor-dination with complementaryinterventions outside the socialfund, or adopting an alternativeinstrument in a particular countrycontext.


x v i i i

EN

GL

ISH

organismos públicos, elsector privado y otros inte-resados, los fondos socialeshan movilizado los recur-sos de la comunidad y esti-mulado la capacidad decontratación privada. Lasrepercusiones para el gob-

ierno central han sido limitadas.En vista de que los organismos delos fondos sociales adoptan deci-siones sobre la asignación de losrecursos, la coordinación conministerios operativos que tam-bién cumplen esa función es devital importancia, en particularpara los fondos sociales de granenvergadura.

Sin embargo, la coordinaciónha resultado difícil, especialmenteen relación con la planificaciónsectorial y el equilibrio de los gas-tos de capital y ordinarios. Lasrepercusiones para los gobiernoslocales han sido más positivas,pero han variado según la índolede la participación del fondosocial, el grado de responsabili-dad delegado en los gobiernoslocales o las comunidades y el ali-neamiento de las operaciones delfondo con el marco de descentra-lización. En las comunidades, laspruebas indican que los fondossociales han tenido efectos dispa-res en el fortalecimiento de lacapacidad y el capital social,cuando se las compara con aque-llas comunidades donde no exis-tió la intervención de un fondosocial. En resumidas cuentas, losfondos sociales han actuadocomo usuarios en vez de produc-tores de capital social.

¿Cuáles son lasconsecuencias?Organizados inicialmente comomecanismos de respuesta ante

mulé la formation de capa-cités de délégation au sec-teur privé. L’impact de cesopérations sur l’administra-tion centrale est limité. Lesfonds sociaux décidant del’allocation des ressources,il est crucial qu’ils coordon-

nent leurs opérations avec lesministères de tutelle qui remplis-sent les mêmes fonctions, surtoutlorsque ces ressources sontimportantes.

Il s’est cependant avéré difficilede coordonner les activités,notamment au niveau de la plani-fication sectorielle et du dosagedes dépenses d’équipement et defonctionnement. Les fonds soci-aux ont eu un impact plus positifau niveau des autorités locales,bien que différent suivant l’objec-tif du fonds, le niveau des respon-sabilités déléguées aux autoritéslocales et aux communautés etl’intégration de leurs activités dansle cadre du processus de décen-tralisation. Il est clair, si l’on com-pare les résultats obtenus par lescommunautés bénéficiant defonds sociaux et ceux des autrescommunautés, que les fonds ontun impact variable au plan durenforcement des capacités et ducapital social. En général, lesfonds sociaux ont opéré pluscomme des utilisateurs quecomme des producteurs de capitalsocial.

Quelles en sont lesimplications ?Initialement conçus pour servir demécanismes d’intervention d’ur-gence, les fonds sociaux ont,après un temps, visé à promou-voir le développement à longterme et à atteindre des objectifsde développement institutionnel.

ES

PA

NO

L

FR

AN

ÇA

IS

Recommendations• Strengthen Integration of

Social Funds into theBank’s Country and Sec-toral Strategies andClient PRSPs. The Bank should not

support a social fundunless priorities are explicitlyidentified and alternativeapproaches to address those pri-orities are weighed in the light ofcountry conditions. The rationaleand objectives of Bank supportneed to be clear and should drivethe choice of instrument ratherthan the other way around.• Give More Attention to Long-

term Impacts.The tradeoffs between speed

and efficiency of subproject pro-cessing and long-term impactsshould be explicitly acknowl-edged and addressed, andreflected in performance indica-tors. Social fund agency roles,responsibilities, and relationshipsshould be aligned with existinginstitutions.• Ensure Efficiency of Resource

Allocation.Strong measures are needed to

ensure that beneficiaries are ade-quately informed and consultedon investment options, costs, andbenefits. The community and/orlocal government concernedshould choose subprojects basedon an articulation of the costs andbenefits of alternatives and theyshould monitor actual benefits inrelation to their expectations.• Develop Policy Requirements.

Policy requirements for supportto social fund projects shouldidentify conditions for their intro-duction or continuation and theirstrategic justification within theCAS and PRSP.


x i x

EN

GL

ISH

alguna emergencia, losfondos han modificado suobjetivo principal parapasar a ocuparse de losefectos del desarrollo alargo plazo y el avanceinstitucional, transiciónque, en algunos casos, es

difícil de llevar a la práctica. Elnuevo objetivo exige cambiosnotables en los incentivos deldesempeño, la dotación de per-sonal y la combinación de espe-cialidades de un organismo. Paramejorar el desempeño es precisodedicar más atención a lossiguientes factores:• Qué utilidad y eficacia tienen

las opciones de los subproyec-tos y el mecanismo de selec-ción de los fondos sociales.

• Cuándo cabe esperar que losfondos sociales contribuyan alos objetivos de desarrollo alargo plazo.

• Cómo se relacionan con otrasinstituciones públicas y con lagestión del gasto público, ycómo las afectan.

• Qué escala y sectores de ope-ración son apropiados.

• Qué estrategias de trans-formación o de salida serecomiendan.

Al ocuparse de estas cuestioneshabrá que modificar los diseñosinstitucionales de los fondossociales, mejorar la coordinacióncon intervenciones complementa-rias ajenas al fondo social o adop-tar un instrumento alternativopara la situación particular de unpaís.

Recomendaciones• Mejorar la integración de los

fondos sociales en las estrate-gias del Banco para países y

Or, ce changement de cap,qui exige que l’organismemodifie dans une largemesure son cadre incitatifainsi que le nombre et lacomposition de ses effec-tifs, est parfois difficile àréaliser. Pour améliorer

l’impact des programmes, il fau-dra accorder une plus grandeimportance à :• La pertinence et l’utilité de la

liste et du mécanisme de sélec-tion des sous-projets de fondssociaux.

• La définition des objectifs : unfonds social doit-il servir àpoursuivre des objectifs dedéveloppement à long termeet, le cas échéant, dans quellesconditions ?

• La place des fonds sociauxdans le contexte des autresinstitutions publiques et l’im-pact qu’ils ont sur ces institu-tions et la gestion desdépenses publiques.

• L’envergure des activités et lessecteurs visés.

• Les stratégies indiquées à desfins de transformation ou dedésengagement.

Pour régler ces problèmes ilpeut être nécessaire de modifierle modèle institutionnel du fondssocial, améliorer la coordinationdes opérations financées par lefonds avec d’autres interventionsou adopter un instrument diffé-rent dans certaines conditions.

Recommandations• Renforcer l’intégration des opé-

rations de fonds sociaux dansles Stratégies d’aide-pays et lesstratégies sectorielles de la Ban-que et dans les DSRP des clientsde la Banque.

ES

PA

NO

L

FR

AN

ÇA

IS


x x

sectores y en el documentode estrategia de lucha con-tra la pobreza preparadopor el cliente.

El Banco no deberíaprestar apoyo a un fondosocial a menos que sedeterminen explícitamente

las prioridades y se evalúen, a laluz de las condiciones del país,otras opciones para abordarlas. Lajustificación y los objetivos delapoyo del Banco deben ser claros y guiar la elección del ins-trumento; no debe ocurrir locontrario.• Prestar más atención a las con-

secuencias a largo plazo.Deberían reconocerse explícita-

mente y abordar las relaciones decompensación entre velocidad yeficiencia en la tramitación desubproyectos y las consecuenciasa largo plazo, y recogerse en indi-cadores de desempeño. Las fun-ciones, responsabilidades yrelaciones de los organismos delos fondos sociales deberíanarmonizarse con las institucionesexistentes.• Asegurar la eficiencia en la

asignación de recursos.Se necesitan medidas firmes

para asegurar que se informa yconsulta adecuadamente a losbeneficiarios sobre opciones deinversión, costos y beneficios. Lacomunidad y/o el gobierno localinteresado deberían elegir lossubproyectos a partir de la articu-lación de los costos y beneficiosde las alternativas, y deberíanvigilar los beneficios reales enrelación con las expectativas. • Imponer exigencias

normativas.Las exigencias normativas para

el apoyo a los proyectos sobrefondos sociales deberían estable-

La Banque ne devraitappuyer aucune opérationfaisant intervenir un fondssocial tant que les prioritésn’auront pas été clairementdéfinies et que les avanta-ges comparatifs du fondssocial et d’autres instru-

ments n’auront pas été examinéscompte tenu des conditions parti-culières du pays considéré. Lesmotifs et les objectifs de l’appuide la Banque doivent être clairs ;ils doivent dicter le choix del’instrument d’intervention et nonêtre dictés par ce dernier.• Accorder une plus grande

importance à l’impact à longterme.Il est important de bien com-

prendre et analyser les compro-mis qui doivent être faits entre,d’une part, la vitesse et l’efficacitédu traitement des dossiers desous-projets et, d’autre part,l’impact à long terme des opéra-tions, et d’établir des indicateursde performance sur cette base. Lerôle, les responsabilités et les rela-tions des fonds sociaux doiventêtre définis dans le contexte desinstitutions existantes.• Assurer une allocation ration-

nelle des ressources.Il importe de prendre des

mesures rigoureuses pour s’assu-rer que les bénéficiaires sont bieninformés et consultés sur lesinvestissements possibles, leurcoût et leurs avantages. La com-munauté et/ou les administrationslocales concernées doivent pou-voir choisir les sous-projets entoute connaissance des coûts etdes avantages de chaque optionet procéder au suivi des opéra-tions pour comparer les avantagesqu’elles obtiennent aux résultatsescomptés.

ES

PA

NO

L

FR

AN

ÇA

IS

cer las condiciones para suintroducción o continua-ción y su adecuada justifi-cación dentro de laestrategia de asistencia alos países y el documentode estrategia de lucha con-tra la pobreza.

• Formuler les grandsprincipes à respecter. Les grands principes

régissant la fourniture d’unappui aux projets de fondssocial doivent faire état desconditions de leur mise enplace ou de leur maintien

ainsi que de leur justification stra-tégique dans le cadre de la SAP etdu DSRP.


x x i

FR

AN

ÇA

IS

ES

PA

NO

L

x x i i i

AIDS Acquired Immune Deficiency SyndromeADF Albanian Development FundAfDB African Development BankAGETIP Agence d’Exécution de Travaux d’Intérêt PubliqueAFR Africa RegionCAS Country Assistance StrategyCBO Community-Based OrganizationDFID Department for International Development (U.K.)EAP East Asia and Pacific RegionECA Europe and Central Asia RegionECRF Eritrean Community Rehabilitation FundESF Emergency Social FundESRDF Ethiopia Social Rehabilitation and Development FundFHIS Honduras Social Investment FundFI Financial IntermediaryFISE Nicaragua or Ecuador Emergency Social Investment FundFONCODES Peru Social Development and Compensation FundFOPAR Argentina Participatory Social Investment FundHD Human Development (Network)HIPC Highly Indebted Poor Country InitiativeICR Implementation Completion ReportIMF International Monetary FundJSIF Jamaica Social Investment FundLAC Latin America and Caribbean RegionMASAF Malawi Social Action FundMNA Middle East and North Africa RegionMPP Micro-planning processMPU Zambia Micro-Projects UnitMSIF Moldova Social FundNGO Nongovernmental organizationNUB Nucleo of Beneficiaries (Argentina)OED Operations Evaluation DepartmentPAD Project Appraisal DocumentPPAR Project Performance Audit ReportPREM Poverty Reduction and Economic Management NetworkPRSP Poverty Reduction Strategy PaperPSA Argentina Programa Social AgropecuarioPSR Project Status ReportPTA Parent-Teacher AssociationPTI Program of Targeted InterventionsQAG Quality Assurance GroupRSDF Romanian Social Development FundSAR Staff Appraisal ReportSAR South Asia RegionSFD Social Fund for Development (Egypt)SIF Social Investment FundZAMSIF Zambia Social Investment Fund

ABBREVIATIONS AND ACRONYMS

x x v

Overview

Bank lending through social fund projects grew significantly in the1990s and is set to expand further. Most social fund projects havegenerated “repeater” operations and continued Bank support, often

accompanied by a shift in objectives from emergency response to the long-term provision of infrastructure assets or the development of local institu-tions. Both the growing volume of social fund operations and their durabilityunderlie the need for this independent review.

This review assesses the performance of theprojects against their stated objectives, usingstandard Operations Evaluation Department(OED) criteria. It especially focuses on the sus-tainability of subprojects financed by socialfunds and on their institutional developmentimpact. It is based on a portfolio assessment, aliterature review, task manager interviews, stake-holder surveys, household surveys of 1,687 ran-domly selected households and community-levelkey informant interviews and focus groups in4 countries, and institutional analysis based onfield research in 7 countries. It also draws on datagathered for six countries for a detailed self-eval-uation of the impact of social funds managed bythe Human Development Network’s (HD’s)Social Protection Unit and the Poverty Reductionand Economic Management Network’s (PREM’s)Poverty Analysis Unit, Social Funds 2000 ImpactEvaluation (Social Funds 2000 hereafter, WorldBank forthcoming).1

The Social Fund ApproachThe distinguishing characteristic of social fundprojects is that rather than implementing invest-ment decisions predetermined at projectappraisal, they allow local stakeholders to deter-mine these decisions through subprojectproposals they submit during project imple-mentation. Social fund agencies expect to achievefour major aspects of development effective-ness using various approaches:• Relevance of investments to the priority needs

of the poor is to be ensured through theparticipation of local stakeholders in selectingsubprojects from a multi-sectoral menu ofchoices. This menu can ensure self-targetingby focusing on investments that benefit thepoor. Proactive measures may also be takento target the poorest districts and communities.

• Efficiency and responsive operating pro-cedures are expected to result from the auto-nomy of the implementing agency and

appropriate staff incentives. The subproject-based approach, with standard procedures forthe subproject cycle, can also contribute totransparency, speed, and cost-effectiveness.The demand-driven mechanism is to achieveallocative efficiency in sectoral and geo-graphic distribution of resources; participationand cost-sharing by beneficiaries can helpensure that the selected investments yieldhigh returns; and other sectoral mechanismsare expected to ensure coordination withnational public expenditure priorities.

• Sustainability of investments is to beenhanced by efforts to ensure the quality ofworks, as well as by building local ownershipthrough local participation in subproject selec-tion, implementation, and costs. This can besupported by measures to build local technicaland institutional capacity for operations andmaintenance, and by coordination with otherrelevant agencies.

• Institutional development impacts for centraland local governments—through learning-by-doing effects, demonstration effects, andcompetitive effects—may result from engag-ing these entities in social fund projects.Social fund engagement at the communitylevel either through participation in projectactivities or through training may enhancelocal capacity for collective action.

The social fund approach has potential ben-efits, but achieving them involves importanttradeoffs. The community demand mechanismallows community participation in subprojectdecisions and management, but can make itdifficult to reach the poorest communities, whichare often the least competitive in preparing pro-posals. Equally, community-level decisionmakingmay help to overcome information asymme-tries by bringing to bear local knowledge, butit does not lend itself to projects that require deci-sions to be made above the local level in orderto deal effectively with externalities or tapeconomies of scale. The subproject-basedapproach, limiting engagement to communityinvolvement in a small, discrete investment,makes it possible to achieve speed and effi-ciencies through standardization, but does not,

in and of itself, ensure that the necessary com-plementary inputs are in place for achievingthe intended welfare and development impacts.The multisectoral scope of social fund projectscan give voice to local actors and encourage localinitiative by allowing communities to determinetheir own investment priorities, but it also placesheavy demands on a single agency to meet thevarying technical and institutional requirementsfor delivery of a wide range of goods andservices. Autonomy of the social fund agencymay enable it to adopt new processes and “rulesof the game” for transparent and responsiveoperation, but these gains may not be readilytransferable to permanent public sector agenciesin the absence of specific incentives and a pol-icy environment conducive to change. Con-versely, autonomy makes necessary the creationof mechanisms to coordinate social fund proj-ects with mainstream government processes inorder to avoid negative implications for themanagement of public expenditure, accounta-bility, and sustainability.

FindingsThis review confirms that social fund projectshave been highly effective in delivering small-scale infrastructure, with slightly more than pro-portional benefits to the poor and the poorest.As a result of the additional infrastructure, facil-ity access and utilization have improved amongthis population. Of the 23 closed social fund proj-ects as of end-fiscal 2000 (out of the total socialfund portfolio of 66 projects examined for thisreview), 96 percent were rated “satisfactory” onoutcome by OED. This rating compares favor-ably with outcome ratings for Program of Tar-geted Interventions (PTI) projects (78 percent)as well as for all World Bank projects (71 per-cent). Social fund projects have disbursed rap-idly and, in difficult conditions, have producedvisible outputs quickly. The vast majority ofbeneficiaries have been satisfied with the result-ing social fund infrastructure. Staffing andequipment for this infrastructure appears to beat least as good as in comparator facilities. Inaddition, social fund projects have experimentedwith increased local decisionmaking, mobilizednongovernment and private sector capacity and


x x v i

community resources, and have had some pos-itive institutional development impacts, mostlyat the local level.

There are concerns, however. These includethe varied results on outcomes and welfareimpacts when comparing social fund commu-nities with non–social fund communities; the sig-nificant number of non-poor beneficiaries; theinsufficiency of complementary inputs; the lackof adequate mechanisms to ensure allocative effi-ciency of social fund investments; the negativeimpacts on public sector management whenthe social fund is operating on a large scale; andthe mixed impact of the social fund participa-tory process in building community capacity orsocial capital. Although projects have givenincreasing attention to the financial and institu-tional requirements for sustainability, the nec-essary mechanisms have not been adoptedacross all projects, and maintenance continuesto be a problem. In sectors where operations andmaintenance are primarily the responsibility ofcentral or local governments, insufficient atten-tion has been given to ensuring from the out-set that the scope and scale of activitiesundertaken by a social fund are guided byassessment of relevant capacity in each of theproposed sectors of social fund intervention.Social fund performance has varied by sector,with economic infrastructure subprojects (roadsand water) performing less well overall thansocial sector subprojects (schools and healthclinics).

OutcomeRelevance. How well have social fund projectsmatched the Bank’s assistance strategies andthe poverty reduction programs of governments?The poverty focus of social fund projects hasbeen broadly consistent with Bank and gov-ernment poverty reduction objectives. However,the discussion of social funds in the Bank’sCountry Assistance Strategies (CASs) has tendedto focus on the activities that the social fund proj-ects aim to finance, rather than on the role ofthe social fund within a strategic framework.

Have subproject investments been relevant tocommunity priorities? Although most benefici-aries are satisfied with the subprojects they

received, OED’s household surveys in four coun-tries suggest that the subproject selection processcould not be counted on to meet the highest pri-ority problem of the majority of communitymembers, even when the relevant investmentswere on the subproject menu.2 Several factorswere found to influence the demand-drivenprocess, including the role of “prime movers,”community leaders who were critical to themobilization of support and preparation of a suc-cessful subproject proposal and whose interestswere determined by the nature of their position.For example, when the “prime mover” was aheadmaster or health worker, a bias was foundtoward subproject investments in schools orhealth facilities. It is natural and appropriatethat prime movers should bring project ideas tothe community. At the same time, mechanismsare needed to ensure that the ideas of the lead-ers are also the most important ones for the com-munity as a whole.

Efficacy. To what extent have social fundprojects met their stated goals? With respect tothe crisis-response function, they have per-formed well overall, especially in post-conflictsituations. They have enhanced public confi-dence by producing visible outputs quickly andhave helped to mobilize donor and communityresources. In the context of structural adjustmentprograms, however, there have been difficultiesin reaching the precise groups that were directlyhurt. Employment effects for private contractorsand consulting firms are positive but relativelysmall. The overall employment impact of socialfund projects has been limited and temporary.

Social fund projects have been highly effica-cious at constructing or rehabilitating small-scalephysical infrastructure, disbursing rapidly, andreaching large numbers of beneficiaries. Thefour most common types of investments havebeen schools, piped water supply systems, healthfacilities, and roads. Evidence on social fund out-comes and welfare impacts—for example, enroll-ment rates, academic achievement, incidence ofdiarrhea, wasting, and infant mortality—show avaried picture when compared with those innon–social fund communities. Evidence fromSocial Funds 2000 shows that some indicatorsof outcome and welfare impact registered a

O v e r v i e w

x x v i i

positive social fund impact, a few showed anegative impact, and several registered no sig-nificant social fund impact compared with thecontrol group. One reason for these results maybe the difficulty of the social fund in ensuringthat the necessary complementary inputs and“software” are forthcoming, both for the oper-ations and maintenance of the facility itself andfor related investments needed to make a sub-stantial impact on key development indicators.

With respect to microfinance activities, socialfund projects have had some isolated successes,but have been generally less successful in thisfunction than in providing public goods. Aboutone in four social fund projects have supportedmicro-enterprise components, but with unevenresults. In at least 2 out of 13 cases, and in verydifferent conditions, these components werehighly successful. Many of the schemes experi-enced problems in adhering to sound financialpractices or providing the necessary technicalassistance.

Performance on poverty targeting has gen-erally improved over time. Evidence from SocialFunds 2000 indicates that social fund projectstypically have had mildly progressive geographictargeting and neutral or very mildly progressivehousehold targeting. Therefore, a significantproportion of social fund resources benefit thenon-poor.

Social funds have faced tradeoffs amongobjectives—for example, among output targets,demand-orientation, employment generation,poverty targeting of benefits, building localcapacity, and subproject sustainability. Whentradeoffs have become apparent during projectimplementation, short-term objectives havetended to prevail. The hierarchy of objectives hasoften not been clearly defined. Monitoring ofproject performance has focused on indicatorsof outputs rather than outcomes or impacts.

Efficiency. How do social fund project costscompare with their benefits, and with the costsand benefits of other similar projects? Data showthat social fund subprojects have had lower unitcosts in infrastructure delivery in some countriesand sectors, and higher costs in others. There issome indication that unit costs tend to be lowerwhere community contributions were high or

there is community management of resourcesand contracting for both social fund and non—social fund projects. In most cases, social fundshad lower overheads than other programs,although there were exceptions. An importantindicator of efficiency would be cost-effective-ness in achieving specific outcomes, but thisaspect has not been systematically monitored.

Available evidence does not provide assurancethat participation and cost-sharing by commu-nities in subprojects as practiced by social fundshas been adequate to ensure selection of invest-ments with the highest net benefits to the com-munity. Subproject selection has been influencedby a variety of factors, and the procedures havenot always ensured that investment alternatives,or opportunity costs, are adequately examinedat the community or other level. In addition,more than a quarter of the social fund projectsrequired no community contribution and, there-fore, had no willingness-to-pay indicator. Littleattention has been given under social fund proj-ects to ensuring that social fund expenditurescontribute positively to the overall balance andefficiency of public expenditure.

Subproject SustainabilityTo what extent will social fund projects continueto produce net benefits as long as intended, oreven longer? Surveys of social fund facilitieshave found that staffing and equipment levelsone to three years after subproject completionwere at least as good as those in comparator facil-ities, although both types of facilities sufferedfrom shortages. Evidence for one of two coun-tries suggests that better staffing of social fundfacilities may have been at the expense ofdeclines in other facilities, raising the questionof the balance between capital and recurrentexpenditures.

Although projects have given increasing atten-tion to the financial and institutional requirementsfor sustainability, including training aimed atdeveloping community capacity, the necessarymechanisms have not yet been adopted acrossall projects. Issues for subproject sustainabilityhave varied between sectors depending on thenature of the goods and services. Overall, the sus-tainability of education and health investments


x x v i i i

has been better than in other sectors, such aswater.

Institutional Development ImpactTo what extent have social fund projects improvedcountries’ capacity to use their human, organi-zational, and financial resources effectively?Social fund implementing agencies have gainedcapacity as effective and innovative organizationsusing transparent procedures to channelresources to communities. Much of this effec-tiveness is attributed to their autonomous status,which gives them independence in recruiting andoperating procedures. They have developedinnovative procedures for project management—including management information systems, useof poverty maps, and procurement and dis-bursement procedures for community-baseddevelopment that have been adopted in otherBank projects. Most of the funds were envisagedas temporary entities, and have continued to befinanced primarily from external donor sources,but have assumed de facto long-term status astheir mandates have been extended, oftenaccompanied by adoption of longer-term objec-tives. Some of the social funds set up with a focuson immediate outputs have not readily reorientedtheir processes and operations to the achieve-ment of longer-term goals and sustainable insti-tutional development impact.

The wider institutional development objectivesof social fund projects, beyond the agency itself,have shifted from the central government tolocal governments and communities. At the cen-tral government level, positive institutionalimpacts have been limited. Coordination hasoften proved difficult, particularly in relation toconsistency with sectoral planning and the cap-ital recurrent expenditure balance. Impacts at thelocal level have been more positive but highlyvariable, depending on the nature of social fundengagement at that level, the degree of respon-sibility delegated to local governments or com-munities, and the alignment of social fundoperations with the decentralization framework.

Social funds have increased awareness ofthe potential of participatory approaches,poverty targeting of investments, public-privatepartnerships, and local decisionmaking. By fos-

tering partnerships between government agen-cies, the private sector, nongovernmental organ-izations (NGOs), and community-basedorganizations, social funds have mobilizedcapacity and community resources, and in somecases have strengthened private sector capac-ity. However, they have had negative institu-tional effects for public planning processes andbudget accountability when they have beeninadequately integrated in these processes atcentral or local levels and when social fund dis-bursements have accounted for a significantshare of public expenditure.3

At the community level, OED’s household sur-veys yielded mixed results on the extent towhich social funds have built capacity or skillsor enhanced social capital when compared withnon–social fund communities. The nature andextent of information and community partici-pation in social fund projects was found to besufficient for effective subproject execution, butnot necessarily to achieve a consistently signif-icant positive impact on community capacity orsocial capital. This is explained, at least in part,by the short engagement of the social fund atthe community level, limited to processing of asingle subproject. Positive impacts were confinedmainly to a small group of leaders in the com-munity, typically members of the subprojectcommittee. Overall, social funds have operatedas users rather than as producers of social cap-ital, relying on the existing stock.

Future Bank Support to Social Funds:Issues to ConsiderThe findings of this review point to five issues,discussed below, that warrant more attention inthe Bank’s support to social fund operations.Depending on the country context, addressingthese issues may be a matter of modifying socialfund operation, of coordination with comple-mentary interventions outside the social fund, orof adopting an alternative instrument. Based onOED’s review, more attention has been given inthe Bank to how problems and new demandscan be managed within the social fund modelthan to whether development objectives andcountry conditions warrant Bank finance througha social fund (and on what scale).

O v e r v i e w

x x i x

Issue 1: A key challenge in project design is tostrike the right balance between “demand-driven” and “supply-driven” features. In practice, social funds include a number ofsupply-driven features. Similarly, supply-drivenprograms may offer beneficiaries choice andinclude elements of community participation. Inmost cases, effective service delivery must com-bine elements of both demand and supply, thebalance depending on the project objectives,nature and scale of the services being provided,and country context. This review has foundthat the demand-driven approach is not neces-sarily synonymous with responding to the high-est priority problem of the community, evenwhen the subproject menu includes invest-ments that would address that problem. Thecommunity-based demand-driven mechanismwas found to allow a bias toward certain sec-tors. Igniting demand for sectors where primemovers or community organization do notalready exist, or addressing information asym-metries that have tended to prevent choice fromthe full range of available options, may warrant“supply” of specific inputs such as outreachand facilitation or a different kind of participa-tory process. For example, in some contexts,subproject identification using participatoryresearch may result in subprojects that are morerelevant to community needs than the simpleinvitation of subproject proposals.

Issue 2: Social funds may not automaticallyadapt to long-term objectives.More attention is needed to providing comple-mentary inputs (“software” as well as “hard-ware”) and to ensuring the institutionalarrangements necessary to achieve long-termimpact of subproject investments. Results, includ-ing tradeoffs with other objectives, need to bewatched closely. The transition from emergencyresponse mechanisms to longer-term develop-ment impact or institutional development objec-tives is proving difficult to implement in somecases. The new focus requires significant changesin the social fund agency’s performance incen-tives, staffing, and skills mix. In taking on thedifficult challenges of long-term developmentimpact, sustainable service delivery, and insti-

tutional development, social funds are liable tomeet the same constraints as other kinds ofinterventions and may lose the strengths onwhich their reputation has been built.

Issue 3: The suitability and design of socialfund projects depends on the nature of thegoods and services to be provided, and on theinstitutional context.The appropriateness of a social fund mechanismin serving a specific sectoral activity deservescareful assessment in relation to alternativedevelopment instruments. The suitability andappropriate design of a social fund depends ona range of contextual factors: the strength of exist-ing institutions and public sector reform; thenational budgeting process, structure of publicexpenditure, and sectoral planning; the extentof political, administrative, and fiscal decentral-ization; and the social structure and capacity ofcommunities. Project characteristics can andshould be modified to meet different countryconditions, and this is reflected in the diversityof social fund design. However, when the coun-try context warrants fundamental modificationsin design and operating procedures of the socialfund—for example, when the decentralizationcontext calls for support to strengthening theintergovernmental fiscal framework—then itmust be asked if the ideal response is still a socialfund. Such contextual factors vary across coun-tries, and over time within a country, as the insti-tutional and policy environment changes, andmust be continually reevaluated.

Issue 4: The development impact of socialfunds and the sustainability of subprojectsdepend on the progress of broader institutionaland public sector reforms.The multisectoral character of social fund proj-ects means that sectoral investments may notalways be subject to the full analysis and tech-nical standards applied in sector projects. Theevidence on sustainability suggests that, in sec-tors where central or local governments carry sig-nificant responsibility, it is unrealistic to expectoperations and maintenance of social fund facil-ities to differ substantially from the countrynorm. In this case, substantial improvements


x x x

would depend on changes in the prevailing fis-cal and institutional environment. The Bankshould be cautious about investing heavily insocial funds unless it is confident that the broaderfiscal and institutional constraints can be dealtwith through other interventions. The scopeand scale of activities undertaken by a social fundshould be guided from the outset by assessmentof operations and maintenance capacity andconstraints in each of the proposed sectors ofsocial fund intervention in the country con-cerned, including the existing composition andbalance of public expenditure and projectedfiscal capacity. When social funds are engagedon a large scale over a long period in rehabili-tating existing infrastructure, rather than engag-ing in new construction, this risks becoming anexpensive substitute for adequate budgetaryprovision for ongoing maintenance of publicfacilities.

Issue 5: The relationship between the socialfund and mainstream public sector institutionsis critical to the institutional developmentimpact of social funds.It is important to define from the outset theappropriate relationship of the social fund withgovernment. This relationship should be reex-amined as the institutional environment changes,with attention to any necessary changes in thesocial fund’s scope and scale of operation or exitstrategy, as relevant. For the majority of socialfunds, no exit strategy exists, nor have theirlonger-term roles and responsibilities been sat-isfactorily clarified relative to other public agen-cies. Social fund agencies have had difficultiesin cooperating with line ministries and localgovernments and existing mechanisms for coor-dination have not proven fully effective. Nega-tive effects for public sector processes havebeen reported for some of the older social fundsthat have allocated a significant share of publicexpenditure in sectors or activities for which lineministries or local governments are account-able. In these circumstances, the de facto per-manent status of autonomous social funds isdifficult to justify. Where such status might bejustified, the level of external support to thesocial fund should decline over time, with a

corresponding increase in local financing. Aslong as donor funding is readily available, theremay be little pressure to phase out the social fundagency, change its status, or to integrate it withmainstream government structures.

Recommendations

Strengthen integration of social fundsinto the Bank’s country assistance andsectoral strategies, and into clients’Poverty Reduction Strategy Papers(PRSPs) where relevant. • The rationale and objectives of Bank support

need to be clear and should drive the choiceof instrument rather than the other wayaround.

• Social fund projects should not displace pol-icy reform and should be designed as part ofa package of Bank support rather than as iso-lated interventions.

• In order to achieve compliance with Bank sec-toral policies and technical standards andconsistency with country sectoral and publicsector management reform strategies, theBank should improve country-team coordi-nation on social fund projects.

Give more attention to long-term impacts.• In design of social fund projects, the trade-

offs between speed and efficiency of sub-project processing and long-term impactsneed to be explicitly acknowledged andaddressed, and reflected in performanceindicators.

• The continuation or extension of Bankfinancing to a social fund should be based onevidence of that project’s development impact.

• The appropriate role and focus of the socialfund agency and its relationship with exist-ing institutions should be anticipated as faras possible from the outset because of the dif-ficulties experienced in changing the orien-tation once the agency is well established.

Ensure efficiency of resource allocation.• Social fund projects should ensure that invest-

ment decisions include a systematic articula-tion of the benefits as well as the costs of

O v e r v i e w

x x x i


x x x i i

alternative investments by the communityand/or local government concerned, whoshould also be charged with monitoring actualbenefits in relation to their expectations atsubproject appraisal.

• Stronger measures are needed to ensure thatbeneficiaries are adequately informed andconsulted on investment options, costs, andbenefits.

• The appropriate scale and scope of socialfund activities should be addressed at theproject design stage, and reassessed regularlyduring implementation, with reference tobudgetary processes and public expenditureanalysis.

Develop policy requirements.• Policy requirements for support to social

fund projects should be developed to provideclear criteria and procedures for Banksupport. These policy requirements shouldidentify conditions that indicate the appro-priateness or inappropriateness of using thesocial fund instrument and its strategic justi-fication within the CAS and PRSP, provideguidance on the country-specific informa-tion and analysis needed to ensure alignmentof the social fund with the institutional con-text, and identify viable transformation or exitstrategies.

1

Introduction

World Bank lending through social fund projects began in 1987and grew significantly through the 1990s. As of end-fiscal 1999,about $2.4 billion had been approved, and $1.3 billion disbursed,

for 66 projects in 42 countries.1 In fiscal year 2000, social funds accountedfor 9 percent of disbursements for the Program of Targeted Interventions(PTI).2 As of May 2001, more than 98 social fund projects operated in 58countries, with a total Bank investment of $3.5 billion (World Bank forth-coming). This growth, based in large part on the performance of social funds

in producing immediate outputs, is set to con-tinue at a rapid pace. As the great majority ofsocial fund projects have been followed up withcontinuing Bank support, often accompanied bya shift in objectives from emergency responseto the long-term provision of infrastructure as-sets and institutional development, this reviewfocuses on the institutional development im-pact of the projects and the sustainability of theinvestments they have financed. The growthand de facto long-term role of social fundsmakes it important to assess how successfullythey have been adapting to new demands.

Social Fund Projects Have DiverseFeatures . . .The distinguishing characteristic of social fundprojects is that they allow local stakeholders todetermine investment decisions. Rather thanbeing predetermined at project appraisal, such

decisions are made through subproject pro-posals they submit during the implementationof the social fund project. Social funds are de-fined as “agencies that finance small projects inseveral sectors targeted to benefit a country’spoor and vulnerable groups based on a partic-ipatory manner of demand generated by localgroups and screened against a set of eligibilitycriteria.”3 This definition implies five key char-acteristics: investments in small subprojects;multisectoral scope; targeted interventions; par-ticipatory, demand-driven features;4 and criteria-based eligibility. A further common characteristichas been a high degree of independence of theagency implementing the social fund from lineministries and civil service regulation.5 The fourmost common activities financed by social fundshave been construction or rehabilitation ofschools, piped water supply systems, healthfacilities, and roads (Annex D). These broad

11

similarities aside, social funds have variedgreatly in their degree of autonomy,6 in the em-phasis placed on targeting or on communityparticipation, and in the nature and range ofeligible subprojects. About one in four projectshas included microcredit components, while afew projects have begun working in new areassuch as infrastructure for street children, AIDS-related activities, and the protection of culturalheritage.

The first social funds supported by the Bankwere created in the Latin America and CaribbeanRegion (LAC) to mitigate the social impact ofstructural adjustment by generating temporaryemployment, primarily in small-scale infra-structure works. Social funds have since spreadto every other Region (Annex C). In Africa theyhave used a community-based approach pri-marily to remedy neglect of small infrastruc-ture. In East Asia (EAP) they were created toachieve rapid impact in response to civil con-flict or financial crisis through a broad range ofactivities. In the Middle East and North Africa(MNA) they have focused on mobilizing non-governmental resources to ease social impactsin the context of rationalization of large and in-efficient public sectors. And in Europe andCentral Asia (ECA), in the context of transitionsfrom heavily centralized systems and unprece-dented increases in poverty, they haveresponded to the pressing need for rehabilita-tion of deteriorating social infrastructure.

. . . And Their Objectives Have Evolved Although the majority of social funds were setup as temporary mechanisms, virtually all are stillin operation. Most projects have taken on mul-tiple objectives. The crisis-response function hascontinued to be evident in some projects, but inmost cases the focus has moved to longer-termobjectives such as improving small-scaleinfrastructure and supporting institutional de-velopment. Within institutional development,objectives have shifted from the central to thelocal government and community levels. The de-sign of social fund projects has evolved to in-clude more poverty targeting and to addressissues for subproject sustainability. At the sametime, social funds have been under pressure to

expand their operations and extend their activ-ities in new areas and directions.

Therefore, the Evaluative FrameworkMust Be BroadThe goals of this OED review were to assess thedevelopment effectiveness of social fund proj-ects, identify best practice examples/elements,and distill lessons and implications for futureBank support to such projects. Evaluation iscomplicated by the evolving objectives, design,and activities of social fund projects and by theabsence of Bank policies that establish standardsto which social fund projects should be held.7

This review assessed the performance of proj-ects against their stated objectives, using stan-dard OED criteria for development effectiveness:• Outcome: To what extent have the projects

been relevant, efficacious, and efficient? Thatis, to what extent have they been effective inproviding services consistent with communitypriorities, country needs, and the Bank’sCountry Assistance Strategy? To what extenthave they met their stated goals? How do so-cial fund project costs compare with theirbenefits, and with the costs and benefits ofother similar projects?

• Sustainability: To what extent will the proj-ect continue to produce net benefits as longas intended or even longer?

• Institutional Development: To what extenthave social fund projects improved coun-tries’ capacity to use their human, organiza-tional, and financial resources effectively?

Particular attention was given to the issues ofsustainability and institutional development,which have been under-researched in the socialfunds literature. In assessing the sustainability ofsubprojects, the review considered both the sus-tainability of the services produced by the in-frastructure and the sustainability of theinfrastructure itself. Assessment of institutionaldevelopment impact included the unintended aswell as intended institutional effects of the so-cial fund, both positive and negative, for cen-tral and local governments, the private sector,nongovernmental organizations, and the com-munity. An in-depth assessment of the social


2

fund participatory process was undertaken inorder to understand and assess institutional de-velopment effects at the community level.

Design of the review included theory-basedmethodology (Weiss 1998a,b) to allow exami-nation of mechanisms on which the social fundapproach depends (box 1.1) as well as projectoutcomes. The social fund approach must beadapted to meet its objectives, and this involvesmaking design tradeoffs (box 1.2). Data weregathered to assess the extent to which social fundprocesses work as intended.

The approach used combined quantitativeand qualitative research tools (described in AnnexB): (i) a literature review included an examina-tion of the conceptual as well as empirical liter-ature on social funds, sustainability, institutionaldevelopment, social capital, and organizationallearning; (ii) a portfolio assessment of the 66 so-cial fund projects in the portfolio as of end-fis-cal 1999 (Annex A) based on project documents,OED, and Quality Assurance Group (QAG) re-views, regional reviews, and task manager in-terviews; (iii) country cases, including householdsurveys, focus groups, and key informant inter-views as well as institutional analysis at the so-

cial fund agency, national, local, and communitylevels in four countries—Jamaica, Malawi,Nicaragua, and Zambia—and field research with-out household surveys in Argentina, Bolivia, andEritrea; (iv) a survey administered to the partic-ipants of the Second International Conference onSocial Funds held in Washington, D.C., in June2000; (v) a survey of Bank country directors andresident representatives in countries with socialfund projects; and (vi) a survey of Bank sectordirectors. The OED household surveys covered3,056 respondents in 845 randomly selectedhouseholds in 17 randomly selected socialfund–assisted communities, and 842 randomly se-lected households in 17 matched communities infour countries. In the former, 1,525 respondentswere interviewed (roughly 2 per household, thehousehold head and, where available, anotherhousehold adult of the opposite sex). Thisamounted to 284 respondents in 4 communitiesin Jamaica, 499 respondents in 5 communities inMalawi, 252 respondents in 3 communities inNicaragua, and 490 respondents in 5 communi-ties in Zambia. Among non–social fund house-holds, 1,531 respondents were interviewed (294in Jamaica, 493 in Malawi, 257 in Nicaragua, 487

I n t r o d u c t i o n

3

• Relevance and Poverty Targeting. The relevance of investmentsto the priority needs of the poor is to be ensured through theparticipation of local stakeholders in subproject selection froma multi-sectoral menu of choices. This menu can ensure self-targeting by focusing on investments that benefit the poor.Proactive measures may also be taken to target the poorestdistricts and communities.

• Efficacy and Efficiency. Autonomy of the implementing agencycan allow social funds to ensure appropriate staff incentivesand establish efficient and responsive operating procedures.The subproject-based approach, with standardization of pro-cedures for the subproject cycle, can also contribute to trans-parency, speed, and cost-effectiveness. Allocative efficiency(in sectoral and geographic distribution of resources) is to beachieved through the demand-driven mechanism (participa-tion and cost-sharing by beneficiaries are to ensure that theselected investments yield high returns) and through mech-

anisms to ensure coordination with national public expendi-ture priorities.

• Subproject Sustainability. Sustainability of investments is tobe enhanced by ensuring the quality of works, as well as en-suring local ownership through local participation in sub-project selection, implementation, and costs. This can besupported by measures to build local technical and institu-tional capacity for operations and maintenance and by coor-dination with other relevant agencies.

• Institutional Development Impact. Engagement with the so-cial fund may have positive impacts for central and localgovernments through, for example, learning-by-doing effects,demonstration effects, and competitive effects. Social fund en-gagement at the community level either through participationin project activities or through training may enhance local ca-pacity for collective action.

H o w S o c i a l F u n d A g e n c i e s E x p e c t t oA c h i e v e D e v e l o p m e n t E f f e c t i v e n e s s

B o x 1 . 1

in Zambia). This review reports all results byrespondent unless indicated otherwise. Thecountry-level percentages refer to responses fromthe sampled households in each of those coun-tries rather than to the population as a whole. The

review also drew from the data gathered for sixcountries for a detailed self-evaluation of theimpact of social funds managed by HD’s SocialProtection Unit and PREM’s Poverty AnalysisUnit, Social Funds 2000.8


4

The design of social fund projects involves a number of trade-offs, and the balance must depend on project objectives, the na-ture of services to be delivered, and the country context.• The local demand mechanism can allow local participation

in subproject decisions and management, but makes it diffi-cult to reach the poorest communities, which are often theleast competitive in preparing proposals. Equally, local de-cisionmaking may help to overcome information asymme-tries by bringing to bear local knowledge, but it does notlend itself to projects that require decisions to be made abovethe local level in order to deal effectively with significant ex-ternalities or to tap economies of scale.

• The subproject-based approach, limiting engagement with acommunity to implementation of a small, discrete investment,can make it possible to achieve speed and efficiencies throughstandardization, but may not, in itself, ensure that the neces-sary complementary inputs are in place for achieving a par-ticular development impact. In addition, learning-by-doing

benefits for any of the entities with which the social fund in-teracts are limited to activities of the subproject cycle.

• The multisectoral scope of social fund projects can give voiceto local actors and encourage local initiative by allowingcommunities to determine their own investment priorities,but it also places heavy demands on a single agency to meetthe varying technical and institutional requirements for de-livery of a wide range of goods and services.

• Autonomy of the social fund agency may enable it to adopt newprocesses and “rules of the game” for transparent and re-sponsive operation, but these gains may not be readily trans-ferable to permanent public sector agencies in the absenceof specific incentives and a policy environment conducive tochange. Conversely, autonomy makes necessary the creationof mechanisms to coordinate social fund projects with main-stream government agencies, in order to avoid negative im-plications for the management of public expenditure,accountability, and sustainability.

S o c i a l F u n d s F a c e T r a d e o f f s T o B eE f f e c t i v e

B o x 1 . 2

5

Outcome

What Have Social Fund Projects Aimed to Achieve?

Of the 66 social fund projects in the Bank’s portfolio, about one-thirdhave aimed to respond to economic or financial crises brought aboutby a structural adjustment program or post-conflict or natural dis-

aster situation.1 About three-quarters of the projects (48 of 66) have includedconstruction or rehabilitation of infrastructure facilities among their statedobjectives. Sixty-one percent of projects (40 of 66) included the objectiveof improving living standards.2 Social fund objectives are illustrated in table2.1, reproduced directly from a recent Bank review of social fund opera-tional manuals.

Project status and evaluation documents forsocial fund projects highlight tradeoffs betweensuch objectives as implementation speed, povertyreduction, participatory development, strength-ening local institutions, quality of physical out-puts, and sustainability. Yet the hierarchy ofobjectives, or the priority to be attached to eachof the multiple objectives, has not been sys-tematically addressed in project design. Whenthese tradeoffs become unavoidable during im-plementation, priority is usually given to meet-ing output targets, which have dominated theperformance indicators.

Of the 66 social fund projects in the portfolio,the appraisal documents for roughly 80 percentincluded performance indicators, predominantlyoutput indicators. Several implementation com-

pletion reports (ICRs) and project performanceaudit reports (PPARs) noted that the definedperformance indicators were not adequate totrack all relevant project objectives: there was afocus on physical outputs, while impact indica-tors of welfare or institutional development im-pacts and sustainability were neglected. Anexamination of project status reports for 26 proj-ects (see Annex B for sample selection) showedthat many of the performance indicators iden-tified in the staff appraisal reports (SARs) orproject appraisal documents (PADs) had notbeen reported. While almost two-thirds of theproject documents for social funds in the port-folio mentioned beneficiary assessments (to beconducted as one-time, periodic, or regularexercises), few details were provided about the

22

issues they would examine or the indicatorsthey would track.3

To What Extent Have Social FundProjects been Relevant, Efficacious, andEfficient? Of the 23 closed social fund projects as of end-fiscal 2000 (out of the total social fund portfolioof 66 projects examined for this review), 96percent were rated “satisfactory” on outcome byOED4 This outcome rating for social fund proj-ects compared favorably with outcome ratingsfor PTI projects (78 percent satisfactory) as wellas all World Bank projects (71 percent satisfac-tory).5 From OED’s survey of country direc-tors/resident representatives—although theresponse rate was low (13 responses of the 51requested)—the majority of respondents ratedsocial fund projects more favorably than otherprojects in the country portfolio on overall de-velopment impact. Responses from sector di-rectors, however (also with a low response rateof 7 responses of the 33 requested, in additionto 4 sector directors who indicated they were notinvolved in the social funds in their Region and,

therefore, were not in a position to answer thequestions), were less positive than those ofcountry directors/resident representatives, es-pecially on the overall development impact ofeconomic infrastructure components of socialfund projects (4 of 6 respondents to this ques-tion rated social fund projects worse than others).

Relevance at the Bank and Government LevelsHow well have social fund projects matched theBank’s assistance strategies and the poverty re-duction programs of governments? Given theirpoverty focus, social fund projects fit into theBank’s corporate priorities as expressed in Coun-try Assistance Strategies (CASs) and into thepoverty reduction programs of governments.6

The review of ICRs, PPARs, and QAG reviewsfound no instances in which overall relevancewas questioned.7 Rather, there were occasionalreservations concerning quality-at-entry issues,such as supporting too many initiatives in diversesectors, attempting to simultaneously addressshort-term and medium-term results, and lack ofclarity on the longer-term role or the phasing outof the social fund agency.


6

Africa Asia ECA LAC MNA

Social fund objective

Economic adjustment X X X X X X

Poverty alleviation X X X X X X X X X X X X X X

Decentralization X X X X X X X

Employment X X X X X X X X X X

Infrastructure development X X X X X X X X X X X X X

Participatory development X X X X X X X X X X X X X X

Strengthen local institutions X X X X X X X X X X X X X X

Increase welfare system coverage X X X X X X X X X X

Government & community partnerships X X X X X X X X X X X X X X

Private sector economic development X X X X X X X X X XSource: Weissman 2001. This document noted that the objectives were identified based on the formal statement of objectives in the operations manuals and a review of the contents ofthe social fund manuals. It also notes that some of the objectives are not stated explicitly, but they appeared in the description of the project menu. OED notes that in reviewing appraisaldocuments for these projects, it found that a majority of these objectives were also mentioned in the respective appraisal documents.

M o s t S o c i a l F u n d s H a v e M a n y O b j e c t i v e sT a b l e 2 . 1

Ethi

opia

Mal

awi

Togo

Cam

bodi

a

Thai

land

Alb

ania

Mol

dova

Rom

ania

Arg

entin

a

Bel

ize

Jam

aica

Nic

arag

ua

Egyp

t

Wes

t Ban

k/G

aza

Yem

en

The recent OED evaluation of Bank efforts tosupport poverty reduction found that CASs wereoften weak at locating Bank activities within astrategy rooted in country context (OED 2000a).For social sectors, undue emphasis was foundin CASs on delivering quantity rather than qual-ity of services—an issue particularly relevantfor social funds that are focused on infrastruc-ture delivery. In examining CASs for this re-view, a tendency was found for CASs to focuson some of the particular activities that the so-cial fund would finance. The questions of howsocial funds would develop infrastructure orsupport other activities strategically within theoverall effort to reduce poverty and improve liv-ing conditions, or whether they were the mosteffective means of reducing poverty, were gen-erally not addressed.8

The strategic rationale for a social fund needsto be debated at the CAS and country-team levelwhere the opportunity costs of lending thougha social fund can be fully considered. The in-creasing emphasis placed on the quality of serv-ice delivery in social sector strategy papersmakes it particularly important to clarify thestrategic role of the social fund. The case for mak-ing a social fund the preferred instrument forpoverty reduction in specific circumstancesneeds to be weighed and argued—not simplyannounced—by the Bank in the CAS, and by thegovernment in the Poverty Reduction StrategyPaper (PRSP), where relevant. More is neededto make the case that social fund projects are pre-ferred to other means of poverty reduction.Overall, country directors/resident representa-tives and sector directors who responded to theOED survey thought that social funds were bet-ter than other projects on short-term poverty re-duction, but worse over the long term.9

Relevance at the Community Level Have subproject investments been relevant tocommunity priorities? OED’s household surveysin four countries found that subproject selectionhas been less a unified expression of commu-nity will than a process in which prime moversusually determine project choice. Even where vir-tually the whole community participated in someaspect of the subproject, the community as a

whole did not necessarily drive project choice.While a vast majority of beneficiaries have beensatisfied with the subprojects financed,10 theOED household surveys found that the sub-project selection process could not be countedon to meet the highest priority problem of themajority of households, even if the relevant in-vestments were on the subproject menu. TheOED household survey data for 284 respon-dents in Jamaica, 499 in Malawi, 252 inNicaragua, and 490 in Zambia showed the fol-lowing results (Annex M, table M.1):• The top problem was addressed for 27 per-

cent of the respondents in Jamaica, 34 per-cent in Malawi, 23 percent in Nicaragua, and22 percent in Zambia.

• When only the problems that could be ad-dressed by an investment on the subprojectmenu were considered, the top priority wasmet for 31 percent of the respondents in Ja-maica, 47 percent in Malawi, 26 percent inNicaragua, and 28 percent in Zambia.

• One of the top three priorities was met for 42percent of the respondents in Jamaica, 52percent in Malawi, 35 percent in Nicaragua,and 38 percent in Zambia.

• OED household survey results by communityshowed that considering only the problemsthat could be addressed by an investment onthe subproject menu, in 9 of 17 communitiesthe top-priority problem was addressed. Infour of the remaining eight communities, thesecond-ranked problem was addressed.

In order to avoid biased household responses,the OED household surveys did not directly askbeneficiaries if the subproject they had alreadyreceived was their priority. Instead, OED asked“in your personal opinion, what were the threebiggest problems facing your community (inthe year prior to the approval of the subproject)?”This open-ended question was asked immedi-ately following questions relating to basic so-cioeconomic and demographic issues, wellbefore the community was asked about the par-ticular subproject financed by the social fund.Then, OED compared the household responsewith the subproject actually financed by thesocial fund in that community. This analysis

O u t c o m e

7

provided an indication of the extent to which thebiggest community problems were (or werenot) addressed by the social fund.11

Several factors were found to influence thedemand-driven process, including informationasymmetries and the role of prime movers, whowere critical to the mobilization of support andpreparation of a successful subproject proposaland whose interests were determined by the na-ture of their positions. It is natural and appro-priate that prime movers should bring projectideas to the community (leaders play a crucialrole in a demand-driven process). At the sametime, mechanisms are needed to ensure thatthe ideas of the leaders are also the most im-portant ones for the community as a whole.

Typically, social fund staff visited the com-munity after they had received communicationfrom the community of its interest in a particu-lar type of subproject or after they had receivedan application. By this time, prime movers hadalready mobilized support for particular sub-projects (Annex O).12 OED field research founda bias in many cases toward subproject invest-ments in sectors where prime movers were ac-tive. Other sources have cited related findingsincluding the importance for subproject selec-tion of existing community organizations in theeducation sector (Frigenti and Harth with Huque1998), or differences in views between com-munity leaders and members (Salmen 1987).The extent to which addressing the biggestproblems is important depends on the realisticpotential of other development agents to addresshigher priorities that are not addressed by thesocial fund.

Many social fund projects have predomi-nantly delivered schools and health clinics.13

The existence of prime movers and establishedlocal organizations in the form of headmasters,parent-teacher associations, health workers, orhealth user committees in education and healthhas been an important factor in making these sec-tors heavily subscribed in social fund projects.14

The requirements of effective subproject for-mulation, submission, and implementation havemeant that enterprising prime movers often be-came the operators of the demand-driven mech-anism. The community model of social funds as

currently practiced has been generally less wellsuited for sectors where there was no clearchampion or existing organization. Effective ex-ecution of water and road projects is likely tobe facilitated by the formation of community-based organizations dedicated to these sectors—for example, road or water associations. Sufficienttime and resources are critical for fostering thegrowth of such organizations, which need to berooted in existing social structures.

Many other factors can affect the extent towhich community priorities are met. For exam-ple, it was noted for the second Malawi socialfund project that the criteria for upfront com-munity contribution have met with problems inthose parts of the country where the natural en-vironment does not allow people to produce fire-baked bricks or acquire materials such as riversand and stones for construction purposes. Insuch places, it has become difficult for these com-munities to apply for their first-choice projectsthat require substantial contribution in local ma-terials; instead they have ended-up with bore-holes, which do not require much materialcontribution. The Zambia social fund faced a sim-ilar situation. Counterpart contribution require-ments meant that communities chose subprojectswith a high unskilled labor component, ratherthan others that would have required a largercash contribution.

It is also possible that the investment to addressthe highest priority problem may be technicallyunviable and may therefore not be provided bythe social fund, even when on the subprojectmenu. Other factors can also influence commu-nity choice, as acknowledged in the social fundsliterature (see Annex O, “Participation and Sub-project Execution”). For one thing, communitiesmay request what they think they are likely to getrather than what they most want and are willingto pay for. In many countries, social funds areknown primarily as builders of schools, so alarge proportion of the community requests theyreceive are for schools. Frigenti and Harth withHuque (1998) note that “since the MPU dependedon organized groups, it funded whatever tech-nically viable proposal came through. The neg-ative side of this was that MPU soon becameknown as an agency that funds schools, starting


8

a cycle of funding education and getting proposalsonly for education.”

The results of OED’s community-level fieldresearch showed how facilitation should work:motivating local structures to take control of theprocess rather than creating parallel arrange-ments not rooted in the existing social structure.Two examples of social funds that aim at im-proving facilitation skills among local leadersare Moldova (Annex N, “Moldova: Training‘Prime Movers’ ”) and Jamaica. In Jamaica, 24leaders were successfully trained as facilitatorsin 1999 and 77 in 2000. These facilitators areemployed by the Social Development Com-mission or by JSIF as the need arises (Dijkstraand Green 2000).

It is possible that wider planning processes(as opposed to exclusively community-centeredones), involving local governments and other ac-tors, may lead to a shift in investments away fromschools. For example, in Albania or Bulgaria,where local governments have played a keyrole in subproject selection, road and water sub-projects have dominated; and in Nicaragua, theintroduction of the municipal-level micro-plan-ning process (see Annex N, “Nicaragua: Broad-based Local Planning”) resulted in the expressionof more varied community demand, includinga higher demand for water than in past years.Whether such a managed investment planningprocess at the local government level leads toenhanced relevance of investments to commu-nity needs merits further analysis. It is also pos-sible that the use of participatory research (as inJamaica) or other traditional modes of deliverymay result in the identification of subprojects thatare more relevant to community needs (Moserand Holland 1997;15 Tendler 2000a; and Raoand Ibanez forthcoming16).

Social Funds 2000 generally addresses theissue of community priorities and preferences byasking about the importance or priority of thesubproject in reference to the already-financedsocial fund subproject. The overall result was thatcommunities considered the investments theyhad received as meeting their most importantneeds (Annex E, table E.4). When OED applieda similar methodology, the results were alsopositive. For example, when asked, in refer-

ence to the chosen social fund subproject,“would you have preferred that another projectwas chosen instead?” a vast majority of respon-dents in the four countries answered “no.” Thisresponse confirms the OED conclusion that a vastmajority of beneficiaries were satisfied with thefinanced subprojects. It is not inconsistent withOED household survey results on the extent towhich the biggest community problems weremet, in that such a response may have been in-fluenced by beneficiary perceptions of whattheir options were and what they assumed theycould realistically have gotten. In its generalconclusion, Social Funds 2000 signals the needto ensure greater relevance to community needsand to move toward more participatory localplanning in order to identify investment priori-ties within the broader goals and activities of acommunity.

Efficacy—To What Extent Have Social FundProjects Met Their Stated Goals?17

Responding to Crises. Several reviews, and no-tably QAG data, have found that social fund proj-ects have been relatively effective in meetingtheir short-term objectives in response to crises.They have not been equally effective in re-sponding to all types of emergencies or to alltheir aspects. In the context of structural ad-justment programs, for example, there havebeen difficulties in reaching the precise groupsthat were directly hurt (for example, miners inBolivia, as noted in Jorgensen, Grosh, and Schac-ter 199218). It has also been contended that thetypes of activities often supported by socialfunds, such as school rehabilitation, have littledirect impact on income poverty, and so little,if any, consumption-smoothing effects.19 Em-ployment effects for private contractors andconsulting firms are positive, but relatively small.The overall employment impact of social fundshas been limited (Cornia 1999), and most of thejobs created by social fund projects have beentemporary.

But social funds have often had a positiveimpact in post-conflict situations. They have im-proved public confidence by producing visibleoutputs quickly. They have helped to mobilizedonor and community resources. The Cambo-

O u t c o m e

9

dia social fund met the challenges of the post-conflict situation well, recognizing the tradeoffsup-front and making project design choices inlight of country conditions (see Annex N, “Cam-bodia: Meeting Post-Conflict Challenges”). In-deed, social fund projects are increasingly beingcalled upon to revert to their crisis-responsefunction even after they have moved on to beingdevelopmental organizations, as in Hondurasand Nicaragua after Hurricane Mitch. In these andother cases, social funds switched goals, rose tothe occasion, and provided a distinctly effectiveemergency response. While this sometimes tooka toll on regular social fund programs and or-ganizational efficiency, governments rarely hada ready institutional alternative to deal with theconsequences of emergencies. Nevertheless,some social funds have tended to develop a path-dependence that has inhibited a quick return tostrict adherence to operational procedures oncethe emergency has ended. Difficulties in re-adjusting to normal discipline should not beunderestimated.

Expanding or Improving Infrastructure.Social fund projects have demonstrated re-markable success in disbursing rapidly for in-frastructure expansion and/or improvement.The evidence is strongest for schools, whichhave been the focus of many social fund proj-ects, but they have also delivered infrastructurein other sectors. The achievement of social fundprojects in delivering infrastructure should notbe underestimated—donors and governmentsoften spend large sums of money on infra-structure investments that do not materializeeven in relatively accessible communities. As aresult of the additional infrastructure, facility ac-cess and utilization have improved (box 2.1;Annex E, table E.1).

However, instances of underutilization of so-cial fund facilities have also been reported, es-pecially with respect to health clinics wheresupplies and equipment shortages have oftenprevented full capacity utilization.20 Accordingto Social Funds 2000 background research forNicaragua, despite the increase in utilization, therates remain far below the capacity of ruralhealth posts and reflect underutilization problemsin FISE and non-FISE posts alike. A majority of

the respondents in OED’s survey of country di-rectors/resident representatives and sectordirectors considered “visible outputs on theground” as one of the top three attractive fea-tures of social fund projects.

Overall, education and health infrastructure,for which design and construction can be read-ily standardized, have experienced fewer im-plementation problems than water and roads. ASocial Funds 2000 summary noted that with re-spect to water and sanitation: “systems appearedto function reasonably well overall, though de-sign issues are critical to system performance,water supply is variable, and maintenance issuesare often problematic.”

Outcomes and Welfare Impacts. What ef-fect has all this infrastructure development had onwelfare? The main source of such data is thebackground research for Social Funds 2000. Someindicators of outcome and welfare impact regis-tered a positive social fund impact (higher primaryenrollments in Nicaragua and reduced infant andchild mortality in Bolivia), a few showed a neg-ative social fund impact, and several indicators reg-istered no significant social fund impact comparedto the control group (box 2.2 and Annex E, tableE.2). Some findings, notably on child malnutrition,are difficult to interpret.21 Overall, the picturethat emerges is a varied one.

One reason for the varied social fund impactson outcomes and welfare may be that socialfunds respond to discrete subproject proposalsfor infrastructure (such as health clinics) ratherthan being driven by the objective of achievinga specific development impact (such as im-proved health) and ensuring that all the neces-sary complementary ingredients to achieve it arein place (either through its own activities orthrough other interventions). A review of socialfund beneficiary assessments (Owen and VanDomelen 1998) noted that several social funds(those in Armenia, Ecuador, and Peru) identifiedthe need for complementary actions within typesof projects (for example, educational materials,equipment). Social Funds 2000 found instancesof incomplete facilities: in Nicaragua, pipedwater was available in less than half of both so-cial fund and control facilities, in Hondurasabout 40 percent of both social fund and con-


1 0

trols did not have piped water, in Bolivia onlyabout one-third of facilities (both social fund andnon–social fund) had access to electricity (withsocial fund facilities having somewhat less ac-cess than the controls). The same evaluation alsofound that in Nicaragua, only 35 percent of so-cial fund schools with piped water installationshad regular water service during school hours.

In addition to the delivery of physical infra-structure, “software” inputs are often necessaryto ensure optimal facility use and impact22 whereoptimal utilization requires behavioral changeson the part of users—for example, in using non-traditional health services or encouraging school-

ing of girls—or significant training and capacitybuilding. So far, such software investments havebeen a minor part of the social funds’ productline, and training and capacity building com-ponents have typically not received adequate at-tention. Social Funds 2000 background researchexplains the lack of impact in specific cases asfollows: the lack of health impact of water in-vestments in Honduras can be largely explainedby the relative focus on rehabilitating existing sys-tems, versus installation of new systems; thelimited impact of sewerage investments on healthcan be attributed to the overall low connectionrates. The evaluation points to linkages between

O u t c o m e

1 1

Schools• The Peruvian social fund, FONCODES, constructed or reha-

bilitated more than 9,000 classrooms during 1992–98.• In Malawi, MASAF has financed the construction of about 1,600

classrooms.• In Honduras, between 1995 and 1998, FHIS built 503 new pri-

mary schools, 2,326 new classrooms, and improved 4,163classrooms.

• In Nicaragua, FISE helped to increase the number of class-rooms from, on average, 4.6 to 6 per school.

• Egypt’s Social Fund for Development established only 7 class-rooms out of the target of 150 in 1994, but 2 years later it es-tablished 1,417 in a single year and more than 1,500 thefollowing year.

• During 1995–99, Cambodia’s social fund project supportedthe building of 4,411 new classrooms and the supply of 67,430pieces of classroom furniture.

Health Facilities• In Honduras, between 1994 and 1998, FHIS built 127 new rural

health posts and 40 new urban health posts, remodeled 90 ruralhealth posts and 108 urban health centers, and built 1 newmother and child clinic and remodeled 8 such facilities. ForFHIS rural health posts this led to an average of 11 visitsdaily, compared with the national average of 9 each day. Forall FHIS health centers, there was an average of 54 consul-tations each day, compared with the national average of 34.

• In Nicaragua, between 1993 and 1997, FISE health posts reg-istered a significantly larger increase in average number of

daily visits, from 11.3 to 17, compared with an increase of be-tween 8.8 and 14 visits in non-FISE posts.

• In Zambia, between 1993 and 1998, the SIF health posts reg-istered an increase in (i) the hours of operation for all serv-ices, (ii) number of cases (declined in non-SIF), (iii) maternaldeliveries and child attendance.

• Cambodia’s social fund financed the construction of 78 healthcenters, 2,767 water wells, 1,535 latrines, 24,935 cubic metersof dikes, irrigation facilities for 3,800 hectares of agricultural land,and rehabilitation of 17,216 square meters of bridges/ culverts.

Water• Under MASAF I and II in Malawi, more than 2,000 boreholes

were sunk.• In Bolivia, between 1993 and 1997, the proportion of households

with piped water increased by 16 percent in Chaco and 10 per-cent in Resto Rural. Frequency of water availability decreasedfrom 21.95 to 19.38 hours in Chaco and increased from 18.49to 21.15 hours in Resto Rural. Distance to water source fell bymore than 50 percent.

• In Honduras, between 1994 and 1997, FHIS built 2,815 new con-nections and upgraded 46,000 connections. FHIS investmentsincreased water availability, reduced average expenditureson water, and reduced time spent collecting water.

Road• The second Madagascar social fund rehabilitated 185 health

centers and 967 kilometers of rural road.• The Albanian Rural Poverty Alleviation Pilot Program financed

278 kilometers of rural roads and 524 meters of footbridges.

Q u a n t i t y , A c c e s s , a n d U t i l i z a t i o n o fP h y s i c a l I n f r a s t r u c t u r e H a s I m p r o v e d

B o x 2 . 1

Source: OED portfolio assessment and Annex E, table E.1. Additional access and utilization indicators are presented in Annex E, table E.1.

household characteristics and outcomes (for ex-ample, the linkage between parents’ educationor household poverty levels and children’s ed-ucational outcomes), and draws the implicationthat demand-enhancing measures such as out-reach are a necessary complement to the pro-vision of infrastructure.23

Although social fund projects are multisectoral,in practice they have not necessarily operated ina holistic24 way in each community. For exam-ple, they usually finance only a single investmentin each community, which may increase thetotal number of subprojects and beneficiaries, butat the expense of depth of involvement and de-velopment impact. The need for a greater focuson development impact is also a policy impli-

cation of Social Funds 2000, which recommends:“Greater attention not just to successful projectimplementation, but to the impacts from thoseinvestments, including involvement of commu-nities in the up-front identification of expectedimpacts” (World Bank 2000h).

Poverty Targeting. Benefiting the poor is theoverarching objective of social fund projects: “so-cial funds establish menus, procedures and tar-geting criteria to support investments benefitingthe poor” (Jorgensen and Van Domelen 1999,p. 7). The poor, poorer, poorest, or poverty arementioned in the objectives of more than three-quarters of social fund projects. Furthermore, thepoor, or some category of poor people, are anexplicit target group in the majority of cases


1 2

These results compare outcomes and welfare impacts in socialfund communities versus non–social fund communities and reportstatistically significant results (at 90 percent confidence level).

Education: In Nicaragua and Zambia (using pipeline match),enrollment rates increased significantly for the social fundschools compared with the control facilities, but social fundschools in Zambia (using propensity score match), Bolivia, Hon-duras, and Peru (Apoyo) registered no significant impact com-pared with control facilities. In Peru (Paxon-Schady), districtsthat received the most social fund expenditures for school im-provements achieved the largest gains in primary school en-rollments, although these enrollment gains could only beconfirmed at the household level among the poorest populationsin the rural survey. The percentage of students in the correctgrade for age was significantly higher for Honduras, Zambia(using pipeline match), and Nicaragua (for students in grade 1).An insignificant difference was noted for Zambia (using propen-sity score match), Nicaragua (students for all grades), and Peru.There was improvement in absenteeism in Peru, but not in Bo-livia or Nicaragua, using pipeline match, but it significantlyworsened in Nicaragua using propensity score match. No sig-nificant impact was noted for student academic achievementmeasured by math and language tests in Bolivia, the only placewhere this was assessed.

Health: The incidence of diarrhea was similar between so-cial fund and control communities in Bolivia, Nicaragua (using

pipeline match), and Zambia. In Nicaragua (using propensityscore match), the incidence of diarrhea was higher for the so-cial fund communities compared to non–social fund communi-ties. The number of social fund beneficiaries (among the sick)that sought primary health services was significantly higher inHonduras, but similar in Bolivia. Infant and child mortality, meas-ured only in Bolivia, registered a significant decline for socialfund communities compared with control communities.

Water: Incidence of diarrhea declined over time for socialfund communities in Bolivia (Chaco region) and Peru. The im-pact on incidence of diarrhea of social fund water subprojectsin Bolivia (Resto rural region), Honduras, and Nicaragua was verysimilar to that with control water subprojects. There was a de-cline in stunting in children under six in Nicaragua. Child mor-tality has registered a significant decline for social fundcommunities since the start of the subproject in Bolivia andPeru.

Latrines: In Honduras (bivariate analysis) and Peru, the in-cidence of diarrhea declined in social fund communities com-pared with control communities. In Nicaragua, an insignificantdifference was noted in the incidence of diarrhea between so-cial fund and control communities.

Sewerage: In Peru and Honduras no differential household-level health impacts were found. In Nicaragua, the increase inaccess to sewerage led to positive health impacts at the com-munity but not household level.

O u t c o m e s a n d W e l f a r e I m p a c t s A r e V a r i e d B o x 2 . 2

Source: Social Funds 2000 data.

(“poor” in 80 percent of projects, “poorest” in 46percent of projects, “vulnerable” in 44 percentof projects, and “low-income” in 10 percent ofprojects).25 Other target groups are “unem-ployed” (in 20 percent of projects), “indige-nous” (in 10 percent of projects), and “women”(in 61 percent of projects).26 However, system-atic data on targeting outcomes at the householdlevel are available for only a small number ofcountries. The available data, based largely onSocial Funds 2000 background work for Arme-nia, Bolivia, Honduras, Nicaragua, Peru, andZambia, allow a number of conclusions with re-spect to targeting, detailed below.

Geographic targeting. The share of poor dis-tricts in social fund resources is greater than theirshare in the population, meaning that socialfunds achieve a progressive allocation of fundsacross districts (figure 2.1). The poorer districts’share of resources has generally increased overtime as a result of better targeting procedures(such as the use of poverty maps and the adop-tion of allocation formulae), more extensiveoutreach activities made possible in part by de-

centralization of the social fund (as word of thesocial fund has spread), and a shift from emer-gency mandates to longer-term goals when in-creased attention was given to targeting.However, district allocations are still best de-scribed as mildly progressive. Peru is the only ex-ception, having succeeded in devoting asignificant share of its resources to the poorestdistricts with continual fine-tuning of the targetingmechanism and a focus on rural areas that ledto improved performance over time. Annex N(“Peru: Achieving Success in Targeting”) dis-cusses some of the factors accounting for Peru’ssuccess in poverty targeting. Armenia has re-gressive targeting among the wealthier districts—the allocation of expenditures is pro-poor amongthe lower 40 percent of districts.

Household-level targeting. Are the poor over-represented among social fund beneficiaries?Data on household incidence largely indicate thisto be the case, although targeting is very mildlyprogressive or neutral and performance variesby country (figure 2.2) and sector. In Honduras,Nicaragua, and Peru (education), the share of

O u t c o m e

1 3

S o c i a l F u n d s H a v e M i l d l yP r o g r e s s i v e G e o g r a p h i c T a r g e t i n g( c o n c e n t r a t i o n c u r v e s a r e c l o s e t ot h e 4 5 - d e g r e e l i n e e x c e p t f o r P e r u )

F i g u r e 2 . 1


poor households benefiting from a social fundinvestment was 58 percent, 55 percent, and 66percent, compared with 54 percent, 48 percent,and 50 percent of that population nationwide,respectively. In Zambia, 15 percent of socialfund beneficiaries were in the poorest incomedecile, but targeting based on the poverty linewas similar to a random distribution. The shareof extremely poor households benefiting fromthe social fund was about the same as the shareof such population nationwide—71 percent ofthe social fund beneficiaries were classified aspoor and, indeed, just over 70 percent of Zam-bians are poor. In some sectors in certain coun-tries, the share of the poor among beneficiariesis less than the share of the poor in the popu-lation as a whole. Latrines and health clinics havetargeted the poor best, education and water in-vestments were pro-poor, and sewerage projectswere regressive.

Why is geographic targeting more progres-sive than household targeting? Available datamakes comparisons difficult. The problem inthese data is twofold. First, household incidenceapplies to a specific year, while the geographic

data are for a number of years. Second, thegeographic data are for expenditure, but house-hold data refer to beneficiaries, imparting an un-known bias, depending on differential unit costsand geographic distribution by sector. But, if thedata are accepted, they illustrate that other thanin one country, the distribution of social fund re-sources is more progressive on a geographic thanon a household basis (see figures 2.1 and 2.2).This discrepancy is particularly striking in Peru—for education investments, the strong progres-sive geographic allocation is not matched at thehousehold level. An explanation in the case ofPeru is that households above the mean districtincome are more likely to benefit from the so-cial fund than those below the mean (Paxson andSchady 1999). The result from Peru and the dis-crepancy between geographical and householdtargeting indicate that poor households are lesslikely to receive resources than poor districts.One of the implications drawn by Social Funds2000 from its poverty targeting analysis is thatbetter local coordination efforts can improveidentification of poor communities within bothwealthier and poorer districts, and that for those


1 4

■

■

■

■■

■

■

■

■

■

●

●

●

●

●

●

●

●

●

●

▲

▲

▲

▲

▲

▲

▲

▲

▲

▲

◆

◆

◆

◆

◆

◆

◆

◆

◆

◆

■

■

■■

■

■

■

■

■

■

0 10 20 30 40 50 60 70 80 90 100

0

10

20

30

40

50

60

70

80

90

100Cu

mul

ativ

e be

nefic

iarie

s

Cumulative population

■ Armenia

● Honduras

▲ Peru

◆ Zambia

■ Nicaragua

Line of equality

S o c i a l F u n d s H a v e N e u t r a l o rV e r y M i l d l y P r o g r e s s i v e H o u s e h o l dT a r g e t i n g

F i g u r e 2 . 2


poor communities that still have difficulty ac-cessing resources, additional proactive meas-ures, including up-front assistance in communityorganization, waiving of community counter-part requirements, and provision of technical ex-pertise may be necessary to decrease theinequality of results between areas and to en-sure that all poor communities can access pro-gram resources.

Although comparisons are difficult given dif-ferences in program objectives and context, so-cial fund targeting performance comparedfavorably or was similar to that of other programswith comparative data. In Bolivia, municipalitieswith higher poverty indices received a highershare of social fund resources than municipal ex-penditures. In Peru, FONCODES achieves a bet-ter geographic targeting performance than twodirected programs, PRONAA and INFES, al-though the former has a better performancethan FONCODES in targeting within districts. TheArmenia social fund performed in the mid-rangeof national social assistance programs, and inHonduras, the incidence of direct beneficiaryhouseholds among the lowest quintile of house-hold income was 37 percent, about middle ofthe range of performance of 30 targeted socialprograms throughout the LAC Region.

Overall, the results show a mixed record withrespect to targeting. Social funds reach both thepoor and the poorest on a district and house-hold basis. A significant proportion of socialfund resources benefit the non-poor, however.The share of social fund investments going tonon-poor households ranges from 29 to 45 per-cent (where the proportion of non-poor is basedon the national poverty line).27

These social fund poverty targeting outcomesare a function of both promotion and outreachand targeting mechanisms. Promotion is vital toensure that all communities are aware of the so-cial fund and so apply for subprojects. Socialfunds have developed promotion through anumber of channels, which can be very effec-tive in creating awareness. The Malawi socialfund has tailored its promotional campaigns tospecific audiences and used various media chan-nels to disseminate information (see Annex N,“Malawi: Effective Promotion and Outreach”).

The OED household survey in Malawi found that98 percent of the households had heard of thesocial fund, compared with only 39 percent inneighboring Zambia, where promotion effortshave been less intensive. But promotion alonemay not be enough, as the poorest communi-ties are very likely to be those least able toapply for assistance. Facilitation of communityapplications is also required.

Targeting may take place through a numberof mechanisms. Social funds are inherently self-targeting in that they finance activities of inter-est to the poor and whose public nature (for mostsubprojects) means the non-poor cannot ex-clusively capture the benefits. But explicit tar-geting mechanisms are also used, usually at thedistrict level. These mechanisms may be eitherproactive or reactive.28 Proactive targeting allo-cates funds in a pro-poor manner either by ex-cluding the better-off (a cut-off) or using aprogressive allocation rule. Reactive targetingwaits for requests from communities, but favorsthose from poor districts by applying eligibilitycriteria, according them greater priority, and/orby reducing the required level of communitycontribution. Reactive targeting can be expectedto have less favorable targeting outcomes thanproactive targeting. But for proactive targetingto work, facilitation may be required to ensurethe uptake of funds. In Panama, slow uptake offunds by poorer regions allowed funds to be re-allocated to better-off areas as a result of polit-ical interference. Experience has shown that aninitial pro-poor allocation of funds may be re-allocated to the benefit of better-off districts. Andif targeting is just at the district level, it remainsreactive within the district, leaving unresolvedthe problem of reaching poorer communities.Some social funds, but not all, have taken stepsto address these issues. Annex F provides a de-tailed discussion of poverty targeting issues andbox 2.3 outlines lessons on allocation rules. So-cial Funds 2000 concluded that a key designissue for social fund projects is how to reduceleakage, and pointed out that this may mean re-ducing access by better-off regions, introducingintra-district targeting procedures to identify thepoorest communities within districts, restrictingitems from the menu that tend to have higher

O u t c o m e

1 5

rates of leakage to the non-poor, and usingother means to increase the concentration of ben-efits to the poor.

Gender Effects. The impact of social fundsdifferentiated by gender seems varied. Thereare some cases in which females appear to havebenefited particularly from social funds. Forexample, Social Funds 2000 background re-search indicates that female utilization of healthservices, either general visits or specific services,increased significantly in Bolivia (prenatal serv-ices), Nicaragua (female visits), and Zambia (de-liveries in maternity wards), the only threecountries that included gender-disaggregateddata on health services. There were also exam-ples in which no significant pro-female social

fund effects were found. In Zambia, social fundinterventions had no significant effect on fe-male enrollment; in Nicaragua there was no sig-nificant difference by gender on the impact ofdeclines in age at first grade in 1997 and onchange in the percentage of girls enrolled in so-cial fund schools between 1993 and 1997.

Social funds have not normally been proac-tive in addressing gender disparities in em-ployment practices, community decisionmakingprocesses, or other areas of their activities (OEDportfolio assessment). Prevailing disparities inemployment and wages have been perpetu-ated, in that women have been less likely to ben-efit from paid employment in constructionfinanced by the social fund and more likely to


1 6

Reliance on reactive targeting makes it more likely that poorareas will be under-served, unless an aggressive outreach strat-egy is pursued. Even with proactive targeting, however, promotionand outreach are necessary to ensure that poor areas reachtheir spending targets. The following rules are relevant.• Cut-off: Should the allocation rule be applied across all dis-

tricts, or should it be combined with a cut-off to exclude theleast-poor districts? If a cut-off is not used and the least-poor districts are not excluded, adequate mechanisms areneeded for targeting within districts (self-targeting may notbe sufficient for this purpose).

• Progressivity: The ZAMSIF formula for district allocationsgives much more weight to the poorest districts than to thosethat are less poor: the poorest of all receive $30 a head andthe least poor only $1.30. This differential could be narrowedto $11 and $2.30 respectively through a small change: raisingthe poverty term in the allocation rule formula to 0.5 from 1.(This would make the allocation less pro-poor, although thiseffect could be balanced by the introduction of a cut-off. Thismight be necessary because of lack of absorptive capacity andbecause there are poor people in non-poor districts.)

• Setting aside a reserve: The poorest districts are likely to beamong those with the least capacity, so spending targets areleast likely to be met there, at least where communities arethe only eligible subproject sponsors. For these reasons it isperhaps unwise to allocate all expenditures with the allo-

cation rule, since this will eventually force a decision to“take” from some districts to “give” to others in order to meetaggregate spending targets (as has happened in Nicaragua).It will be better to allocate, say, 70 percent by the rule, keep-ing 30 percent in a pool to allocate to districts that reachtheir expenditure target (perhaps restricting the pool to poorerdistricts). Strong facilitation among the poorer districts wouldimprove their chances of accessing resources from the pool.Multiyear allocations in poor districts may be an option, al-though this may not solve the problem unless facilitation isalso provided.

• Facilitation: To ensure that poor districts can absorb as manyresources as possible, promotion and outreach efforts needto be intensified in these areas, including assistance in sub-project preparation.

• Choice of indicators: The targeting literature suggests that theindicator used to allocate resources should be related to thepurpose of the funds. Targeting rules based on income-poverty,as used in Nicaragua and Zambia, for example, may not be thebest way to allocate what are mostly educational resources.In the case of education and health infrastructure, a facilitiesmap may be just as important as a poverty map.

• Incorporating lessons learned: Social fund projects and otherprojects in several countries are developing some of these tar-geting rules. A review of targeting mechanisms would beuseful to enable all social fund projects to adopt good practice.

L e s s o n s f r o m E x p e r i e n c e : A l l o c a t i o nR u l e s f o r R e a c h i n g t h e P o o r

B o x 2 . 3

Source: OED analysis.

provide the voluntary labor required as part ofthe community contribution (ILO 1997).

In the participatory process, in almost allcases, women have been under-represented inproject committees, even where social fundshave tried to impose quotas. OED household sur-veys indicated some gender differences inMalawi and Zambia, but few in Nicaragua andJamaica. For example, in Malawi and Zambia,compared with men in the social fund–assistedcommunities, a statistically smaller percentage ofwomen had knowledge about the social fund intheir countries, and about the person(s) re-sponsible for operation and maintenance of thefacility. Women also spoke less compared withmen in the social fund–assisted communities inJamaica, Malawi, and Zambia and experiencedlower capacity building in both Malawi andZambia (Annex M, table M.26).

Some completion reports note the lack of agender perspective and difficulty in reaching thepoorest and most marginalized groups. Evenwhere the weakest groups benefit from socialfund investments, this is not synonymous withtheir having a voice in subproject selection or theoverall decisionmaking processes. A beneficiaryassessment for the Zambia Social Recovery Pro-ject II, for example, noted continued difficulty inincorporating women and youth in the formal de-cisionmaking structures of the community.

With regard to respondents in the bottomquartile in each community, OED householdsurveys indicated that they were less likely thanrespondents from the top three quartiles to haveheard of the social fund in Malawi and Zambia,and they spoke less in the meetings in Jamaicacompared with the respondents from the topthree quartiles. The respondents from the bot-tom quartiles experienced lower capacity build-ing in Malawi and Zambia, perhaps indicatinglower levels of participation in subproject ac-tivities (Annex M, table M.27).

Efficiency—Do Social Fund Project BenefitsExceed Their Cost? More So Than OtherProjects?Comparison of costs and benefits. Data arelimited for the assessment of the economic effi-ciency of social fund projects. At the project ap-

praisal stage, the range and composition of sub-projects is uncertain and rates of return are usu-ally not estimated. Similarly, the question ofallocative efficiency has rarely been addressed inimplementation completion reports. Standard rateof return analysis is not readily applicable to so-cial fund projects, because they are not intendedto undertake systematic planning of investmentsand because of the large number of small and dis-parate subprojects they finance. Nevertheless,the lack of attention given under social fundprojects to conducting ex-ante and ex-post eco-nomic analysis of a sample of subproject invest-ments is worrisome.29 At the same time, there isno assurance that an adequate process is in placeat the relevant level (usually the community orlocal government) for identifying and comparingbenefits of alternative investments.

In theory, social funds ensure high net ben-efits of their investments by selecting subprojectson the basis of demand at the community level,where the best knowledge exists of local con-ditions, needs, and priorities, and by requiringbeneficiaries to contribute to costs. There are rea-sons to question how effectively this mechanismhas been working: • Social funds’ investment decisions are driven

by supply as well as demand factors. Thefunds use eligibility criteria, targeting mecha-nisms, and appraisal criteria; often have alimited menu of subproject options; and havemore technical expertise in some sectors thanothers.

• Subproject selection within the social fundmenu is influenced by a variety of factors andthere is no assurance that the investment al-ternatives, or opportunity costs, of a selectedsubproject are rigorously examined at thecommunity or other level. OED’s commu-nity-level field research for 845 randomly se-lected households and 1,525 respondents in17 randomly selected social fund–assistedcommunities in 4 countries found that whilethe vast majority of beneficiaries were satis-fied with the chosen subproject, the subproj-ect selection process was led by prime moversin the community. Even within the optionsthat were offered by the social fund, the se-lected subproject did not always address the

O u t c o m e

1 7

highest priority problem of the majority ofcommunity members. In many cases, com-munity members were not well informedabout the options.

• More than one in four social fund projects (14projects of 52) required no community con-tribution and, therefore, had no willingness-to-pay indicator. It may be noted that inpractice, not all community contributionsmade, including contributions in kind, arerecorded. At the same time, however, whencommunity contributions are formally re-quired, they are not always made.

Even if community interests and subprojectbenefits are accurately identified, the efficiencyof social fund investments also depends onmechanisms for coordination with district, re-gional, and national planning processes. De-pending on the nature of the investment,decisions taken on the basis of community in-terests may neglect spillover effects that makethem costly from a broader perspective, andsubproject benefits may be compromised byfactors beyond the knowledge or control of thecommunity (box 2.4). Most projects have es-tablished coordination arrangements with rele-vant line ministries and/or local governments,including mechanisms for sector ministries to re-view and provide “no objections” on each sub-

project proposal and for sector ministry repre-sentatives to be on the board of directors of thesocial fund. This allows them direct access to set-ting policies, determining sectoral approachesand appraisal criteria, and approving specific in-vestments. In practice, cases where the sectorministry representatives actively debate and dis-cuss investment options, priorities, and alloca-tion rules with other members of the Board arenot common, and their role has often been lim-ited primarily to ensuring that obligations tomeet ongoing costs of operating and maintain-ing social fund facilities are acknowledged bythe relevant public agencies. It has been notedthat “Formal approval of microprojects is gen-erally secured, yet the way this is done does notguarantee a coherent inclusion of social fund–initiated microprojects within national or re-gional sector plans. In practice, some socialfunds have been lax in this area, ignoring theneed for formal arrangements to ensure thatsector ministries are represented in provincialconsultative committees. This lack of coordina-tion led to the construction of understaffedfacilities—especially in the health sector—that didnot respect sociodemographic norms (minimumnumber of inhabitants for the catchment area)and service-level norms (for health post, healthcenter, etc.)” (Frigenti and Harth with Huque1998). Social fund projects have paid little at-


1 8

When asked by a community to solve a transport problem, thewhole access issue needs to be analyzed and addressed, not justsimply the construction of roads, tracks, paths, and footbridges.It requires finding the best way to allow people to get from onepoint to the other. Ensuring access requires attention to threeelements: rural transport services and intermediary means oftransport; location and quality of facilities the community wantsto reach; and rural transport infrastructures that will bear thetravel. Indeed, if the building or rehabilitation of rural transportinfrastructures is the first obvious answer, it may not be suffi-cient in itself, or it may be cheaper or more sustainable to in-

vest in one of the other two elements. For example, it may some-times be cost-effective to relocate the goal of the trip closer tothe community rather than to build a new transport infrastruc-ture, or to influence the transport mode, as improving transportinfrastructure is relevant only if it improves the existing trans-port mode or allows the operation of new modes. It is importantto keep in mind that rural transport infrastructure is part of thetransport network, and it is therefore impossible to considertransport infrastructure piecemeal. No road is isolated from an-other, nor can be thought of independently of what is happen-ing in the rest of the network.

S o l v i n g T r a n s p o r t P r o b l e m s I s N o t J u s t aM a t t e r o f B u i l d i n g R o a d s

B o x 2 . 4

Source: Lebo and Schelling 2001; Desmarchelier 2000.

tention to the aggregate impact of their activi-ties on the overall balance and efficiency ofpublic expenditure at national or local levels,within and between sectors and regions, and be-tween capital and recurrent expenditure.

Cost-effectiveness. The indicators of effi-ciency monitored most consistently by socialfund projects have been: (i) unit costs of con-struction or rehabilitation of infrastructure and(ii) the share of overhead expenses in total pro-gram costs. While important, these indicators arenot sufficient. Also needed is some concept ofthe unit costs of social funds in producing spe-cific outcomes (such as increases in school en-rollment rate for girls, in vaccination rates, or inaccess to clean water, employment creation) incomparison with other programs.

Social Funds 2000 background research foundthat cost-effectiveness of social fund projects, asmeasured by unit costs of construction, washighly variable across countries and sectors(Annex H). There is some indication that unitcosts tend to be lower where community con-tributions were high and/or there is communitymanagement of resources and contracting (Ar-menia, Peru, Zambia) for both social fund andnon–social fund projects. In Zambia, the evalu-ation explicitly assessed community-based ap-proaches (including the social fund) versuscontractor-based approaches in nine nationalschool infrastructure programs. Unit costs weresimilar among the various community-basedprograms. The contractor-based approach wasmore than twice as expensive, but delivered in-frastructure faster. Overall, Social Funds 2000 didnot find any consistent advantage in cost effec-tiveness of infrastructure delivery between so-cial funds, local governments, other centralagencies, and nongovernmental organizations(NGOs) across all countries.

The same evaluation found that, in general,overhead expenses for social funds were simi-lar across social funds, within a range of 7 to 14percent of total program costs.30 It also foundthat, in most cases, social funds had lower over-heads than other programs, although there weresome exceptions. Some project and evaluationreports have warned against putting too muchemphasis on overhead costs as an indicator of

efficiency, since this may discourage efforts toimprove construction quality, increase outreachfor better poverty targeting, and/or invest timeand resources in participatory processes.

What is the Record of Bank Performancein Social Funds?OED ratings on overall Bank performance for so-cial fund projects compared favorably with thosefor other Bank projects. Of the 23 closed socialfund projects as of end-fiscal 2000 (out of thetotal social fund portfolio of 66 projects exam-ined for this review), 94 percent were rated sat-isfactory on overall Bank performance.31 Thefigure for PTI projects was 75 percent and forall Bank projects 75 percent.32 The managementof social fund projects within the Bank has in-cluded effective knowledge management and hasfocused recently on impact evaluation in sixcountries. Cross-sectoral coordination on the is-sues for social fund projects has not been strong,but is beginning to receive more attention: forexample, the Africa Region has formed a spe-cial, cross-sectoral working group; measures arebeing taken in some ECA countries to ensure thatsocial fund expenditures support relevant sec-toral restructuring programs; and some socialfund projects, particularly in LAC, are paying in-creasing attention to the decentralization context.The MNA Region has designed social funds inparticularly challenging political contexts. TheEAP Region has moved quickly with social fundsin response to economic and civil crisis. QAGratings for quality of supervision of social fundprojects were somewhat above the average forall Bank projects between 1997 and 1999. QAGassessments found the quality of supervision offiduciary aspects to be strong.

Reporting on safeguard compliance in proj-ect status reports (PSRs) for social fund projectsappears similar to that in PSRs for PTI projects,but seems low overall, especially since a num-ber of safeguard issues are likely to be relevantfor social fund activities (Annex I). The envi-ronmental category assigned to social fund proj-ects may also be an issue, as noted by QAG(Annex I). Of the 16 available social fund QAGreviews (quality at entry and quality of super-vision), 10 were rated “satisfactory” for envi-

O u t c o m e

1 9

ronmental aspects while 6 were rated “not ap-plicable.” There are many examples of socialfund projects where efforts have been made todevelop micro-project environmental assess-ment mechanisms and adequate provisions havebeen made for environmental screening (and forsubproject analysis as necessary). The actualapplication of these provisions appears weak,however, and has not been systematically mon-itored.33 There are indications that the issuewarrants greater attention, especially with regardto environmental safeguards. No social fundhas been given a Category A environmentalclassification, but close to half (43 percent of arandom sample of 30 social fund projects) havebeen classified as Category B. The rest havebeen assigned to Category C. QAG reviews havequestioned if there is not excessive use of Cat-

egory C for the environmental impact of socialfund projects, implying that environmental issueshave been given inadequate attention.

The Bank has recently introduced a new en-vironmental category, “financial intermediary”(FI), which can be applied to social fund proj-ects. For all practical purposes, an FI categoryis similar to a B category—a process is definedto screen “not as yet defined subproject activi-ties” that may be identified (by project man-agement during project implementation) asneeding an environmental screening. The ef-fectiveness with which environmental issues areaddressed in Bank-financed social fund projectswill depend on how well projects are identifiedand screened. The Bank should be assured thatthese functions are adequately performed byproject management (Annex I).


2 0

2 1

Subproject Sustainability

To what extent will social fund projects continue to produce net benefitsas long as intended or even longer? Of the 23 closed social fund projectsas of end-fiscal year 2000 (out of the total portfolio of 66 projects

examined for this review), OED rated sustainability “likely” for 43 percent ofthe projects.1 This compared unfavorably with sustainability ratings for PTIprojects (50 percent) and for all Bank projects (51 percent).2 This chapter looksat social fund subproject sustainability, which most implementation completionreports indicate is an issue requiring greater attention. Issues relating to sus-tainability of social fund agencies themselves are discussed in Chapter 4.

Subproject sustainability can be assessed intwo ways: (i) directly measuring facility use andmaintenance at a specific point after subprojectcompletion and (ii) measuring the extent towhich the conditions for sustainability are inplace for the continued flow of benefits over theproject’s life. Direct measures give a picture ofthe state of the infrastructure and service flowsat the time they are measured. Since social fundsubprojects tend to be relatively recent, this ap-proach may not accurately assess the probabil-ity of sustaining benefits over the intended lifeof the subproject and beyond.

Subproject Sustainability Can BeAssessed by Direct Observation . . . Three elements of subproject sustainability canbe directly observed: (i) technical quality; (ii) cur-

rent staffing and equipment; and (iii) current lev-els of maintenance. The main source of these di-rect measures of sustainability is Social Funds2000 background research3 (Annex E, tableE.3). Other sources include OED’s household andfacility surveys, portfolio assessment, and stake-holder survey.

Technical Quality. The OED portfolio as-sessment found the technical quality of socialfund infrastructure to be variable across coun-tries and between sectors. Social Funds 2000background work entailed collecting data onconstruction quality through facility surveys byengineers and focus group discussions. Thefindings were mixed, although the general re-sult was that social fund facilities performed atleast as well as other facilities on technical qual-ity (Annex E, table E.3). It was noted that there

33

is high variability in technical quality, but fewmajor concerns over outright poor quality in anyprograms. While beneficiaries are generally sat-isfied with the quality of subprojects, some prob-lems were identified. For example, in schools inHonduras, because contractors were not re-quired to make mains connections for electric-ity and water closets, 28 percent of theinstallations built by FHIS remained uncon-nected compared with 4 percent of those builtby other agencies, and the dissatisfaction withoperational problems within the control groupwas much lower than in the FHIS group. Theseproblems were underscored by the school sur-vey respondents, 31 percent of whom reportedproblems in project construction or installation.FHIS has since taken measures to address thisissue.

Results varied across sectors with regard toconstruction quality. The water sector faced par-ticular problems. For example, in Honduras, sitesurveys found that the overall construction qual-ity of FHIS schools was about the same as non-FHIS schools (83 percent “regular” or “good”versus 87 percent for comparators), much bet-ter for health centers (97 percent “regular” or“good” versus 62 percent for comparators), andslightly worse for water systems (93 percent ver-sus 99 percent for comparators). The record wassimilar for sewerage systems (100 percent “good”or “regular” for both), and much better for latrines(100 percent “good” or “regular” for social fundsversus 69 percent for comparators). About 23 per-cent of households in the area of influence of thewater projects reported problems with the qual-ity of the water system, mainly due to designfaults and the use of poor-quality materials,which made water projects the most frequentlycriticized type of social fund investment withrespect to the quality of the physical infrastruc-ture built by FHIS. For Malawi’s social fund, it wasnoted that, with respect to its water supply andsanitation components, “while it is too early toassess whether subprojects implemented by theprojects are sustainable, a recent review of tech-nical quality revealed that a large number of fa-cilities had not been built to the required standard.MASAF has now contracted supervision con-sultants to assist them in certification of techni-

cal quality and providing inputs during subpro-ject supervision” (Kariuki 1998).

Giving responsibility to communities for sub-project implementation, without adequate sup-port, has been associated with problems intechnical quality.4 A World Bank paper hasnoted that “community-based organizations havethe most difficulty conforming to project stan-dards because they lack the technical skillsamong them or cannot afford to hire skilledlabor” (Khadiagala 1995). The Bank’s review ofbeneficiary assessments noted that in Ecuador,projects requested by grassroots organizations re-ceived the lowest quality ratings (Owen andVan Domelen 1998, p. 32). The Zambia SocialRecovery Project, which emphasized communityparticipation in the construction of the physicalinfrastructure, also faced quality problems, al-though measures have since been taken to ad-dress this issue (box 3.1).

Staffing and Equipment. Social Funds 2000background research indicates that nearly allsocial fund facilities surveyed were in operationand were staffed and equipped one to threeyears after their completion. Staffing and equip-ment levels were better than or equal to thosein comparator facilities (not controlling for ageof facilities), but both types of facilities sufferedshortages. Thus, inadequacy of inputs rather thantheir complete absence was the issue (Annex E,table E.3).

Of the two countries for which such infor-mation was available, there was evidence inone, Nicaragua (and to a lesser extent in Zam-bia), that social fund subprojects may draw staffaway from other facilities. Social fund health sub-projects in both countries have seen increasesin staff, while non–social fund facilities have ex-perienced declines.5 In Nicaragua, the frameworkagreement between FISE and line ministriesstipulates that FISE subprojects will not lead toincreases in the line ministries’ recurrent budg-ets, so that any necessary personnel are eitheralready on the payroll or facility expansions arestaffed through reallocations of personnel. Underthe circumstances, it is possible that staffing thenew or expanded social fund health facilities inNicaragua may have been at the expense ofstaffing elsewhere.


2 2

OED’s analysis of direct measures broadlyconfirmed the results from the Social Funds 2000background research. The respondents to theOED household survey6 were asked about theirperception of adequacy of staffing, their view onfacility supplies, improvements (since the baseyear7) in staff attendance,8 and improvements inthe availability of supplies (since the base year).9

For all four countries, a larger proportion of therespondents from the social fund–assisted com-munity than from the matched community hada favorable view of each of these issues, al-

though the absolute percentage of respondentsperceiving adequate staffing in social fund–assisted facilities was low, except in Nicaragua.In addition, a large percentage thought the fa-cility needed improvement.10 Nicaragua wasagain an exception (table 3.1, Annex M, tableM.16). The OED stakeholder survey found that62 percent of the respondents thought that thesocial fund–financed infrastructure was at leastas well maintained as infrastructure financed bythe government, and a similar percentage alsothought that social fund facilities had service

S u b p r o j e c t S u s t a i n a b i l i t y

2 3

Giving communities substantial responsibility in the implemen-tation of subprojects has advantages, including building owner-ship, but they sometimes come at the cost of construction quality.The 1992 Social Recovery Project Annual Report noted that oneof the most important issues for project management was the dif-ficulty of maintaining an acceptable standard of constructiongiven a self-help approach and the limited capacity of socialfund staff to supervise implementation of the projects.

In response to these problems, technical audits of projectswere undertaken to identify technical inputs that could be pro-vided to communities. In addition, all project committees wererequested to engage a full-time, paid foreman. Simple, repro-

ducible designs were sent to communities for construction ofclassroom blocks, health clinics, staff housing, and other in-frastructure. Communities that were not engaging an experiencedcontractor or foreman were asked to request the District Coun-cil to visit the facility and assess construction at critical stages,and seminars were organized at the provincial level to acquaintdistrict staff with the program’s objectives and procedures.While these measures were somewhat helpful, a number ofprojects were not receiving the inputs of skilled labor or tech-nical knowledge they required to produce good-quality con-struction. Further measures have since been taken to addressthese problems.

Z a m b i a : T a c k l i n g T r a d e o f f s B e t w e e nC o m m u n i t y S e l f - H e l p a n d C o n s t r u c t i o nQ u a l i t y

B o x 3 . 1

Question Jamaica Malawi Nicaragua Zambia

Do you feel there are enough staff at the facility? [Yes] 43 69 91 36

How would you describe the facility’s supply of desks, textbooks,

medicines, etc.? [Reasonably good/extremely good] 46 33 69 33

How does staff attendance compare to the situation [in the base

year?] [Better] 34 69 73 44

How does availability of supplies compare to the situation [in the

base year]? [Better] 39 45 51 28

How often is the facility clean and well maintained?

[Always/usually] 72 88 94 67

Do you think the facility needs physical improvements? [Yes] 67 42 17 89Source: OED household survey.

H o u s e h o l d P e r c e p t i o n o f O p e r a t i o n s a n dM a i n t e n a n c e P e r f o r m a n c e ( p e r c e n t )

T a b l e 3 . 1

Source: 1992 Social Recovery Project data and OED portfolio assessment.

flows at least as good as similar facilities fi-nanced by the government (Annex J).

Maintenance. Overall, maintenance prob-lems in social fund subprojects appear to bemore pronounced than problems with opera-tions. According to Social Funds 2000 back-ground research for Honduras, engineering sitesurveys rated maintenance of works as “good”for 53 percent, 91 percent, and 80 percent ofFHIS water, sewerage, and latrine projects re-spectively; for schools and health clinics, datais available for cleanliness, which was rated“good” for 53 percent of FHIS schools and 36percent of FHIS health centers. However, ac-cording to the qualitative evidence in the samereport, none of the cases studied showed fullysatisfactory results for maintenance of works. Thereport noted that: “there is a gap between thecommunity declaring itself the owner of theproject and understanding in theory the basicrules of maintenance, and putting all this knowl-edge into practice.”11 While the vast majority ofrespondents to the OED household surveys feltthat the facility was clean and well maintained,the survey also showed that only 35 percent ofthe respondents in Jamaica, 39 percent in Malawi,29 percent in Nicaragua, and 10 percent in Zam-bia perceived that required repairs were donequickly.12 OED’s stakeholder survey showedthat 39 percent of the respondents felt that so-cial fund–financed infrastructure is better main-tained than similar types of infrastructurefinanced by the government. The level of main-tenance may be expected to vary by country de-pending on the state of recurrent finances, butalso on the particular maintenance culture in thecountry.

. . . Or by Examining Conditions forSustainabilityThe second means of examining sustainability isto determine if the conditions for sustainabilityare met. OED’s literature review identified the fol-lowing conditions for sustainability:13 (i) arrange-ments for ensuring technical quality; (ii) clarityabout operations and maintenance responsibil-ities and awareness among those responsible oftheir obligation; (iii) willingness to fulfill opera-tions and maintenance obligations; (iv) the abil-

ity (technical, financial, and institutional) to meetoperations and maintenance obligations effec-tively; and (v) arrangements for ensuring that sec-tor specificity is adequately addressed.

Condition 1: Arrangements for EnsuringTechnical QualityA World Bank paper notes, “Concern for thequality of fund-financed projects challenges allsocial funds and is a recurrent issue in supervi-sion reports. The need to disburse funds rapidlyand the lack of technical capacity among projectsponsors contribute to the low quality of proj-ects” (Khadiagala 1995). Social funds haveadopted procedures to ensure that constructedfacilities are of good quality, including the useof standard facility designs and supervision dur-ing construction.

Standard facility designs are used by morethan half of social funds (14 of a sample of24).14 Often they have used ministry designsand standards, but these were sometimes inap-propriate. In Honduras, for example, the stan-dard school design provided by the Ministry ofEducation was not well-suited to multi-gradeclassrooms, and it rigidly specified the length forthe perimeter wall, so some social fund schoolsleft these incomplete. Social funds have devel-oped their own designs and standards wherethese were unavailable or inappropriate. Theyhave helped ministries improve their designs orstandards, as in Eritrea and Honduras. In orderto ensure supervision of subproject construction,social funds have used external supervisors, so-cial fund staff, or community volunteers. Butthese arrangements have encountered prob-lems, as indicated in project documents and byOED field research. In the Honduras FHIS andNicaragua FISE there were common complaintswithin communities about inadequate supervi-sion that sometimes allowed contractors to cutcorners or use substandard materials.15 An im-plication drawn by Social Funds 2000 with re-spect to water projects is that although themajority of the water projects evaluated are ofacceptable quality, many present problems thatcall for increased oversight regarding the initialconstruction and sustainability of the physical in-frastructure. In most countries, reports were


2 4

received of low-grade material being used in con-struction, leading to system deterioration.

Condition 2: Clarity and Awareness of Rolesand ResponsibilitiesOED’s portfolio assessment found several in-stances where the operations and maintenanceresponsibilities and obligations of the partieswere not adequately specified or handover andownership issues were not explicitly addressedup front.16 It has been noted that “The social fund[agency] and project beneficiaries need to haveaccurate information on the recurrent cost im-plications of every project if the feasibility andsustainability of the project is to be assessed andcommitments to sustain the project after com-pletion are to be honored. This has been oneof the weakest elements of some social fund op-erations” (Weissman 2001). Lack of clarity ofresponsibilities was a more common problemwith respect to maintenance than in operations.Operations responsibilities were usually out-lined in framework agreements or case-by-caseagreements between the social fund and therelevant agencies. Maintenance responsibilities,in contrast, were left to be agreed at a laterstage, typically at subproject completion, andmost operational manuals consequently ad-dressed maintenance issues in a cursory man-ner. The recent use of maintenance manuals insome social fund projects can be expected tobring improvements.

At the community level, the Bank’s review ofsocial fund beneficiary assessments found “asignificant and fairly universal problem with thelack of information and/or misunderstandingon the part of beneficiaries about the role andrules of the game of the social funds” (Owen andVan Domelen 1998). It noted that in Peru, aboutone-third of the beneficiaries interviewed wereunaware of the community’s commitment tomaintain the works and observed that informa-tion gaps were closely associated with negativeoutcomes, including lower levels of participationand sustainability. This would be a particular con-cern for services that depend significantly oncommunity management, such as water sup-ply. Social funds have taken measures to clar-ify and improve awareness about roles and

responsibilities. In Zambia, these efforts—whichincluded the distribution of project manuals tobeneficiaries and mandatory project-launchworkshops that bring together social fund staff,project committees, local officials, and benefi-ciaries—were particularly successful (Owen andVan Domelen 1998).

OED’s household surveys found that inNicaragua and Malawi only a small proportionof households were unaware of the allocationof responsibilities for facility repairs (9 and 18percent, respectively).17 Forty-two percent ofthe respondents in Jamaica and 40 percent inZambia were unaware of who was responsiblefor facility repairs.18 Furthermore, in Jamaica 60percent of the respondents had no idea who wasresponsible for paying for repairs, in Malawi thiswas 43 percent, in Zambia 42 percent, and inNicaragua only 14 percent.19 In Malawi, an ad-ditional 9 percent of the respondents indicatedthat “no-one” was responsible for paying for re-pairs (Annex M, table M.16). With respect to theOED stakeholder survey, about one-third of therespondents thought that, at the time they madetheir subproject proposal, their communitieswere fully aware of the responsibilities theywould assume by receiving a social fund–financed investment. Overall, the recent OED re-sults are suggestive of improvements over time.The efforts needed to improve community-levelclarity and awareness of roles and responsibil-ities will depend on the intended nature and ex-tent of the roles of the community and otheragents in operations and maintenance.

Condition 3: Willingness to UndertakeOperations and Maintenance ObligationsGovernment entities have generally been willingto meet their obligations, as demonstrated throughframework and subproject-specific agreements.When the relevant agencies are not party to so-cial fund decisions about specific investments thatwere chosen for financing, willingness (as wellas ability) may become an issue. In Jamaica,parish councils were not involved in the identi-fication and implementation of road projects andwere unwilling to maintain them.

At the community level, the OED householdsurveys broadly supported the view that the vast


2 5

majority of beneficiaries were satisfied with thechosen investment. It follows that communitieswere generally willing to keep the investmentrunning. The OED household surveys foundthat, of the respondents who said that the facil-ity needed physical improvements, 88 percentwere willing to pay for these improvements inJamaica, 85 percent in Malawi, 71 percent inNicaragua, and 90 percent in Zambia. The com-munity-level qualitative data also confirmed theidea that communities felt responsible for socialfund facilities—for example, “we also use theborehole carefully to avoid damage since we hadcontributed a lot” (Malawi). This was confirmedby OED’s institutional analysis based on field re-search in Jamaica, where communities were en-thusiastic about maintaining JSIF investments.

In general, communities were found to havea positive attitude toward sustaining social fundinfrastructure. An indicator from the experiencein Albania was the reported success of commu-nities in protecting social fund facilities fromvandalism during civil riots. This review could notestablish the degree of willingness to operateand maintain investments in social fund projectsoverall, but one would expect a close relation-ship to the extent to which the subproject ad-dresses the highest priorities of the community.

Condition 4: Ability to Undertake Operationsand Maintenance ObligationsEnsuring subproject sustainability requires threetypes of capabilities on the part of the entitiesresponsible for operations and maintenance: fi-nancial, technical, and institutional ability.

At the government level, financial ability islinked to two issues: the overall policy environ-ment for sectoral investments and the balance be-tween capital and recurrent budgets. Lack ofgovernment budgets for operations and mainte-nance often reflects the financial constraints ingeneral for government. In Nicaragua, the healthministry commits itself to meeting recurrent costs,while at the same time admitting there will beno money for it (Dijkstra 2000). Fiscal constraintscan be severe in both absolute terms and in thebalance between capital and recurrent costs.

In order to reduce recurrent cost problems,social fund projects have often focused on the

rehabilitation of existing facilities rather thanthe construction of new ones. In some cases,however, even this limited (rehabilitation) strat-egy may encounter problems. First, as shown bythe beneficiary assessment for JSIF, while therewere fewer operational problems when the sub-project was for facility rehabilitation, the result-ing benefits were also smaller or less appreciatedby communities. Second, as Social Funds 2000background research has shown, social fundprojects have not overcome the problems ofunderstaffing and underequipping of facilities.Significant improvements in operations andmaintenance depend on systemic fiscal and sec-toral reforms in the country that are beyond thescope of the social fund. Third, financing reha-bilitation over a long period risks becoming anexpensive option for deferred maintenance,since the necessary institutional improvementsmay not occur if external resources are readilyforthcoming for rehabilitation.

Social funds have also introduced mainte-nance funds as a way to address the sustainabilityproblem. Nicaragua’s maintenance fund is a newapproach that sets aside money from governmentand communities to finance maintenance costsof (social fund, and recently other) investments,and is administered by municipalities in responseto community requests. This approach should bemonitored for potential replicability. Depend-ing on the design of the maintenance fund andthe country context, it may not be the best op-tion. For example, a maintenance fund that ear-marks resources for specific facilities or isoff-budget can raise issues about bypassing reg-ular budgetary systems and giving preferentialtreatment to selected infrastructure.

Social funds have increasingly tried to improvecoordination with ministries or local govern-ments to ensure that subprojects are consistentwith their policies and operating budgets. Whilethis can be expected to produce positive resultsfor sustainability, it may be inadequate to ensurethat social fund facilities regularly operate to ca-pacity. For investments that depend heavily ongovernment for recurrent cost financing, sus-tainability will remain closely tied to the avail-ability of government budgets and may call forsystemic fiscal and sectoral reform.


2 6

At the community level, OED’s portfolio as-sessment showed that communities have oftenfallen short on the financial, institutional, ortechnical wherewithal to ensure that facilities areoperated and maintained either by the commu-nity itself or by accessing the necessary assis-tance. The Social Funds 2000 backgroundresearch for Honduras noted that “in many cases,[maintenance] problems remained unresolvedand often escalated, reflecting the community’spractical inability to resolve them either directlyor through appeal to the competent authority.This incapacity arose from some combination oflack of initiative, lack of organization and lackof economic resources.”20

The extent to which financial systems are al-ready in place for regular collection of feesvaries greatly between countries and sectors. So-cial Funds 2000 noted that the long-term sus-tainability of water and sanitation subprojectsmay be adversely affected by inadequate cost re-covery. In none of the countries where datawas collected did all households report payingfor services; the general perception was thatthe amount collected was insufficient to coveroperations and maintenance. This was true bothof urban systems managed by central agencies,municipal governments, or neighborhood waterassociations, as well as rural systems managedby local water committees. A recent evaluationof social funds by a bilateral donor agency alsonoted that financial issues in operations andmaintenance have proved particularly difficult inwater and other economic infrastructure projects:

government utilities tend to be largely absent inrural areas and there are no established tariff reg-ulations to finance operations costs. Similar in-stitutional and financial issues also apply torural power supply and road construction.21

Community capacity for operations and main-tenance can be supported by the requirement toform maintenance committees and by the provi-sion of maintenance manuals and training. Wherethese mechanisms are lacking, subproject sus-tainability has suffered. For example, for the PeruFONCODES, it was noted that more emphasis wasneeded on training beneficiaries in operationsand maintenance, especially in water supply andsanitation. Recent social fund projects are givingmore attention to community training for opera-tions and maintenance, and other community-level arrangements are also being put in place toensure sustainability, indicating a learning curve(box 3.2). In the Bolivia social investment fund,improvements in water quality were less than ex-pected because there had been inadequate at-tention to “software” aspects. This was followed-upby the provision of additional training through thePROSABAR project, however, which producedfavorable results for sustainability. Best practicesremain to be replicated across all projects. Sus-tainability mechanisms are being supported inthe Ethiopia and Moldova social funds, includingmore attention to community contribution and tofollow-up in the community by social fund staffafter subproject completion (see Annex N,“Ethiopia, Moldova: Incorporating Mechanismsfor Sustainable Service Delivery”).


2 7

The number of projects that required a maintenance contractbetween the community and at least some of the responsible en-tities increased from none in the period 1987–93 to 32 percentin 1994–99.

The number of projects with maintenance manuals forthe responsible entities increased from 40 percent in the period1987–93 to 64 percent in 1994–99.

The number of projects that required maintenance/usercommittees increased from 40 percent in the period 1987–93to 74 percent in 1994–99.

The number of projects that provided small, substantial, or verysubstantial amounts of training or capacity-buildingactivities forcommunities in operations and maintenance activities increasedfrom 40 percent in the period 1987–93 to 79 percent in 1994–99.

L e a r n i n g C u r v e : T r e n d s i n S u s t a i n a b i l i t yM e c h a n i s m s

B o x 3 . 2

Note: N = 24

Source: OED data.

Condition 5: Arrangements for Ensuring thatSector Specificity is Adequately Addressed OED’s portfolio assessment confirmed the dif-ferences among sectors in the issues of partic-ular concern for sustainability. Budgetary supportto subproject operation was most likely to beprovided in the education and health sectors, al-though staffing problems were more commonfor health facilities. Water and sanitation projectstended to experience the greatest problems inoperation from the outset because of poor or in-appropriate technical design or lack of localorganizational or technical capacity. Poor main-tenance of roads and bridges was not onlybrought about by lack of funds, but often to un-clear ownership and responsibility as well. In ad-dition, maintenance problems have been severefor water and roads facilities, since there is notalways clear institutional responsibility or mech-anisms for levying tariffs or collecting resources.

While the 1997 UNDP/World Bank RuralWater Program global study of rural water sys-tems found that the social funds in the samplehad average to above-average sustainability,controlling for the country, the same study foundthat for both countries for which comparativedata were collected, the average sustainabilityscore was worse for the social fund than for theother water project in the same country (Katz andSara 2000). By contrast, in education and health,the facility provides a potential focal point forcollecting fees, although this potential is not al-ways realized. Where economic infrastructure ormicroenterprises were failing, this was generallydue in part to lack of the necessary attention toeconomic or financial viability at the appraisalstage. Sustainability of microcredit schemes, in-cluded in about a quarter of social fund projects(13 of 53 projects),22 was much more uneven:from the two notably successful examples (Al-bania and Honduras), important factors were en-suring village ownership, good technicalcapacity, and experience in the administeringagency.

In Nicaragua FISE III, road projects werefound to be the highest community priority dur-ing the emergency (Hurricane Mitch), but therewas concern about the sustainability of the roadprojects executed during this period. Results

from OED’s qualitative review of 2 JSIF-supported road projects and 2 JSIF-supportedwater projects in Jamaica (involving 18 key in-formant interviews and 9 focus group inter-views) found the same pattern for every road andwater project examined. The Parish Council (forroads) or National Water Commission (for water)had refused to assume responsibility for opera-tion and maintenance of the facilities, while thesponsoring community organization had willinglyassumed this responsibility, with limited resourcesand training. This was likely to raise seriousquestions about the future sustainability of theseprojects unless the communities’ continued rolewas regularized and their ability to undertake thenecessary operations and maintenance activitieswas ensured. The Bank has recently increasedattention and focus on operations and mainte-nance issues relating to JISIF investments, in-cluding the provision of significant trainingactivities aimed at local officials and trainersand the dissemination of information on main-tenance practices.

Arrangements for sustainable service deliverymust reflect the nature of goods and servicesbeing delivered (box 3.3). By being classified asmultisectoral, social fund projects may escape thescrutiny applied to sector projects. This issue hasbeen noted with regard to financial sustainabil-ity standards in the water sector (OED 2001). Forthe road sector, interviews with Bank staff in-dicated that social funds do not adequately dis-tinguish between community roads and tertiaryroads and use similar arrangements for both, withimplications for sustainability. Furthermore, sinceeconomic analysis is not typically used to assesssocial fund subprojects, social fund projects fi-nance community roads of higher unit cost thanwould be justified by sector-financed tertiaryroads programs, and alternatives are not ade-quately assessed. Issues related to sustainabilitycan become particularly acute when the lion’sshare of financing of tertiary roads is providedby the social fund and there is no credible sec-tor strategy for maintenance of these roads. Thiswas the case in Malawi, where the road sectorinvestment program refrained from financingtertiary roads because the social fund was pro-viding extensive resources for them. In this case,


2 8

the sector program assumed responsibility foralso developing a strategy for tertiary roads andaverted adverse effects. In Argentina, some ofthe subprojects that FOPAR is financing through

grants are the same investments that the Mu-nicipal Development Program is financingthrough loans—that is, sidewalks and curbs. Al-though both Bank programs have worked in dif-


2 9

The government role and government level required in the plan-ning, production (investment and O&M), financing, or regula-tion/policy setting for any type of good or service demanded willdepend on the following characteristics of the goods and serv-ices: their public versus private nature (is the good or serviceby its nature available to all in the community, or are access andavailability limited?); the range of impact area and externalities(are there positive or negative effects on nonusers?); the extentof natural monopoly and potential for competition in production;the social value placed on ensuring a minimum level of con-sumption (for merit goods); technical coordination requirements,especially for physically networked services; and informationrequirements for users and producers to realize the desiredquality of services.

From these characteristics, certain implications follow re-garding the most appropriate roles (“who does what”) for the com-munity of beneficiaries and for the next level or levels ofgovernment. That is, for services with mainly private impacts, in-dividuals or households can choose what they want to consumeand pay prices or user charges, so demand and supply can be de-termined through the market mechanism. For services withmainly neighborhood-level impacts, a local community group (orits agents) should plan and finance (through local benefit taxesor levies or user charges, for example). For services with inter-community or citywide impacts, local (or district) government orits agencies should plan investments, ensure technical/systemcoordination, and organize financing. Fiscal transfers (from na-tional government to local, local to community, or to households)may still be justified to ensure minimum essential service lev-els, counteract externalities, or correct for high start-up costs.

Types of subprojects that create recurrent obligations ongovernment budgets for operations (such as staffing schoolsand clinics) also need to be consistent with sectoral plans inthese areas. When users and suppliers have unequal access tocritical information (on safe water quality, for example), andwhen there are important technical coordination requirements

for service levels to be achieved (such as consistent standardswithin networked infrastructure systems), there is a specific ra-tionale for a party beyond the user groups to perform some over-sight, information-sharing, or regulatory functions.

The key implication for social funds is that the effective-ness and sustainability of subprojects requested by communi-ties may depend on various institutional arrangements being inplace for the sharing of responsibilities and functions betweenthe community, the respective local government, and some-times higher levels of government or its agencies; the appropriateinstitutional arrangements will differ with the type of subproject,depending on the technical, economic, financial, and environ-mental characteristics and the scale as described above. Gen-erally, non-networked water projects, for example, can be moreappropriately provided within a community than inter-villageroads. The following points are important in considering to whatextent and in what ways the community requesting/sponsoringa social fund subproject will need a broader framework of gov-ernment support to ensure sustainable results: • Who should provide each type of infrastructure service, and

at what level (community, municipality, district), depends onthe impact area, network or coordination requirements, andexternalities. Appropriate solutions often require sharing ofplanning, financing, and oversight responsibilities amongcommunities and across levels of government.

• A community may act as the “lowest local government” forits members in providing some types of services, but good re-sults still require that the need for accountability, appropri-ate financing mechanisms, coordination or connectivity ofservices, and organization be addressed.

• If the community acts “like a government” by performingthese functions for the provision of local goods or services,it must be legally, formally constituted to collect resources,own assets, and constrain private action (for example, tocharge and collect user fees, protect the commons, maintainthe infrastructure).

W h e n S h o u l d S o c i a l F u n d s S u p p o r tC o m m u n i t i e s “ A c t i n g A l o n e ” t o P r o v i d eG o o d s a n d S e r v i c e s , a n d W h e n S h o u l dG o v e r n m e n t B e I n v o l v e d ?

B o x 3 . 3

Source: Kessides 2000.

ferent municipalities, they could coincide in thefuture, in which case this inconsistency wouldbecome a more serious problem (Serrano 2000).In OED surveys of both country directors/resi-dent representatives and sector directors, theeconomic infrastructure components (road, watersupply and sanitation, and energy components)of social fund projects were rated worse thanother projects in the country/sector with respectto compliance with sector standards for economicanalysis, as well as for financial issues.

In general, it is important to clarify what canand should be expected of social funds with re-spect to sustainability of subprojects. The socialfund can supervise and support technical qual-ity; ensure that subprojects conform with thepolicies and operating budgets of line min-istries and/or local governments; ensure that allparties are aware of, and committed to, carry-ing out their responsibilities; and support theability of communities to play their part in op-

erations and maintenance. For subprojects withoperation and maintenance within the remit ofthe community itself, the social fund can sup-port community capacity building. Many typesof subprojects, however, depend on governmentfor operations and maintenance and are subjectto prevailing fiscal and sectoral constraints in thecountry. In these cases it may be unrealistic toexpect that operations and maintenance of so-cial fund facilities would differ substantiallyfrom the norm. Significant improvements woulddepend on systemic reforms beyond the scopeof the social fund or the community. The scopeand scale of activities undertaken by a socialfund should be guided from the outset by as-sessment of operations and maintenance ca-pacity and constraints in each of the sectors ofsocial fund intervention in the country con-cerned, including the existing composition andbalance of public expenditure and projected fis-cal capacity.


3 0

3 1

Institutional DevelopmentImpact

To what extent have social fund projects improved countries’ capacityto use their human, organizational, and financial resources effectively?Generalizations about the institutional impact of social fund projects

are difficult because the projects vary widely in their country conditions, ob-jectives and design, and nature and scale of activities, and because of changesin these factors over time. Furthermore, institutional impacts have not beensystematically monitored and, as a result, empirical data are limited. OED’sfield research and review of the Bank’s portfolio provide examples, outlinedbelow, of both positive and negative institutional effects of social funds.

The net impact depends on the constraints and op-portunities in the particular institutional setting, andon how well the design of the social fund isaligned with this setting. This chapter examines theintended and unintended effects that social fundprojects have had on institutional development.1

Of the 23 closed social fund projects as ofend–fiscal year 2000 (out of the total social fundportfolio of 66 projects examined for this review),OED rated 65 percent “substantial” on institu-tional development impact.2 This compared fa-vorably with the ratings for PTI projects (43percent) and all Bank projects (36 percent) overa comparable period.3 These ratings were heav-ily influenced by good organizational perform-ance of the social fund agency. Completion andaudit reports found more weaknesses in per-formance on wider institutional objectives, in-

cluding weaknesses in delivering on capacitybuilding components.4

Of the 66 projects in the portfolio, 47 percenthad explicit institutional development objec-tives, to build the capacity of central or local gov-ernments, the private sector, NGOs, and/orcommunity organizations. The most significantchange over time has been the shift in attentionfrom building capacity at the central level to anincreasing focus on local governments and com-munities (table 4.1).

Social fund projects have often had morethan one institutional objective. Typical exam-ples are: to strengthen central capacity forpoverty monitoring and analysis; to strengthenlocal community organizations and municipalgovernments; to contribute to the government’sdecentralization strategy by building local ca-

44

pacity in planning, execution, operations, andmaintenance of small-scale infrastructure proj-ects; to improve the capacity of communities andNGOs to plan, appraise, manage, and maintaininvestment activities through experience gainedin the project; and to strengthen the capacity ofprivate, small-scale contractors and other mi-crobusinesses. Even in projects without suchstated objectives, the expectation of institutionalbenefits has commonly been reported in proj-ect appraisal documents.

Performance monitoring practices suggestthat most projects have attached a relatively lowpriority to institutional development, withoutexplicitly addressing the tradeoffs with otherobjectives. Relevant performance indicators wereincluded in appraisal documents for 31 of the66 social fund projects, and most of those con-sisted of tracking inputs, such as quantity of train-ing. A review of project status reports for arandom sample of 18 of the 31 projects foundthat only one in 6 projects was reporting progresson these indicators. Capacity building compo-nents of projects often have not been fully im-plemented. A few of the newest social fundprojects have a clearer institutional develop-ment focus. The Moldova MSIF and Romania so-cial fund, for example, give priority to buildingcapacity at the community level.

Views on appropriate objectives for socialfund projects cover a wide range. In OED’sstakeholder survey (including government offi-

cials, social fund agency staff, and NGO anddonor representatives), 36 percent of the re-spondents said that the primary function of so-cial funds should be to finance capacity buildingfor infrastructure or service delivery, 18 percentthat it should be to finance delivery of infra-structure, and 6 percent that it should be tosupport institutional reform. Asked whether thesocial fund was currently focusing on the pre-ferred function, only 19 percent thought it wasfully focused and 64 percent considered it par-tially focused (Annex J).

Most Social Fund Agencies AreIndependent and Competent . . .All the Bank’s social fund projects have financedinvestment in building the capacity of the socialfund implementing agency—through staff train-ing, technical assistance, study tours, and othermeans—with some impressive results for orga-nizational effectiveness and innovation in proj-ect management. Agency staff members havetypically been concentrated in engineering, pro-curement, or financial management specialties,often recruited from the private sector. Throughbuilding teams of results-oriented professionals,many social fund agencies have distinguishedthemselves as learning organizations, adaptingto changing country conditions, respondingquickly and effectively to emergencies, estab-lishing transparent operating procedures, puttingsophisticated management information systems


3 2

1987–95 1996–99Objectives (N=24) (N=42)

Decentralization 8 14

Capacity building:

Social fund agency 17 10

Central government 33 14

Local government 8 29

Private sectora 4 10

NGOs 13 12

Communities 29 48a. Includes private contractors, consultants, artisans, skilled labor, farmers, local technicians, and entrepreneurs.

Source: Bank data.

I n s t i t u t i o n a l D e v e l o p m e n t O b j e c t i v e sH a v e B e c o m e M o r e L o c a l ( p e r c e n t )

T a b l e 4 . 1

in place, and developing innovative approachesto community contracting (Annex L) and out-reach.5 Some of the tools and processes devel-oped for management of social fundprojects—including the use of operational man-uals, management information systems, and pro-curement and disbursement proceduresappropriate for community contracting—havehad wider application and benefit within Bankoperations. For example, the proposed Chil-dren and Youth Project in Macedonia is draw-ing on social fund experience in the use ofoperational manuals.

Implementation completion reports attributemuch of this effectiveness to the autonomy ofthe social fund agency. Of the projects for whichthe data have been reported (ranging from 38to 56 projects, depending on the question), themajority of social fund agencies were created aslegal entities, independent of line ministries (89percent), and accountable directly to the presi-dent or prime minister (64 percent). Most havebeen exempt from government regulations in hir-ing staff (95 percent), staff terms and condi-tions (93 percent), and government procurementand disbursement procedures (83 percent and74 percent).6 Although these arrangements havebeen key to the strengths of social funds in pro-ducing immediate outputs and in attracting ex-ternal finance, they have also presenteddifficulties of coordination and accountability rel-ative to mainstream central, regional, and localpublic sector agencies.7

The issues involved are similar to those con-cerning the use of Project Implementation Unitsin general,8 but are particularly important for so-cial funds for three reasons. First, social fundsusually have a much higher degree of inde-pendence from line ministries and sectoral budg-ets. Second, they make decisions on allocationof resources among alternative investments—both across and within sectors and Regions.Third, they have acquired de facto long-term sta-tus, as their mandates continue to be extendedon the strength of external funding (typically sus-tained at 70 percent to more than 90 percent ofexpenditures),9 with continuing accountability,usually to donors as much as to national stake-holders (Dijkstra 2000).

Views differ concerning the justification ofcontinuing the autonomy of social funds whenthey are no longer geared to emergency objec-tives. Social funds are sometimes characterizedas simple executing agencies for central or localgovernments, complementing the line ministries’role of planning and policymaking by special-izing in delivery of physical infrastructure. ManyAGETIPs now play this role, as contractors tolocal or central governments.10 Social funds,however, by controlling the purse strings, havebeen filling a parallel function in deciding theallocation of public resources—based on a com-bination of local demand and poverty mapping(Dijkstra 2000; Dijkstra and Green 2000). An al-ternative approach to defining the role of au-tonomous social funds vis-à-vis other agenciesis that they are filling gaps in the country’s so-cial protection framework. This is indeed the roleof some social funds that are integrated in gov-ernment and primarily government-financed,such as Chile’s FOSIS, whose activities do notinclude social infrastructure. However, in Bank-financed social funds, the bulk of investment hasbeen in public infrastructure, especially schoolsand health facilities, for which the quality, cov-erage, and sustainability of service delivery de-pends on the policies and performance of centraland local governments.

As a result of these ambiguities, the role ofsocial funds in relation to line ministries has beeninherently unclear, and coordination problemshave been reported frequently in project statusand completion reports. Many of these problemshave been attributed to weaknesses in the min-istries concerned, but, at the same time, thespecial status and mandate of social funds has,inter alia, made it difficult for them to havemuch impact on the deficiencies of mainline serv-ices (Nelson 1999).11

The success of social funds in attracting donorfinance has been an important strength and hasalso imposed a challenge for coordination. So-cial fund agencies have experienced difficultiesin managing simultaneously the different re-porting and accounting requirements and proj-ect objectives of multiple donors, unless strongefforts have been made by donors to reducethese problems. The Egypt social fund effectively

I n s t i t u t i o n a l D e v e l o p m e n t I m p a c t

3 3

managed a multidonor partnership with as manyas 17 donors (see Annex N, “Egypt: MultidonorPartnership and Coordination”).

Social funds have attracted significant politi-cal attention. The Winner-Loser Matrix (AnnexG) shows that social fund projects can generatea wide range of costs and benefits—most stake-holders can gain, lose, or both, depending onproject design and context. Political objectivesare not necessarily inconsistent with develop-ment objectives and poverty targeting; after all,the poor comprise a large constituency in manycountries. In Peru, FONCODES was the flagshipproject of the Fujimori administration for reach-ing out to the poor. However, when courting thepoor leads to political favoritism or clientilism,it becomes incompatible with the institutional de-velopment of government since it violates thefairness principle, which lies at the core of gov-ernment performance (Serrano 2000). Socialfunds can be subjected to undesirable politicalpressures. For example, it was found that polit-ical interference in Ecuador adversely affectedFISE’s operations and credibility after mid-1996.Similarly, at a time when elections were immi-

nent, the Panama FIS was threatened by politi-cal influence on staff appointments and sub-project allocations. Related issues have beenencountered in many social fund projects, for ex-ample, in Albania, Guatemala, Madagascar, andTajikistan, and QAG reviews have emphasizedthe importance of protecting social fund agen-cies from political pressures.

. . . And Their Wider Institutional EffectsHave Been LimitedSuperior performance of the social fund agency,when it is achieved, does not in itself imply asignificant institutional development impact.This depends on the extent to which socialfunds promote wider adoption (of values, be-haviors, methods, processes, and skills) in na-tional and local institutions through theirinteraction with other agents. (The channelsavailable to social funds to impact the wider in-stitutional environment, both positively and neg-atively, are outlined in box 4.1.)

Little attention has been given by projects tomonitoring or documenting these impacts, whichare difficult to measure. OED’s stakeholder sur-


3 4

Direct effects: A social fund project includes an explicit insti-tutional development component, providing training or techni-cal assistance to other organizations.

Demonstration and catalytic effects: New approaches, pro-cedures, or methods demonstrated by the social fund are adoptedby other agencies, and where the social fund prompts newthinking and approaches.

Learning-by-doing effects: Other agencies gain experiencein using new approaches, procedures, or methods through en-gagement in subproject decisionmaking/planning or imple-mentation processes.

Competitive effects: Other agencies improve their own effi-ciency in order to stay in the competition for resources andpower; or negatively, where other institutions withdraw fromtasks and responsibilities that the social fund is taking on.

Demand effects: Social fund activities generate additional de-mand at the community level for the services of governmentagencies, stimulating enhancement of government capacity to

meet that demand; or where they stimulate broader demand ef-fects for the services of private contractors or NGOs.

Resource mobilization and allocation effects: The socialfund mobilizes previously unutilized community/local capacityand nongovernmental resources, and contributes to a rational-ization of resource use by government (for example, if govern-ment resources are freed-up to improve quality of services as aresult of the social fund investment in buildings); and/or nega-tively, the social fund agency attracts the most competent staffand critical resources away from government agencies, thusweakening them.

Systemic planning, budgeting, and accountability effects:The social fund contributes to transparent and efficient publicsector processes through close integration with them; or neg-atively, where social funds bypass central or local budgetary andsectoral planning processes, undermine accountability of pub-lic agencies, inhibit fiscal reform, and distort the budgetary al-location process.

I n s t i t u t i o n a l D e v e l o p m e n t E f f e c t s H a v eM a n y C h a n n e l s

B o x 4 . 1

vey12 found that 48 percent of the 160 respon-dents thought the social fund had had a posi-tive impact on the institutional development ofline ministries, compared with 55 percent forlocal government. Forty-six percent thought thatthe social fund had a positive impact on de-centralization in the country, while only 13 per-cent saw no impact and only 2 percent thoughtthe impact had been negative.

Effects on Central GovernmentSocial fund engagement with central governmentas it relates to institutional development hastaken five forms: direct training or technical as-sistance for central government staff; direct en-gagement of central agencies in social fundprocesses as eligible subproject sponsoring agen-cies; demonstration of new approaches; sec-ondment of central government staff to thesocial fund; and cooperation between the socialfund agency and central government on tech-nical standards, investment planning, targeting,and other functions.

Some of the early social fund projects (6 ofthe first 18 funded by the Bank)13 had explicitobjectives aimed at building capacity at the cen-tral government level. Many without explicitobjectives included components to finance train-ing and technical assistance, sometimes becausethey were conceived as temporary agencies thatwould transfer their responsibilities to line min-istries as soon as the ministries had the neces-sary capacity. Of the 24 projects for which datawere available, more than 60 percent includedsome training for central government officials—half of these provided substantial or very sub-stantial training, while 37 percent providedtechnical assistance.14 JSIF in Jamaica offers anexample of successful capacity building of thecentral agency responsible for communitydevelopment.

Overall, progress either in building capacityor in transferring responsibilities at the centrallevel has not met expectations. Recent projectshave been less ambitious in this respect. Weak-nesses in the design and delivery of training wereidentified in some implementation completionreports.15 However, demand factors beyond thecontrol of the social fund—such as frequent

staff transfers, lack of motivation, and lack of in-centives among government staff to receivetraining—have also been important in explain-ing the limited success of these activities.

The scope for learning-by-doing effectsthrough direct engagement as intermediary agen-cies in social fund processes has been small atthe central level and has declined substantially.In only a few countries have central governmentagencies been among the entities eligible toapply for social fund resources.16 Of the 24 proj-ects, the share with more than 10 percent of theirresources going to central government declinedfrom 60 percent before 1994 to only 11 percentfrom 1994 to 1999.

Demonstration effects have been more sig-nificant, but still limited. Ten of 24 projects re-ported carrying out regular dissemination ofinformation to government staff on new devel-opment approaches. The practice of secondingcentral government staff to the social fund hasbecome more prevalent (reported for one in fiveprojects since 1994). Reported impacts have in-cluded changes in procurement and contractingprocedures, adoption of competitive biddingprocedures by government agencies, and, in afew cases, revision of public sector procure-ment regulations.

In some instances, wider use sometimes hasbeen made of poverty maps and managementinformation systems developed for the socialfund. In Guatemala, a pilot program to traincommunity members in teacher selection and su-pervision was adopted nationally by the Ministryof Education.17

Such examples are not widespread, how-ever. OED field research, portfolio assessment,and stakeholder interviews indicate that lineministry staff tend to resent the independenceand perceived privileges of the social funds andmay not be convinced they have anything tolearn from them. When the fund is reputed tobe achieving better results and efficiency, min-istry staff are inclined to attribute this to thegreater resources of the social fund rather thanto differences in working methods.

OED found some evidence of both positiveand negative impacts from competition with thesocial fund. For example, in Nicaragua, com-


3 5

petition between FISE and INIFOM (the institutecharged with the development of municipalities)appears to have had some effect in promptinggreater efforts to improve effectiveness. ButFISE’s entry into water system delivery, whichbrought it into competition with ENACAL (theexisting agency responsible for delivery of watersystems), involved coordination problems, inthat FISE had difficulty in matching all the tech-nical standards of ENACAL.

The main form of interaction at the centrallevel has been through mechanisms to coordi-nate social fund activities with sectoral recurrentbudgets and technical standards. This coor-dination typically depends on: (i) line ministryrepresentation on social fund steering boards; (ii)framework agreements between the social fundand line ministries defining cooperative arrange-ments at various stages of the project cycle; and(iii) the requirement for prior approval of sub-projects by the relevant line ministry.

Of the 24 projects for which task managersprovided detailed information on institutional is-sues, 16 have required a formal coordinationagreement specifying respective roles, and 3have included procedures for regular review ofthe agreement. Sixteen of the projects had aworking-level liaison person in the relevant sec-tor ministries, increasing from 20 percent before1994 to 79 percent since then. Twelve requiredscreening, clearance, or approval of subprojectproposals by central government, but this sharedeclined from 80 percent in 1987–93 to 42 per-cent in 1994–99. Fifteen required clearance of so-cial fund infrastructure norms and technicalstandards.

Overall, these mechanisms appear to havebeen sufficient to at least secure central gov-ernment acknowledgement, where relevant, ofrecurrent cost obligations for the facilities fi-nanced by the social fund (although budgetconstraints have often prevented them from ful-filling these obligations).18 They have usually notbeen sufficient, however, to ensure that socialfund investments are integrated in sound sectoralpolicies and public investment strategies. Thereis little evidence that they have prompted sig-nificant improvements or new approaches on thepart of ministries, although new design and

construction standards developed by social fundsfor school buildings or health facilities havesometimes been adopted nationally. The re-quirement for ministry approval of subprojectshas usually been honored, but often in a rou-tine manner. When cooperation agreementshave been ambitious, ministries have had in-sufficient capacity, resources, or motivation tocomply fully with all the requirements.

Nevertheless, social funds have increasedgeneral awareness of the feasibility of new ap-proaches in the public sector: approaches to in-vestment decisions and service delivery that aremore client-responsive; poverty targeting of in-vestments; and the potential for public agenciesto work with NGOs and the private sector. Forexample, in Nicaragua, FISE seems to have madethe central level more aware of the contributionslocal levels can make in decisionmaking, espe-cially in the education sector (Dijkstra 2000). InBolivia, the emergency social fund improved therelationship between the central governmentand NGOs. In Thailand, the social fund is facil-itating central government collaboration withcivil society organizations (see Annex N, “Thai-land: Fostering Partnerships”). In Egypt, the so-cial fund has been successful in promotingpartnerships between central government, NGOs,and the private sector, despite strong initial re-sistance. In Argentina, national and subnationalgovernments’ skepticism toward the feasibilityof community-based contracting programs di-minished after FOPAR’s pilot phase, paving theway for other programs to adopt a similar ap-proach (Serrano 2000).

Negative institutional impacts have also beenreported, for sectoral planning, efficiency of re-source allocation, or budgetary transparency,when the social fund accounts for a substantialshare of public expenditure—as it does, for ex-ample, in Bolivia, Honduras, and Nicaragua.OED field research did not find evidence of so-cial funds displacing skilled manpower on anysignificant scale from government agencies—thenumbers involved were not significant and so-cial fund staff were often drawn from the pri-vate sector. However, concern that a social fundinhibits sectoral planning and may displace orundermine sectoral reform efforts has been


3 6

raised in a variety of contexts. An evaluation ofthe first emergency social fund project for Bo-livia, for example, flagged the danger that if asocial fund program was not limited to a fixedperiod, it was liable to dilute other reform effortsand allow procrastination. Similar points weremade in evaluations of social fund projects inGuyana and Egypt.

One of the difficulties has been the potentialof large social funds to distort resource alloca-tion in favor of infrastructure investment in theirsectors of specialization. Despite this danger, theimpacts of social funds on the balance of pub-lic expenditure have not been monitored. If na-tional sectoral strategies are weak, or if socialfund activities are not well aligned with sectoralpriorities, a “successful” social fund may attractincreasing donor assistance at the expense of lineministries, distort the balance of expenditurebetween sectors or between capital and recur-rent expenditure, and delay quality improve-ments in the services themselves. When socialfunds are responsible for allocating a significantshare of investment resources in the educationor health sectors, the ability of the relevant min-istries to allocate capital resources according tosectoral criteria (such as enrollment and atten-dance rates, projections of pupil numbers, dataon school performance, and epidemiologicalconditions) is diminished. In addition, sincesocial funds typically finance only investmentcosts, the success of a social fund in attractingdonor resources may exacerbate any existingbias toward capital over current expenditure inthe social sectors (Dijkstra 2000). This tendencyis in opposition to the Bank’s strategies forassistance to the social sectors, which point tothe need for more attention to the quality ofservice and outcomes relative to investment inbuildings.19

Particular difficulties arise when major re-structuring is needed or is under way in the so-cial sectors, as in the Europe and Central AsiaRegion, where strategies for the education andhealth sectors include, or should include, ra-tionalization of unsustainable excess infrastruc-ture. Unless the social fund’s targeting criteria areclosely coordinated with sectoral rationalizationplans at the relevant level, the fund’s activities

risk inhibiting, or obscuring the need for, effortsto plan and implement these difficult reforms.For example, in the absence of clear plans or acommitment to rationalization of schools, butwith strong community demand for school ren-ovation subprojects, the Georgia GSIF facedthree options: to fund education subprojects onthe basis of community demand and risk in-vesting in schools that would later need to beclosed; to halt funding of education subprojectsuntil the ministry or regional authority providedthe necessary guidelines, based on a viable ra-tionalization plan; or to develop their own cri-teria (Schmidt and Marc 1998). By choosing thelast option, the social fund was meeting impor-tant short-term needs, but was failing to supportefforts toward improving sustainability of theoverall system.

Social funds in the Region are now givingmore attention to these issues and are respondingin different ways. In Romania, for example, thesocial fund is not financing health or educationfacilities, leaving this to sectoral projects. InMoldova, the social fund has contributed to in-novations in school heating and space rational-ization, and it invests in village health postsonly after ascertaining that they are not slatedfor restructuring. In Armenia, where major re-forms of school finance and governance arebeing implemented, the social fund is aiming toalign its activities with these reforms by work-ing through the newly elected school councils.

Project documentation indicates that rela-tively little attention has been paid in socialfund projects to the impact of social funds withinthe overall framework of public expenditureplanning, or compliance with national budgetand accounting requirements, although largesocial funds may upset the transparency, ac-countability, and comprehensiveness20 of thebudget process (or efforts to reform this process).As for other kinds of projects, government al-location to the social fund is usually included inthe budget, but donor finance often is not. Ofthe 42 projects for which task managers providedthe information, about one-quarter reported thatall social fund revenue was included in the gov-ernment budget (table 4.2). However, evenwhen all social fund revenue is shown in annual


3 7

budget estimates for parliamentary authorization,this does not include how the money will bespent, which prevents integration with sectoralplanning and budget estimates. The Eritrea Com-munity Fund is unusual in coordinating with thegovernment expenditure planning and budget-ing processes by gathering eligible subprojectproposals for the coming fiscal year, in collab-oration with regional government staff, and con-solidating them in an annual work program andbudget that is submitted to the Ministry of Fi-nance, line ministries, and donors for approvaland financing (see Annex N, “Eritrea: Coordi-nating with the Government’s Investment Plan-ning Process”).

Once again, these impacts become particu-larly important for large-scale social funds. Thepotential significance depends also on the kindsof goods and services financed by the socialfund—whether or not these are public goods andservices for whose provision government agen-

cies are, or need to be, held accountable—andalso on the nature of budgeting practices in thecountry concerned.

Effects on Local Government Social fund engagement with local governmenthas varied greatly between projects. It can takethree forms: direct training or technical assistancefrom the social fund to local government offi-cials; local government participation as an eli-gible subproject sponsor; and participation bylocal government staff in social fund subprojectcycle functions such as identification, appraisal,and supervision.

Fourteen of the 66 Bank-financed social fundprojects have had specific objectives for build-ing the capacity of local governments,21 while9 of the 24 projects for which detailed data areavailable included provision of substantial train-ing to local government staff. Results have beenmixed, but more positive than at the central


3 8

Yes No Don’t know Does not OtherQuestion N= (%) (%) (%) apply (%) (%)

Social fund budget is included in the

government budget 43 82a 21

If included in government budget

It captures all social funds

expenditure 42 24 14 21 5

It captures only those locally

financed 36

Social fund is covered in the auditor

general’s report to parliament 42 60 21 19

In scrutinizing government accounts,

the public accounts committee

extends its scrutiny to social fund 39 56 18 26

If scrutiny is extended to social fund,

an accounting officer or equivalent

is responsible for answering back to

parliament on proper use of social

fund resources 39 51 3 26 21a. Sixty-three percent as a line item, 19 percent in some other form; Honduras FHIS IV included in both (depends on sector), therefore, total exceeds 100.

Note: N is the number of social fund projects, not the number of social funds.

Source: Survey of social fund task managers.

S o c i a l F u n d B u d g e t a r y A r r a n g e m e n t sT a b l e 4 . 2

level. Local governments have benefited fromtraining in participatory planning; project ap-praisal; and in subproject implementation, op-eration, and maintenance. But formal trainingand other specific capacity-building activitieshave often fallen short of plans. The need to givehigher priority to support through appropriatetraining is recognized in a number of current ornew projects, including Yemen and Zambia (seeAnnex N, “Yemen, Zambia: A Shift Toward ‘Soft-ware’: Changing Social Fund Activities?”).

Effects of “learning-by-doing” have dependedon the extent to which responsibilities have beendelegated to local governments, as contractors forsubproject implementation, and/or through arole in subproject identification, appraisal, andsupervision. Of 60 social fund projects, 60 per-cent included local governments among the en-tities eligible to apply for social fund resources,but only 8 percent worked exclusively throughlocal governments, while 23 percent workedonly with communities.22 Eleven of 23 projectsfor which the information was available reportedno social fund resources at all going to local gov-ernments. Although many social funds workthrough both communities and local govern-ments, projects have given little attention to thequestion of which decisions are most appropri-ately taken at which level for the different typesof investments they are financing.23

Whether local governments or communityorganizations act as the subproject sponsor, therecent trend among social fund projects hasbeen to involve elected local governments and/orlocal line ministry representatives more in iden-tifying, appraising, and supervising subprojectsand in providing technical support. Twelve outof 24 projects have required screening of sub-project proposals by local government, while 7have included direct local government partici-pation in all phases of the subproject cycle. So-cial funds in Sub-Saharan Africa have typicallyworked directly with communities, but this ap-proach is increasingly being adjusted to supportthe growing national commitment to decentral-ization or deconcentration of government func-tions. In Eritrea, Malawi, and Zambia, teams oflocal administration staff carry out desk andfield appraisals of subprojects, while the social

fund officer provides support. Some districtsare now using these teams to manage other ac-tivities. In Zambia, the social recovery projectprovided needed resources at the local leveland the opportunity for the existing district de-velopment coordinating committees (comprisingrepresentatives of the sector ministries at the dis-trict level) to participate in planning the use ofthose resources.

However, as the government’s decentraliza-tion efforts move ahead, far greater coordinationwith district councils (comprising locally electedrepresentatives) as well as the district staff of cen-tral ministries will be needed in allocating proj-ect resources. Otherwise, there is the danger ofmissing not only an opportunity to increaselocal responsibility and accountability but of ac-tually weakening district councils by preventingelected representatives from being held ac-countable for local decisions. The NicaraguaFISE is piloting a micro-planning process as thebasis for subproject selection, giving responsi-bility to municipalities to organize communityworkshops and prioritize proposals in a Mu-nicipal Investment Plan, for financing by FISE andother domestic and international agencies. Sim-ilar efforts are also under way in Honduras.

When local officials have not been given a sig-nificant role, this has usually been justified in termsof the lack of local government capacity in apoorly defined decentralization framework. Nev-ertheless, the Albanian Development Fund wassuccessful in allocating responsibility for sub-project selection and implementation from the out-set to newly created local governments in difficultconditions, making a substantial impact on theircapacity to interact with community leaders, man-age small engineering and construction projects,and conduct competitive bidding (see Annex N,“Albania: Clear Division of Roles and Responsi-bilities, Key to a Successful Partnership”). In othercases, such as Peru and Jamaica, the social fundlargely bypasses the local elected authorities andworks directly with communities, leaving otheragencies to support the local governments.

The question of whether, how, and in whatcircumstances social funds can support decen-tralization is receiving increasing attention. A re-cent comparative study suggests that social funds


3 9

can do this most effectively when appropriatedecentralization policies are in place, when thefund works through local governments insteadof setting up its own structures for communityparticipation, and when the planning processstarts with an open menu of choices rather thana limited subproject menu. When there is no de-centralization strategy, the early experience inBolivia suggests that social funds can still playa positive role in enhancing local governance bydemonstrating the feasibility and the potentialrole of local actors in decisionmaking (Parker andSerrano 2000).

The few projects that have given significantresponsibilities and decisionmaking power tolocal governments have reported some positiveimpacts in building local government capacity.These effects include: increased interaction be-tween the population and local government;strengthened technical, administrative, and fi-nancial management capacity; and, when localdelegates of line ministries are involved, supportto deconcentration of government by strength-ening the position of local staff relative to thecentral ministries. In most cases, the involvementof local governments is mainly in administrativeand technical functions rather than in planningand decisionmaking.

Even when the social fund works directlywith communities and bypasses local govern-ments, some researchers have noted that build-ing capacity at the community level can producepositive “demand effects,” helping to make localgovernments more accountable and responsiveto communities (Parker and Serrano 2000). It iscommon for social fund project beneficiaries toreport increases in expectations (Dijkstra 2000;Dijkstra and Green 2000; Serrano 2000).

The evidence of response-effects on the partof local governments is much more limited. Thefollowing examples indicate the potential forsuch impacts:24 in Armenia, when a member ofa subproject management committee was sub-sequently elected as mayor of the local munic-ipality; in Moldova, where some elected localgovernments are signaling transparency in theiroperations by choosing to work through com-munity project committees that were set up forthe social fund; and, in Argentina, where a priest

took an interest in social mobilization, partici-pated in FOPAR activities, and was subsequentlyelected councilor. The sustainability of theseeffects depends on the decentralization frame-work and on other measures to improve the abil-ity of local and central governments to respond.

Negative effects on decentralization mirrorthose at the government level on planning,budgeting, and accountability. When significantfunctions have been devolved to elected localgovernments, social funds that work directlywith communities diminish a local government’saccountability to the population and inhibit itsbudgeting function (as in Peru and Jamaica).25

JSIF in Jamaica, which has mostly bypassedelected local governments and focused its train-ing activities on NGOs, communities, and a cen-tral agency working with communities, has atbest been missing the opportunity to promotelocal government capacity and accountabilityto local citizens. This tradeoff does not appearto have been explicitly addressed by JSIF. By-passing local governments may have jeopardizedsustainability of some JSIF investments that arein sectors within the Parish Councils’ responsi-bility, such as local roads.26

Working through local governments may notalways be sufficient to ensure that the social fundis supporting, or at least not undermining, localgovernment capacity and accountability. De-pending on the nature of decentralization andthe fiscal framework, social funds can limit thescope for systematic planning and efficient re-source allocation at the local level. This prob-lem emerges when the resources channeled bythe social fund are large relative to the untiedrevenues of the local government. Social fundscan undermine the intergovernmental fiscalframework by introducing a “soft budget con-straint” and inhibiting effective planning. This be-came an issue in Bolivia, for example, wherelocal leaders did not have to prioritize within agiven resource envelope, and thus were notheld accountable for their decisions and couldpass responsibility upward. The social investmentfund appears to have contributed to non-transparency in resource allocation (by trans-ferring resources through discretionary channels)and weak accountability in intergovernmental fis-


4 0

cal relationships. The proliferation of socialfunds within a country, as in Guatemala,27 in-creases this impact as well as the transaction costsfor local governments.

Effects on CommunitiesSocial funds aim to have effects at the commu-nity level through a number of mechanisms:project committee members receive training andgain experience through learning-by-doing; othercommunity members may learn new skillsthrough participation in subproject activities;and the capacity of the community for collec-tive action may be improved through the ex-perience of working together on a subproject.Impacts of the last could include increases incommunity cohesion and the capacity for futurecommunity-based initiatives (“bonding socialcapital”), and improvements in the community’slinks with outside (“bridging social capital”).

Thirty-two percent (21 of 66) of social fundprojects in the portfolio included communitycapacity building among their objectives. Com-munity empowerment was mentioned among so-cial fund objectives in 12 percent of projects. Fivepercent of projects mentioned increasing socialcapital28 and social cohesion among their ob-jectives.29 The Bank’s strategy for social protec-tion proposes further increasing the emphasis onsocial capital in social fund projects.

Capacity Building and Skills Development.Half of the 24 projects for which the data areavailable aimed to provide substantial trainingfor communities—in subproject identification, im-plementation, accounting, organizational man-agement, or maintenance—and 10 providedmoderate or substantial technical assistance.Project supervision and completion reports in-dicate that the planned community training hasoften not been fully implemented, while bene-ficiary assessments have reported requests formore training at the community level. OED re-search (Annex O) in 10 randomly selected com-munities in 2 countries30 found an insignificantimpact of social funds (compared with matchedcommunities) in developing capacity and skillsat the community level. Key informant interviewswith subproject committee members mentionedsome new skills among committee members.

Social Capital Impacts. With respect to socialcapital effects, comparing social capital changesin social fund communities with those in non–so-cial fund communities, OED findings were mixed(box 4.2 and Annex O). A statistically significantpositive impact was found for both bondingand bridging social capital in Jamaica, while inNicaragua no significant impact was found forbonding social capital and a significant negativeimpact was found for bridging social capital.31

In Zambia, there were mixed bonding social cap-ital effects (positive on one variable, negative onthe other), and there were no significant bridg-ing social capital effects. In Malawi, no signifi-cant bonding or bridging social capital effectswere found.

Qualitative data from community-level fieldwork—based on the views of a small number ofself-selected community members—presented amore positive picture and provided examples ofspecific cases where social funds had led to fur-ther collective action and raised hopes and ex-pectations for future development activity. Inkey informant interviews, subproject committeemembers reported that their capacity to dealwith contacts outside their community had beenimproved. The qualitative studies pointed tosome instances of negative effects on social cap-ital through social fund activities (such as alle-gations of corruption relating to subprojectcommittee members or cases of conflict and dis-cord created within the community), but thesewere not common.

An OED institutional analysis based on qual-itative research in Argentina,32 comparing the so-cial capital impacts of two projects, found thatthe social fund project contributed to bondingwithin the community and to the creation of newand better connections between individual com-munities and formal organizations. For example,more and better-articulated demands to localgovernments, increased capacities to contract andoversee technical assistance, and increased abil-ity to deal with banking institutions were noted.The fund was less successful in generatingchanges in the way formal organizations relatedto poor communities because of the scant lever-age the poor can exert when acting as individ-ual communities (Serrano 2000; Annex O). The


4 1

sectoral (agricultural) project in Argentina, in con-trast, increased this leverage by organizing alarge number of the rural poor around a singleidentity shared across communities—that is,being small farmers. This greater leverage al-lowed the poor to pressure national and sub-national governments to take them into accountin policy and budgetary decisions. The creationof intercommunity ties was facilitated not onlyby organizing around a common identity, butalso by involving beneficiaries in state-levelmanagerial structures, accompanying benefici-ary groups for periods of three years or more(compared to an average of eight months for thesocial fund), and adopting a flexible and de-centralized managerial strategy.

Overall, the nature and extent of information-sharing and participation by community mem-bers in social fund subprojects was found to besufficient to allow successful subproject execu-

tion, but not to achieve a consistently significantpositive impact on community capacity or socialcapital (Annex O). These findings on the socialcapital impacts of social fund projects are con-sistent with other sources, which suggest that along-term, internally motivated participatoryprocess is needed to support significant self-development of communities (for example,Narayan and Ebbe 1997). Commentators havealso pointed out that generous grant funding mayactually be an impediment to social capital de-velopment (Ellerman 2001). Among the fourfieldwork countries, the sampled communitiesin Jamaica were the only ones showing a sig-nificantly positive social fund effect for bridgingsocial capital. Possible reasons for the positiveimpacts in Jamaica may have been the pre-dominance of “new” construction in the sampledsocial fund–assisted communities versus reha-bilitation in other cases and the use of partici-


4 2

This analysis is based on OED’s household surveys of 1,687households and 3,056 respondents. Only results that are statis-tically significant on multivariate analysis and are confirmed bythe difference-in-difference approach (t-statistics and chi-square test) are reported here (details are provided in Annex O).

JamaicaBonding social capital: Respondents’ perception of change inthe level of trust and community cooperation between peoplefrom different backgrounds and ethnic groups was significantlygreater for the social fund–assisted communities. Their per-ception of change in the ease of participation in groups and as-sociations of people outside the immediate household wassignificantly higher for the social fund–assisted communities.

Bridging social capital: Their perception of change in the ef-fectiveness of government’s responsiveness to their needs wasalso significantly higher.

MalawiThere were no significant differences between the social fund–assisted communities and the non–social fund communities with

respect to changes in bonding or bridging social capital.

NicaraguaBonding social capital: No significant differences were foundbetween the social fund–assisted communities and the non–social fund communities.

Bridging social capital: Respondents’ perception of changein the effectiveness of the responsiveness of government andcommunity leaders to their needs was significantly lower for thesocial fund–assisted communities.

ZambiaBonding social capital: Respondents’ perception of change ingetting the whole community to agree on a decision was sig-nificantly higher for the social fund–assisted communities. How-ever, their perception of change in the level of trust andcommunity cooperation between people from different back-grounds and ethnic groups was significantly lower.

Bridging social capital: No significant differences werefound between the social fund–assisted communities and thenon–social fund communities.

S o c i a l C a p i t a l I m p a c t s i n F o u r S u r v e y e dC o u n t r i e s H a v e B e e n M i x e d

B o x 4 . 2

Note: The base year was the year shortly before the start of the subproject. In Jamaica and Nicaragua, the base year was 1995 across all communities. In Malawi and

Zambia, the base year varied by community depending on the year of subproject approval.

Source: OED household survey.

patory research in Jamaica, which had high-lighted the importance of communal facilities inaddressing the problem of violence (Moser andHolland 1997;33 Rao and Ibanez forthcoming34),among others.

The findings have implications for the typesof investments that can be successfully sup-ported by social funds, as well as for the appli-cability of the social fund model to serve theobjective of building social capital. The OEDhousehold surveys and institutional analysisshowed that to prepare successful proposals, mo-bilize community contributions, and managesubproject execution, the subproject processhas depended on prime movers in the commu-nity, whose interests are influenced by their po-sition—for example, in the education or healthsectors. At the same time, the relatively brief en-gagement of the social fund with a commu-nity—for planning and execution of a single,small subproject—has meant that the process de-pends largely on the existing organization.

Overall, social funds have operated as usersrather than producers of social capital. Buildingon existing institutions to develop collective ca-pacity in functions that are entirely new to thecommunity concerned (for example, when users’committees are needed to manage water supplyor roads maintenance) takes more time and dif-ferent staff skills than have typically been pro-vided through a social fund. Some recent projectsare including stronger measures to address theseissues, such as allowing early engagement of thesocial fund with the community and allocatingmore resources to training and facilitation. Theresults, including the tradeoffs with other ob-jectives, need to be monitored closely.

Effects on the Private Sector Social fund engagement with the private sectorhas taken three forms: direct support through mi-crocredit or training; creating demand for serv-ices of private business; and changing theenvironment for small business by establishingthe use of competitive procurement and ten-dering procedures where these previously didnot exist. The institutional development impacthas been substantial in countries where thesesectors were particularly weak, and where the

use of force account has been replaced by com-petitive contracting.

About one in four social fund projects havesupported microcredit components,35 with un-even results. Many of the schemes experiencedproblems traditionally associated with such pro-grams, including lack of financial sustainability,not fitting into the country’s financial system, andweak administrative capacity. At least twoschemes in very different country conditionshave been particularly successful. In Albania,strong ownership at the village level made it pos-sible to sustain high repayment rates, eventhrough the growth and collapse of the pyramidschemes and the ensuing civil crisis. By givinghigh priority to sustainability and transferringmanagement of the scheme to dedicated finan-cial institutions, the social fund has been in-strumental in improving Albania’s weak capacityin financial intermediation. Success of the pro-gram in Honduras was attributed to the capac-ity and experience of administering NGOs.

Creation of business opportunities has oc-curred primarily through the increase in de-mand for the work of construction contractors,financial intermediaries, and consulting firms fi-nanced through social funds. In Africa, workingcapital constraints to starting small constructionbusinesses have been overcome through con-tracting at the community level with materialspurchased by the social fund. The picture isdifferent in parts of Latin America where the pri-vate construction sector is already well estab-lished. There is even concern that the incentivesof enterprising contractors can dominate theparticipatory process and bias community de-mand (a concern reflected in project documentsand in the literature). While political leaderssuch as mayors were found to be important, theOED household surveys in four countries did notfind evidence of contractors leading the sub-project identification process. Improvements inthe environment for small and medium-sizebusiness have been greatest where the existingenvironment was particularly weak. Social fundsin some transition countries, such as Armenia andAlbania (and to some extent in Nicaragua), havehad a particular impact in introducing transpar-ent competitive procurement practices, devel-


4 3

oping the capacity of local organizations toapply these practices; facilitating the creation ofnew private contracting enterprises; and, througheligibility requirements, encouraging legal reg-istration of companies.36

Projects that have used both private contrac-tors and the direct hire of local labor for con-struction subprojects have noted a tradeoff.While the use of private contractors can supportprivate sector development and may achievespeed and higher quality of construction, it alsohas a smaller impact on employment within thebeneficiary community.

Effects on NGOs Social fund engagement with NGOs as it relatesto institutional development has taken threeforms: direct support through training or tech-nical assistance; NGO participation as eligiblesubproject sponsors or intermediaries; and sub-contracting of specific activities, such as outreachor service delivery, to NGOs. Eight of the 66 proj-ects included specific objectives related to build-ing the capacity of NGOs.37 NGO staff havereceived training to fulfill their functions eitheras project sponsors or as outreach agents, oftenthrough a specific capacity-building componentof the project (as in Yemen, for example). Themost significant impacts, however, have beenthrough participation as intermediaries or sub-contractors.

Interaction of social funds with NGOs has var-ied greatly between countries and Regions. Ofthe 60 social fund projects for which informa-tion is available, 36 projects (60 percent) haveincluded NGOs as eligible sponsors for sub-projects.38 Typically, in LAC and MNA coun-tries, NGOs have been eligible to apply forsubprojects. African funds, in contrast, haverarely afforded an intermediary role to NGOs,preferring to work directly with communities,which has created competition between NGOsand the fund, prompting criticism by NGOs.The few examples of social funds working withNGOs in Africa are those in Benin, Madagascar,and Malawi.

The degree of success has also been highlyvariable, depending largely on the nature andstrength of the NGO sector, but also on the pre-

paredness of the social fund to work through dif-ficulties, ensure clear mechanisms for NGO par-ticipation, and be adaptable in responding tocapacity-building needs. In Malawi, despite theassessment that local NGOs were not well-suitedto play a significant role, MASAF has now de-veloped an innovative and participatory ap-proach to cooperate with NGOs through onecomponent, giving NGOs a role in sponsoringsubprojects for AIDs victims, street children, or-phans, and other vulnerable groups. In someother countries, the NGOs with the best recordin participatory development projects have beenunwilling to work with what was perceived asa government structure or to accept the time anddisbursement pressures of social funds. In Bo-livia, however, the autonomous status and trans-parent procedures of the social fund made itpossible to establish credibility and working re-lationships with organizations that were ideo-logically opposed to adjustment policies anddistrustful of government programs.

Social Funds’ Institutional RoleContinues to ChangeAlthough most social funds were conceived astemporary organizations with short-term objec-tives, their mandates have been extended. Theyhave continued to attract donor finance, refineoperating processes, and expand their activities.Of the 66 projects in the Bank portfolio, 24have been follow-on projects with existing so-cial fund agencies. There has been little move-ment to reduce their heavy dependence onexternal funding or to integrate them as main-stream government agencies. The lack of clar-ity in the fund’s role relative to other agenciesbecomes a serious issue as the scale and scopeof social fund operations expand.

Unless the social fund is confined to niche ac-tivities, its ultimate institutional developmentsuccess would be demonstrated by either mak-ing itself redundant by building the capacity ofpermanent institutions to take over its functionsor by transforming itself into an institution in-tegrated with mainstream institutions and with-out special status. Early social funds wereexpected to eventually transfer their functions tocentral government institutions, but did not.


4 4

The majority of subsequent projects have had noclear strategy to overcome the anomaly of so-cial fund status as the scale of their activities in-creased in a de facto long-term role, often underpressure from donors and/or politicians. Somestrategies are emerging, however, to limit the au-tonomy of the social fund from the outset, or toallow at least theoretically for their eventualphasing out or integration. For example:• Argentina’s FOPAR has been more integrated

from the outset in the structure of governmentthan other social funds, financing micro-scale,niche subprojects, with a focus on buildingcommunity capacity through community con-tracting, and not overlapping with the corefunctions of other ministries (Serrano 2000).

• The Albania ADF spun off its successful mi-crocredit scheme to dedicated financial in-termediaries.

• In Zambia, the long-term plan under theZAMSIF project is to build the capacity of localgovernments to the point where the socialfund is not needed, with the eventual aim thatfunds will be passed in a block grant from thecenter to districts to manage.

• The first social fund, in Bolivia, is the furthestadvanced in this process (see Annex N, “Bo-livia: From Social Fund to Cofinancing Mech-anism”). It is due to be integrated with similarfunds in a new national municipal fund thatprovides matching grants to local govern-ments for poverty-oriented expenditures.

The viability of these strategies in differentconditions has not yet been proven. Where ef-forts have been made to move the primary focusof some of the older social funds from infra-structure delivery to enhancing development

impact, capacity building, and/or subprojectsustainability, this has proved difficult in somecases (as in Albania, Armenia). The shift re-quires a change in performance incentives,staffing, and skills mix in the implementingagency. It may risk losing the support of donorswho continue to expect the same speed and pre-dictability of disbursements and an explicit linkbetween external funding and tangible assetcreation.

It is not clear that social funds can succeedas instruments for local institutional developmentwithout a longer and more intensive form of en-gagement with local organizations—one thatputs more emphasis on facilitating and foster-ing the growth of capacity for collective actionand self-development. Furthermore, social fundsare constrained by the difficulties they experi-ence in coordinating with other central agencies.Decentralization and privatization measures can-not be expected to work unless they are matchedwith redirected and more efficient roles for cen-tral administration. Effective complementary ac-tions at the central level are essential to thesuccessful impact of the social fund and, evenwhen the social fund focuses on working withlocal governments, it must be aligned with cen-tral government initiatives that aim to improvethe regulatory and fiscal environment for de-centralization. If the social fund bypasses cen-tral regulation of local decisionmaking—forexample, a ministry of local government—it isliable to undermine the intergovernmentalframework on which the decentralization processdepends. The appropriate strategy can only beassessed case by case, depending on the insti-tutional constraints and opportunities in thecountry.


4 5

4 7

Future Bank Supportto Social Funds:Issues to Consider

community-based organizations) and theadvantages of public-private-civil sector part-nerships. In some contexts, social funds cansupport capacity development in local govern-ment when they delegate responsibilities tolocal governments in alignment with the coun-try’s decentralization framework. Examples ofbest practice in social funds (Annex N) illustratesome of the improvements that have been madeover time within the social fund model,especially with respect to the development ofnew processes and procedures for projectmanagement.

However, the creation of new social funds—and increases in the scale of support to exist-ing social funds—requires caution and carefulanalysis of the tradeoffs between specific ob-jectives in a particular country context. Thefindings of this review point to a number of is-sues that warrant more attention in the Bank’ssupport to social fund operations: appropriate-

ness and effectiveness of the social funds’demand-driven mechanism; whether or whensocial funds can be expected to serve long-term development objectives; how they de-pend on, and affect, other public institutions;what scale and sectors of operation are appro-priate; and what transformation or exit strate-gies are indicated. Depending on the countrycontext, addressing these issues may be a mat-ter of modifying social fund operation, of co-ordination with complementary interventionsoutside the social fund, or of adopting an al-ternative instrument. OED reviews of socialfund literature, Country Assistance Strategies, andproject documents suggest that more attentionhas been given in the Bank to how problemsand new demands can be managed within thesocial fund model than to the question of whendevelopment objectives and country conditionswarrant Bank finance through a social fund(and on what scale), and when they do not.

55

Social funds have a clear and significant role in some lending circum-stances, particularly in emergency or post-conflict conditions, whenexisting institutions are ineffective, when the need for flexibility and

speed is paramount, and when deficiencies in small-scale infrastructure area significant constraint to development. They have a comparative strengthin piloting innovation. In the context of centralized planning, they candemonstrate both the feasibility of local decisionmaking (by working with

Issue 1: A key challenge in project design is tostrike the right balance between demand-driven and supply-driven features. In practice, social funds combine both demand-and supply-driven elements, the latter includingeligibility criteria, targeting mechanisms, ap-praisal criteria, and project menus. Similarly,supply-driven programs may offer beneficiarieschoice and include elements of consultationand participation in responding to communityneeds. In most cases, effective service deliverymust combine elements of both demand andsupply, the balance depending on the countrycontext and on the nature and scale of the serv-ices being provided.

The effectiveness of the demand-driven ap-proach as used in social funds has been variable.This review has found that the approach is notnecessarily synonymous with responding to thehighest priority problem of the community, evenwhen the subproject menu includes investmentsthat would address that problem. The commu-nity-based demand-driven mechanism was foundto allow a bias toward certain sectors, in part be-cause of the important role of prime movers insubproject formulation, submission, and imple-mentation. Igniting demand for sectors whereprime movers or community organization do notalready exist, or addressing information asym-metries that have tended to prevent choice fromthe full range of available options, may warrant“supply” of specific inputs, such as outreachand facilitation and/or a different kind of par-ticipatory process. For example, in some con-texts, it is possible that subproject identificationthrough the use of participatory research orother traditional modes of delivery may result insubprojects that are more relevant to communityneeds than a demand-driven, simple invitationfor subproject proposals.1

Issue 2: Social funds may not automaticallyadapt to long-term objectives.While social fund projects have been success-ful in channeling substantial external resourcestoward local development, disbursing rapidly andachieving their physical output targets, their im-pacts on outcomes and welfare variables, andon community capacity building and social cap-

ital, have been mixed. Yet social fund projectsare moving toward these latter, longer-term ob-jectives, and some newer projects give explicitpriority to community development objectives.To this end, more attention would be needed tothe complementary inputs (“software” as well as“hardware”), and to ensuring the institutionalarrangements necessary to achieve long-termimpact from investments. Even if these issues canbe effectively addressed by social funds, it is im-portant to avoid the presumption that buildinginfrastructure is necessarily the first thing thatshould be done to help people develop.2 Theresults, including the tradeoffs with other ob-jectives, need to be watched closely.3

Many social funds that were initially set upas emergency response mechanisms are nowbeing called upon to shift their focus to longer-term development impact and institutional de-velopment objectives, but this transition isproving difficult to implement in some cases.4

The new focus requires significant changes inthe social fund agency’s performance incen-tives, staffing, and skills mix. In taking on thedifficult challenges of long-term development im-pact, sustainable service delivery, and institutionaldevelopment, social funds are liable to meetthe same constraints as other kinds of inter-ventions and may lose the strengths on whichtheir reputation has been built. For example,building capacity and social capital at the com-munity level are time- and human resource–intensive processes, making disbursementspotentially slower and less predictable.Experience suggests that the constraints to ac-complishing this transition should not be un-derestimated and the tradeoffs should beexplicitly addressed.

Issue 3: The suitability and design of socialfund projects depends on the nature of goodsand services to be provided and on theinstitutional context.Nature of goods and services. Coping with thesectoral differences among subprojects is oneof the main challenges of multisectoral pro-grams. Investments in the various sectors requireparticipation of differing natures and degrees,as well as different definitions of beneficiary


4 8

groups, different degrees and sources of co-financing, and different technical expertise. Forexample, the beneficiary group is more difficultto define and user charges are less relevant forroads than for water projects. Standardization ofthe subproject cycle may not lend itself easilyto addressing all the requirements. Social fundsthat have included components for support tomicroenterprise have usually been less suc-cessful in this function, and have experienceddifficulties in combining this activity with grantfinancing for public investments. In practice, so-cial funds have specialized in the delivery ofsmall-scale infrastructure, with a bias in manycases toward the education and health sectors,where the community-level requirements fortechnical, institutional, and financial sustain-ability are usually less complex. The appropri-ateness of a social fund mechanism in servinga specific sectoral activity deserves careful as-sessment in relation to alternative developmentinstruments.

Even when it is considered a high priority toallow communities to choose investments froma multisectoral menu, such a selection processdoes not necessarily imply that delivery (“pro-vision”) of the infrastructure or service itselfhas to be managed by a multisectoral agency.In some cases, technical or other requirementsmay warrant a sector-specific rather than a mul-tisectoral approach to “provision.”

Institutional context. Judgments on the suit-ability of the social fund instrument, or on theappropriate design of a social fund, can only bemade case by case, depending on country teamknowledge and debate of actual constraintsand opportunities at the country level. Contex-tual factors to be considered are outlined in box5.1. Project characteristics can and should bemodified to meet different country conditions,and this is reflected in the diversity of social funddesign. However, when the country contextwarrants fundamental modifications in designand operating procedures of the social fund,then it must be asked if the ideal response isstill a social fund. For example, when the de-centralization framework makes the buildingof local government capacity a high priority, thesocial fund may respond by modifying its pro-

cedures to delegate more of its responsibility tolocal government, but, depending on the extentof delegation, the possible tradeoff with speedand efficiency must be considered. Even whensocial funds work through local governments,in some contexts, such as when the most sig-nificant needs of local government are in the de-velopment of systemic planning and fiscalsystems, the social fund may not be the best in-strument to support decentralization.5 Such con-textual factors vary across countries, and overtime within a country, as the institutional andpolicy environment changes. They must there-fore be continually reevaluated.

Issue 4: The development impact of socialfunds and the sustainability of subprojectsdepend on the progress of broader institutionaland public sector reform.The multisectoral character of social fund proj-ects means that sectoral investments may not al-ways be subject to the full analysis and technicalstandards applied in sector projects. An impor-tant factor in the long-term development impactof social funds is the extent to which they areintegrated in sound strategies for their sectors ofoperation. Where such strategies exist, efforts areneeded at the project design stage to ensure thatsocial fund activities are fully consistent withthem. Where no such strategy exists, the de-velopment impact of social funds will dependon other initiatives to develop sound sectoralstrategies, and on the consistency of social fundactivities with them.

The evidence on sustainability suggests that,in sectors where central or local governmentscarry significant responsibility, it is unrealistic toexpect that operations and maintenance of so-cial fund facilities would differ substantially fromthe country norm. In this case, substantial im-provements would depend on changes in theprevailing fiscal and institutional environment.The Bank should be cautious about investingheavily in social funds unless it is confident thatthe broader institutional constraints can be dealtwith through other interventions. The scopeand scale of activities undertaken by a social fundshould be guided from the outset by assessmentof operations and maintenance capacity and

F u t u r e B a n k S u p p o r t t o S o c i a l F u n d s : I s s u e s t o C o n s i d e r

4 9

constraints in each of the proposed sectors of so-cial fund intervention in the country concerned,including the existing composition and balanceof public expenditure and projected fiscal ca-pacity. When social funds are engaged on alarge scale over a long period in rehabilitating ex-isting infrastructure rather than engaging in newconstruction, they risk becoming an expensivesubstitute for adequate budgetary provision forongoing maintenance of public facilities.

Issue 5: The relationship between the socialfund and mainstream public sector institutionsis critical to the institutional developmentimpact of social funds.The autonomous status of the social fund canhave benefits, but also negative implications. So-cial fund agencies have experienced difficultiesin cooperating with line ministries and localgovernments, and existing mechanisms for co-ordination have not proven fully effective. This


5 0

Strength of Existing Institutions and Public Sector Reform. Au-tonomous social fund agencies can compensate in the shortrun for failure or inertia in the public sector by establishingtransparent procedures, mobilizing resources in the private andcommunity sectors, and focusing on the speedy delivery of vis-ible outputs, allowing time and helping to build public supportfor implementation of systemic reforms. The precise degree ofautonomy—legal, administrative, procedural, financial—thatis warranted will depend on the state of existing institutions andgovernment commitment to reform. The creation of autonomousagencies should be questioned when promising reforms areunder way in a weak but viable public sector, or when publicinstitutions are quite strong. In such cases, it may be better toprovide support directly to agencies that are integrated withinthe public sector, as is the case with Chile’s FOSIS.

National Budgeting Process, Structure of Public Expenditures,and Sectoral Planning. When there is some degree of trans-parency and accountability in national budgetary processes,this is undercut by a social fund that is responsible for allocat-ing significant expenditure without full integration in the gov-ernment’s budgetary process. The existing structure and balanceof public expenditure (inter- and intra-sectoral) should be takeninto account to ensure that social fund expenditures do not ex-acerbate any imbalances within and between sectors and, in par-ticular, toward capital over recurrent expenditure. When reformsto restructure sectoral expenditure are under way, measures areneeded to ensure that social fund expenditures support, or at leastdo not undermine, these reforms. If the social fund operates ona significant scale in relation to sectoral expenditures, the abil-ity of the relevant (central or local) government agency to allo-cate resources according to sectoral and other criteria isdiminished, as the social fund operates its own parallel processof allocating resources.

Extent of Political, Administrative, and Fiscal Decentraliza-tion. The nature and extent of decentralization in a countryshould influence whether or how a social fund is used. Whereno local governments exist, social funds introduce a mechanismfor responding to local needs. Where local governments existbut have no clear mandate or resources for financial manage-ment, social funds can improve their capacity to manage smallinfrastructure projects and to respond to communities. Wherethere is growing national commitment to decentralizationprocesses, however, and where local governments are viable,social funds can undermine the development of local governmentby working directly and primarily with communities for activi-ties that are within the responsibility of local authorities. Whenthe capacity building needs of local governments are best metthrough gaining experience in project management, the socialfund may be appropriate. However, when development of sys-temic planning and fiscal systems at the local level is the higherpriority, other options should also be considered.

Community Social Structure and Capacity. The scope for ef-fective community-level action through a one-time social fundsubproject intervention may be limited to communities that havesome existing organization and are relatively homogenous andunstratified. The current orientation of social fund projects to-ward subproject processing may not be suitable to achieve sig-nificant capacity and social capital enhancements, or nurturea participatory process in communities or sectors where exist-ing local organization is weak, where there is no prime moverto take the lead in seeing the subproject through, or where theprime mover does not represent the interests of the majority ofcommunity members. A different and/or more in-depth and long-term participatory process may be needed to ensure adequatecoverage of poorer communities with weak skills and organi-zational capacity.

W h a t K e y F a c t o r s I n f l u e n c e t h e S u i t a b i l i t ya n d D e s i g n o f S o c i a l F u n d s ?

B o x 5 . 1

can be attributed to weaknesses within line min-istries and/or the lack of priority given by thesocial fund. Negative effects for public sectorprocesses have been reported for some of theolder social funds that have allocated a signifi-cant share of public expenditure in sectors or ac-tivities for which line ministries or localgovernments are accountable (Honduras andNicaragua). In these circumstances, the de factopermanent status of autonomous social funds isdifficult to justify.

Although many social funds supported bythe Bank were originally conceived as tempo-rary institutions, virtually all are still operatingand growing. This is attributed primarily to theirstrong performance in the quick delivery of vis-ible outputs. In addition, there are pragmatic ad-vantages in bypassing bureaucratic constraintsto project implementation and in disbursing rap-idly. Experience suggests a natural inclination tocontinue supporting the operation of the socialfund when substantial capacity has been builtin the social fund agency. As long as donorfunding continues to be readily available, theremay be little pressure for change, either to phaseout the social fund agency or to change its sta-tus and integrate it in mainstream governmentstructures. Within line agencies, institutional-ization of lessons learned has been rare. Exitstrategies have not been defined for the major-ity of social funds, nor have longer-term rolesand responsibilities been satisfactorily clarifiedrelative to other public agencies. Some of thepossible options are as follows:• Social fund maintains semi-autonomous sta-

tus but is confined to micro-scale, niche,“gap-filling” activities.

• Social fund maintains semi-autonomous sta-tus, but as an executing or technical supportagency for central or local governments incompetition with other agencies.

• Social fund maintains semi-autonomous sta-tus on a temporary basis with adequate meas-ures to ensure coordination with otheragencies and the transfer of knowledge, skills,and functions by the end of the period.

• Social fund is transformed into an agencyproviding block grants and technical supportto local governments.

• Social fund is integrated in permanent gov-ernment structure.

A few projects have attempted to confrontthese questions by aiming to transfer increasingresponsibility to local governments, by spin-ning-off specialized activities to permanent in-stitutions, or transforming the social fund into aco-financing mechanism. While promising, noneof these approaches is yet proven, and theyshould be closely monitored to gauge successand potential replicability.

It is important to define from the outset theappropriate nature of the relationship of the so-cial fund with government. This relationshipshould be reexamined as the institutional envi-ronment changes, with attention to any neces-sary changes in the social fund’s scope and scaleof operation or exit strategy as relevant. If socialfunds are to move toward becoming permanentinstitutions in a country, their level of externalsupport should decline over time, with a corre-sponding increase in local financing. This wouldcomprise one measure of the value attached bythe country to the activities of the social fund.

F u t u r e B a n k S u p p o r t t o S o c i a l F u n d s : I s s u e s t o C o n s i d e r

5 1

5 3

Recommendations

Aparticular strength of social funds, and of Bank support to social funds,is their capacity for innovation, flexibility, and learning from expe-rience. This review finds that the performance of social fund proj-

ects has improved over time in many respects, including procurement andcommunity contracting; poverty targeting; attention to sustainability; and, mostrecently, attention to the potential impact of social funds in decentralizationprocesses. Strong dialogue among social fund practitioners ensures contin-uing attention to improvement and to the dissemination of good practices.

Less attention has been paid to the suitability androle of the social fund within a strategic frame-work, to the importance of complementary in-terventions, or to the long-term developmentimpact of Bank support to expand the scaleand extend the mandate of social funds. Im-provements would be achieved by implement-ing the following recommendations.

Strengthen Integration of Social Funds intothe Bank’s Country Assistance and SectoralStrategies, and into Clients’ PRSP whereRelevant.The rationale and objectives of Bank support needto be clear and should drive the choice of instru-ment, rather than the other way around. TheBank should not support a social fund unless pri-orities are explicitly identified and alternative ap-proaches to address those priorities are weighed

in the light of country conditions. This is partic-ularly important for social funds because new ra-tionales are being developed for extension of theirlives under changing conditions. The choice ofa social fund should be clearly justified on thebasis of: (i) the strength of existing institutions andpublic sector reform; (ii) the extent of political,administrative, and fiscal decentralization; (iii)the national budgeting process, structure of pub-lic expenditures and sectoral planning; (iv) com-munity social structure and capacity; (v) thenature and scale of goods and services to be de-livered; and (vi) the capacity of entities respon-sible for operations and maintenance.

Social fund projects should not displace policyreform and should be designed as part of a pack-age of Bank support rather than as isolated in-terventions. Bank support through social fundsshould not crowd-out more policy-intensive ini-

66

tiatives in which the Bank has a comparative ad-vantage in the country concerned, especiallywhen these have a significant prospect of success.At the same time, impact of the social fund proj-ect depends on the success of complementary ini-tiatives. Social funds typically focus on the buildingor rehabilitation of physical infrastructure. For thisinfrastructure to yield a sustainable flow of serv-ices, concurrent programs and policies that en-sure complementary inputs are essential toachieving desired impacts on service delivery, inline with the Bank’s sectoral strategies.

In order to achieve compliance with Banksectoral policies and technical standards andconsistency with country sectoral and publicsector management reform strategies, the Bankshould improve country-team coordination onsocial fund projects. The importance of sector-specific factors to the success or failure of so-cial fund activities argues for more coordinationwithin the country team and sector specialistinput (including social and economic sector,public sector management, private sector de-velopment) in the design and implementation ofsocial funds. Such input is needed to ensure thetechnical viability and sustainability of subpro-jects. It is also needed to ensure that social fundprojects support, or at least do not undermine,sectoral reform strategies and follow sector poli-cies.1 The option of using a sector-specific ap-proach should not be precluded when thetechnical or institutional challenges are particu-larly great. Specialist input and country-teamcoordination are doubly important when thefund accounts for a significant proportion ofpublic expenditure.

Give More Attention to Long-Term Impacts.In design of social fund projects, the tradeoffs be-tween speed and efficiency of subproject pro-cessing and long-term impacts need to beexplicitly acknowledged and addressed, and re-flected in performance indicators.

The continuation or extension of Bank fi-nancing to a social fund should be based on ev-idence of that project’s development impact.Higher priority is needed during project designand implementation of each project to collect-ing data on outcomes, sustainability, and de-

velopment impact. The value of the informationprovided through beneficiary assessments couldbe enhanced through systematic implementationof recommendations already made in the Bank’sreview of social fund beneficiary assessments,including improvements in methodology.

The appropriate role and focus of the socialfund agency and its relationship with existing in-stitutions should be anticipated to the extent pos-sible from the outset, because of the difficultiesexperienced in changing the orientation once theagency is well-established. The use of a socialfund as an emergency response may be sufficientjustification for the autonomy of the imple-menting agency but, when the social fund shiftsto long-term objectives, the autonomy may beless appropriate and should be reexamined.The shift requires clarification of the social fund’saccountability and role relative to other agencies,as well as changes in the staff skills mix and in-centives in the social fund agency, all of whichhave proved difficult to implement.

Ensure Efficiency of Resource Allocation.Social fund projects should ensure that investmentdecisions include a systematic articulation ofthe benefits as well as the costs of alternative in-vestments by the community and/or local gov-ernment concerned,who should also be chargedwith monitoring actual benefits in relation to theirexpectations at subproject appraisal.

Stronger measures are needed to ensure thatbeneficiaries are adequately informed and con-sulted on investment options, costs, and benefits.Depending on the institutional and policy con-text, some alternative or supporting mechanism(such as up-front participatory research, eco-nomic analysis on samples of subprojects) maybe needed to ensure that resources are chan-neled to efficient uses.

The appropriate scale and scope of social fundactivities should be addressed at the project de-sign stage, and reassessed regularly during im-plementation, with reference to budgetaryprocesses and public expenditure analysis. Theissues of allocative efficiency and of sectoralcoordination become important when socialfund disbursements represent a substantial shareof total public expenditure.


5 4

Develop Policy Requirements.Policy requirements for support to social fundprojects should be developed to provide clear cri-teria and procedures for Bank support to socialfunds. The policy requirements should: • Identify conditions that indicate appropri-

ateness or inappropriateness of using thesocial fund instrument and its strategic justi-fication within the CAS and PRSP.

• Specify measures needed for coordinationwithin the country team to ensure that socialfund projects benefit fully from sectoral ex-pertise and support sectoral strategies.

• Provide guidance, from lessons of experi-ence, on the tradeoffs between differentproject objectives and between differentproject design features, and on ways ofmanaging these tradeoffs in varying countryconditions.

• Specify the country-specific information andanalysis needed to (i) establish whether the

institutional and policy context is adequateand appropriate to enable a social fund toachieve its intended development impactand, if not, whether complementary inter-ventions can be expected to overcome theconstraints; (ii) assess the likely positive andnegative institutional development impactsof a proposed project at all levels; and (iii)ensure that project design is aligned with theinstitutional and policy context.

• Specify measures to strengthen monitoring ofimpacts—on the welfare of beneficiaries,including gender effects; on institutionaldevelopment; and on the balance and effi-ciency of public expenditure—during projectimplementation.

• Identify viable transformation or exit strate-gies for social funds in different institutionalcontexts and triggers that will help to avoideither the scale or institutional role of the so-cial fund becoming inappropriate.

R e c o m m e n d a t i o n s

5 5

5 7

ANNEX A: SOCIAL FUNDS PORTFOLIO

ANNEXES

IDA/IBRD ProjectBoard commit. cost

Country Project name FY approval ($M)* ($M)

SUB-SAHARAN AFRICA

Angola Social Action Project 1996 12/21/1995 24 52

Benin Social Fund 1998 5/19/1998 17 20

Burundi Bi-Social Action Project-Twitezimbere 1993 5/13/1993 10 16

Comoros Population & Human Resources 1994 12/14/1993 13 16

Comoros Comoros Social Fund 1998 3/1/1998 12 14

Djibouti Public Works/Social Development 1999 5/25/1999 15 16

Eritrea Community Fund 1996 2/29/1996 18 50

Ethiopia Ethiopia Social Rehabilitation & Development Fund 1996 4/9/1996 108.5 242

Madagascar Social Fund 1 (Food Security & Nutrition Project) 1993 3/18/1993 21 32

Madagascar Social Fund II 1996 9/14/1995 40 45

Madagascar Social Fund III 1999 3/23/1999 15 18

Malawi Social Action Fund 1996 5/9/1996 56 65

Malawi Social Action Fund II 1999 10/15/1998 66 70

Mali Grassroots Initiatives to Fight Hunger & Poverty 1998 4/7/1998 22 22

Rwanda Food Security and Social Action Project 1992 6/1/1992 19 46

Rwanda Community Reintegration & Development Project 1999 10/15/1998 5 5

São Tomé Multisector Project 1989 6/14/1989 5 8

São Tomé Second Multisector Project 1991 6/27/1991 6 6

Togo Pilot Social Fund 1999 4/14/1999 5 6

Zambia Social Recovery Project 1991 6/19/1991 20 49

Zambia Social Recovery Project II 1995 6/28/1995 30 65

Zimbabwe Community Action Program 1998 5/19/1998 60 60

EAST ASIA AND PACIFIC

Cambodia Social Fund 1995 6/8/1995 20 22

Cambodia Social Fund II 1999 3/25/1999 25 25

Philippines SZOPAD Social Fund 1998 3/24/1998 10 15

Thailand Social Investment Project 1999 7/9/1998 300 462

S o c i a l F u n d P r o j e c t s A p p r o v e d b y t h eB o a r d B e t w e e n F i s c a l 1 9 8 7 a n d E n d - F i s c a l1 9 9 9


5 8



EUROPE AND CENTRAL ASIA

Albania Rural Pilot Poverty Alleviation 1993 2/23/1993 2 5

Albania Rural Development Project 1995 2/14/1995 6 12

Albania Urban Works & Microenterprise Pilot 1996 8/1/1995 4 4

Albania Community Works Project 1999 1/19/1999 9 17

Armenia Social Investment Fund 1996 11/9/1995 12 20

Bosnia-

Herzegovina Emergency Public Works and Employment Project 1997 7/30/1996 10 45

Bulgaria Bulgaria Regional Initiative Fund 1999 11/20/1998 5 6

Georgia Social Investment Fund 1998 11/12/1997 20 28

Moldova Social Investment Fund 1999 2/16/1999 15 15

Romania Social Development Fund 1999 1/19/1999 10 27

Tajikistan Pilot Poverty Alleviation Project 1997 4/10/1997 12 13

LATIN AMERICA AND CARIBBEAN

Argentina Social Protection Project 1996 11/21/1995 152 652

Belize Social Investment Fund 1997 3/18/1997 7 12

Bolivia Emergency Social Fund Phase I 1987 6/23/1987 10 0

Bolivia Emergency Social Fund II 1988 3/1/1988 27 55

Bolivia Social Investment Fund 1990 4/24/1990 20 96

Bolivia Social Investment Fund II 1993 6/29/1993 40 70

Ecuador Third Social Development Fund: Emergency SIF 1994 2/22/1994 30 120

Guatemala Social Investment Fund 1993 11/24/1992 20 80

Guatemala Social Investment Fund II 1999 11/12/1998 50 64

Guatemala Reconstruction & Local Development Project 1999 7/28/1998 30 38

Guyana Health, Nutrition, and Water Project 1993 8/7/1992 10 14

Haiti Economic and Social Fund 1991 1/17/1991 11 23

Honduras Social Fund 1991 2/28/1991 20 68

Honduras Social Investment Fund II 1992 6/25/1992 10 68

Honduras Social Investment Fund III 1996 7/11/1995 30 113

Honduras Social Investment Fund IV 1999 7/14/1998 45 137

Jamaica Social Investment Fund 1997 9/5/1996 20 50

Nicaragua FISE I 1993 11/17/1992 25 68

Nicaragua FISE II 1996 7/11/1995 30 102

Nicaragua FISE III 1999 11/10/1998 45 166

Panama FES 1997 6/17/1997 28 80

Peru Social Development Fund 1994 12/16/1993 100 497

Peru FONCODES II 1997 7/16/1996 150 150


A n n e x e s

5 9



MIDDLE EAST AND NORTH AFRICA

Algeria Social Safety Net 1996 4/25/1996 50 79

Egypt Emergency Social Fund 1991 6/21/1991 140 400

Egypt Social Fund II 1996 5/21/1996 120 775

Egypt Social Fund III 1999 6/1/1999 50 65

Yemen Social Fund for Development 1997 5/22/1997 30 85

SOUTH ASIA

Pakistan Poverty Alleviation Fund 1999 6/17/1999 90 90Note: The West Bank and Gaza Community Development Project (included in the Social Protection Unit list) is omitted as the project was financed from a trust fund and was not an IDA-or IBRD-financed project. The more recent Social Protection Unit list of 11/28/2000 includes three additional projects: Sri Lanka Poverty Alleviation Project, Argentina Fourth Social Pro-tection Project, and Ghana Community-Based Poverty Reduction Project. It excludes one project: Rwanda Community Reintegration & Development Project. The figures for IDA/IBRD com-mitments represent total project costs, even though in some cases the social fund was only one component of the project. Source: World Bank data.


6 1

Literature ReviewThe literature review included an examinationof the conceptual as well as empirical literatureon social funds, sustainability, institutional devel-opment, social capital, and organizational learn-ing. The analysis was informed, in particular, bytwo recent Bank studies on social funds: SocialFunds 2000 and Promoting Good Local Gover-nance through Social Funds and Decentraliza-tion (Parker and Serrano 2000)

Portfolio AssessmentThe portfolio assessment was based on the fol-lowing analyses: (i) in order to trace the diver-sity of social fund features across countries andthe evolution in their design and implementa-tion, task managers were requested to fill in anOED basic information questionnaire, sustain-ability matrix, and institutional developmentmatrix. The basic information questionnaireelicited responses for 41 projects (out of 66 forwhich the information was requested). The sus-tainability and institutional development matri-ces elicited responses for 24 projects (out of 38for which this information was requested). Wheretask manager input was not available, data wereobtained from SARs/PADs, operational manualswhere available, and project files; (ii) reviews ofproject status reports for 26 social fund projects1

and available mid-term reviews; (iii) reviews of16 social fund projects undertaken by the Qual-ity Assurance Group (QAG); (iv) analysis ofimplementation completion report reviews for23 closed social fund projects; (v) review ofperformance audit reports produced by OED for7 closed social fund projects; (vi) regionalreviews of social fund projects in East Asia andthe Pacific, Europe and Central Asia, Latin Amer-ica and the Caribbean, the Middle East and

North Africa, South Asia, and Sub-Saharan Africa;and (vii) ratings for social fund projects assignedby OED and QAG.

Country Cases Field research was conducted in Jamaica, Malawi,Nicaragua, and Zambia. These countries were cho-sen for their long-standing and substantial socialfund experience and for the diversity of imple-mentation arrangements they represent. Second-tier field research (without household surveys)was undertaken for Argentina, Bolivia, and Eritrea.

National, Local, and Social Fund Agency-level Institutional Analysis (comprising semi-structured interviewing). In most cases, twomissions of two weeks each by two internationalexperts, working in some cases with a localresearcher, were undertaken in the four coun-tries to analyze the effects of social funds at thegovernment level and the sustainability of theirbenefits. Social fund effects at the level of NGOsand the private sector were also assessed. Themissions conducted semi-structured interview-ing of stakeholders—officials in the social fundagency, ministries, district governments, NGOs,other members of civil society, beneficiaries,and donor representatives. The respondentswere selected purposively and by “snow balling”(each respondent was asked to name otherpotential respondents). In order to maintaincomparability across the four countries, theexperts were guided by the Social Funds Eval-uation Toolkit. Differing perspectives weregathered on the same issue from various stake-holders. Data collection included both arrange-ments in actual practice as well as on paper,noting reasons for any divergence.

Community-level Analysis (comprising quan-titative and qualitative surveys). Two local

ANNEX B: REVIEW INSTRUMENTS

research firms—one each in Africa and LatinAmerica—were commissioned to collect com-munity-level data regarding the sustainability,social capital, and capacity building effects ofsocial fund projects. A number of survey toolswere used: household questionnaire, key inform-ant interviews guidelines, focus groups guide-lines, and facility survey questionnaire. Thesurvey tools were developed by OED (except forthe facility survey, which was taken from SocialFunds), commented upon by Bank and socialfund staff, and pilot tested in the field. OED’scommunity-level field research comprised house-hold surveys (through which a total of 3,056respondents in 34 communities were inter-viewed) and qualitative research including keyinformant interviews and focus groups in 4countries. The household surveys were con-ducted in 3 to 5 randomly chosen socialfund–assisted communities and 3 to 5 matchednon–social fund communities in each of the4 countries and was administered on approxi-mately 50 randomly chosen households in eachcommunity for a total of 1,687 randomly-selectedhouseholds (845 social fund–assisted house-holds and 842 households in the matched com-munities). In the former 1,525 respondents wereinterviewed (roughly 2 per household, house-hold head and where available another house-hold adult of opposite sex). This amounted to284 respondents in 4 communities in Jamaica,

499 respondents in 5 communities in Malawi, 252respondents in 3 communities in Nicaragua, and490 respondents in 5 communities in Zambia.Among non–social fund households, 1,531respondents were interviewed (294 in Jamaica,493 in Malawi, 257 in Nicaragua, 487 in Zambia).All results are reported by respondent unless indi-cated otherwise. The country-level percentagesrefer to responses from the sampled householdsin each of those countries rather than from thepopulation as a whole. The matched non–socialfund communities were identified using datarelating to geographic, socioeconomic, and cul-tural characteristics that were provided by socialfund agency staff, key informants, and the localresearch firms. Details of the number of surveysconducted in each of the four countries are pro-vided in table B.1. The purpose of this com-munity-level field research was to understandsocial fund decisionmaking and participatoryprocesses at work on the ground. Conclusionsrelating to the portfolio are based on a triangu-lation of information from OED’s various datasources, described in this Annex.

Stakeholder SurveyThe World Bank Institute administered the OEDstakeholder survey to 120 participants of the Sec-ond International Conference on Social Fundsheld in Washington, D.C., in June 2000. Stake-holders represented at the conference included


6 2

Jamaica Nicaragua Malawi Zambia

Number of social fund subprojects/communities 4 3 5 5

Number of matched communities/projects 4 3 5 5

Number of households interviewed in each community 48–52 49–50 48–51 47–50

Number of individuals interviewed 578 509 992 977

Number of individuals interviewed within each householda 1–2 1–2 1–2 1–2

Number of focus groups in each community 1–2 1–3 5–9 5–9

Number of key informant interviews in each community 2–4 2–5 4–6 4–5

Number of facility surveys in the social fund–assisted communities 3 2 5 5a. In each household, the surveyors were instructed to conduct two interviews: the first with the principal respondent—the household head or other knowledgeable household member—and the second with an adult (above 16 years of age) of the opposite sex chosen randomly from all adults in the household. The purpose of randomly selecting a household member ofthe opposite sex is to pick up any gender differences. In practice, the second interview was not always possible because there was no member of the opposite sex living in the household.

S a m p l e S i z e f o r C o m m u n i t y S t u d i e s( T o t a l N u m b e r o f I n d i v i d u a l s I n t e r v i e w e d :3 , 0 5 6 )

T a b l e B . 1

social fund agency staff, government officials,NGO staff, and donor agency staff. The same sur-vey was administered by telephone to 40 addi-tional stakeholders from stakeholder groups andregions underrepresented among the confer-ence participants. These additional stakeholderswere identified from project files or by socialfund task managers. Survey questions coveredviews about the role and function of socialfunds, their impact on government and com-munities, and the sustainability of their benefits.The survey was completed by 160 of roughly 300conference participants. Thirty-five percent of therespondents were from social fund agencies,17 percent from NGOs, 14 percent from gov-ernment, 12 percent from the World Bank, and8 percent from other donors agencies; 14 per-cent were affiliated with some other organiza-tion2; and 1 percent did not indicate institutionalaffiliation. Nineteen percent of the respondentsworked on countries in the Africa Region, 6percent in EAP, 23 percent in ECA, 19 percent

in LAC, 14 percent in MNA, 3 percent in SouthAsia, and 16 percent in other regions or had miss-ing entries.

World Bank Country Directors/ResidentRepresentatives SurveyA survey was administered to Bank countrydirectors and resident representatives in coun-tries with social fund projects. The questions cov-ered the performance of social fund projectscompared with other projects in the countryportfolio along several dimensions. The surveywas sent to 51 country directors/resident rep-resentatives, 13 of whom completed the survey.

World Bank Sector Director SurveyA survey was administered to Bank sector direc-tors concerning their views on the performanceof social fund projects compared with otherprojects in the sector portfolio in their Region.The survey was sent to 33 sector directors, 7 ofwhom completed the survey.

A n n e x e s

6 3

6 5

The Regional distribution of social fund projectsis dominated by Latin America and the Caribbeanand Sub-Saharan Africa, but the share of socialfund disbursements in other Regions hasexpanded (figure C.1). The earliest funds werein Latin America, starting with Bolivia in 1987.This Region has continued to consume a largethough declining share of social fund disburse-ments. Sub-Saharan Africa has also been animportant recipient of social fund projects,

beginning with São Tomé and Principé in fiscal1989. In the Middle East and North Africa, dis-bursements have been dominated by three proj-ects to support the Egyptian social fund sincefiscal 1991. Other Regions have come to socialfunds more recently, with a steady growth ofsocial funds across Europe and Central Asia,beginning with Albania in fiscal 1993, and expan-sion in East Asia, beginning with Cambodia infiscal 1995.

ANNEX C: REGIONAL DISTRIBUTION OF SOCIAL FUNDS

Perc

ent o

f soc

ial f

und

disb

urse

men

ts

South AsiaMiddle East and North Africa

Europe and Central Asia

Latin America and Caribbean

100

90

80

70

60

50

40

30

20

10

0

East Asia and Pacific

Africa

1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

T h e R e g i o n a l D i s t r i b u t i o n o fS o c i a l F u n d P r o j e c t s S h i f t e d a st h e A p p r o a c h S p r e a d

F i g u r e C . 1

Note: These data include total Bank disbursements to projects included in the Social Protection Unit’s database of social fund projects, even when the social fund isonly one of the components. Because of the inclusion of a large project in Thailand that had a social fund component, the growth of disbursements in the East Asiaand Pacific Region is overstated by roughly 50 percent.

6 7

ANNEX D: SOCIAL FUND FINANCING BY ACTIVITY

Activity Projects (%) Activity Projects (%)

School construction/rehabilitation 94 Culverts 38

Construction and/or rehabilitation of piped water Footpaths 38

supply systems 91

Construction/rehabilitation of health facilities 85 Provision of teaching supplies and/or educational

furniture 53

Road construction/rehabilitation 83 Markets 34

Bridge construction/rehabilitation 60 Erosion control 32

Drainage works 57 Forestry 30

Construction and/or rehabilitation of wells,

handpumps 55 Flood control 28

Irrigation works 53 Microcredit 25

Training, capacity building, technical assistance 53 Housing/dwellings for children, elderly 23

Solid waste disposal, sanitation works 45 Agroprocessing 21

Latrine construction/rehabilitation 43 Infrastructure for street children, homeless 19

Provision of medicines/equipment 42 Electrification 15

Sports complexes, community centers, day care

centers 40Note: N=53. Source: Task manager responses to the basic information questionnaire and, where these were unavailable, SARs/PADs and operational manuals.

6 9

ANNEX E: SOCIAL FUNDS 2000 RESULTS

Country Indicator Finding

Bolivia Schools 1994–1998: US$65.5m disbursed for building and improving schools.

Health facilities 1994–1998: US$21.7m disbursed for building and improving health posts.

Water/sanitation US$39.1m disbursed for water and sanitation system building and improvement.

1993–1997: Increase in percentage with piped water by 16% in Chaco and 10% in Resto Rural.

Increase in percentage with sanitary services by 3% in Chaco and 44% in Resto Rural.

Frequency of water availability decreased from 21.95 to 19.38 hours in Chaco and increased

from 18.49 to 21.15 hours in Resto Rural. Distance to water source reduced by more than 50%.

Honduras Schools 1995–98: 503 new primary schools built (58% of all new primary schools), and 2326 new

classrooms (which is 61% of all new classrooms). Improved 4163 classrooms (15% of the

national stock).

On average for the previous week, 32% of the morning and 17% of the afternoon shift lost due

to staff leave, holidays and parent strikes.

Health facilities 1994–98: 127 new rural health posts built (72% of all new rural health posts), and 90

remodeled (45% of previously existing stock); 40 new urban health posts constructed (56% of

all new urban health posts), and 108 improved (17% of previously existing stock); one mother

and child clinic constructed, 8 improved.

For rural health posts, on an average 11 visitations per day compared to the national average of

9 per day. Adding FHIS health centers, there were an average of 54 consultation per day

compared to the national average of 34 consultations per day.

“Alarming incidence of closures” (page 47)—1.9 and 1.5 days in previous week for FHIS and

non-FHIS projects respectively; reasons provided in the report are meetings with local

government, holidays, and health committee meetings.

Water 1994–1997: 2815 new connections built (1.5% of all new connections) and 46000 connections

upgraded (5% of all).

92% of FHIS households have piped water compared to 87% non-FHIS, and FHIS investments

reduced the average expenditures on water and reduced time spent collecting water

(42 minutes per month).

Water availability for 25.3 days per month (12.7 hours per day) for FHIS compared to 18 days

per month (10.9 hours per day) for non-FHIS.

89% of the households with FHIS supplied water drink it (directly or after boiling) compared to

63% of the households with the water supplied in the control group.

A c c e s s a n d U t i l i z a t i o n : S o c i a l F u n d s2 0 0 0 R e s u l t s

T a b l e E . 1


7 0


Sewer/Latrines 1994–1997: 132,000 sanitary services constructed (17% of all new sanitary services).

92% of FHIS households had a sewer in the street (compared to 7% in non-FHIS). 40% of FHIS

households were actually connected to the sewer system.

65 percent non-FHIS households had no sanitary facility compared to zero percent FHIS

households. 100 percent usage by everyone in FHIS compared to 98 percent in the control group

households.

Nicaragua Schools Increase in number of classrooms from 4.6 to 6 in FISE compared to an increase from 3.5 to 4 at

national level. Increase in school size (enrollment) in FISE schools of 21% compared to zero

percent in the control group.

Health facilities 1993–1997: Increase in average daily number of visits to a FISE post from 11.3 to 17 compared

to an increase from 8.8 to 14 in non-FISE posts.a The increase in visitation was larger among

women in FISE facilities.

Water About a fourth more households have access to piped water in FISE investment areas. Distance

to nearest water source reduced by 600 meters.

Connection rates are around 90% of systems’ capacity, and half the systems are functioning at

above 100% capacity.

Sewer/latrines Net increase in access to flush toilet of almost 32% and decrease of households without

latrines by almost 20% (1993–1998) in FISE compared to the control group. Connection rates for

FISE are low, but differ greatly between systems within and outside Managua.

Peru Schools Average increase of 34 students, 1 classroom and 1.6 grades per school.

(APOYO) Water 7% of the systems were not functioning (higher for older projects). On an average the system

did not function 4 times in the previous year.

The water availability described as permanent varies between 67% and 80% of households by

region. The quality of water is perceived to have improved.

Household connection saved on average of 79 minutes per day and public stand pipes saved 63

minutes per day.

Sewer/latrines 27% of the households reported that the system was not working (higher in more recent years

reflecting low connection rates)

89% of the households use latrines as intended.

Zambia Schools Increased demand for school. Enrollment in SIF schools increased by 17% between 1992–98

versus decrease in 1.8% in the control group.

Health facilities 1993–1998: Hours of operation increased in SIF for all services, declined or stagnated in non-

SIF for all but family planning. Increase in hours of service for family planning four times as high

in SIF facilities as compared to the control group facilities.

Average increase of 567 cases seen for SIF, drop of 271 cases seen for the control group. Also,

23% increase in maternal deliveries and 26% increase in child attendance in SIF centers.a. While utilization of both FISE and non-FISE health centers increased significantly, a slightly greater and significant increase was observed in the FISE health posts.

A c c e s s a n d U t i l i z a t i o n : S o c i a l F u n d s2 0 0 0 R e s u l t s

T a b l e E . 1

A n n e x e s

7 1


Intermediate indicators

Bolivia EDUCATION (1993–97)

Chaco Enrollment rate (5–12)d Insignificant difference between social fund and the control group.

Regularity of attendanced Insignificant difference between social fund and the control group.

Repetition rated Insignificant difference between social fund and the control group.

Drop-out rated Significant reduction when measured from household data and insignificant using

administrative data.

Mean test scoresd Insignificant difference between social fund and the control group.

Chaco Registered studentsc Negative social fund impact.

& Effective studentsc Positive social fund impact.

Resto Repeatersc Insignificant difference between social fund and the control group.

Rural Drop-out ratec Positive social fund impact.

HEALTH (1993–97)

Pre-natal controlc Positive social fund impact.

Births attended by trained

personnelc Insignificant difference between social fund and the control group.

Use of health services

conditional on illnessc Insignificant difference between social fund and the control group.

Treatment of diarrheac Insignificant difference between social fund and the control group.

Treatment of coughc Negative social fund impact.

Diarrhea & cough incidencec Insignificant difference between social fund and the control group.

Infant and child mortalitya Positive social fund impact.

WATER (1993–1997) Results based only on social fund subprojects

Knowledge and use of ORT Increase overtime.

Level of fecal contamination Reduction over time.

Infant and child mortalityb Significant decline overtime.

Diarrhea incidence Decline overtime in Chaco. No change overtime in Resto Rural.

Honduras EDUCATION

Gross primary enrollment

(6–12)a Insignificant difference between social fund and the control group.g

Grade for agea Positive social fund impact.

HEALTH

Among the sick, those that

sought professional helpa Positive social fund impact.

WATER

Diarrhea incidenceb Insignificant difference between social fund and the control group.h

LATRINESi

Diarrhea incidencea Significant social fund impact using bivariate analysis, insignificant social fund

impact using multivariate analysis.

SEWERAGE

Diarrhea incidenceb Insignificant difference between social fund and the control group.

O u t c o m e s a n d W e l f a r e I m p a c t : S o c i a lF u n d s 2 0 0 0 R e s u l t s

T a b l e E . 2


7 2


Nicaragua EDUCATION

Net enrollment rateb Positive social fund impact.

Grade for ageb Insignificant difference between social fund and the control group (but age in first grade

significantly lower).

Absenteeismb Weak negative—negative social fund impact using propensity control and insignificant

social fund impact using pipeline control.

Education gapb Positive social fund impact.

Repetitionb Weak positive—positive social fund impact using propensity match and insignificant

using pipeline control.

Days not assisted classes Weak negative—negative social fund impact using propensity control and insignificant

last monthb social fund impact using pipeline control.

HEALTHj

Contact rate under sixb Weak positive—positive social fund impact using propensity score match and insignif-

icant difference between FISE assisted and the control group communities using

pipeline control.

Contact rate above fiveb Weak negative—no significant difference using propensity score match and negative


Contact rate Weak positive—positive social fund impact using propensity score match and insignif-

icant difference between FISE assisted and the control group communities using

pipeline control.

Contact rate under six

with diarrheab Positive social fund impact using both propensity match and pipeline control.

Pre-natal careb Weak negative—negative social fund impact using propensity score match and insignif-

icant difference using pipeline control.

Institutional birthsb Weak positive—no significant difference using propensity score match and positive


Vaccine coverageb Insignificant difference between social fund and the control group.

Diarrheab Weak negative—negative social fund impact using propensity and insignificant social

fund impact using pipeline control match.

Respiratory infectionsb Weak negative impact—negative social fund impact using propensity score match;

Insignificant difference between FISE assisted and the control group communities

using pipeline control.

Stuntingb Positive social fund impact.

Wastingb Weak positive impact—positive social fund impact using pipeline control and insignif-

icant social fund impact using propensity match.

Underweightb Weak positive impact—positive social fund impact using pipeline and insignificant social

fund impact using propensity score match.

WATER

Diarrheab Insignificant difference between social fund and the control group

Stuntingb Positive social fund impact.

Wastingb Insignificant difference between social fund and the control group.


T a b l e E . 2

A n n e x e s

7 3


LATRINEk

Diarrheab Insignificant difference between social fund and the control group.

Stuntingb Insignificant difference between social fund and the control group.


SEWERAGE

Diarrheab Insignificant difference between social fund and the control group.

Stuntingb Insignificant difference between social fund and the control group.


Peru EDUCATION

Attendance rate (6–11)e Positive social fund impact.

Grade for agee Insignificant difference between social fund and the control group.

(APOYO) Enrollment ratee Insignificant difference between social fund and the control group.

Attendancee Insignificant difference between social fund and the control group.

Absenteeism due to illness

over last monthe Significant reduction.f

Accumulated years of

educatione Significant improvement.f

WATER

Child Mortality ratee Significant reduction.f

Incidence of diarrheae Reduced significantly for households with house connections.f

Incidence of dysenterye Reduced significantly.f

SEWER

Incidence of diarrheae Insignificant difference between social fund and the control group.f

LATRINE

Incidence of diarrheae Reduced significantly for households with house connections.f

Incidence of dysenterye Reduced significantly.f

Zambia EDUCATION

Attendance rate, 7–12b Weak positive—positive impact using pipeline match and insignificant difference

using propensity score match.

Attendance in gradeb Weak positive—positive impact using pipeline match and insignificant difference

using propensity score match.

Education share of Social fund households spend significantly higher proportion of their expenditure on

household expenditureb education as compared to the control group households.

Expenditures of households Overall, an insignificant difference between social fund and the control group—with

with primary-school-age significantly larger expenditures in rural areas and weak positive in urban areas

childrenb (lower using pipeline and insignificant using propensity score).

HEALTH

Household member sickb Significantly higher percentage from social fund communities reported sick as com-

pared to the control group communities.l

Treatment (if sick)b Significantly lower percentage from social fund communities sought treatment as

compared to the control group communities.


T a b l e E . 2


7 4


Went to hospital (if sick)b Significantly lower percentage from social fund communities went to hospital as com-

pared to the control group communities.

Went to health facility Significantly higher percentage from social fund communities went to local health post

(if sick)b as compared to the control group communities (using pipeline match); Insignificant dif-

ference using propensity score match.

Health share of household Insignificant difference between social fund and the control group, but significant

expenditureb difference in rural areas.

Vaccination/Childb Weak positive impact for some vaccination types; Insignificant differences between social

fund and the control group for most vaccinations.

Diarrhea incidence

(all households)b Insignificant difference between social fund and the control group.

Diarrhea incidence (if sick)b Weak positive—positive social fund impact using propensity score match and insignif-

icant social fund impact using pipeline match.

Stuntingb Insignificant difference between social fund and the control group (overall); but posi-

tive social fund impact for rural Zambia

Wastingb Negative social fund impactNote: Only results significant at 90 percent confidence level have been reported.a. Result holds for multivariate analysis.b. Bivariate analysis of ex-post difference with control (that is, not double difference). c. Difference-in-difference impact.d. Eligibility on randomization (only Bolivia).e. It is not clear if the results are from bivariate or multivariate analysis.f. It is not clear if this is a before/after comparison, with/without comparison, or difference-in-difference approach.g. According to the research on Honduras for Social Funds 2000, the study team did not expect to find an impact in enrollment due to the already high primary school enrollment rate,which would have required a much larger sample size to draw conclusions about this variable.h. Honduran research explains these findings by the fact that FHIS did not emphasize increased coverage in water but rather rehabilitated run-down systems (so the main benefits wereeconomic gains to houses that already had potable water).i. According to research on Honduras for Social Funds 2000, the fact that FHIS placed latrines were none existed previously and that sanitary services radically reduce diarrhea leads tothe inescapable conclusion that the FHIS program in this area has had a clear and positive impact.j. The researchers working on the Nicaraguan data consider the health results inconclusive.k. The researchers explain the lack of impact of latrines on health on the high existing access to latrines among the control group.l. The researchers in Zambia noted that health interventions increased awareness of health issues, so that people more frequently report themselves sick when they have minor maladiesthat previously they would not have considered worthy of mention.Source: Social Funds 2000 background research; World Bank data.


T a b l e E . 2

A n n e x e s

7 5

Bol

ivia

Hon

dura

sN

icar

agua

Peru

(APO

YO)

Zam

bia

Sam

ple

size

SCH

OO

L FA

CILI

TY

Sam

ple

size

Qual

ity o

f

faci

litie

s

Su

st

ai

na

bi

li

ty

:

So

ci

al

F

un

ds

2

00

0R

es

ul

ts

Ta

bl

e

E.

3

7,30

0 ho

useh

old

surv

eys.

66 S

IF &

72

non-

SIF

scho

ol s

ur-

veys

; 940

SIF

& 1

,020

non

-SIF

hous

ehol

d su

rvey

s.

For

Chac

o, a

nd C

haco

& R

esto

Rura

l (co

mbi

ned)

, SIF

scho

ols h

ad

signi

fican

tly h

ighe

r per

cent

age

of

clas

sroo

ms i

n go

od co

nditi

on a

nd

signi

fican

tly be

tter s

anita

tion s

erv-

ices

than

the

non-

SIF s

choo

ls. S

IF

scho

ols h

ad b

ette

r acc

ess t

o el

ec-

trici

ty (2

0%) c

ompa

red

to n

on-S

IF

scho

ols

(17%

).

2,59

0 ho

useh

old

surv

eys.

12 FH

IS (a

ll op

erat

ing)

& 1

2 no

n-FH

IS sc

hool

surv

eys;

320

SIF

& 3

20 n

on-S

IF h

ouse

hold

surv

eys.

Proj

ect s

urve

y: F

HIS

scho

ols

had

bette

r con

-

stru

ctio

n, f

urni

ture

ava

ilabi

lity,

deco

ratio

n

and

clean

lines

s tha

n no

n-FH

IS fa

ciliti

es. H

ow-

ever

, som

e 28 p

erce

nt of

the w

ater

clos

ets a

nd

elec

tric

inst

alla

tions

bui

lt by

FHI

S w

ere

not

wor

king

, com

pare

d w

ith 4

per

cent

of t

hose

built

by o

ther

age

ncie

s. 7

5% o

f FHI

S sc

hool

s

repo

rted

prob

lem

s w

ith p

iped

wat

er c

om-

pare

d to

83%

in n

on-F

HIS

scho

ols.

How

ever

,

25%

cite

d fa

ulty

inst

alla

tion

(com

pare

d to

15%

in n

on-F

HIS)

, 33%

secu

rity (

com

pare

d to

15%

in n

on-F

HIS)

, 25%

lack

of s

pace

(com

-

pare

d to

4%

in n

on-F

HIS)

as a

pro

blem

.aFI

SE

scho

ols w

ere

wor

se th

an th

e no

n-FIS

E sc

hool

s

for

(i) a

cces

s to

saf

e w

ater

(77%

vs

91%

)

and

(ii) a

cces

s to

ele

ctric

ity (5

4% v

s 64

%).

Bene

ficia

ry H

ouse

hold

: Qua

lity o

f FHI

S w

orks

was

ass

esse

d as

“go

od”

by 7

3%; “

regu

lar”

by 1

9% a

nd b

ad b

y 6%

. And

supe

rvis

ion

was

asse

ssed

as

“effi

cien

t” b

y 58

%.

1,31

0 ho

useh

old

surv

eys

& 4

040

natio

nal s

urve

ys.

24 FI

SE &

24

non-

FISE

scho

ol su

rvey

s;

240

SIF

& 2

40 n

on-S

IF h

ouse

hold

surv

eys.

Facil

ity su

rvey

: FIS

E sc

hool

s wer

e sig

-

nific

antly

bet

ter

than

the

non

-FIS

E

scho

ols f

or (i

) acc

ess t

o ra

in d

rain

age

(38%

vs

7%),

(ii)

wat

er t

hrou

gh

pipe

line

(68%

vs 3

7%),

(iii)

acce

ss to

safe

wat

er (8

7% v

s 64

%),

(iv) e

lec-

trici

ty a

vaila

ble

durin

g sc

hool

hou

rs

(56%

vs 2

3%) a

nd (v

i) com

mun

ity w

ith

elec

trici

ty (8

1% vs

57%

). Th

ere

wer

e

insig

nific

ant d

iffer

ence

s in

num

ber o

f

latri

nes p

er cl

assr

oom

and

latri

nes i

n

good

con

ditio

n.

Pare

nts o

f stu

dent

s fro

m FI

SE re

porte

d

grea

ter i

mpr

ovem

ents

in sc

hool

con-

ditio

n ov

er p

ast 5

yea

rs c

ompa

red

to

pare

nts

of s

tude

nts

from

non

-FIS

E.

How

ever

, no

sign

ifica

nt d

iffer

ence

was

fou

nd in

atte

ntio

n to

mai

nte-

nanc

e be

twee

n FI

SE &

non

-FIS

E

scho

ols.

5,12

0 ho

useh

old

surv

eys

70 S

IF &

71 no

n-SI

F sch

ool

surv

eys;

760

SIF

& 7

70

non-

SIF h

ouse

hold

surv

eys.

Proj

ect s

urve

y: 68

% o

f SIF

clas

sroo

ms

are

in g

ood

cond

ition

(vs

44 in

non

-

SIF)

. 67

%

of

the

SIF

scho

ols

have

acc

ess

to

safe

wat

er (v

s 32

in n

on-

SIF)

.

2,95

0 ho

useh

old

surv

eys

&

13,5

00 n

atio

nal s

urve

ys.

43 S

IF &

25

non-

SIF s

choo

l sur

-

veys

; 1,0

20 S

IF &

1,0

20 n

on-S

IF

hous

ehol

d su

rvey

s.

Hous

ehol

d be

nefic

iary

: SI

F

scho

ols h

ad si

gnifi

cant

ly fe

wer

stud

ents

per

ven

tilat

ed p

it

latri

ne th

an th

e no

n-SI

F sch

ools.

SIF

scho

ols

had

sign

ifica

ntly

low

er

perc

enta

ge

repo

rting

unsa

fe w

ater

(81%

vs 7

6%), a

nd

high

er p

erce

ntag

e re

porti

ng

wat

er a

nd e

lect

ricity

ava

ilabi

l-

ity (6

0% v

s 32

%) a

s co

mpa

red

to n

on-S

IF s

choo

ls.

And

SIF

scho

ol h

ad 6

8 st

uden

ts p

er V

IP

as c

ompa

red

to 7

00 p

er V

IP in

non-

SIF.

Ther

e w

as n

o si

gnifi

-

cant

diff

eren

ce b

etw

een

SIF a

nd

non-

SIF f

acili

ties i

n cla

sses

hel

d

outs

ide,

stu

dent

s pe

r re

gula

r

latri

nes,

stu

dent

s pe

r flu

sh to

i-

let,

and

scho

ols w

ith e

lect

ricity

alw

ays

avai

labl

e.


7 6

Bol

ivia

Hon

dura

sN

icar

agua

Peru

(APO

YO)

Zam

bia

Com

plem

enta

ry In

puts

Staf

fing

Supp

lies,

equi

pmen

t

and

furn

iture

Fina

ncin

g

& c

ost

reco

very

Su

st

ai

na

bi

li

ty

:

So

ci

al

F

un

ds

2

00

0R

es

ul

ts

Ta

bl

e

E.

3

For C

haco

& R

esto

Rur

al (c

ombi

ned)

,

the

stud

ent-t

each

er ra

tio w

as lo

wer

for S

IF-b

uilt

scho

ols

as c

ompa

red

to

the

non-

SIF

scho

ols

(20

vs 2

2).

For C

haco

SIF

sch

ools

wer

e si

gnifi

-

cant

ly be

tter t

han

the

non-

SIF s

choo

ls

for:

(i) n

umbe

r of

bla

ckbo

ards

, (ii)

blac

kboa

rds

per

clas

sroo

m a

nd (i

ii)

desk

s per

teac

her p

er cl

assr

oom

. For

Chac

o &

Res

to R

ural

(com

bine

d), t

he

SIF

faci

litie

s sh

owed

sig

nific

ant

impr

ovem

ent b

etw

een

1993

–97

and

as co

mpa

red

to th

e no

n-SI

F for

(i) t

he

text

s per

stud

ent,

(ii) M

2 pe

r stu

dent

,

(iii)

stud

ents

per

cla

ssro

om a

nd (i

v)

stud

ents

per

teac

her.

Proj

ect s

urve

y: Th

e le

vel o

f per

sonn

el

assi

gnm

ent

was

gen

eral

ly s

atis

fac-

tory

with

34

stud

ents

per

teac

her i

n

both

FHI

S an

d no

n-FH

IS s

choo

ls.

Bene

ficia

ry h

ouse

hold

: 33%

cite

d la

ck

of p

erso

nnel

as a

pro

blem

(onl

y 7%

for

non-

FHIS

)

Proj

ect s

urve

y: FH

IS sc

hool

s wer

e be

t-

ter t

han

non-

FHIS

in d

esk p

er st

uden

t

(1.1

2 co

mpa

red

to 0

.88)

but

no

dif-

fere

nce

was

foun

d in

text

book

s pe

r

stud

ent.

58%

cite

d la

ck o

f equ

ipm

ent

or m

ater

ial (

com

pare

d to

26%

in n

on-

FHIS

) as

a pr

oble

m.

FHIS

scho

ols a

re m

ore

likel

y to

rely

on

bene

fact

ors’

sup

port,

less

like

ly t

o

char

ge fe

es, a

nd sh

owed

sim

ilar u

se

of fu

ndra

isin

g.

Facil

ity su

rvey

: Tea

cher

num

bers

up by

1.48

in FIS

E and

by 0.

15 in

non-

FISE s

choo

ls. FI

SE

scho

ols s

how

sign

ifica

ntly

grea

ter i

ncre

ase

in nu

mbe

r of t

each

ers,

adm

inist

rato

rs, a

nd

othe

r sta

ff as

com

pare

d to

non

-FIS

E. T

he

stud

ents

per

teac

her r

atio

was

24

in b

oth

FISE a

nd no

n-FIS

E sch

ools.

Inte

rvie

ws w

ith

teac

hers

sug

gest

that

FIS

E co

ordi

nate

d

wel

l with

the E

duca

tion M

inist

ry to

ensu

re

that

new

scho

ols w

ould

be

staf

fed.

No si

gnifi

cant

diff

eren

ces w

ere

obse

rved

betw

een

FISE

and

non-

FISE

scho

ols.

Both

are

wel

l equ

ippe

d w

ith te

achi

ng m

ater

i-

als.

But

, ben

efic

iarie

s pe

rcei

ved

bette

r

furn

iture

in F

ISE

as c

ompa

red

to n

on-

FISE

. And

in n

on-F

ISE,

the

bene

ficia

ries

also

pe

rcei

ved

wor

seni

ng

phys

ical

infra

stru

ctur

e.

FISE

scho

ols

are

mor

e ac

tive

in fu

ndra

is-

ing,

but

are

sim

ilar t

o no

n-FIS

E sc

hool

s in

term

s of t

heir

supp

ort f

rom

the m

inist

ry an

d

thei

r app

licat

ion

of s

tude

nt fe

es. 4

2% o

f

FISE s

choo

ls an

d 47%

of no

n-FIS

E sch

ools

char

ge th

eir s

tude

nts a

tuiti

on fe

e.

The

stud

ent-t

each

er ra

tio

was

low

er fo

r SIF

scho

ols

as c

ompa

red

to th

e no

n-

SIF

scho

ols

(28

vs 3

2).

Staf

fing

at S

IF f

acili

ties

was

mor

e tha

n non

-SIF,

alth

ough

that

was

true

even

befo

re S

IF in

volve

-

men

t. Th

e st

uden

t-tea

cher

ratio

was

sl

ight

ly

low

er

for

SIF

scho

ols a

s com

pare

d to

the

non-

SIF

scho

ols

(47

vs 4

8).

SIF

scho

ols

wer

e si

gnifi

cant

ly

mor

e lik

ely

to h

ave

desk

s fo

r

stud

ents

and

teac

hers

than

the

non-

SIF

faci

lity,

espe

cial

ly in

rura

l are

as.

SIF s

choo

ls w

ere

mor

e lik

ely t

o

pay P

TA-fe

es, G

PF, m

aint

enan

ce

fees

than

non

-SIF

scho

ols.

And

SIF s

choo

ls w

ere

likel

y to

attra

ct

supp

ort b

eyon

d PT

A th

an n

on-

SIF.

A n n e x e s

7 7

Bol

ivia

Hon

dura

sN

icar

agua

Peru

(APO

YO)

Zam

bia

HEA

LTH

FA

CILI

TY

Sam

ple

size

Qual

ity o

f

faci

litie

s

Su

st

ai

na

bi

li

ty

:

So

ci

al

F

un

ds

2

00

0R

es

ul

ts

Ta

bl

e

E.

3

90 S

IF &

87

non-

SIF h

ealth

pos

ts;

1,92

1 SI

F & 1

,921

non

-SIF

hou

se-

hold

sur

veys

.

SIF h

ealth

pos

ts h

ad m

ore

patie

nt

room

s, le

ss a

cces

s to

wat

er a

nd

elec

trici

ty th

an n

on-S

IF p

osts

.

12 FH

IS (A

rura

l FHI

S he

alth

pos

t sam

pled

was

inop

erat

ive du

e to l

ack o

f per

sonn

el) &

12 no

n-

FHIS

hea

lth p

osts

; 386

SIF

& 1

88 n

on-S

IF

hous

ehol

d su

rvey

s.

Proj

ect

surv

ey: V

ery

little

diff

eren

ce w

as

foun

d re

gard

ing

the

deco

ratio

n an

d cl

eanl

i-

ness

bet

wee

n th

e tw

o gr

oups

. How

ever

, 67%

of th

e FH

IS co

nstru

cted

room

s wer

e qu

alifi

ed

as “g

ood”

as co

mpa

red t

o 4%

of th

e non

-FHI

S.

Wat

er cl

oset

s and

ele

ctric

inst

alla

tions

wer

e

prob

lem

atic

in b

oth

FHIS

and

non

-FHI

S fa

cil-

ities

: 34

perc

ent

and

38 p

erce

nt w

ere

not

wor

king

in F

HIS

and

non-

FHIS

hea

lth c

en-

ters

, res

pect

ivel

y.

Bene

ficia

ry h

ouse

hold

s: Qu

ality

of F

HIS

wor

ks

was

ass

esse

d as

“go

od”

by 8

8%; “

regu

lar”

by 9

% a

nd b

ad b

y 0%

. And

sup

ervi

sion

was

asse

ssed

as

“effi

cien

t” b

y 77

%.

20 F

ISE

& 2

0 no

n-FI

SE h

ealth

pos

ts;

199

FISE

& 1

99 n

on-F

ISE

hous

ehol

d

surv

eys.

Faci

lity

surv

ey: A

ll he

alth

pos

ts la

ck

acce

ss to

bas

ic in

frast

ruct

ure.

How

-

ever

, FIS

E he

alth

pos

t hav

e be

tter:

(i)

acce

ss to

safe

wat

er (9

1%),

(ii) b

ath-

room

s (2.

2), (i

ii) flu

sh to

ilets

(1), (

iv) ra

in

drai

nage

faci

lity

(4%

), (v

) ele

ctric

ity

durin

g ho

urs o

f ser

vice

(64%

), an

d (vi

)

pipe

d w

ater

dur

ing

hour

s of

ope

ra-

tion

(26%

). N

one

of th

ese

are

sign

if-

ican

tly

diffe

rent

fro

m

non-

FISE

infra

stru

ctur

e. H

owev

er, F

ISE

heal

th

post

s ar

e le

ss li

kely

to re

ly s

olel

y on

Min

istry

of H

ealth

fund

s fo

r mai

nte-

nanc

e, re

ceivi

ng e

xtra

hel

p fro

m co

m-

mun

ity, a

nd FI

SE p

roje

cts h

ave

grea

ter

prev

enta

tive

mai

nten

ance

for

san

i-

tary

& e

lect

rical

sys

tem

s.

— —

16 S

IF &

14

non-

SIF

heal

th

post

s; 2

150

FISE

& 2

150

non-

FISE

hou

seho

ld s

urve

ys.

SIF

heal

th p

osts

impr

oved

sig

-

nific

antly

mor

e th

an th

e no

n-SI

F

faci

litie

s fo

r: (i)

del

iver

y an

d

med

ical

room

s, (i

i) flu

sh to

ilets

and

vent

ilate

d pi

t lat

rines

, (iii

)

hand

pum

ps, b

oreh

ole,

sup

ply

line,

and

(iv)

acc

ess

to e

lect

ric-

ity. I

n ad

ditio

n, S

IF f

acili

ties

wer

e fo

und

to b

e in

sign

ifica

ntly

bette

r ph

ysic

al c

ondi

tion

than

non-

SIF

com

para

tors

. 31%

of

SIF

faci

litie

s w

ere

note

d as

bein

g in

ver

y go

od s

hape

com

-

pare

d to

7%

of t

he c

ompa

rato

r

grou

p. 6

% o

f SIF

faci

litie

s wer

e

note

d as

bei

ng in

poo

r sh

ape

com

pare

d to

29%

of

non-

SIF

faci

litie

s.


7 8

Bol

ivia

Hon

dura

sN

icar

agua

Peru

(APO

YO)

Zam

bia

Com

plem

enta

ry In

puts

Staf

fing

Supp

lies,

eq

uipm

ent

and

furn

iture

Fina

ncin

g&

cos

t re

cove

ry

Su

st

ai

na

bi

li

ty

:

So

ci

al

F

un

ds

2

00

0R

es

ul

ts

Ta

bl

e

E.

3

SIF

heal

th p

osts

had

mor

e be

ds(1

.9 vs

1.2

) tha

n th

e no

n-SI

F pos

ts.

For t

he S

IF fa

cilit

ies,

ther

e w

as a

62%

incr

ease

in th

e in

dex o

f ava

il-ab

le m

edic

al s

uppl

ies

(199

3–97

)in

crea

sed

com

pare

d to

23%

inno

n-SI

F fa

cilit

ies.

For

all h

ealth

pos

ts s

taffi

ng w

as d

efic

ient

and

belo

w M

inist

ry st

anda

rds.

Only

one

FHIS

heal

th p

ost h

ad a

doc

tor a

nd 4

had

no

doc-

tor o

r qua

lifie

d nu

rse.

How

ever

, 63%

of F

HIS

heal

th p

osts

had

NGO

sup

port

vs. 4

5% o

fno

n-FH

IS. F

urth

er F

HIS

had

mor

e vo

lunt

eers

(on

aver

age

19 v

ersu

s 8)

.

Proj

ect s

urve

y: V

ery

little

diff

eren

ce re

gard

-in

g th

e pr

ovis

ion

of f

urni

ture

bet

wee

n th

etw

o gr

oups

. 42%

and

53%

of r

ecom

men

ded

med

icin

es a

nd e

quip

men

t w

ere

avai

labl

e(co

mpa

red

to 4

4% a

nd 5

9% in

non

-FHI

S fa

cil-

ities

, res

pect

ivel

y).

Focu

s gro

ups:

The

nega

tive

asse

ssm

ent f

rom

the

facil

ity su

rvey

wer

e co

ntra

dict

ed b

y pos

i-tiv

e rea

ctio

ns of

the f

ocus

grou

ps. H

owev

er, th

em

ost f

requ

ent c

ompl

aint

was

lack

of m

edi-

cines

. (Ex

plan

atio

n pr

ovid

ed: F

HIS

inte

rven

-tio

n ha

s in

crea

sed

expe

ctat

ions

resu

lting

ingr

eate

r diss

atisf

actio

n w

ithin

an

obje

ctive

lysim

ilar s

ituat

ion)

.Pr

ojec

t su

rvey

: 23%

of

FHIS

had

mon

thly

inco

mes

from

co-p

aym

ents

of m

ore

than

L.50

0co

mpa

red

to 1

8% o

f non

-FHI

S.

Facil

ity su

rvey

: FIS

E he

alth

pos

ts w

ere

bette

r st

affe

d, in

clud

ing

volu

ntee

rsth

an n

on-F

ISE

(18.

6 vs

13.

1 to

tal s

taff

and 2

.6 vs

1.7 p

rofe

ssio

nal s

taff)

. Tot

alst

affin

g in

crea

sed

for F

ISE

(21%

) and

decl

ined

for

non

-FIS

E (2

4%)

for

1993

–199

7. B

ut b

oth

rem

ain

belo

wM

inis

try n

orm

s.

Facil

ity su

rvey

: Ava

ilabi

lity o

f med

icine

belo

w m

inist

ry n

orm

s in

both

FISE

and

non-

FISE f

acilit

ies (

abou

t 45%

are o

ften

not

avai

labl

e).

In F

ISE

25%

of

the

requ

ired m

edici

nes w

ere n

ever

avail

able

(com

pare

d to

31%

in n

on-F

ISE)

. FIS

Ehe

alth

post

s wer

e bet

ter e

ndow

ed th

anno

n-FIS

E in

sofa

r as

avai

labi

lity

of k

eym

edica

l sup

plie

s (5

9% v

s 42

%),

and

equi

pmen

t (65

% vs

60%

).

Facil

ity su

rvey

: Mai

nten

ance

fund

s are

gene

rally

lack

ing,

par

ticul

arly

in F

ISE

facil

ities

.

Non

-FIS

E he

alth

pos

ts a

re a

lmos

ttw

ice

as li

kely

to

ask

for

volu

ntar

yco

ntrib

utio

ns fr

om u

sers

(75%

non

-FI

SE v

s. 3

9% F

ISE)

. Of t

he 4

0 po

sts

surv

eyed

onl

y on

e (a

FIS

E po

st)

char

ges

man

dato

ry fe

es. T

he m

ajor

-ity

of F

ISE

post

s pro

vide

thei

r ser

vices

for f

ree.

— — —

SIF

heal

th p

osts

wer

e be

tter s

taffe

d th

anno

n-FIS

E (14

.6 vs

7.7 t

otal

staf

f and

4.2 v

s 2.6

prof

essio

nal s

taff)

. Bet

wee

n 19

93-1

998

the

tota

l sta

ffing

incre

ased

by 12

% fo

r SIF

and 7

%fo

r non

-SIF,

prof

essio

nal s

taffi

ng in

crea

sed b

y7%

for S

IF an

d dec

reas

ed by

10%

for n

on-S

IFan

d the

regi

ster

ed nu

rses

decli

ned b

y 35%

for

SIF a

s com

pare

d to

89%

for n

on-S

IF.As

com

pare

d to

non

-SIF

hea

lth p

osts

, the

SIF h

ealth

pos

ts h

ad 2

-3 ti

mes

mor

e tro

lleys

,ha

ngin

g w

eigh

ing,

sph

ygom

eter

s, e

xam

i-na

tion

couc

hes,

tabl

es, c

hairs

, cup

boar

ds,

& b

eds.

Also

, SIF

heal

th p

osts

hav

e sli

ghtly

high

er a

vaila

bilit

y of

11

esse

ntia

l med

i-ci

nes

and

cond

oms

(79

vs 7

7). S

imila

rlyth

ey a

re si

gnifi

cant

ly le

ss a

t ris

k to

run

out

of C

hlor

oqui

ne th

an n

on-S

IF fa

cilit

ies.

Diffe

renc

es in

fees

char

ged

by S

IF an

d no

n-SI

F fa

cilit

ies

are

inco

nclu

sive

. Whi

le g

en-

eral

cons

ulta

tion

fees

are

foun

d on

ave

rage

to b

e le

ss a

t soc

ial f

und

faci

litie

s (K

W28

0vs

. KW

530)

, boo

king

and

regi

stra

tions

fees

are

high

er a

t SI

F fa

cilit

ies

(KW

450

vs.

KW30

0). 8

0% o

f SIF

faci

litie

s an

d 70

% o

fno

n-SI

F fa

cilit

ies

char

ge o

ther

fees

.

But,

SIF f

acili

ties a

re m

ore

likel

y to

rece

ive

addi

tiona

l su

ppor

t fro

m

NGO

s th

anco

mpa

rato

rs.

A n n e x e s

7 9

Bol

ivia

Hon

dura

sN

icar

agua

Peru

(APO

YO)

Zam

bia

WAT

ERSa

mpl

e si

ze

Qual

ity o

ffa

cilit

ies

Su

st

ai

na

bi

li

ty

:

So

ci

al

F

un

ds

2

00

0R

es

ul

ts

Ta

bl

e

E.

3

No

proj

ect s

urve

y; 1,

235

base

line

and

1,10

9 fo

llow

-up

hous

ehol

d su

rvey

sfro

m 1

8 pr

ojec

ts.

No

evid

ence

of c

hlor

inat

ion,

lack

of

train

ing

of th

e pe

rson

nel d

esig

nate

dfo

r mai

nten

ance

in e

arlie

r sta

ges

ofpr

ojec

t im

plem

enta

tion,

with

impr

ove-

men

t in

late

r sta

ges.

12 F

HIS

(all

oper

atin

g) a

nd 1

2 no

n-FH

IS w

ater

sys

tem

s. 3

24 h

ouse

hold

surv

eys e

ach

from

FHIS

and

non

-FHI

Sco

mm

uniti

es.

Proj

ect s

urve

y: N

on-F

HIS

had

bette

ror

igin

al c

onst

ruct

ion,

how

ever

, the

FHIS

had

bet

ter

mai

nten

ance

. 50%

of FH

IS fa

ciliti

es m

entio

ned

low

wat

erpr

oduc

tion

(com

pare

d to

21

in n

on-

FHIS

), 25

% m

entio

ned

netw

ork p

rob-

lem

s (co

mpa

red

to 2

1% in

non

-FHI

S),

0% w

ith lo

w p

ress

ure

(com

pare

d to

13%

in n

on—

FHIS

), 19

% w

ith so

urce

cont

amin

atio

n (c

ompa

red

to 1

3% in

non-

FHIS

) as

maj

or p

robl

ems.

Bene

ficia

ry h

ouse

hold

: Ser

vice

pro

-du

ctio

n w

as a

pro

blem

, alth

ough

the

mai

nten

ance

and

wat

er q

uant

ity &

qual

ity w

as b

ette

r for

FHIS

faci

litie

s.46

% a

sses

sed

FHIS

sol

utio

ns a

s“g

ood”

and

54%

agr

eed

that

sup

er-

visi

on w

as e

ffici

ent.

How

ever

, 23%

asse

ssed

FHIS

solu

tions

as “

bad”

and

15%

sa

id

that

th

ere

was

no

supe

rvis

ion.

5 su

cces

sful

(with

suf

ficie

nt w

ater

) & 5

unsu

cces

sful

(with

insu

ffici

ent

wat

er)

FISE

sys

tem

s. 9

5 ho

useh

old

surv

eys.

Proj

ect s

urve

y: T

he s

ucce

ssfu

l sys

tem

s(m

aint

aine

d by

INAA

) gen

eral

ly u

se su

b-te

rrane

an w

ater

sou

rces

, hav

e 10

0%ne

cess

ary m

aint

enan

ce a

nd 8

8% a

cces

sto

tec

hnic

al a

ssis

tanc

e. U

nsuc

cess

ful

ones

(mai

ntai

ned

by m

unici

palit

y or l

ocal

grou

ps) u

se s

urfa

ce w

ater

, 67%

hav

ene

cess

ary m

aint

enan

ce a

nd 6

0% a

cces

sto

tech

nica

l ass

ista

nce.

Wat

er a

vaila

bil-

ity is

thre

e tim

es m

ore

in th

e su

cces

sful

than

in th

e un

succ

essf

ul sy

stem

s. W

ater

is ra

tione

d in

the

latte

r sys

tem

s, av

aila

ble

on a

vera

ge t

hree

day

s a

wee

k. W

ater

from

all

ten

syst

ems

is d

escr

ibed

to b

eba

cter

ia-fr

ee, h

owev

er, o

ne th

ird o

f the

unsu

cces

sful

syst

ems a

re n

ot ch

lorin

ated

.Si

x w

ater

sys

tem

s ha

ve “

acce

ptab

le”

drai

nage

syst

ems,

the

othe

r 4 d

rain

into

the

grou

nd o

r a

river

bed

. 40%

of

the

unsu

cces

sful

syst

ems r

epor

t det

erio

ratio

nin

wat

er s

uppl

y an

d le

aks

com

pare

d to

11%

det

erio

ratio

n in

wat

er su

pply

and

nole

aks

in s

ucce

ssfu

l sys

tem

s.

166

SIF

and

166

non-

SIF

proj

ects

; 1,

176

SIF

and

1,17

6 no

n-SI

F ho

useh

old

surv

eys.

Ho

useh

olds

per

ceiv

ed a

nim

prov

emen

t in

the

qual

-ity

of w

ater

.

— —


8 0

Bol

ivia

Hon

dura

sN

icar

agua

Peru

(APO

YO)

Zam

bia

Com

plem

enta

ry In

puts

Staf

fing

Supp

lies,

eq

uipm

ent

Fina

ncin

g&

cos

t re

cove

ry

Su

st

ai

na

bi

li

ty

:

So

ci

al

F

un

ds

2

00

0R

es

ul

ts

Ta

bl

e

E.

3

—

Poor

qua

lity o

f mat

eria

l (tub

eset

c.) i

n so

me

syst

ems.

Lack

of

met

ers

to m

easu

re w

ater

quan

tity.

—

Proj

ect s

urve

y: S

taffi

ng w

as w

eak

(on

aver

age

2.34

em

ploy

ees

per

1000

con

nect

ions

in F

HIS

and

2.33

in

non-

FHIS

, w

ith n

o en

gine

ers

empl

oyed

. Rur

al s

yste

ms

depe

nd o

n vo

lunt

eers

(1.0

7 pe

r 100

0 co

nnec

tions

in FH

IS a

s com

pare

dto

0.7

8 in

non

-FHI

S).

Proj

ect s

urve

y: 1

9% o

f FHI

S fa

cilit

ies m

entio

ned

lack

of

acce

ssor

ies

(com

pare

d to

4%

in n

on-

FHIS

).

Proj

ect s

urve

y: “R

even

ue p

erfo

rman

ce w

as g

en-

eral

ly m

edio

cre.

” Th

e av

erag

e m

onth

ly re

venu

epe

r con

nect

ion

for F

HIS

was

L. 1

0 w

hich

is b

et-

ter t

han

non-

FHIS

(L. 5

.6) b

ut b

elow

com

mun

ityw

ater

boa

rd (L

.11.

6) a

nd m

unici

pal (L

. 14.

2). H

ow-

ever

, 83%

rep

orte

d th

at t

heir

reve

nue

did

not

cove

r ope

ratio

nal c

osts

.

Bene

ficia

ry h

ouse

hold

: On

aver

age

hous

ehol

dspa

id L

. 10.

9 fo

r pip

ed w

ater

last

mon

th in

FHI

S(c

ompa

red

to L

. 9.3

in n

on-F

HIS)

; FHI

S in

vest

-m

ent r

educ

ed a

vera

ge e

xpen

ditu

re p

er m

onth

s on

wat

er f

rom

alte

rnat

ive

reso

urce

s to

L. 1

5.6

inFH

IS (c

ompa

red

to L

. 33.

8 in

non

-FHI

S)

Apar

t fro

m 1

sys

tem

all

syst

ems

empl

oyed

few

er t

han

6 pe

ople

. All

syst

ems

have

at

leas

t one

adm

inis

trato

r or a

bra

nch

man

ager

.Th

ree

of th

e 5

unsu

cces

sful

pro

ject

s ha

ve a

plum

ber a

nd o

nly o

ne o

f the

succ

essf

ul o

nes

has a

plu

mbe

r. 60

% o

f the

uns

ucce

ssfu

l sys

-te

ms

repo

rt la

ck o

f per

sonn

el a

s a

prob

lem

. Pr

ojec

t sur

vey:

100

% o

f the

suc

cess

ful s

ys-

tem

s ha

ve e

lect

ric p

umps

com

pare

d to

20%

of th

e un

succ

essf

ul o

nes.

80%

of t

he u

nsuc

-ce

ssfu

l cite

lack

of t

ools

, equ

ipm

ent,

parts

,an

d m

ater

ials

for O

&M

. Av

erag

e hou

seho

ld co

nnec

tion c

ost i

s US$

37.7

in su

cces

sful

com

pare

d to

US$

8.45

in u

nsuc

-ce

ssfu

l sys

tem

s. Us

ers

pay

US$4

.4 m

onth

lyfe

es i

n su

cces

sful

sys

tem

s co

mpa

red

toUS

$1.4

5 in u

nsuc

cess

ful s

yste

ms.

None

thel

ess,

paym

ents

are

mor

e tim

ely

in s

ucce

ssfu

l sys

-te

ms.

Lack

of fu

nds i

n the

unsu

cces

sful

syst

ems

lead

s to

grea

ter s

yste

m d

eter

iora

tion.

46%

repo

rt pa

y-in

g fo

r ser

vice

s.

— — —

A n n e x e s

8 1

a. A

ccor

ding

to th

e au

thor

s of

the

Hond

uras

Soc

ial F

unds

200

0im

pact

eva

luat

ion,

the

abov

e le

vel o

f dis

satis

fact

ion

coul

d po

ssib

ly b

e ex

plai

ned

by “

incr

ease

d ex

pect

atio

ns”

due

to F

HIS

inte

rven

tion.

Sour

ce:

New

man

and

oth

ers

2000

; Wal

ker a

nd o

ther

s 19

99, 2

000;

Pax

son

and

Shad

y 19

99; W

orld

Ban

k da

ta.

Bol

ivia

Hon

dura

sN

icar

agua

Peru

(APO

YO)

Zam

bia

SEW

ER &

LAT

RIN

E FA

CILI

TYSa

mpl

e si

ze

Qual

ity o

f fa

cilit

ies

Fina

ncin

g&

cos

t re

cove

ry

Su

st

ai

na

bi

li

ty

:

So

ci

al

F

un

ds

2

00

0R

es

ul

ts

Ta

bl

e

E.

3 — — —

8 FH

IS (1

inop

erat

ive

due

to la

ck o

f con

nect

ion

to s

ewer

mai

n) a

nd4

non-

FHIS

sew

erag

e sy

stem

s an

d 6

FHIS

and

5 n

on-F

HIS

latri

nes.

162

hous

ehol

d su

rvey

s ea

ch fr

om F

HIS

and

non-

FHIS

.

Proj

ect s

urve

y: Bo

th FH

IS a

nd n

on-F

HIS

sew

ers s

ubpr

ojec

ts h

ave

very

posi

tive

eval

uatio

n fo

r orig

inal

con

stru

ctio

n. B

ut F

HIS

mai

nten

ance

was

ass

esse

d as

“goo

d” b

y 91%

, “re

gula

r” b

y 5%

and

“bad

” by 5

%co

mpa

red

to 1

00%

ass

esse

d as

“re

gula

r” fo

r non

-FHI

S se

wer

s.

FHIS

latri

nes

com

pare

wel

l with

non

FHI

S fo

r orig

inal

con

stru

ctio

nan

d pr

esen

t sta

te o

f mai

nten

ance

and

clea

nlin

ess.

And,

FHIS

latri

nes

had

bette

r qua

lity

door

s an

d se

ats

than

non

-FHI

S.

Bene

ficia

ry h

ouse

hold

s: Q

ualit

y of

FHI

S se

wer

s w

as a

sses

sed

as“g

ood”

by 61

%, “

regu

lar” b

y 15%

. And

supe

rvisi

on w

as ef

ficien

t by 8

8%.

Qual

ity o

f FHI

S la

trine

s was

ass

esse

d as

“goo

d” b

y 78%

; “re

gula

r” b

y15

% an

d bad

by 3%

. And

supe

rvisio

n was

asse

ssed

as “e

fficie

nt” b

y 67%

.Bu

t, 11

% la

trine

s had

pro

blem

s in

FHIS

com

pare

d to

8%

in n

on-F

HIS.

FH

IS s

ewer

age

proj

ects

had

hig

her

acce

ss a

nd c

onne

ctio

n ra

tes

than

non

-FHI

S co

mm

uniti

es. F

HIS

2 fo

und

44%

con

nect

ion

rate

.

10 F

ISE

sew

erag

e an

d 23

FIS

Ela

trine

s (a

ll IN

AA m

aint

aine

d). 7

4ho

useh

old

surv

eys f

or se

wer

s and

234

for l

atrin

es.

The

5 se

wer

syst

ems i

n M

anag

uala

ck w

ater

trea

tmen

t and

dis

pose

raw

sew

age

in th

e la

ke. T

he s

ys-

tem

s re

port

few

pro

blem

s an

dm

aint

enan

ce a

ctiv

ities

are

carri

edou

t reg

ular

ly. H

owev

er, t

he m

ajor

-ity

expe

rienc

e pro

blem

s in t

he ra

iny

seas

on. T

he s

yste

ms

have

led

toim

prov

emen

ts in

roa

d co

nditi

on,

stre

ams,

and

pud

dles

.

Conn

ectio

n co

sts

vary

bet

wee

nsy

stem

s, w

ith th

ose

in M

anag

uaav

erag

ing

US$3

0. M

onth

ly f

ees

aver

age

US$2

.00

and

are

mos

tlyin

clud

ed in

the

wat

er b

ill. E

xcep

tfo

r 30%

of u

sers

in M

anag

ua w

hodo

not

pay

the

ir fe

es, a

ll ot

her

user

s hav

e pa

id fe

es to

pre

sent

. In

60%

of

the

syst

ems

fund

s co

l-le

cted

are

not

suffi

cien

t for

O&

M.

21 S

IF &

13

non-

SIF

sew

ers;

86

SIF

& 8

3 no

n-SI

F la

trine

s. 2

24ho

useh

olds

eac

h fo

r SIF

& n

on-

SIF s

ewer

s; 51

0 ea

ch fo

r lat

rines

.

— — —


8 2

Armenia: In Armenia, 79 percent of community members interviewed said that the implemented microproject had solved

the most important problem in the communitya (Armenia 1999 sociological study).

Honduras: The water projects were the preference of 64 percent of beneficiaries; education projects, of 47 percent; health

projects, of 35 percent; latrines projects, of 4 percent; and sewerage projects, of 7 percent. (Ex-post Evaluation of the Hon-

duras Social Investment Fund, FHIS II).b

Nicaragua: Seventy-two percent of the respondents answered that the FISE projects were those that had most benefited

the communityc (1998 qualitative evaluation of FISE beneficiaries).

Peru: When asked if the project selected was the one that the community most needed, 90 percent responded that it was

the highest priority investmentd (APOYO 1995).

Zambia: The majority of projects were the community’s first priority based on intensive ranking exercisese (1994 benefici-

ary assessment).a. Based on 1,190 persons, 20 project sites, focus group interviews with beneficiary groups, members of implementing agencies, construction companies, andlocal government representatives. b. Management notes that the findings of the Honduras household survey were that where FHIS funded water, education, and health projects, these projectsrepresented the top priority of beneficiary households. For communities that invested in water projects, 64 percent of respondents stated that as the numberone priority (the top vote getter), in communities that implemented education projects 47 percent of respondents stated that education was the number one pri-ority (the top vote getter), and in communities that invested in health projects 35 percent of respondents stated that health was the number one priority (the topvote getter). Since these three project types represent 98 percent of FHIS portfolio, 98 percent of FHIS investments reflect the top priority of household mem-bers. For sewerage and latrines, households in communities where these investments were made place sewerage and latrines much lower on the list of priorities.c. Based on 256 individual surveys (12 percent local government officials, 27 percent line ministry staff, 13 percent former workers on FISE projects, 40 percentlocal beneficiaries, and 9 percent contractors), and 24 focus groups on projects implemented during 1993–96. This question compared the FISE subproject withother projects in the community—it is not clear how the comparator projects were chosen. This result does not address the question of whether FISE met thetop priority of the community. The respondents were not randomly selected.d. Based on interviews of 2,800 community members, beneficiaries, project committee members, FONCODES staff.e. Based on 28 focus group interviews, 30 ordinary group discussions, and 45 semi-structured interviews. The respondents were not randomly selected.

C o m m u n i t y P r i o r i t i e s : S o c i a l F u n d s2 0 0 0

T a b l e E . 4

8 3

Benefiting the poor is the overarching objectiveof social fund projects: “social funds establishmenus, procedures, and targeting criteria to sup-port investments benefiting the poor” (Jorgensenand Van Domelen 1999). Accordingly, the “poor”(or “poorer,” “poorest,” “poverty”) are mentionedin the objectives of more than three-quarters ofsocial fund projects.1 Furthermore, the poor, orsome category of poor people, are mentioned asan explicit target group in the majority of cases(“poor” in 80 percent of projects, “poorest” in 46percent of projects, “vulnerable” in 44 percent ofprojects, and “low-income” in 10 percent of proj-ects). Other target groups are “unemployed,” in20 percent of projects; “indigenous,” in 10 per-cent of projects; and women, in 61 percent ofprojects. This annex looks at the means employedby social funds to reach poor people, and theirsuccess in doing so.

Promotion and Outreach MechanismsSocial fund resources are disbursed to commu-nities and intermediaries (central government,local government, and NGOs) based on sub-project proposals made by them. This assumesthat communities and intermediaries are awareof the social fund’s existence and application pro-cedures, and make viable subproject proposals.

Social fund projects use two mechanisms tomake this happen: (i) “promotion,” which aimsat ensuring that poor communities hear of theopportunities provided by the social fund; and(ii) “outreach” (including facilitation), whichaims at helping poor communities to prepare andsubmit viable subproject proposals in the formatrequired by the social fund agency. A review ofproject documents (staff appraisal reports orproject appraisal documents, and operationalmanuals where available) found promotion and

outreach mentioned in 47 out of 66 social fundprojects (71 percent). A variety of media are usedfor promotion and outreach activities2 using dif-ferent channels.3

Given the possibility that better-off commu-nities are more advantageously placed to applyfor subprojects from the social fund, some socialfund projects include special promotion andoutreach measures specifically directed at poorercommunities. Outreach in poorer communitiestypically consists of training or technical assis-tance to help them prepare subproject propos-als, or assisting them to meet pre-investmentcosts related to subproject preparation. Forexample, in the second Peru FONCODES proj-ect, the social fund’s Promotion Office assistedthe poorest communities in preparing and sub-mitting proposals by providing technical assis-tance. Bolivia SIF II financed pre-investmentcosts of subprojects in poor communities toenable them to subcontract third parties toundertake necessary studies.

Targeting MechanismsTo reach the poor, social fund projects gener-ally use one of three targeting mechanisms: • Self-targeting: provision of benefits that will

only be of interest to the target group, suchas basic (rather than high-end) services.

• Indicator targeting: directing resources togroups, households, or individuals in accor-dance with an indicator (such as unemployedor malnourished).

• Geographic targeting: directing resources tospecific areas.

Self-TargetingSocial fund projects intrinsically contain an ele-ment of self-targeting. About two-thirds of social

ANNEX F: PROMOTION, OUTREACH, AND TARGETING

fund projects include a positive list of eligiblesubproject activities, and another 22 percent anegative list of ineligible subproject activities. Eli-gible activities are meant to be those from whichthe poor will benefit. For example, staff notedthat for FISE in Ecuador, the type of subprojectseligible for FISE financing—including primaryhealth and education services, basic infrastruc-ture (e.g., feeder roads, standpipes for water sup-ply), and support for small-scale productivecommunity investments—by their nature tend tobe oriented to the needs of the poor.

Some subprojects in social fund menus (suchas roads) may also be of interest to the non-poor,but these are usually public goods for whichrivalry is low, so the poor also benefit. Somesocial fund expenditure is on private goods ofinterest to the non-poor (principally microcre-dit), though these account for only a small frac-tion of total expenditure.

Indicator TargetingSocial fund projects also use indicator targeting.While generally the indicator in social fundprojects refers to the poor, it may also refer tothe vulnerable or non-poor, such as unemployedgraduates in the first Egypt social fund project.

There are two prerequisites for indicator tar-geting to be a viable means of targeting. First,there has to be a clear definition of the indica-tor and the indicator has to be measurable. ABank review found that a majority of socialfund projects refer to broad and general cate-gories of poor populations such as “poor andvulnerable,” “low-income,” or “poorest amongthe population” without defining them more pre-cisely. Only a few social fund projects definetheir target group in measurable terms—forexample, populations with infant mortality rateshigher than a specified level. Second, there hasto be a mechanism to ensure that benefits doindeed reach people from that group. For exam-ple, the Peruvian Second Social Developmentand Compensation Fund supplemented its geo-graphic poverty map with information on thedistribution of the indigenous population andthe ethno-linguistic status of the applying com-munities in order to target more resources toindigenous people.

Geographic TargetingThe main means of targeting in social fund proj-ects has been geographic, used in 92 percent ofthe projects (54 of 59 projects). Geographic tar-geting mechanisms can be classified as pro-active and reactive. Using data on the geographicdistribution of poverty, usually combined withpopulation data, proactive targeting allocates aspecified quantity of resources up front to eacharea. By contrast, once applications have beenreceived, reactive targeting uses poverty data inprioritizing them to ensure that resources reachpoor areas (table F.1).

The level of aggregation of the targeting unitis important. Ideally, community-level data areavailable that allow targeting resources to thepoorest communities. More usually, data areonly available at district (or equivalent) level.Proactive targeting by district means that intra-district allocation is still on a reactive basis.Social funds may use alternative means to tar-get within this geographical unit. For example,in Zimbabwe local experts and key informantsidentify the poorer communities after a generaldistrict-level allocation has taken place. Withoutsuch additional local expert identification mech-anisms within districts, there is a danger that theless-poor communities within districts will dom-inate the use of funds.

To What Extent Have Social FundProjects Reached the Poor?Quantitative targeting data are available for justa handful of countries. Data on both a geo-graphic and a household basis from Social Funds2000 studies for six countries show overall thatsocial fund projects have delivered slightly morethan proportional benefits to the poor and thepoorest.

On a geographic basis, i.e. the cumulativeshare of expenditures going to poor districts, thedistribution was mildly progressive for all coun-tries other than Peru, where it was very pro-gressive. The share of the bottom 20 percentranges from 20.4 percent (Zambia) to 29.1 per-cent (Nicaragua), but was 46.5 percent for Peru(see figure F.1). For the bottom 40 percent, thesefigures are 43.9 percent (Zambia) to 53.3 percent(Nicaragua), and 78 percent for Peru. Armenia


8 4

has regressive targeting among the wealthierdistricts—the allocation of expenditure is pro-poor among the lowest 40 percent of districts.

Social fund targeting performance comparedfavorably in some cases with that of other pro-grams. In Bolivia, municipalities with higherpoverty indices received a higher share of socialfund resources than they did of municipal expen-ditures. In Peru, FONCODES achieved a bettergeographic targeting performance than twodirected programs, PRONAA and INFES. TheArmenia social fund performed in the mid-rangeof national social assistance programs, and inHonduras, the distribution of beneficiary house-holds was about average compared with 30 tar-geted social programs throughout the LAC Region.

The data for the household level classify ben-eficiaries by income decile. These data are thusnot directly comparable with those at the geo-graphic level as the unit of analysis is different

(expenditure versus beneficiary), and since theyrefer to different years (the household data arefor a single year). Nonetheless, it is striking thatthe conclusion that social funds are “mildly pro-gressive” holds even more strongly at the house-hold level, the concentration curve for none ofthe countries showing a marked deviation fromthe line of equality (figure F.2). The main excep-tion is Honduras with each of the bottom twodeciles being above the line of equality.4 InArmenia, household targeting is mildly regressivefor the bottom half of the population, but pro-gressive when upper income deciles are included.

Hence the available data, although imper-fect, suggest that the distribution of social fundresources is more progressive on a geographicbasis than on a household basis (except forHonduras). This discrepancy is particularly strik-ing in Peru, where for education investments, thestrong progressive geographic allocation is not

A n n e x e s

8 5

Proactive Progressive: an allocation rule is used Zimbabwe Community Action Project resources are allocated to districts based

that accords up-front more resources on a measure of deprivation and population size. In Honduras, FHIS allocates

per capita to poorer areas. resources to departments according to population size and a poverty index; a

1 percent increase in the poverty index gave a US$0.14 increase in funds per

person. The recent Zambian Social Investment Fund (ZAMSIF) allocates funds to

districts using a progressive targeting rule based on the poverty headcount. Peru

previously used a province-level poverty map and, since 1996, has used a

district-level one. Nicaragua uses an income-poverty map to set investment

ceilings for districts classified into four poverty levels.

Cut-off: resources are directed only to The Albanian Rural Development Project aimed at supporting activities in 14

areas or communities with a certain districts, representing the poorest and most mountainous of Albania’s 36 districts

level of poverty. (SAR 13156-ALB). The Second Bolivian Social Investment Fund limited itself to the

country’s 33 poorest districts, identified through a poverty map using social indicators.

Reactive Prioritization: once applications are Under the Zambia Social Recovery Project, the Living Conditions Monitoring

received, they are prioritized, the Surveys identified specific areas and communities where extreme poverty was

poverty of the area being one criterion present. These communities were given priority in the project selection process. The

used in this process. Health, Nutrition, Water and Sanitation Project in Guyana determined eligibility and

priority of proposals on the basis of poverty maps and other indicators such as mal-

nutrition and infant mortality rates.

Eligibility: only applications from In Madagascar, the Third Social Fund financed applications only from rural areas

communities satisfying a poverty where 80 percent of the population is considered poor.

criterion are considered.

T y p o l o g y o f G e o g r a p h i c T a r g e t i n g i nS o c i a l F u n d P r o j e c t s

T a b l e F . 1


8 6

C o m p a r i s o n o f G e o g r a p h i c a lT a r g e t i n g A c r o s s C o u n t r i e sF i g u r e F . 1

●

■

●■

■

■

■

■

■

■■

■

■

■

■

●

●

●

●

●

●

●

●

●

●

●

▲

▲ ▲

◆

◆

◆

◆

◆

◆

◆

◆

◆◆

◆

■

■

■

■

■

■

■

■■

❍

❍

❍

❍

●

❍

●

●❍

0 10 20 30 40 50 60 70 80 90 100

0

10

20

30

40

50

60

70

80

90

100Cu

mul

ativ

e ex

pend

iture

s fro

m fu

nd in

cept

ion


▲

▲

▲

▲

▲▲

▲▲

■ Armenia

● Honduras

▲ Peru

◆ Zambia

■ Nicaragua

● Bolivia

Line of equality

Note: Management notes that the geographic targeting data presented is cumulative since the inception of each social fund, including the early years under emer-gency mandates when the primary benefit was considered the employment generated and the social funds did not employ geographic targeting. Using data fromonly the most recent years would show improvement in the geographical targeting of resources.Source: Social Funds 2000 data.

H o u s e h o l d T a r g e t i n g O u t c o m e s b yI n c o m e D e c i l e sF i g u r e F . 2

■

■

■

■■

■

■

■

■

■

●

●

●

●

●

●

●

●

●

●

▲

▲

▲

▲

▲

▲

▲

▲

▲

▲

◆

◆

◆

◆

◆

◆

◆

◆

◆

◆

■

■

■■

■

■

■

■

■

■

0 10 20 30 40 50 60 70 80 90 100

0

10

20

30

40

50

60

70

80

90

100

Cum

ulat

ive

bene

ficia

ries


■ Armenia

● Honduras

▲ Peru

◆ Zambia

■ Nicaragua

Line of equality


matched by a markedly progressive distributionat the household level. The results are explainedin Peru (the only case for which there are suchdata) by data showing that households above dis-trict mean income are more likely to benefit fromthe social fund than those below mean income.The authors of the Peru study suggest theseresults show that “there was essentially no (pos-itive) intra-district targeting of FONCODESresources” (Paxson and Schady, 1999: p. 13).5

The result from Peru and the discrepancybetween geographical and household targetingindicate that poor households are less likely toreceive resources than poor districts. One of theimplications drawn by Social Funds 2000 fromits poverty targeting analysis is that better localcoordination efforts can improve identification ofpoor communities within both wealthier andpoorer districts, and that for those poor com-munities that still have difficulty accessingresources, additional proactive measures, includ-ing up-front assistance in community organiza-tion, waiving of community counterpartrequirements, and provision of technical expert-ise may be necessary to decrease the inequalityof results among areas and to ensure that all poorcommunities can access program resources.6

Main findings for Nicaragua and Peru arethat:7

• FISE in Nicaragua reached areas in whichthe poor live. Potential beneficiaries (that is,those living in project catchment areas) weredisproportionately from the lower end of thecountry’s income distribution, other than inthe case of sewerage.

• Self-targeting within subproject catchmentareas worked for sewerage in Nicaragua(though the share of poor households remainsfar below their population share) and workedslightly for education.

• The poorest 10 percent (household basis) gotaround 8 percent of the benefit from healthand education expenditures in Nicaragua.8

• In Peru, the poorest 10 percent (district basis)got around one-third of the benefit of edu-cation spending in 1998.9 The better target-ing performance in Peru allows for lesssectoral variation in performance. Even inthe sectors that did least well (i.e., education,10

transport, health), the bottom 40 percentreceived about 60 percent of the benefits in1995; for the other sectors this figure wasaround three-quarters.

Overall, at the household level, latrines andhealth clinics targeted the poor best, educationand water investments were pro-poor, and sew-erage projects were regressive.

What Should the Type II Error Be?The percent of resources reaching the non-pooris called a Type II error. The degree of Type IIerror partly depends on the level of poverty.Despite improvements over time, data are con-sistent with the possibility of poorer communi-ties within districts being missed (Type I error).Social funds spend significant resources that ben-efit the non-poor (Type II error), with non-poorhouseholds receiving from 29 to 45 percent ofsocial fund investments.11 In Zambia, 28 percentof the population is non-poor and they accountfor 29 percent of social fund beneficiaries.12 TypeII error is large whether targeting results arereported at either the geographic or householdlevel, with the notable exception of Peru at thegeographic level, where an overwhelming major-ity of resources reach poor districts.

Given the way municipal and district povertymaps are devised, any investments in the largerurban areas will appear as a Type II error sincelarge numbers of poor live in these cities. WhileType II errors can be a misleading indicator forprograms that are national in scope and someType II error is unavoidable for community-level projects, especially in urban areas, thequestion remains as to the acceptable degree oferror and the extent to which the mildly pro-gressive geographic targeting shown by socialfunds indicates they are succeeding in theirpoverty targeting objective.

Two issues arise from this finding: (i) howsocial fund targeting performance comparesagainst social funds’ own objectives; and (ii)the desirable or acceptable level of Type IIerror. The data presented here refer only to thedistribution of funds by income group. Data arenot available on the distribution of funds toother target groups, such as women.

A n n e x e s

8 7

Given the objective of reaching the poor, ifa significant portion of funds reach the non-poor,this must be seen as a failure to attain that objec-tive. Social fund projects are not designed to min-imize Type II error, although their statedobjectives might suggest this to be the case. Asdescribed earlier, some social fund projects usea cut-off so that no funds reach non-poor areas,but most social fund projects do not do this. Pro-gressive targeting rules allocate some funds evento the wealthiest areas. The first two social fundprojects supported by the Bank in Honduras,FHIS I and II, aimed “to improve the standardof living of marginal population groups,” the thirdproject aimed “to improve the standard of liv-ing in poor communities,” and the fourth proj-ect aimed “to increase access among the poorto small-scale social and economic infrastruc-ture.”13 Yet, FHIS expenditure plans allocated halfof FHIS resources in non-poor municipalities, i.e.,municipalities with “deficient” (which excludes“very poor” and “poor” categories), “regular”, and“acceptable” levels of poverty (accounting for 59percent of the population).14

The targeting literature shows that the costsof targeting rise with the accuracy of targeting;avoiding all leakage can be prohibitively expen-sive. Besides there are reasons for not com-pletely omitting the non-poor on politicaleconomy grounds as well as because those justabove the poverty line may also be vulnerable.So the target Type II error lies somewherebetween zero and one.

However, in a number of circumstances itwould be reasonable to expect the allocation ofsocial fund resources to be progressive; that is,reaching the poor is not sufficient, they shouldbenefit “more than average.” As seen above,social fund projects only just achieve this. Forexample, in Zambia just over 70 percent of thesocial fund beneficiaries are poor. Indeed, justover 70 percent of Zambians are poor, so theresult is the same as a random allocation offunds. The Zambian results refer to a time whentargeting in Zambia was dependent upon the suc-cess of promotion and outreach activities ratherthan geographic targeting—suggesting that pro-motion and outreach alone are not sufficient toensure that the poor benefit disproportionately.

In the other five Social Funds 2000 countries,the representation of the poor among socialfund beneficiaries is larger than their share in thenational population, although the difference isslight. While the high levels of poverty in Zam-bia (over 70 percent of the population) mean thatthe overall risk of Type II errors is considerablylower than for other countries, the errors stillexist.

A DFID (Department for International Devel-opment, U.K.) study contends that safety nets forparticularly vulnerable groups cannot be left tolargely demand-driven initiatives, but rather haveto be based on more active (supply-driven)attempts at inclusion.15 Where social funds havetargeted the most vulnerable sections of thepopulation, for example, through the sponsoredsubprojects component of MASAF in Malawi,their success depends on the extent to whichintermediaries (e.g., NGOs) who “sponsor” sub-projects on behalf of AIDS victims, the elderly,the disabled, and orphans represent the latter’svoice and interests.

Trends over TimeTargeting has generally improved over time asa result of better targeting procedures, such asthe use of poverty maps and the adoption ofallocation formulae; more extensive outreachactivities made possible in part by decentral-ization of the social fund, as word of the socialfund has spread; and a shift from emergencymandates to longer-term goals when increasedattention was given to targeting.

Better Poverty Targeting ProceduresImprovements in targeting procedures arereported in the ICRs for Ecuador, Egypt, Guate-mala, Haiti, Honduras, and Peru. In Nicaragua,there has been a steady refining of the povertymap that is used to classify the poverty level ofcommunities as “extreme,” “high,” “medium,”and “low.” Expenditure plans for each categoryaim to ensure higher per capita expenditure inpoorer communities. This poverty map hasevolved both to update the data and to improvethe targeting mechanism. The first map used abasic needs composite indicator calculated frominfant mortality (40 percent), access to drinking


8 8

water (40 percent), proportion of displaced per-sons (20 percent), and population weights fromthe 1971 census. The resulting scheme was biasedin favor of densely populated municipalities.16 Anew map introduced in 1998 used data from the1993 LSMS and 1995 census to estimate income-poverty measures for all areas. A third wave ofrefinement would be to base the poverty map onsome combination of income and basic needs aswell as census data, and differentiate the target-ing method based on the specific type of serv-ice being delivered (income poverty may not bethe best means to target what are essentiallyinfrastructure facilities such as water or educa-tion).17 In Sub-Saharan Africa, several social fundsare now developing poverty maps to enableprogressive targeting of resources to poorer dis-tricts; examples include Malawi, Zambia, andZimbabwe.

Other changes have also helped implementa better geographic allocation of resources.Within the social fund these include both morevigorous promotion and outreach (discussedabove), but also the decentralization of the struc-ture of the social fund, together with greateroperational integration with local government.The latter channel will be most effective wheregovernment decentralization is also taking place.

Improved PerformanceData from Zambia already show some improve-ment in targeting during the 1990s. In the period1991–94, 20 percent of resources went to theleast-poor districts (accounting for one-fifth ofthe population). By 1995–98 this figure hadfallen to 10 percent,18 and it is expected to fallfurther in the future using the progressive tar-geting rule adopted under the new ZAMSIFprogram.19

The changes in Nicaragua have also led toimproved poverty targeting outcomes. Theexpenditure share of municipalities in Nicaraguaclassified as extreme poor (which account for 18percent of the population) has improvedmarkedly, rising from 11 to 34 percent between1991 and 1998.20 In Peru, the share in educationspending of the bottom 20 percent of the pop-ulation (district basis) rose from 25 to 59 percentbetween 1992 and 1998. For the bottom 40 per-

cent, these figures are 52 and 93 percent respec-tively. By 1998, no resources at all were goingto the top quintile.21 The success of the Peruviansocial fund in reaching the poor is the combi-nation of a number of features: using a pro-gressive targeting rule with community-levelpoverty data to prioritize applications, an activePromotion Office that helps poor communitieswith their applications, and targeting only ruralareas (in later years).

Politics and TargetingDoes success at reaching the poor mean thattraditional political economy constraints thatresult in elite capture do not apply in socialfund projects? Not necessarily. Politics canwork for, or against, a pro-poor allocation ofresources.

Political motives can lead to resources beingdirected to the poor. This seems to have beenthe case in Peru, where overall social fund spend-ing increased before the 1993 election and therewas an attempt to “buy support” in the alloca-tion of funds.22 But this was also compatiblewith a pro-poor allocation of resources, whichhas disproportionately favored poorer areas.

In Nicaragua and Panama, political pressureshave played a part in offsetting a progressiveallocation rule. In Nicaragua, a project statusreport observed that poorer areas were failingto spend their allocations, and funds were beingdiverted to the better-off areas. It noted that theBank reviewed the ex-ante distribution of FISEfunds according to the poverty map and theactual figures of approved projects. Some munic-ipalities, especially those in the Atlantic coastregion, have overdrawn their allocations con-siderably whereas others, especially extremelypoor municipalities, are far from executing theirallocations. This points to weaknesses in theplanning and promotion department of FISE, butit also shows the increasing (political) pressureon FISE to finance projects outside the ex-anteallocations. “We urged FISE management tomake sure that the overdraw is not made at theexpense of the poorest municipalities.”23 Panamahas been similar: “as for political pressures forthe allocation of FES subprojects, the agreedpoverty criteria do offer some protection to the

A n n e x e s

8 9

FES against such pressures. This should beactively monitored, however, as requests andapprovals for projects from the two poorestgroups (Groups 1 and 2) have fallen short oftheir agreed share while those for the two bet-ter off groups (Groups 4 and 5) have exceededtheir agreed allocation.”24

The examples show that planned allocationsmay deviate from what is actually allocated andthat this deviation can be to the detriment of thepoor. But the examples also show the importanceof facilitating demand from poor communitiesto ensure that they use the resources allocatedto them.


9 0

9 1

Winners (likely to support the social fund)

Donors. Because social funds have special status, social fund

projects are generally easy and satisfying to work on. Social

funds have the power to implement the suggestions of donors,

the management style is familiar, and the staff are highly pro-

fessional. Within donor organizations (including the World

Bank and IDB, according to Tendler) social fund loan approval

is placed on a fast track.

Donors and borrower governments. Social funds help “sell”

austerity reforms.

Borrower governments. “Donors often suggest social funds

as part of a larger lending package, borrower governments

tend to favor their creation as a surefire way of obtaining donor

financing” (Tendler).

Governments/ruling parties. Social funds allow them to gain

political support of power groups (e.g. education and health

workers, contractors, politicians, and important geographically

defined groups) while addressing the issues of poverty

(Schady).

Social fund professionals. They are exempt from civil serv-

ice regulations, enjoy a professional environment, are com-

petent, and receive a great deal of support from donors.

Line ministries. They feel relieved not to have to deal with

infrastructure.

NGOs. In many countries, NGOs have sprung up in response

to the social fund’s desire to work through them. Skeptics point

to NGOs run by the urban, upper-middle class, who see social

funds as a way to obtain funding for their own private gain.

Losers (not likely to support the social fund)

Donors. If they have concerns regarding political uses, e.g.,

for partisan purposes.

Opposition parties. If the ruling party uses social funds to its

political advantage.

Politicians. If they cannot manipulate social fund expenditures

(while they may have more control over line ministries) (e.g.

Bolivia. Subbarao and others, p. 142).

Bureaucratic actors outside of the social fund (for example,

line ministries). They resent the social fund’s special status

and freedom from civil service and procurement regulations

and may perceive competition for funds and/or status.

Line ministries. They may feel the social fund (especially

because it is demand driven) takes away their ability to set

nationwide priorities (including targeting resources where they

believe the need lies).

NGOs. Those that are not selected for social fund projects and

feel discriminated against, or if social fund does not work with

NGOs at all.

ANNEX G: SOCIAL FUND WINNERS AND LOSERS


9 2

Winners (likely to support the social fund)

Building contractors, project design firms, equipment suppli-

ers. They gain contracts and may be able to determine project

choice (especially if communities are unaware of the range of

menu options or rely on these actors for help in application.

Local education and health workers. They can improve their

work environment and be perceived to “bring” the project to

the community (especially if the community is unaware of the

range of menu options or their own ability to apply or if they

rely on these workers for the application process).

Mayors, legislators, ward chairs, etc. They can be perceived

to give the project to the community (especially if communi-

ties are unaware of their own ability to apply or if they lack

capacity). Local administration officials whose capacity to ful-

fill their functions is enhanced.

Municipalities. Those that were targeted for political reasons

(e.g., political support of the ruling party) (Schady).

Workers. Those that gain employment on social fund projects.

Less-poor communities. They are able to obtain benefits

despite the social fund’s poverty reduction goals. This will be

the case especially if the poverty map is not very localized and

if the screening process does reject applications from less-

poor communities.

The non-poor within a community. Because social funds pro-

vide public goods to geographically defined communities,

the non-poor within a community benefit. (To the extent that

geographic units are heterogeneous with respect to income,

social funds are not narrowly targeted at the poor.) The non-

poor within a community may be able to determine the proj-

ect type if the poorer members lack capacity or time to

participate in decision making.

The poor within a community. They benefit from the public

goods provided.

Community. If they previously had no access to social fund-

type infrastructure or if they had no access to similar match-

ing funds. (e.g. Social Funds 2000 study found that FISE was

the cheapest source of funds for communities in Nicaragua.)References: Graham and Kane 1998; Nelson 1999; Schady 1999. Subbarao and others 1997; Tendler 2000a.

Losers (not likely to support the social fund)

Building contractors, project design firms, equipment suppliers.

Those that have “inside track” to line ministries and who may

lose business, particularly large contractors.

Local governments. Many social funds bypass local govern-

ment. Those that do work through local governments may often

do so conditionally, based on their perceived capability and

seriousness or based on their party affiliation.

Municipalities controlled by the opposition (e.g. Mexico’s

PRONASOL, see Subbarao and others, p. 143; and Schady).

Workers. If they compare social funds to workfare programs

that pay a higher wage (Subbarao and others, p. 135). Local

workers who resent contractors’ bringing labor from outside.

Less-poor communities. If they cannot gain access to social

fund resources, they may feel a competition for the resources.

(Subbarao and others, p. 135).

The poorest communities. Those that do not have capacity or

time to apply for and make community contributions of unpaid

time (this problem is mitigated if the social fund initiates con-

tact with poor communities, assists in the application process,

and reduces the required community contribution.

The poor among the community. If they perceive social fund

benefits flowing mainly to the better-off.

Community. If they believe that they should not be expected

to provide the “voluntary contribution” (which less-poor com-

munities may not have to provide in order to obtain educa-

tion or health services).

9 3

Because of difficulties in controlling for construction quality and varying cost accounting method-ologies, the differences in costs between programs in this Annex should be regarded as indica-tive rather than conclusive.

ANNEX H: COST DATA

Country Sector Comparator Per unit Per beneficiary

Armenia Education Central agency Lower (3.76) Lower (1.56)

Education NGO Lower (2.97) Higher (0.91)

Water (rural) NGO Higher (0.69 -0.85) Mixed (0.83 -1.07)

Bolivia Education NGO Higher (0.90) n.a

Honduras Education Central Agency Higher (0.74) Lower (1.20)

Water Central Agency Mixed (3.04 -0.31) Higher (0.12 -0.99)

Water Local Government Higher (0.87 -0.42) Higher (0.19 -0.47)

Water NGO Higher (0.30) n.a.

Sewerage Central Agency Higher (0.84) Higher (0.84)

Sewerage Local Government Lower (1.13 -1.45) Mixed (0.90 -1.01)

Health Central Agency Mixed (0.81 -2.05) n.a.

Nicaragua Education Central Agency Higher (0.64 -0.69) Higher (0.66 -0.73)

Education Local Government Higher (0.57 -0.83) Higher (0.24 -0.42)

Water Central Agency Higher (0.12) Higher (0.46)

Water Local Government Higher (0.17 -0.51) Higher (0.11 -0.26)

Water NGO Higher (0.03 -0.42) Higher (0.39 -0.61)

Sewerage Central Agency Lower (5.42 -6.42) Lower (5.47 -6.50)

Sewerage Local Government Same (1.00) Same (1.00)

Latrines Local Government Lower (1.02) Lower (2.27)

Latrines NGO Higher (0.72) Higher (0.63)

Health Local Government Higher (0.48 -0.77) Higher (0.30 -0.83)

Health NGO Higher (0.42 -0.68) Higher (0.18 -0.50)

Peru Education Central Agency Lower (1.65) Lower (1.41)

Water NGO Lower (1.69) Lower (1.05)

Latrines NGO Lower (1.14) n.a.

Health Central Agency n.a. Lower (2.95)

Zambia Education Central Agency n.a. Lower (1.11 -3.02)Source: Social Funds 2000 data provided by Social Protection Unit, 2001.

S o c i a l F u n d C o s t s V e r s u s C o m p a r a t o r s( I n d e x S F U n i t C o s t = 1 , c o m p a r a t o r i n p a r e n t h e s e s )


9 4

Average Unit Cost for Construction of Schools(cost including building, furniture, and transportation of material) 1999–2000

UNDP SFKCSchool (no.) Cost (M2)a School (no.) Cost (M2)b

2 rooms 14 72.65 2 66.13

3 rooms 5 61.38 46 68.36

4 rooms 46 65.46

5 rooms 80 65.49

Average 69.68 66.25In Siem Reap Province (US$)UNDP (Carere) Social fund

Estimated totalcost after

competitive EstimatedSchool m2 process total cost Cost per m2 m2–cost m2

First comparison

144 m2 9,250.00 9,273.00 64.40 216 – 70.58

144 m2 9,470.00 9,517.00 66.09 288 – 63.19

144 m2 9,107.00 8,925.00 61.98 360 – 60.58

144 m2 9,311.00 9,125.00 63.37 432 – 71.68

144 m2 9,427.00 9,238.00 64.13

144 m2 8,468.00 8,298.00 57.63

144 m2 9,200.00 9,898.00 68.74

Average 63.77 66.51

Second comparison

216 65.53 216 – 70.58

216 60.19 288 – 63.19

216 60.19 360 – 60.58

216 68.74 432 – 71.68

216 52.26

Average 61.38 66.51a. The unit costs are for Siem Reap province only.b. The unit costs for all the provinces.

PAMR AGDSD (semi-All in CFAF AGeFIB (Bank project) PAGER (AfDB) autonomous agency)

3 room school storeroom, office tables/chairs 10,000,000 12,500,000 12,000,000 14,700,000

One market stall 750,000 1,800,000

Storeroom 5,000,000 6,500,000 12,000,000

Clinic/supplies 5,000,000 6,000,000

Wells (linear meter) 200,000 280,000 180,000Source: Task Team, June 2001.

B e n i n

C a m b o d i a

A n n e x e s

9 5

Average costs proposed by the various institutions fordelivery of a rural water system

UNICEF $ 9,918

Municipalities $25,600

MINSA, Health Ministry (contracted to

NGOs and private firms) $28,841

FIS (contracted to NGOs and private firms) $42,850a

a. FIS has claimed much better design and quality and further analysis of costs is beingcarried out. Accounting differences may exist between the programs. Another source pres-ents the cost differential between programs as: MINSA (in-house): $35,200; NGOs:$25,000; UNICEF: $9,300; the figures for FIS are the same through either source.Source: Latin America and the Caribbean Regional Office, World Bank, January 2001.

P a n a m a

For the Social Development Fund Project, Phase I (SDF I), anex-post cost analysis of small infrastructure subprojects wascarried out on a case-studies basis. The project-by-project datacollection and analysis found that the Romanian Social Devel-opment Fund’s (RSDF) small rural infrastructure subprojectsare executed for less than two-thirds (62.2 percent) of thecost of comparable non-RSDF works. The largest variance withnon-RSDF financed execution is found for building rehabili-tation (community center) at 34 percent of non-RSDF costs. Pro-ject design for RSDF small rural infrastructure subprojects iscontracted out at slightly over half (52.3 percent) of the designcost of comparable non-RSDF financed projects. The largestvariances with non-RSDF financed projects are found amongroad projects at 50.4 percent. In the area of site supervision,RSDF financed subprojects are also substantially cheaper(57.3 percent) than comparable non-RSDF projects.

R o m a n i a

Health CentersUnit cost by m2 (including costs of building, latrine, well, etc.) 1998, 1999, 2000

Number Average cost Number Average costADB ADB SFKC SFKC

Square design 149 23,656.32 64 21,458.74Long design 9 24,378.89 41 23,559.22Source: Supervision Mission, June 2000.

C a m b o d i a

Cost comparisons between FONCODES and other institutionsType of projectSchools FONCODES INFES

Cost per m2 198.90 352.97

Cost per classroom 10,820.00 41,567.67

Cost per student 328.00 495.51

Irrigation systems FONCODES PRONAMACHCS

Cost per meter 18.28 26.29

Cost per Ha 241.80 307.47Potable water system FONCODES CARE

Cost per family 269.73 331.08Source: FONCODES Cost-effectiveness study (Social Funds 2000).

P e r u


9 6

Average cost in selected activities for education subprojectsUnit of Ministry of

Item measure Social fund Public works Construction

Excavation of usual earth (for foundation) Cubic meter 2.05 2.4 7.5

Foundations Cubic meter 98.72 118.42 183.3

Roofs Cubic meter 110.58 126.19 195.8

Bazelt stone walls Cubic meter 29.46 39.26 52.08

Squared stone walls Cubic meter 40.76 57.86 Na

Walls Square meter 2.25 2.44 3.75

Ceiling Square meter 1.98 2.09 3

Average cost for civil works in education projects, US$, 1999Education sector

investment BasicItem SFD Public works project education

Cost per square meter 145 170 170 188

Cost per square meter, double floor 122–174 Na 149 Na

Average cost of equipment for health units by different agencies, US$SFDa Ministry of Healthb

September 1998 13,957 15,372

April 1999 6,959 13,555

October 1999 13,583 24,806

Average cost per unit, SFD vs. Government, Water Supply Projects (US$), 1999Item Unit SFD GARW

4” dia medium LM 12 21

3” dia medium LM 6.5 17.2

2.5” dia medium LM 7.3 14.4

2” dia medium LM 5.6 12.5

1.5” dia medium LM 4.34 9.4

Water tanks M3 128 187.5

Pump House M2 191 218

Pumping unit (pump + engine)

(head 100m–150m & discharge 4–6 lit/s) Per unit 13,470 14,062a. Including lab equipment.b. Excluding lab equipment.Source: World Bank 2001.

Y e m e n

9 7

Overall Bank performance is measured by theBank’s performance on identification, appraisal,and supervision. OED ratings on overall Bankperformance for social fund projects comparefavorably with those for other Bank projects. Ofthe 23 closed social fund projects as of end fis-cal year 2000, 94 percent were rated “satisfac-tory” on overall Bank performance. The figurefor PTI projects worldwide was 75 percent andfor all World Bank projects, 75 percent.1 Byabout mid-2000, QAG had conducted 9 quality-at-entry assessments and 7 quality-of-supervisionassessments for social fund projects. Sixty-sevenpercent of the quality-at-entry reviews containeda “satisfactory” rating (6 of 9 reviews) and 33 per-cent a “marginal” rating (3 of 9 reviews). Withrespect to quality of supervision, 14 percent ofthe reviews contained a “highly satisfactory” rat-ing (1 of 7 reviews), 57 percent a “satisfactory”rating (4 of 7 reviews), and 29 percent a “mar-ginal” rating (2 of 7 reviews). Ninety-one per-cent of the active social fund projects in June1999 were rated “not at risk.”2

Safeguard and Environmental Issues. Analy-sis of the project status reports for a random sam-ple of 25 social fund projects found that, for most,safeguards are not rated, not applicable, or leftblank. Only 15 (6 percent) ratings were givenin the 25 reports. More than half (9 ratings) ofthese were for environmental assessment, withtwo each for indigenous people and involuntaryresettlement, and one each for forestry and inter-national waters. This overall coverage is com-parable to that found in a random sample ofother PTIs (drawn from the same countries), forwhich 12 (5 percent) of the possible 250 ratingswere given. This does not mean that safeguardsare being adequately applied, however. Specif-ically, environmental assessment, natural habi-

tats, indigenous peoples, and cultural propertyall appear likely to be relevant in a larger num-ber of cases than have included them thus far(e.g., indigenous people were a target group inat least six social fund projects).

Of the 16 available social fund QAG reviews(quality at entry and quality of supervision), 10were rated “satisfactory” for environmentalaspects while 6 were rated “not applicable.”There are many examples of social fund proj-ects where efforts have been made to developmicroproject environmental assessment mecha-nisms and adequate provisions have been madefor environmental screening (and for subprojectanalysis as necessary). The application of theseprovisions in practice appears weak, however,and has not been systematically monitored.There are indications that the issue warrantsgreater attention, especially with regard to envi-ronmental safeguards. No social fund has beengiven a Category A environmental classification,but close to half (43 percent of a random sam-ple of 30 social fund projects) have been clas-sified as Category B. The rest have been assignedCategory C. QAG reviews have questioned ifthere is not excessive use of Category C forenvironmental impact of social fund projects,implying that environmental issues have beengiven inadequate attention.3 In one SIF, QAGnoted an inconsistency between the Category Cenvironment designation adopted for social fundprojects and the treatment of similar activitieswhen they are freestanding or sector-type proj-ects. In another social fund project, the QAGreview pointed out an inappropriate assignmentof the environment category for the opposite rea-son: although the project was rated Category Cfor environmental impacts, it included substan-tial efforts dedicated to environmental control

ANNEX I: BANK PERFORMANCE

and mitigation—water and sanitation, sustainableagricultural practices, monitoring use of agro-chemicals—with staff members responsible forenvironmental oversight, environmental criteriafor subproject evaluation, and a proposed envi-ronmental management plan. The same pointcan be made in relation to other safeguards. Inone project , the QAG review questioned theproject team’s indication that the Bank’s Oper-ational Directive on resettlement did not applyto the project, and recommended that a Bankresettlement specialist advise on the potentialproblems identified. The Bank has recently intro-duced a new environmental category “Financial

Intermediary” (FI) that can be applied to socialfund projects. For all practical purposes, an FIcategory is similar to a B category—a process isdefined to screen “not as yet defined subprojectactivities” that may be identified (by projectmanagement during project implementation) asneeding an environmental screening. The effec-tiveness with which environmental issues areaddressed in Bank-financed social fund projectswill depend on how well projects are identifiedand screened. The Bank should be assured thatthese functions are adequately performed byproject management.4


9 8

9 9

ANNEX J: STAKEHOLDER SURVEY RESULTS

Government NGO Social fund Other donor World Bank Other Missing TotalN=22 N=27 N=56 N=13 N=19 N=22 N=1 N=160

What impact has the social fund

had on the institutional capacity

of the local governments?

Positive impact 55 33 66 46 63 55 0 55

Negative impact 0 0 0 0 5 14 0 3

No impact 14 30 7 15 0 5 0 11

Do not know 14 37 18 8 11 9 100 18

Missing 18 0 9 31 21 18 0 13

Total 100 100 100 100 100 100 100 100

What impact has the social fund

had on the decentralization

process in the country?

There is no decentralization process

in the country 14 19 7 15 5 0 0 9

Positive impact 45 37 55 31 32 59 0 46

Negative impact 0 4 0 8 0 5 0 2

No impact 14 22 14 8 5 9 0 13

Do not know 9 11 14 8 16 9 100 13

Missing 18 7 9 31 42 18 0 17

Total 100 100 100 100 100 100 100 100

At the time they make their

proposal, to what extent are

communities aware of the full

range of eligible subprojects?

Fully aware 32 19 48 8 16 9 0 28

Partially aware 32 59 29 62 32 64 100 43

Unaware 5 15 2 8 5 5 0 6

Do not know 5 7 9 0 5 5 0 6

Missing 27 0 13 23 42 18 0 18

Total 100 100 100 100 100 100 100 100


1 0 0

Government NGO Social fund Other donor World Bank Other Missing TotalN=22 N=27 N=56 N=13 N=19 N=22 N=1 N=160

At the time they make their

proposal, to what extent are

communities aware of the

responsibilities they will have

to meet as a result of receiving a

social fund-financed investment?

Fully aware 45 30 45 0 32 18 100 34

Partially aware 27 52 34 69 16 55 0 39

Unaware 0 7 0 8 5 5 0 3

Do not know 5 11 9 0 5 5 0 7

Missing 23 0 13 23 42 18 0 17

Total 100 100 100 100 100 100 100 100

To what extent are subproject

choices made with the active

participation of the community?

Choice is usually made with the

active participation of diverse

groups in the community 41 37 54 8 0 36 0 36

Choice is usually made with the

active participation of the

community, although some groups

have little voice 23 33 20 23 42 23 100 26

Choice is inappropriately influenced

by NGOs, private contractors,

elected or unelected community

leaders 9 26 4 15 5 14 0 11

Do not know 5 4 7 0 11 9 0 6

Other 0 0 4 31 0 0 0 4

Missing 23 0 13 23 42 18 0 17

Total 100 100 100 100 100 100 100 100Note: The category “Other” included six consultants, one international NGO, one social development partnership organization, one networking institute, one UN specialized agency, andone university.

1 0 1

Source: Owen and Van Domelen: Getting anEarful: A Review of Beneficiary Assessments ofSocial Funds.

According to a Bank review of beneficiary assess-ments for Armenia, Bolivia, Chile, Ecuador,Malawi, Peru, Senegal, and Zambia, beneficiaryparticipation in the identification of projectsvaried widely across funds, with better per-formance where there is both a formal mecha-nism and where beneficiary committees areeligible to execute projects directly. In all socialfunds, community participation was higherduring identification than in the actual design ofprojects.

Beneficiaries usually felt that projects reflectedtheir needs and priorities even when they hadnot been involved in identification. For instance,in Peru, where 96 percent of respondents saidthe project was a community priority, only 66percent of the beneficiaries said the project wasprioritized by the community itself, 7 percent bythe mayor, and 6 percent by either the promoter,project sponsor, or contractor. On the whole, theinsertion of intermediaries, be they local gov-ernments, NGOs, promoters, or private con-tractors, did not adversely affect the relevanceof the project to the community. This may bedue, in part, to the basic types of investmentseligible for financing by social funds and the largerange of unmet needs in poor communities.

The characterization that communities organ-ize, enter into direct communication with thesocial fund, and prepare their own projects wasnot fully borne out in the beneficiary assess-ments. Beneficiaries perceived a more complexweb of actors inserted between the social fundand the community in terms of promoting andpreparing projects.

Beneficiary communities participated to ahigh degree in the execution of projects. Of thevarious types of participation, beneficiaries oftendefined participation only in terms of resourcesgiven. The most frequent participation was man-ual labor, followed by money, and then mate-rials. In rural areas, participation often involvedlabor rather than money, whereas cash contri-butions were more common in urban areas.

Beneficiary views on community contributionstended not to be solely positive, pointing outproblems like skewing the choice of project,regressive effects between communities, andnegative effects such as less time tending fieldsand perception of unequal cost sharing. Despitesocial funds’ reputation for efficient execution,beneficiaries commonly cited problems experi-enced during execution, many arising fromsocial fund capacity or procedures, especiallydelays in disbursements and lack of adequatesupervision.

In general, social investments tend to reachpoor sectors that exhibit active conduct in search-ing for solutions to their problems, and themore active communities are thus better able toaccess and implement social fund projects,regardless of poverty levels. Community lead-ership was a crucial factor in being able to pres-ent and execute projects, and several funds havebegun to address community leadership as partof local capacity building.

Methodological IssuesThe focus of the beneficiary assessments is nat-urally beneficiaries, but the various levels ofinformants, such as social fund regional staffand service providers, are not always included.Nor did all the assessments include a controlgroup of non-project areas to measure net effects.

ANNEX K: HIGHLIGHTS OF BENEFICIARY ASSESSMENT FINDINGS ONPARTICIPATION

The choice of who to interview as beneficiariesinfluences results; asking only direct beneficiar-ies and project committee members about com-munity needs rather than community membersat large is likely to yield biased results. The par-ticular questions asked in beneficiary assess-ments are also likely to influence the results. Forexample, the types of questions regarding com-munity needs indicate an explicit or implicit biastoward what is eligible for financing, thus ignor-

ing other needs. In cases where the communitylacks access to most basic services and infra-structure, the list of priorities may be extensive.Furthermore, from the final beneficiary assess-ment reports, it is often unclear how particularquestions were posed and upon what under-standing and interpretation responses were given.Finally, the sample sizes, in some cases as fewas two per project site, may raise questions aboutthe representativeness of the assessment findings.


1 0 2

1 0 3

Source: Based on Samantha de Silva, Community-based Contracting: A Review of StakeholderExperience.

Among the social funds that use community-based contracting, several different models arefound. Funds may be channeled to and managedentirely by the community (e.g., Malawi SocialAction Fund, Zambia Social Recovery Project,Peru FONCODES, Bolivia social fund), or chan-neled through an intermediary agency workingclosely with the community (e.g., Ethiopia SocialRehabilitation and Development Fund), or acombination of approaches may be used,depending on the institutional capacity of thecommunity, as in the Eritrea Community Devel-opment Fund.

What Are the Benefits? • Community contracting can be more effi-

cient, more transparent, and result in lowercosts compared with centralized bidding.

• Community contracting can encourage a senseof ownership among beneficiary communities,assuring better operation and maintenance.

• The participatory process can be institution-alized through partnerships betweencommunities and local governments, and thedecentralization process strengthened,when social funds work closely with localgovernments.

What Are the Risks?• Poor technical execution and manipulation by

contractors can occur when community lead-ership is not sufficiently accountable to users,or when communities lack relevant businessor technical skills. Training and active supportcan mitigate these risks.

• The interests of marginalized groups may beignored when leaders do not represent theinterests of the very poor, women, or indige-nous groups. Pre-investment support in usercommittee formation can help to mitigatethis risk.

• Project selection is biased in favor of small,community-level projects that fit the com-munity contracting criteria. This bias can bereduced if NGOs and local government unitsdemonstrate capacity in managing larger andmore complex projects on the community’sbehalf.

• Economies of scale may be lost when com-munities cannot benefit from volume dis-count or high-quality technical supervision.

When Is it Appropriate?• When users are a clearly identifiable group

of households in the same community, withsome degree of social cohesion.

• When government supports communitymanagement.

• When community user-based organizationshave legal status and responsibility for O&M.

• When mechanisms for community capacitybuilding are available.

ANNEX L: COMMUNITY-BASED CONTRACTING

1 0 5

The Comparison Group MethodologyThe OED Review adopted the comparison groupmethodology to identify corresponding matchedcommunities for the 17 randomly-selected socialfund–assisted communities in the 4 field researchcountries based on input from local researchers,social fund staff members, community leaders,and local government officials. Seven broadcriteria were used to match each socialfund–assisted community to a correspondingmatched community: (a) Similar geographic, socioeconomic and cul-

tural characteristics(b) Similar type of community (urban or rural)(c) Similar poverty level (non-poor/poor/

extremely poor)(d) Similar population density of the community(e) Similar type of facility financed by social

fund investment(f) Community without social fund investment (g) Accessibility to the communities.

The quality of the match was tested and con-firmed for each of the social fund–assisted com-munities using information from the OEDhousehold surveys on: (a) The household characteristics(b) The average economic status(c) The religious make-up(d) The socioeconomic problems faced by the

community(e) The ways of interaction of community

members(f) The level of participation by community

members in collective action(g) The role of community leaders and govern-

ment officials(h) The level of social network.

The analysis resulted in some matching prob-lems. These problems were addressed as follows:• In Nicaragua, a pair of social fund–assisted

community and its match was droppedbecause the subproject selected was a proj-ect that was implemented in 1995, beforeFISE’s emphasis on community participation.The retained subprojects are 1998 or post-1998approvals, when community participationwas an explicit emphasis of FISE.

• In Jamaica, the road subproject selected didnot have a comparable subproject in thematched community. Thus, the match was notconsidered for the analysis of the sustain-ability-related issue.

• In Malawi, three communities selected asmatched communities had received socialfund financing for another type of subproject,and thus these communities could not beused for comparison relating to changes incapacity building and social capital resultingfrom the social fund. These three matchedcommunities were dropped for purposes ofthe capacity building and social capital analy-sis. Such elimination of (three of five) com-munities can lead to a bias, since the fivesocial fund–assisted communities were com-pared to two matched communities. A sim-ple t-test was conducted to see the meandifference in the two groups of social fund–assisted communities (group 1 with a suitablematch and group 2 with no match). The resultindicated that for all but three variables (test-ing over 15 relevant variables) the differ-ences were insignificant. For three variablesthe mean of group 2 was larger than themean of group 1. Thus, a comparisonbetween the social fund–assisted communi-

ANNEX M: OED HOUSEHOLD SURVEY RESULTS

ties and the matched communities couldresult in a bias, and the bias, if any, wouldbe in favor of social funds.

Two techniques were used to analyze thedata. First, t-statistics for difference of means ondisaggregated data (based on gender and eco-nomic status within social fund–assisted com-

munities) and difference-in-difference between thetreatment and the control group. Second, probitor ordered probit multivariate regressions wereused to analyze the impact of social fund invest-ment. A Heckman two-step procedure—two-stepconsistent estimators and maximum likelihoodestimates—were also conducted, and all the mul-tivariate analyses gave broadly similar results.


1 0 6

Jamaica Malawi Nicaragua Zambia Total

Top priority addressed 27% 34% 23% 22% 27%

Top priority (within social fund menu) addressed 31% 47% 26% 37% 34%

One of top three priorities addressed 42% 52% 35% 38% 43%

Total number of social fund–assisted communities’

respondents 284 499 252 490 1, 525Note: The problem “No school building” could refer either to the school building being too far away, the lack of appropriate grades in the school, the lack of classrooms, or to the actualabsence of a school. Since the social fund project financed rehabilitation or expansion of a school building in the community, the analysis was “adjusted” to account for the fact that thecommunity mentioned “No school building” as a problem when in fact there was a school in the community. For each respondent who stated “No school building” as a problem in thebase year but did not mention it as a problem in the year of the survey (2000), it was assumed that his/her problem was addressed.

Similar adjustment methodology adopted for clinics.In Jamaica, the problem identified by community #4 was “Lack of safety or sports field.” The social fund financed a sports stadium in that community. The M2 table considers this to

be a good match between the problem and the financed subproject—participatory research by Moser and Holland in 1997 had identified the importance of community facilities (such assports complexes) for addressing issues of violence in Jamaica.

P r i o r i t y P r o b l e m s o f t h e C o m m u n i t yi n t h e B a s e Y e a r

T a b l e M . 1

A n n e x e s

1 0 7

Question Base Jamaica Malawi Nicaragua Zambia

Who had the most important role in the decision of Heard of SFs 53 78 71 32what type of project to apply for? [leader]

Did any of the following encourage you to attend Knew of meeting 39 53 36 60the meeting? [leader]

What was the most important factor in deciding Heard of SFs 12 72 4 44how much contribution each household should make? [leader]

What happened to households that did not make Heard of SFs 3 70 2 35a contribution? [they could not use the facility, or other punitive action]

Have you heard of [the social fund]? All SFAC 65 98 97 47All MC 47 92 79 30

At the time you first heard about [the social fund] All SFAC 15 89 71 37what type of projects did you think you could apply for? [school, clinic, water or road/bridge]

At the time you first heard about [the social fund] All SFAC 7 49 23 21what did you think the community would have to do to obtain the funds? [contribute cash, labor, raw material]

Did the community have a meeting to decide about All SFAC 39 79 60 33the project? Heard of SFs 60 81 62 71

Did you attend the meeting? All SFAC 28 58 27 23Heard of SFs 43 59 28 50

Knew of meeting 72 73 45 71

Did you say anything at the meeting? All SFAC 13 16 12 14Heard of SFs 20 17 12 31

Knew of meeting 34 21 19 44Attend meeting 48 28 42 62

Were you or your household asked to make All SFAC 25 91 10 50contributions of time, money, or materials during Heard of SFs 37 92 10 84the construction?a

Did your household contribute the following items All SFAC 14 91 10 49during the construction period? [cash and/or kind Heard of SFs 21 91 10 83and/or sand or stones, unpaid time]a Contribute 56 100 100 98

How did you first hear of [social fund]? [media] All SFAC 18 45 31 3All MC 28 46 31 4

How did you first hear of [social fund]? [leader] All SFAC 13 31 18 24All MC 3 17 7 10

Note: (i) All SFAC: All respondents in the social fund-assisted communities; (ii) All MC: All respondents in the matched communities; (iii) Heard of SFs: All respondents who have heardof social fund within the social fund–assisted communities; (iv) Knew of meeting: All respondents who knew about the community meeting held for selection of the social fund subproject;(v) Attend meeting: All respondents who attended the community meeting that was held for selection of the social fund subproject; and (vi) Contribute: All respondents who said that wereasked to contribute.a. The results for these questions are reported by household (rather than by respondent) because of the nature of the questions (where the focus is on household—not individual—behavior).

P a r t i c i p a t i o n ( p e r c e n t )T a b l e M . 2


1 0 8

Jamaica Malawi Nicaragua Zambia

No=0 57% 41% 72% 50%

Yes=1 43% 59% 28% 50%

Total 186 489 245 230

D i d Y o u A t t e n d t h e M e e t i n g ?T a b l e M . 3

Jamaica Malawi Nicaragua ZambiadF/dx dF/dx dF/dx dF/dx

Dummy for community #1 0.349b -0.144a 0.086 -0.228a

Dummy for community #2 0.375b -0.171b -0.210c -0.301b

Dummy for community #3 0.364c 0.061 -0.344b

Dummy for community #4 0.025 -0.198

Dummy for a SFAC’s woman -0.060 -0.072 0.099a -0.051

Economic status -0.071b 0.004 -0.035a 0.011

Size of the household 0.052 0.004 -0.003 -0.020

Number of school-going children -0.058 -0.011 -0.019 0.016

Highest education level achieved in the household 0.011 -0.009 0.021a 0.024

Dummy for a Christian household -0.118 -0.027 -0.105 0.226

Dummy for a household following another religion 0.124 -0.184

Dummy for a young respondent (age below 25) 0.268 -0.170a 0.201 -0.201

Dummy for a respondent who never married -0.226b 0.223 0.000

Age of the respondent -0.001 -0.002 0.029b -0.014

Age square 0.000 0.000 0.000a 0.000

Dummy for traditional ways of interaction (base year) 0.170 -0.114 0.128 -0.075

Dummy for social ways of interaction (base year) -0.022 0.053 -0.035 -0.138

Dummy for communal ways of interaction (base year) -0.003 0.098a 0.102 -0.130

Dummy for top problem being a public good 0.061 0.017 0.123a -0.048

Number of leaders known within community (base year) 0.091b 0.078c 0.003 0.138c

Number of leaders known outside community (base year) 0.070b -0.006 -0.055 -0.053

Level of participation in meetings (base year) 0.136b 0.070b 0.368c 0.166b

Dummy for blue collar skills (base year) 0.006 0.002

Dummy for managerial skills (base year) 0.075 0.028

Number of observations 177 455 241 189

Wald Chi square (18/22) 60.85c 79.44c 94.95c 31.10c

R-squared 0.300 0.126 0.407 0.172Note: Assuming that respondents who hadn’t heard of social fund project did not attend the meeting. (i) Annex M, table M.25 for explanations for independent variables used in the multi-variate analysis.a. 90% confidence level.b. 95% confidence level.c. 99% confidence level.

A n n e x e s

1 0 9


No=0 80% 83% 88% 69%

Yes=1 20% 17% 12% 31%

Total 186 489 245 230

D i d Y o u S p e a k a t t h e M e e t i n g ? T a b l e M . 4


Dummy for community #1 0.266b -0.045 -0.013 -0.042

Dummy for community #2 0.181 -0.053 -0.023 -0.168

Dummy for community #3 0.371c 0.056 -0.052

Dummy for community #4 -0.079a 0.087

Dummy for a SFAC’s woman -0.156c -0.087a -0.016 -0.130a

Economic status 0.020 0.006 -0.020b 0.032

Size of the household 0.035 0.016 -0.018c -0.005

Number of school-going children -0.060 -0.020 0.003 0.000

Highest education level achieved in the household -0.001 0.002 0.007 0.021

Dummy for a Christian household 0.033 -0.017 -0.121b -0.039


Dummy for a young respondent (age below 25) 0.281 0.022 -0.024 -0.062

Dummy for a respondent who never married 0.030 0.057 0.004

Age of the respondent 0.016 0.007 0.005 0.001

Age square 0.000 0.000 0.000 0.000

Dummy for traditional ways of interaction (base year) -0.077 -0.039 -0.012 -0.019

Dummy for social ways of interaction (base year) -0.099 0.011 -0.052b -0.057

Dummy for communal ways of interaction (base year) 0.124 -0.042 0.054a -0.077

Dummy for top problem being a public good -0.099 -0.087b -0.045 -0.211a

Number of leaders known within community (base year) 0.008 0.037c -0.020 0.080b

Number of leaders known outside community (base year) 0.029 -0.008 -0.006 -0.024

Level of participation in meetings (base year) 0.094b 0.028 0.087c 0.176c

Dummy for blue collar skills (base year) 0.080b 0.024

Dummy for managerial skills (base year) 0.021 -0.009



R-squared 0.213 0.195 0.270 0.213Note: Assuming that respondents who hadn’t heard of social fund project did not speak at the meeting. (i) Annex M, table M.25 for explanations for independent variables used in themultivariate analysis.a. 90% confidence level.b. 95% confidence level.c. 99% confidence level.


1 1 0


No=0 63% 10% 89% 17%

Yes=1 37% 90% 11% 83%

Total 186 489 245 230

W e r e Y o u o r Y o u r H o u s e h o l d A s k e d t oM a k e C o n t r i b u t i o n s o f T i m e , M o n e y , o rR a w M a t e r i a l D u r i n g t h e C o n s t r u c t i o n( f o r R e s p o n d e n t s W h o H a v e H e a r d o fS o c i a l F u n d s ) ?

T a b l e M . 5


Dummy for community #1 0.570c -0.112c -0.047 -0.281b

Dummy for community #2 0.325a 0.016 -0.024 -0.198a

Dummy for community #3 0.594c 0.007 -0.268a


Dummy for a SFAC’s woman -0.129 0.028 -0.025 0.045

Economic status 0.019 -0.005 0.007 -0.010

Size of the household 0.046 0.011 -0.014b 0.005

Number of school-going children 0.008 -0.019 0.040c -0.009

Highest education level achieved in the household 0.005 -0.010c 0.002 0.009

Dummy for a Christian household 0.045 0.029 -0.206c 0.773c


Dummy for a young respondent (age below 25) 0.413a 0.000 -0.004 0.054

Dummy for a respondent who never married 0.216b -0.022

Age of the respondent 0.045b -0.001 0.001 -0.006

Age square 0.000a 0.000 0.000 0.000

Dummy for traditional ways of interaction (base year) 0.085 0.116b 0.000 0.245b

Dummy for social ways of interaction (base year) -0.166 0.033 -0.008 0.081a

Dummy for communal ways of interaction (base year) 0.079 0.026 -0.051c -0.019

Dummy for top problem being a public good 0.065 0.054b 0.043

Number of leaders known within community (base year) 0.046 0.011 -0.013 0.031

Number of leaders known outside community (base year) 0.021 -0.003 -0.015 0.012

Level of participation in meetings (base year) 0.021 0.043c 0.072c 0.112c

Dummy for blue collar skills (base year) 0.034 0.186c

Dummy for managerial skills (base year) 0.013 -0.067



R-squared 0.235 0.215 0.341 0.262Note: (i) Annex M, table M.25 for explanations for independent variables used in the multivariate analysis.a. 90% confidence level.b. 95% confidence level.c. 99% confidence level.

A n n e x e s

1 1 1

Jamaica Malawi Nicaragua ZambiaSFAC Non-SF SFAC Non-SF SFAC Non-SF SFAC Non-SF

No=0 87% 97% 69% 83% 82% 93% 76% 90%

Yes=1 13% 3% 31% 17% 18% 7% 24% 10%

Total 284 294 499 195 252 257 490 487

H o w D i d Y o u F i r s t H e a r o f [ S o c i a l F u n d ] ?[ L e a d e r ]

T a b l e M . 6


Dummy for social fund–assisted community 0.050c 0.115b 0.128c 0.188c

Dummy for community #1 0.101b -0.078 -0.010 -0.075

Dummy for community #2 0.044 -0.125b 0.035 -0.008

Dummy for community #3 0.177c -0.165c -0.092b

Dummy for community #4 -0.119b -0.006

Dummy for a SFAC’s woman 0.038b 0.087 0.025 -0.055a

Dummy for SFAC’s bottom quartile respondent -0.022 -0.039 -0.068b -0.016

Economic status -0.005 -0.063c -0.030c 0.007

Size of the household -0.002 0.016 -0.008 -0.003

Number of school-going children 0.012a -0.010 0.030b 0.028b

Highest education level achieved in the household 0.000 -0.003 0.018c 0.010a

Dummy for a Christian household 0.007 -0.066 -0.037 0.038

Dummy for a household following another religion -0.010 0.062

Dummy for a young respondent (age below 25) 0.034 -0.147b -0.040 -0.072

Dummy for a respondent who never married 0.000 0.162 -0.015

Age of the respondent 0.000 -0.021b -0.003 0.002

Age square 0.000 0.000c 0.000 0.000

Dummy for traditional ways of interaction (base year) -0.015 0.024 0.002 0.026

Dummy for social ways of interaction (base year) 0.041b -0.035 -0.043 0.037

Dummy for communal ways of interaction (base year) -0.018 -0.029 -0.032 -0.030

Dummy for top problem being a public good 0.003 -0.059 0.024 0.062b

Number of leaders known within community (base year) 0.010b 0.001 -0.005 0.000

Number of leaders known outside community (base year) 0.004 -0.012 -0.011 0.021

Level of participation in meetings (base year) 0.022c -0.002 0.077c -0.006


Dummy for managerial skills (base year) -0.031 0.070b





1 1 2

Base Jamaica Malawi Nicaragua Zambia

Have you become able to engage in blue collar All SFAC 36% 21%skillsd or has your level of ability improved? All MC 44% 25%

t-statistics -0.08a -0.04

Have you become able to engage in managerial All SFAC 25% 23%skillse or has your level of ability improved? All MC 28% 28%

t-statistics -0.03 -0.05a

Compared to [base year], is it more difficult or easy All SFAC 55% 61% 43% 42%to participate in groups and associations of All MC 44% 66% 53% 37%people outside your immediate household? t-statistics 0.10b -0.05 -0.09b 0.05[Much more easy/more easy]

How often last year did you participate in All SFAC 15% 22% 7% 18%collective action, for example attended a All MC 11% 17% 18% 11%community meeting, etc? [compared to] t-statistics 0.04 0.04 -0.11c 0.07c

How often did you participate in collective action[in base year]? [more]

Compared to [base year], is the level of trust and All SFAC 45% 7% 37% 11%community cooperation between people from All MC 35% 13% 36% 20%different backgrounds and ethnic groups in your t -statistics 0.10b -0.07c 0.01 -0.09c

community better or worse or the same? [more]

Compared to [base year], is it more difficult or All SFAC 50% 61% 49% 43%easier to get the whole community to agree on All MC 44% 66% 57% 34%a decision? [easy/very easy] t-statistics 0.05 -0.04 -0.08a 0.10c

Does the government respond more effectively to All SFAC 25% 50% 18% 7%your needs now than it did [in base year]? All MC 16% 46% 24% 6%

t-statistics 0.10c 0.08a -0.07a 0.02

Do you feel your local leadershipf responds more All SFAC 29% 30% 26% 23%effectively to your needs now than [in base year]? All MC 21% 46% 35% 20%

t-statistics 0.07a - 0.16c -0.12c 0.03Note: (i) All SFAC: All respondents in the social fund–assisted communities; (ii) All MC: All respondents in the matched communities; and (iii) t-statistics: The value in this row indicatesthe difference in the mean value of the SFAC and the MCs. Notes a–c indicate level of statistically significance using student’s difference-in-means test (t-statistics).a. 90% confidence level.b. 95% confidence level.c. 99% confidence level.d. Blue collar skills include carpentry, masonry, brick making, or other skilled labor.e. Managerial skills include managing construction/maintenance/upkeep, bookkeeping/accounting, resolving disagreements, etc.f. Local leadership includes village chief, village headman, local elected official, ward chairperson, etc.

S o c i a l C a p i t a lT a b l e M . 7

A n n e x e s

1 1 3


No=0 64% 56% 79% 75%

Yes=1 36% 44% 21% 25%

Total 499 195 490 487

H a v e Y o u B e c o m e A b l e t o E n g a g e i n[ C a r p e n t r y , M a s o n r y , B r i c k M a k i n g , o rO t h e r S k i l l e d L a b o r ] o r H a s Y o u r L e v e lo f A b i l i t y I m p r o v e d ?

T a b l e M . 8


Dummy for social fund–assisted community -0.003 0.013


Dummy for community #2 -0.003 0.007

Dummy for community #3 -0.034 -0.120c

Dummy for community #4 -0.064 -0.017

Dummy for a SFAC’s woman -0.038 -0.019

Dummy for SFAC’s bottom quartile respondent -0.115a -0.079a

Economic status -0.012 -0.009

Size of the household -0.019 0.002

Number of school-going children 0.024 -0.021

Highest education level achieved in the household -0.004 0.017c

Dummy for a Christian household 0.001 0.137b

Dummy for a household following another religion -0.087

Dummy for a young respondent (age below 25) 0.089 0.036

Dummy for a respondent who never married -0.151

Age of the respondent 0.007 -0.003

Age square 0.000 0.000

Dummy for traditional ways of interaction (base year) -0.040 -0.074

Dummy for social ways of interaction (base year) 0.009 -0.026

Dummy for communal ways of interaction (base year) -0.004 -0.069b

Dummy for top problem being a public good -0.130c 0.050

Number of leaders known within community (base year) 0.066c 0.020

Number of leaders known outside community (base year) 0.036b -0.020

Level of participation in meetings (base year) 0.064b -0.008

Dummy for blue collar skills (base year) 0.415c 0.430c

Dummy for managerial skills (base year) 0.025 0.126c

Number of observations 648 757

Wald Chi square (21/26) 163.95c 256.70a

R-squared 0.217 0.378Note: (i) Annex M, table M.25 for explanations for independent variables used in the multivariate analysis.a. 90% confidence level.b. 95% confidence level.c. 99% confidence level.


1 1 4


No=0 75% 72% 77% 72%

Yes=1 25% 28% 23% 28%

Total 499 195 490 487

H a v e Y o u B e c o m e A b l e t o E n g a g e i n[ M a n a g i n g C o n s t r u c t i o n / M a i n t e n a n c e /U p k e e p o f F a c i l i t i e s o r B o o k k e e p i n g /A c c o u n t i n g o r R e s o l v i n g D i s a g r e e m e n t se t c . ] o r H a s Y o u r L e v e l o f A b i l i t y I m p r o v e d ?

T a b l e M . 9


Dummy for social fund–assisted community -0.043 0.013

Dummy for community #1 0.066 0.071




Dummy for a SFAC’s woman 0.132c 0.016

Dummy for SFAC’s bottom quartile respondent -0.035 -0.060

Economic status -0.009 -0.001

Size of the household -0.012 0.009

Number of school-going children 0.026 -0.032b

Highest education level achieved in the household 0.010a 0.018c

Dummy for a Christian household -0.017 0.073

Dummy for a household following another religion -0.022

Dummy for a young respondent (age below 25) 0.055 0.080

Dummy for a respondent who never married -0.099

Age of the respondent 0.001 0.012

Age square 0.000 0.000

Dummy for traditional ways of interaction (base year) -0.068 -0.039

Dummy for social ways of interaction (base year) 0.022 -0.109c

Dummy for communal ways of interaction (base year) 0.055a -0.066a

Dummy for top problem being a public good -0.050 0.050

Number of leaders known within community (base year) 0.047c 0.024a

Number of leaders known outside community (base year) 0.015 -0.004

Level of participation in meetings (base year) 0.022 0.034

Dummy for blue collar skills (base year) -0.013 0.115c

Dummy for managerial skills (base year) 0.580c 0.476c

Number of observations 648 757

Wald Chi square (21/26) 234.03c 280.71c

R-squared 0.443 0.443Note: (i) Annex M, table M.25 for explanations for independent variables used in the multivariate analysis.a. 90% confidence level.b. 95% confidence level.c. 99% confidence level.

A n n e x e s

1 1 5


Much more difficult=1 8% 8% 3% 2% 4% 1% 3% 5%More difficult=2 12% 26% 8% 7% 6% 19% 10% 10%Same=3 23% 19% 27% 26% 46% 28% 42% 47%More easy=4 36% 35% 52% 55% 37% 53% 29% 28%Much more easy=5 19% 9% 8% 10% 6% 0% 13% 10%Don’t know/missing= . 1% 3% 0% 0% 0% 0% 2% 1%Total 284 294 499 195 252 257 490 487Pearson Chi-square 24.74 c 2.43 53.32 c 4.22

C o m p a r e d t o [ B a s e Y e a r ] , I s I t M o r eD i f f i c u l t / E a s y t o P a r t i c i p a t e i n G r o u p sa n d A s s o c i a t i o n s o f P e o p l e O u t s i d e Y o u rI m m e d i a t e H o u s e h o l d ?

T a b l e M . 1 0


Dummy for social fund–assisted community 0.400b -0.292b 0.190 0.222b

Dummy for community #1 -0.705c -0.093 -0.079 0.158Dummy for community #2 -0.940c 0.376b -0.428c -0.030Dummy for community #3 -1.079c -0.344b 0.452c


Dummy for a SFAC’s woman -0.030 0.143 -0.171 -0.035Dummy for SFAC’s bottom quartile respondent 0.011 -0.255a 0.131 0.115Economic status -0.027 -0.031 -0.067a -0.032Size of the household 0.006 -0.060 0.018 0.046a

Number of school-going children -0.023 0.078 0.051 -0.088b

Highest education level achieved in the household 0.009 -0.013 0.026 0.031a

Dummy for a Christian household 0.055 0.024 -0.414b -0.035Dummy for a household following another religion 0.550b -0.332Dummy for a young respondent (age below 25) -0.466b -0.210 -0.629c 0.078Dummy for a respondent who never married -0.177a -0.521 -0.185 0.093Age of the respondent -0.024 -0.037b -0.076c 0.008Age square 0.000 0.000b 0.001c 0.000Dummy for traditional ways of interaction (base year) 0.185 -0.133 0.045 -0.390b

Dummy for social ways of interaction (base year) -0.156 0.080 0.092 -0.276c

Dummy for communal ways of interaction (base year) 0.122 -0.007 0.056 -0.449c

Dummy for top problem being a public good -0.076 -0.342c -0.046 0.135Number of leaders known within community (base year) 0.001 0.169c 0.233c -0.026Number of leaders known outside community (base year) -0.024 -0.043 0.013 0.057Level of participation in meetings (base year) 0.171c 0.054 -0.090 0.054Dummy for blue collar skills (base year) 0.285c 0.393c

Dummy for managerial skills (base year) 0.029 0.279c

Number of observations 523 648 496 747Wald Chi square (21/26) 119.22c 94.16c 72.14c 266.05c



1 1 6


Less=1 17% 20% 16% 13% 11% 12% 10% 7%

Same=2 68% 69% 63% 70% 83% 70% 72% 82%

More=3 15% 11% 22% 17% 7% 18% 18% 11%

Total 284 294 499 195 252 257 490 487

Pearson Chi-square 2.84 3.20 15.04 c 15.34 c

H o w O f t e n L a s t Y e a r D i d Y o u P a r t i c i p a t ei n C o l l e c t i v e A c t i o n , f o r E x a m p l e A t t e n d e da C o m m u n i t y M e e t i n g , e t c . ? [ C o m p a r e d t o ]H o w O f t e n D i d Y o u P a r t i c i p a t e i nC o l l e c t i v e A c t i o n [ i n B a s e Y e a r ] ?

T a b l e M . 1 1


Dummy for social fund–assisted community 0.107 -0.144 -0.076 0.005Dummy for community #1 0.343 -0.260 -0.404c -0.254Dummy for community #2 0.385b 0.073 -0.165 -0.471b

Dummy for community #3 0.166 -0.341a -0.625c


Dummy for a SFAC’s woman -0.005 0.090 -0.209 0.256a

Dummy for SFAC’s bottom quartile respondent -0.183 0.003 -0.314a -0.269a

Economic status -0.059 -0.033 -0.056 -0.102c

Size of the household -0.060 -0.040 -0.039 0.009Number of school-going children 0.161b 0.116a 0.024 -0.043Highest education level achieved in the household 0.048b 0.042b 0.031 0.039b

Dummy for a Christian household -0.236 0.061 0.050 0.534a

Dummy for a household following another religion 0.038 0.286Dummy for a young respondent (age below 25) 0.241 0.230 -0.674b 0.081Dummy for a respondent who never married 0.166 -0.620 -0.189 -0.067Age of the respondent 0.035 0.001 -0.013 -0.011Age square 0.000 0.000 0.000 0.000Dummy for traditional ways of interaction (base year) -0.248a -0.108 -0.273 0.292Dummy for social ways of interaction (base year) 0.513c 0.136 -0.137 -0.149Dummy for communal ways of interaction (base year) 0.035 0.284c 0.071 0.018Dummy for top problem being a public good -0.091 -0.108 -0.268 0.069Number of leaders known within community (base year) 0.135c 0.161c 0.142b 0.185c

Number of leaders known outside community (base year) -0.001 0.007 0.030 -0.035Level of participation in meetings (base year) -0.861c -1.091c -1.089c -0.798c


Dummy for managerial skills (base year) 0.146 0.211a



A n n e x e s

1 1 7


Worse=1 17% 29% 42% 50% 2% 4% 40% 31%

Same=2 37% 33% 51% 37% 61% 60% 46% 47%

Better=3 45% 35% 7% 13% 37% 36% 11% 20%

Don’t know/missing=. 1% 3% 1% 0% 0% 0% 3% 1%

Total 284 294 499 195 252 257 490 487

Pearson Chi-square 14.40 c 14.78 c 1.62 19.70 c

C o m p a r e d t o [ B a s e Y e a r ] , I s t h e L e v e l o fT r u s t a n d C o m m u n i t y C o o p e r a t i o n B e t w e e nP e o p l e f r o m D i f f e r e n t B a c k g r o u n d s a n dE t h n i c G r o u p s i n Y o u r C o m m u n i t y B e t t e ro r W o r s e o r t h e S a m e ?

T a b l e M . 1 2


Dummy for social fund–assisted community 0.602c -0.082 0.167 -0.489c

Dummy for community #1 -0.756c -0.269a 0.097 0.213Dummy for community #2 -1.043c -0.491c -0.202 0.260Dummy for community #3 -1.061c 0.148 -0.258Dummy for community #4 0.030 0.952c

Dummy for a SFAC’s woman -0.168 0.172 -0.087 0.141Dummy for SFAC’s bottom quartile respondent -0.173 0.225a 0.269 0.057Economic status -0.046 -0.034 0.016 0.062a

Size of the household -0.030 0.016 0.029 0.049a

Number of school-going children 0.067 -0.004 0.053 -0.017Highest education level achieved in the household 0.033a 0.046c -0.003 -0.023Dummy for a Christian household 0.367b 0.062 -0.430a 0.192Dummy for a household following another religion -0.030 0.645c

Dummy for a young respondent (age below 25) -0.269 0.095 0.101 -0.107Dummy for a respondent who never married 0.026 -0.092 0.312 -0.054Age of the respondent 0.003 -0.009 0.048a -0.011Age square 0.000 0.000 -0.001b 0.000Dummy for traditional ways of interaction (base year) 0.145 0.046 -0.023 0.785c

Dummy for social ways of interaction (base year) -0.040 0.085 -0.028 0.056Dummy for communal ways of interaction (base year) 0.114 0.132 0.146 0.113Dummy for top problem being a public good -0.262 0.159 -0.498c -0.231b

Number of leaders known within community (base year) 0.037 -0.150c 0.294c 0.000Number of leaders known outside community (base year) -0.043 0.080a 0.037 -0.047Level of participation in meetings (base year) 0.154b -0.095 -0.148a 0.040Dummy for blue collar skills (base year) -0.276c -0.310c

Dummy for managerial skills (base year) 0.131 0.024Number of observations 522 646 496 749Wald Chi square (21/26) 101.86c 71.34c 72.33c 140.77c



1 1 8


Much more difficult=1 12% 7% 2% 3% 4% 3% 3% 4%

More difficult=2 13% 27% 10% 14% 9% 16% 10% 15%

Same=3 22% 16% 26% 17% 37% 24% 41% 46%

More easy=4 34% 23% 50% 51% 44% 56% 27% 24%

Much more easy=5 16% 20% 11% 14% 5% 2% 16% 10%


Total 284 294 499 195 252 257 490 487

Pearson Chi-square 28.22 b 8.21 a 19.99 c 14.85 c

C o m p a r e d t o [ B a s e Y e a r ] , I s I t M o r eD i f f i c u l t o r E a s i e r t o G e t t h e W h o l eC o m m u n i t y t o A g r e e o n a D e c i s i o n ?

T a b l e M . 1 3


Dummy for social fund–assisted community 0.164 0.046 0.020 0.332c

Dummy for community #1 -0.193 -0.045 -0.084 0.230Dummy for community #2 -0.986c 0.352b -0.488c -0.093Dummy for community #3 -0.873c 0.037 0.426b

Dummy for community #4 0.028 -0.964c

Dummy for a SFAC’s woman -0.036 -0.175 -0.022 -0.045Dummy for SFAC’s bottom quartile respondent 0.019 -0.332b 0.037 0.063Economic status -0.007 -0.007 -0.004 -0.052Size of the household -0.018 0.001 -0.007 0.019Number of school-going children 0.131a -0.008 0.063 -0.053Highest education level achieved in the household 0.013 -0.036b -0.014 0.032a

Dummy for a Christian household 0.013 -0.101 -0.377a -0.039Dummy for a household following another religion 0.035 -0.337Dummy for a young respondent (age below 25) -0.151 0.107 0.112 -0.127Dummy for a respondent who never married 0.060 -0.658a 0.146 0.490c

Age of the respondent 0.034 0.005 0.010 -0.011Age square 0.000a 0.000 0.000 0.000Dummy for traditional ways of interaction (base year) 0.114 -0.202 0.076 -0.897c

Dummy for social ways of interaction (base year) 0.013 0.020 0.167 -0.253c

Dummy for communal ways of interaction (base year) 0.324c 0.002 -0.034 -0.547c

Dummy for top problem being a public good -0.474c -0.198a 0.178 0.033Number of leaders known within community (base year) 0.043 0.179c 0.254c 0.042Number of leaders known outside community (base year) -0.035 -0.123c 0.026 0.010Level of participation in meetings (base year) 0.065 0.040 -0.142a 0.093Dummy for blue collar skills (base year) 0.289c 0.359c

Dummy for managerial skills (base year) 0.020 0.225b


R-squared 0.069 0.038 0,055 0.110Note: (i) Annex M, table M.25 for explanations for independent variables used in the multivariate analysis.a. 90% confidence level.b. 95% confidence level.c. 99% confidence level.

A n n e x e s

1 1 9


No=0 66% 77% 41% 52% 77% 67% 78% 82%

Yes=1 25% 16% 50% 46% 18% 24% 7% 6%


Total 284 294 499 195 252 257 490 487

D o e s t h e G o v e r n m e n t R e s p o n d M o r eE f f e c t i v e l y t o Y o u r N e e d s N o w T h a nI t D i d [ i n B a s e Y e a r ] ?

T a b l e M . 1 4


Dummy for social fund–assisted community 0.091a -0.067 -0.156c 0.025

Dummy for community #1 -0.050 -0.227c 0.144c -0.025

Dummy for community #2 -0.051 -0.061 0.120b 0.083a

Dummy for community #3 -0.100a -0.267c 0.025


Dummy for a SFAC’s woman -0.042 0.043 0.019 -0.013

Dummy for SFAC’s bottom quartile respondent -0.034 -0.076 0.091 -0.012

Economic status 0.004 -0.010 0.021 -0.001

Size of the household -0.025a -0.025 -0.009 -0.005

Number of school-going children 0.017 0.021 0.036a 0.009

Highest education level achieved in the household 0.002 0.009 0.004 0.002

Dummy for a Christian household -0.039 0.002 -0.079 -0.009

Dummy for a household following another religion -0.134a -0.166

Dummy for a young respondent (age below 25) 0.234b -0.004 0.219b -0.032

Dummy for a respondent who never married -0.039 -0.206 -0.047

Age of the respondent 0.015 -0.012 0.030c -0.005

Age square 0.000 0.000 0.000c 0.000

Dummy for traditional ways of interaction (base year) -0.025 -0.061 0.047 -0.157c

Dummy for social ways of interaction (base year) -0.007 -0.024 -0.003 -0.033a

Dummy for communal ways of interaction (base year) -0.014 -0.040 -0.115c -0.046b

Dummy for top problem being a public good 0.034 0.007 -0.028 0.000

Number of leaders known within community (base year) -0.025 -0.015 -0.023 0.005

Number of leaders known outside community (base year) 0.043c 0.060c 0.018 0.026c

Level of participation in meetings (base year) 0.010 -0.041 -0.058a -0.008

Dummy for blue collar skills (base year) 0.116b 0.022

Dummy for managerial skills (base year) -0.115c 0.026


Wald Chi square (21/26) 33.47c 69.16c 55.45c 62.42



1 2 0


No=0 62% 64% 64% 50% 71% 56% 63% 66%

Yes=1 29% 21% 30% 46% 26% 35% 23% 20%


Total 284 294 499 195 252 257 490 487

D o Y o u F e e l Y o u r L o c a l L e a d e r s h i p ( V i l l a g eC h i e f , H e a d m a n , W a r d C h a i r p e r s o n , e t c . )R e s p o n d s M o r e E f f e c t i v e l y t o Y o u r N e e d sN o w t h a n [ i n B a s e Y e a r ] ?

T a b l e M . 1 5


Dummy for social fund–assisted community 0.043 -0.032 -0.189c 0.023

Dummy for community #1 -0.266c 0.441c -0.235c 0.269c

Dummy for community #2 -0.183c 0.243c -0.065 0.327c

Dummy for community #3 -0.188c 0.041 0.150b


Dummy for a SFAC’s woman -0.034 -0.003 0.057 -0.026

Dummy for SFAC’s bottom quartile respondent -0.001 -0.155c -0.047 0.072

Economic status -0.033 0.016 0.045c -0.013

Size of the household -0.028a 0.002 -0.002 0.022b

Number of school-going children 0.027 -0.034 0.008 -0.034b

Highest education level achieved in the household 0.020b 0.012 -0.001 0.009

Dummy for a Christian household -0.047 0.046 0.018

Dummy for a household following another religion -0.040 -0.068 -0.190b

Dummy for a young respondent (age below 25) 0.252b -0.034 0.272b

Dummy for a respondent who never married 0.039 -0.170 0.056 0.051

Age of the respondent 0.020a 0.002 0.040c 0.015a

Age square 0.000a 0.000 0.000c 0.000

Dummy for traditional ways of interaction (base year) -0.042 -0.077 -0.070 -0.048

Dummy for social ways of interaction (base year) 0.040 0.054 -0.088a -0.086b

Dummy for communal ways of interaction (base year) 0.025 0.001 -0.129c -0.055

Dummy for top problem being a public good 0.049 -0.173c -0.022 -0.017

Number of leaders known within community (base year) 0.033 0.032b 0.011 0.018

Number of leaders known outside community (base year) 0.045c 0.050c 0.030 -0.025a

Level of participation in meetings (base year) 0.019 0.055a -0.038 -0.014

Dummy for blue collar skills (base year) 0.076a 0.047

Dummy for managerial skills (base year) -0.036 0.019


Wald Chi square (21/26) 56.15c 103.99c 59.83c 129.58


A n n e x e s

1 2 1

Base Jamaica Malawi Nicaragua Zambia

Do you feel there are enough staff at the facility? Heard of SFs 43% 69% 91% 36%All MC 33% 58% 63% 17%

How would you describe the facility’s supply of Heard of SFs 46% 33% 69% 33%supplies? [reasonably well/extremely well] All-MC 26% 21% 48% 11%

How does staff attendance compare to the situation Heard of SFs 34% 69% 73% 44%[in base year]? [more] All-MC 21% 73% 55% 10%

t-statistics 0.06 -0.03 0.10b 0.10b

How does supply availability compare to the Heard of SFs 39% 45% 51% 28%situation [in base year]? [more] All-MC 23% 41% 16% 4%

t-statistics 0.18c 0.06a 0.44c 0.29c

How often is the facility clean and well maintained? Heard of SFs 72% 88% 94% 67%[always/usually]

Do you think the facility needs physical improvements? Heard of SFs 67% 42% 17% 89%

Who makes repairs—fixes windows, furniture and Heard of SFs 42% 18% 9% 40%the roof—when they are needed? [don’t know] All-MC 57% 17% 30% 53%

Who pays for repairs? [don’t know] Heard of SFs 60% 43% 14% 42%[no one] Heard of SFs 9%

All MC 61% 48% 33% 56%

Would you be willing to pay for the improvements? SFAC-Need 88% 85% 71% 90%[yes/maybe] Improvement

When repairs are needed, how long does it take for Heard of SFs 35% 39% 29% 10%them to get done? [quickly] All MC 23% 40% 14% 9%

When repairs are needed, how long does it take for Heard of SFs 41% 4% 9% 13%them to get done? [always long] All MC 47% 3% 18% 8%

Note: (i) All MC: All respondents in the matched communities; (ii) Heard of SFs: All respondents who have heard of social fund within the social fund–assisted communities; (iii) SFAC-Need Improvement: All respondents from the social fund-assisted communities who felt that the facility funded by social fund needed improvement; (iv) t-statistics: The value in this rowindicates the difference in the mean value of the SFAC and the MCs. Notes a–c indicate level of statistical significance using student’s difference-in-means test (t-statistics).a. 90% confidence level.b. 95% confidence level.c. 99% confidence level.

S u s t a i n a b i l i t yT a b l e M . 1 6


No=0 16% 14% 24% 33% 4% 18% 47% 53%

Yes=1 43% 33% 69% 58% 91% 63% 36% 17%


Total 186 222 489 493 245 257 230 487

D o Y o u F e e l T h e r e A r e E n o u g h S t a f fa t t h e F a c i l i t y ?

T a b l e M . 1 7


1 2 2


Dummy for social fund–assisted community -0.092 0.096b 0.172c 0.154b

Dummy for community #1 0.266c -0.235c -0.060 -0.356c

Dummy for community #2 0.074 0.294c -0.127c 0.032

Dummy for community #3 0.373c 0.057 -0.321c

Dummy for community #4 0.121c -0.363c

Dummy for a SFAC’s woman 0.063 0.044 -0.013 0.062

Dummy for SFAC’s bottom quartile respondent -0.035 -0.022 0.056 0.124

Economic status 0.062a 0.020 -0.015a 0.013

Size of the household 0.018 0.002 -0.006 0.019

Number of school-going children 0.012 0.008 0.021 0.005

Highest education level achieved in the household -0.009 -0.014b 0.007 -0.006

Dummy for a Christian household 0.149 -0.060 0.051 -0.087


Dummy for a young respondent (age below 25) 0.086 -0.053 -0.001 0.068

Dummy for a respondent who never married 0.157b 0.094 -0.135b

Age of the respondent 0.013 -0.006 0.000 0.016

Age square 0.000 0.000 0.000 0.000

Dummy for traditional ways of interaction (base year) 0.068 -0.040 0.019 -0.174

Dummy for social ways of interaction (base year) 0.010 -0.023 -0.033 0.003

Dummy for communal ways of interaction (base year) -0.074 -0.073b 0.023 -0.013

Dummy for top problem being a public good 0.142 -0.099c 0.166c -0.183c

Number of leaders known within community (base year) 0.027 -0.012 -0.011 -0.036

Number of leaders known outside community (base year) -0.055b 0.000 0.009 -0.001

Level of participation in meetings (base year) -0.097b 0.011 -0.028 -0.031

Dummy for blue collar skills (base year) -0.006 0.032

Dummy for managerial skills (base year) -0.017 -0.033




A n n e x e s

1 2 3


Very poor=1 1% 4% 10% 14% 1% 1% 15% 18%

Poor=2 4% 10% 26% 36% 5% 25% 31% 35%

Reasonably good=3 30% 18% 23% 18% 52% 45% 30% 11%

Extremely good=4 16% 8% 10% 3% 17% 3% 3% 0%


Total 186 222 489 493 245 257 230 487

H o w W o u l d Y o u D e s c r i b e t h e F a c i l i t y ’ sS u p p l y o f S u p p l i e s ?

T a b l e M . 1 8


Dummy for social fund–assisted community 0.971c 0.371c 1.060c 0.443c

Dummy for community #1 -0.149 0.152 -0.295a -1.125c

Dummy for community #2 0.454 -0.082 -0.551b

Dummy for community #3 -0.693b -0.533c -1.082c

Dummy for community #4 0.263b -0.711c

Dummy for a SFAC’s woman -0.098 0.019 -0.177 0.062

Dummy for SFAC’s bottom quartile respondent -0.463 0.012 0.480b 0.035

Economic status -0.106 -0.013 0.050 0.003

Size of the household 0.148b 0.004 0.014 -0.104c

Number of school-going children -0.201 0.013 0.043 0.164c

Highest education level achieved in the household -0.113c -0.006 0.003 -0.020

Dummy for a Christian household -0.362 -0.165 0.577c 0.242

Dummy for a household following another religion -0.258 -0.024

Dummy for a young respondent (age below 25) 0.331 -0.023 -0.008 0.069

Dummy for a respondent who never married -0.204 -0.161 0.043

Age of the respondent 0.027 -0.001 -0.007 0.040

Age square 0.000 0.000 0.000 0.000

Dummy for traditional ways of interaction (base year) -0.333 -0.377b -0.187 0.517a

Dummy for social ways of interaction (base year) -0.717c 0.098 -0.241a 0.206

Dummy for communal ways of interaction (base year) 0.194 -0.004 -0.036 -0.202

Dummy for top problem being a public good 0.523 -0.151 0.468b -0.153

Number of leaders known within community (base year) -0.068 0.004 0.121a -0.003

Number of leaders known outside community (base year) -0.001 0.063 0.085 -0.081

Level of participation in meetings (base year) -0.012 0.050 0.056 -0.093

Dummy for blue collar skills (base year) -0.161a 0.212

Dummy for managerial skills (base year) -0.196b -0.198





1 2 4


Worse=1 3% 8% 5% 4% 1% 2% 12% 18%

Same=2 22% 11% 15% 14% 18% 23% 27% 40%

Better=3 34% 21% 69% 73% 73% 55% 44% 10%


Total 186 222 489 493 245 257 230 487

Pearson Chi-square 11.95 c 0.91 6.15 b 84.37 c

H o w D o e s S t a f f A t t e n d a n c e C o m p a r et o t h e S i t u a t i o n [ i n B a s e Y e a r ] ?( B e t t e r = 3 , S a m e = 2 a n d W o r s e = 1 )

T a b l e M . 1 9


Dummy for social fund–assisted community 0.571a -0.065 0.475b 0.918c

Dummy for community #1 -0.544 0.210 -0.336a -0.485b

Dummy for community #2 -0.306 -0.399c -0.616c 0.318Dummy for community #3 -1.040c -0.311b -0.498b


Dummy for a SFAC’s woman 0.441a 0.060 -0.241 0.090Dummy for SFAC’s bottom quartile respondent 0.126 -0.234 0.276 -0.003Economic status -0.185b 0.104b -0.059 -0.014Size of the household 0.027 -0.059a -0.153c -0.032Number of school-going children -0.043 0.063 0.251c 0.071Highest education level achieved in the household 0.080b 0.009 0.078c 0.033Dummy for a Christian household -0.032 -0.152 -0.108 0.246Dummy for a household following another religion 0.499 -0.043Dummy for a young respondent (age below 25) -0.006 -0.051 -0.201 0.237Dummy for a respondent who never married -0.009 -0.185 -0.108Age of the respondent 0.047 0.016 -0.047 0.027Age square 0.000 0.000 0.000 0.000Dummy for traditional ways of interaction (base year) 0.525b 0.003 -0.066 -0.578b

Dummy for social ways of interaction (base year) -0.432b -0.090 -0.136 -0.014Dummy for communal ways of interaction (base year) -0.127 -0.023 -0.022 -0.121Dummy for top problem being a public good 0.107 -0.201a 0.344a 0.256a

Number of leaders known within community (base year) -0.133 0.074a 0.056 0.020Number of leaders known outside community (base year) 0.112 0.047 0.181b -0.061Level of participation in meetings (base year) 0.203 0.089 -0.013 0.153Dummy for blue collar skills (base year) 0.158 -0.289b

Dummy for managerial skills (base year) -0.318c 0.238a



A n n e x e s

1 2 5


Worse=1 1% 5% 10% 11% 1% 4% 29% 44%

Same=2 9% 9% 14% 17% 22% 44% 20% 13%

Better=3 39% 23% 45% 41% 51% 16% 28% 4%


Total 186 222 489 493 245 257 230 487

Pearson Chi-square 11.02 c 3.18 68.33 c 77.78 c

H o w D o e s S u p p l y A v a i l a b i l i t y C o m p a r e t ot h e S i t u a t i o n [ i n B a s e Y e a r ] ?

T a b l e M . 2 0


Dummy for social fund–assisted community 1.101c 0.170 1.369c 1.199c

Dummy for community #1 6.963c 0.415c -0.314a -1.142c

Dummy for community #2 -0.699a -0.035 -0.631b

Dummy for community #3 -1.457c -0.358c -0.939c


Dummy for a SFAC’s woman -0.145 0.143 -0.330 -0.343Dummy for SFAC’s bottom quartile respondent -0.364 -0.126 0.292 0.101Economic status -0.188a 0.021 -0.082 0.034

Size of the household -0.043 -0.034 0.009 0.005

Number of school-going children 0.171 -0.056 0.105 -0.016

Highest education level achieved in the household 0.058 -0.017 0.038 -0.046

Dummy for a Christian household 0.384 -0.085 0.176 -0.626a


Dummy for a young respondent (age below 25) 0.227 0.002 -0.007 0.279

Dummy for a respondent who never married -0.312 -0.078 0.286

Age of the respondent 0.038 -0.020 0.007 0.055

Age square 0.000 0.000 0.000 -0.001

Dummy for traditional ways of interaction (base year) 0.063 -0.525b 0.075 -0.123

Dummy for social ways of interaction (base year) 0.015 -0.196a -0.214 -0.027

Dummy for communal ways of interaction (base year) -0.331 0.124 -0.428c -0.234

Dummy for top problem being a public good -0.227 0.094 0.168 0.296a

Number of leaders known within community (base year) 0.098 0.118c 0.015 0.053

Number of leaders known outside community (base year) -0.088 -0.044 0.293c -0.052

Level of participation in meetings (base year) 0.245 0.078 -0.091 0.238b

Dummy for blue collar skills (base year) 0.060 -0.157






1 2 6


No=0 33% 57% 82% 9%

Yes=1 67% 42% 17% 89%

Don’t know/missing=. 0% 1% 2% 2%

Total 186 489 245 230

D o Y o u T h i n k t h e F a c i l i t y N e e d s P h y s i c a lI m p r o v e m e n t s ?

T a b l e M . 2 1


Dummy for community #1 -0.302b -0.303c -0.037 -0.092

Dummy for community #2 -0.223 -0.448c 0.226b 0.018

Dummy for community #3 -0.213a -0.294c -0.055

Dummy for community #4 -0.406c -0.003

Dummy for a SFAC’s woman -0.108 -0.011 0.034 0.015

Economic status 0.000 0.042b 0.011 -0.021c

Size of the household -0.006 0.001 0.047c 0.006

Number of school-going children 0.007 -0.008 -0.043a 0.003

Highest education level achieved in the household -0.005 0.016 -0.013 0.011a

Dummy for a Christian household 0.118 0.004 0.034

Dummy for a household following another religion 0.108 0.003

Dummy for a young respondent (age below 25) 0.205 -0.048 0.064 0.000

Dummy for a respondent who never married 0.004 -0.121 -0.085a

Age of the respondent 0.028 -0.011 0.014 -0.020

Age square 0.000 0.000 0.000 0.000

Dummy for traditional ways of interaction (base year) -0.029 -0.217b -0.120

Dummy for social ways of interaction (base year) -0.071 0.083 -0.035 -0.007

Dummy for communal ways of interaction (base year) 0.120 0.015 0.055 -0.017

Dummy for top problem being a public good 0.029 -0.177c -0.107 0.025

Number of leaders known within community (base year) -0.016 0.069c -0.050b 0.013

Number of leaders known outside community (base year) 0.021 -0.062c 0.063c -0.010

Level of participation in meetings (base year) -0.017 0.008 0.067b 0.005

Dummy for blue collar skills (base year) 0.149c 0.010

Dummy for managerial skills (base year) 0.170c 0.012




A n n e x e s

1 2 7


Worse=1 3% 8% 5% 4% 1% 2% 12% 18%

Always long=1 41% 47% 4% 3% 9% 18% 13% 8%

Usually long=2 7% 2% 13% 15% 14% 26% 16% 13%

Usually quickly=3 17% 12% 16% 17% 26% 16% 15% 13%

Quickly=4 35% 23% 39% 40% 29% 14% 10% 9%


Total 186 222 489 493 245 257 230 487

W h e n R e p a i r s A r e N e e d e d , H o w L o n g D o e sI t T a k e f o r T h e m t o G e t D o n e ?

T a b l e M . 2 2


Dummy for social fund–assisted community 0.104 0.141 0.539c -0.059Dummy for community #1 -0.899c -0.182 -0.022 0.176Dummy for community #2 -0.630c 0.774c 0.429b -0.479Dummy for community #3 -0.505b -0.237a -0.989c


Dummy for a SFAC’s woman 0.028 -0.075 0.122 0.093Dummy for SFAC’s bottom quartile respondent 0.057 -0.232 0.266 -0.071Economic status -0.075 -0.035 -0.039 0.030Size of the household -0.043 -0.058 0.027 -0.069a

Number of school-going children -0.003 0.073 0.199c 0.081Highest education level achieved in the household -0.015 0.016 0.058b 0.044Dummy for a Christian household -0.178 0.083 0.209 -0.110Dummy for a household following another religion -0.232 0.127Dummy for a young respondent (age below 25) 0.217 -0.042 -0.097 0.135Dummy for a respondent who never married -0.104 -0.405 -0.094Age of the respondent 0.033 -0.020 -0.070b -0.015Age square 0.000 0.000 0.001b 0.000Dummy for traditional ways of interaction (base year) -0.320b 0.198 -0.144 0.220Dummy for social ways of interaction (base year) -0.014 0.013 0.012 0.193Dummy for communal ways of interaction (base year) 0.037 -0.096 0.091 -0.100Dummy for top problem being a public good 0.165 -0.046 0.158 -0.409b

Number of leaders known within community (base year) 0.111a 0.003 0.062 -0.034Number of leaders known outside community (base year) -0.019 -0.025 0.050 0.129a

Level of participation in meetings (base year) 0.198b 0.117a 0.104 0.349c

Dummy for blue collar skills (base year) 0.008 0.307a

Dummy for managerial skills (base year) 0.148 -0.531c




1 2 8


Know=0 58% 43% 82% 83% 91% 70% 60% 47%

Do not know=1 42% 57% 18% 17% 9% 30% 40% 53%

Total 186 222 489 493 245 257 230 487

W h o M a k e s R e p a i r s — F i x e s W i n d o w s ,F u r n i t u r e a n d t h e R o o f — W h e n T h e y A r eN e e d e d ? [ D o n ’ t K n o w ]

T a b l e M . 2 3


Dummy for social fund–assisted community -0.388c -0.029 -0.279c -0.031

Dummy for community #1 0.446c 0.256c 0.023 0.130

Dummy for community #2 0.367c -0.052 0.151c 0.071

Dummy for community #3 0.332c -0.001 0.215b


Dummy for a SFAC’s woman 0.092 0.111c 0.001 0.166b

Dummy for SFAC’s bottom quartile respondent 0.086 -0.011 0.198b -0.143

Economic status 0.054a -0.019b 0.017 -0.016

Size of the household -0.003 -0.004 -0.003 -0.006

Number of school-going children -0.021 0.003 -0.001 -0.009

Highest education level achieved in the household 0.008 0.000 -0.010 -0.014

Dummy for a Christian household 0.025 0.003 -0.009 -0.126

Dummy for a household following another religion 0.173 0.120a

Dummy for a young respondent (age below 25) -0.144 0.018 0.019 -0.058

Dummy for a respondent who never married 0.076 -0.016 -0.020

Age of the respondent -0.013 0.007 -0.008 0.001

Age square 0.000 0.000 0.000 0.000

Dummy for traditional ways of interaction (base year) -0.080 0.016 -0.046 0.149

Dummy for social ways of interaction (base year) -0.214c 0.009 -0.088b 0.074

Dummy for communal ways of interaction (base year) -0.229c -0.024 -0.040 0.154c

Dummy for top problem being a public good -0.011 0.028 0.000 -0.062

Number of leaders known within community (base year) -0.058a -0.022c 0.024 -0.082c

Number of leaders known outside community (base year) -0.049a -0.014 -0.041b -0.033

Level of participation in meetings (base year) -0.063 0.008 -0.058b -0.080b

Dummy for blue collar skills (base year) -0.074c -0.131c





A n n e x e s

1 2 9


Know=0 40% 39% 57% 52% 86% 67% 58% 44%

Do not know=1 60% 61% 43% 48% 14% 33% 42% 56%

Total 186 222 489 493 245 257 230 487

W h o P a y s f o r R e p a i r s — F i x e s W i n d o w s ,F u r n i t u r e a n d t h e R o o f — W h e n T h e y A r eN e e d e d ? [ D o n ’ t K n o w ]

T a b l e M . 2 4


Dummy for social fund–assisted community -0.076 -0.150c -0.189c 0.012

Dummy for community #1 0.191a 0.424c -0.149c 0.146

Dummy for community #2 0.176b 0.033 0.049 0.063

Dummy for community #3 0.120 0.148c 0.179a

Dummy for community #4 0.139b -0.197b

Dummy for a SFAC’s woman 0.085 0.169c -0.070 -0.010

Dummy for SFAC’s bottom quartile respondent 0.095 0.017 0.050 -0.168a

Economic status 0.023 0.010 0.019 -0.040c

Size of the household 0.015 -0.014 0.008 0.015

Number of school-going children -0.015 0.020 -0.008 -0.016

Highest education level achieved in the household -0.003 -0.006 -0.008 -0.037c

Dummy for a Christian household 0.076 -0.021 -0.012 0.083

Dummy for a household following another religion 0.155 0.033

Dummy for a young respondent (age below 25) -0.130 0.075 0.001 0.125

Dummy for a respondent who never married 0.059 0.084 0.071

Age of the respondent -0.008 0.012 -0.007 0.029b

Age square 0.000 0.000 0.000 0.000b

Dummy for traditional ways of interaction (base year) 0.125b 0.119a -0.132c 0.161

Dummy for social ways of interaction (base year) -0.170c -0.009 -0.080b 0.079

Dummy for communal ways of interaction (base year) -0.119a -0.005 0.009 0.074

Dummy for top problem being a public good -0.059 0.100b -0.010 -0.077

Number of leaders known within community (base year) -0.067b -0.059c 0.020 -0.084c

Number of leaders known outside community (base year) -0.022 0.043c -0.057c -0.012

Level of participation in meetings (base year) -0.020 0.003 -0.135c -0.140c

Dummy for blue collar skills (base year) -0.081b -0.087a

Dummy for managerial skills (base year) -0.118c -0.050





1 3 0

Independent variable Definition

Base year The base year was the year shortly prior to the start of the subproject. In Jamaica and

Nicaragua, the base year was 1995 across all communities. In Malawi and Zambia,

the base year varied by community depending on the year of subproject approval

Dummy for a household following another religion Rastafarians in Jamaica; Muslims in Malawi

Dummy for traditional ways of interaction (base year) Traditional/cultural festivals, weddings, funerals, birthdays, religious festivals and

services

Dummy for social ways of interaction (base year) Independence/freedom day, sports/games, cultural; outings, water source and market/

grinding mill

Dummy for communal ways of interaction (base year) Local government, political rallies, community meetings, social clubs, women’s groups,

communal works

Number of leaders known within the community (local) Village chief, local elected official, project committee chairperson, PTA chairperson,

and [base year] health advisory committee chairperson

Number of leaders known outside the community MP, SF official, junior official in ruling party, policeman, NGO staff, contractor, and senior

[base year] police official

Dummy for blue collar skills (base year) Carpentry, masonry, brick making, or other skilled labor

Dummy for managerial skills (base year) Managing construction, maintenance/upkeep, bookkeeping/accounting, resolving

disagreements

D e f i n i t i o n o f t h e I n d e p e n d e n t V a r i a b l e sU s e d i n t h e M u l t i v a r i a t e A n a l y s i s

T a b l e M . 2 5

A n n e x e s

1 3 1

Jamaica Malawi Nicaragua ZambiaFemale Male Female Male Female Male Female Male

Have you heard of [the social fund]? 0.623 0.692 0.964 1.000c 0.958 0.985 0.363 0.582c

Who had the most important role in the decisionof what type of project to apply for? [leader] 0.175 0.085b 0.361 0.263b 0.202 0.165 0.188 0.297c

Did you attend the meeting? 0.534 0.586 0.599 0.654 0.319 0.260 0.470 0.528Did you say anything at the meeting? 0.205 0.329a 0.114 0.237c 0.107 0.134 0.193 0.382c

Were you or your household asked to make contributions of time, money, or materials during the construction? 0.333 0.411 0.904 0.902 0.088 0.125 0.866 0.864

Have you become able to engage in blue collar skills or has your level of ability improved? — 0.290 0.429c 0.141 0.281c

Have you become able to engage in managerial skills or has your level of ability improved? — 0.246 0.255 0.141 0.305c

Compared to [base year], is it more difficult or easy to participate in groups and associations of people outside your immediate household? 3.404 3.515 3.540 3.534 3.286 3.406 3.353 3.411

How often last year did you participate in collective action, for example, attended acommunity meeting, etc? [compared to] How often did you participate in collective action [in base year]? 1.968 2.008 2.067 2.057 1.941 1.977 2.115 2.043

Compared to [base year], is the level of trust and community cooperation between people from different backgrounds and ethnic groups in your community better or worse or the same? 2.232 2.357 1.675 1.610 2.319 2.376 1.730 1.664

Compared to [base year], is it more difficult or easier to get the whole community to agree ona decision? 3.228 3.381 3.532 3.626 3.356 3.376 3.365 3.498

Does the government respond more effectively to your needs now than it did [in base year]? 0.234 0.314 0.578 0.518 0.209 0.175 0.062 0.104

Do you feel your local leadership responds more effectively to your needs now than [in base year]? 0.290 0.353 0.302 0.335 0.304 0.234 0.234 0.294

Do you feel there are enough staff at the facility? 0.754 0.712 0.796 0.693c 0.946 0.959 0.485 0.398How does staff attendance compare to the

situation [in base year]? 2.574 2.500 2.743 2.702 2.725 2.822a 2.403 2.384How would you describe the facility’s supply of

supplies? 3.125 3.250 2.539 2.389 3.119 3.160 2.385 2.207How does supply availability compare to the

situation [in base year]? 2.727 2.848 2.555 2.468 2.639 2.714 1.968 1.991Do you think the facility needs physical

improvements? 0.625 0.722 0.407 0.445 0.204 0.141 0.927 0.902Who makes repairs—fixes windows, furniture

and the roof—when they are needed? [don’t know=1; otherwise=0] 0.458 0.389 0.240 0.126c 0.088 0.084 0.518 0.338c

Who pays for repairs? [don’t know=1; otherwise=0] 0.646 0.556 0.517 0.336c 0.114 0.160 0.459 0.393When repairs are needed, how long does it take

for them to get done? 2.469 2.433 3.196 3.306 3.011 2.894 2.395 2.372Number 154 130 252 247 119 130 234 230a. 90% confidence level.b. 95% confidence level.c. 99% confidence level.

D i s a g g r e g a t i o n b y G e n d e r w i t h i n S o c i a lF u n d – a s s i s t e d C o m m u n i t i e s

T a b l e M . 2 6


1 3 2

Jamaica Malawi Nicaragua ZambiaBottom Top 3 Bottom Top 3 Bottom Top 3 Bottom Top 3

Have you heard of [the social fund]? 0.651 0.657 0.949 0.994c 0.948 0.983 0.387 0.517c

Who had the most important role in the decisionof what type of project to apply for? [leader] 0.105 0.146 0.372 0.290a 0.156 0.194 0.224 0.254

Did you attend the meeting? 0.525 0.573 0.592 0.640 0.324 0.272 0.552 0.491Did you say anything at the meeting? 0.150 0.311b 0.179 0.178 0.155 0.107 0.293 0.320Were you or your household asked to make

contributions of time, money, or materials during the construction? 0.339 0.385 0.899 0.905 0.111 0.107 0.909 0.850

Have you become able to engage in blue collar skills or has your level of ability improved? 0.321 0.373 0.132 0.251c

Have you become able to engage in managerial skills or has your level of ability improved? 0.182 0.276b 0.178 0.249a

Compared to [base year], is it more difficult or easy to participate in groups and associations of people outside your immediate household? 3.553 3.413 3.467 3.564 3.494 3.286a 3.455 3.351

How often last year did you participate in collective action, for example, attended a community meeting, etc? [compared to] How often did you participate in collective action [in base year]? 2.000 1.980 2.088 2.052 1.857 2.006c 2.007 2.109b

Compared to [base year], is the level of trust and community cooperation between people fromdifferent backgrounds and ethnic groups in your community better or worse or the same? 2.286 2.291 1.696 1.622 2.416 2.320 1.651 1.716

Compared to [base year], is it more difficult or easier to get the whole community to agree on a decision? 3.349 3.275 3.441 3.631b 3.429 3.339 3.531 3.391

Does the government respond more effectively to your needs now than it did [in base year]? 0.230 0.288 0.520 0.559 0.167 0.201 0.085 0.084

Do you feel your local leadership responds more effectively to your needs now than [in base year]? 0.384 0.293 0.219 0.355c 0.151 0.318c 0.308 0.247

Do you feel there are enough staff at the facility? 0.679 0.753 0.737 0.747 0.986 0.939 0.420 0.433How does staff attendance compare to the

situation [in base year]? 2.733 2.463b 2.625 2.757b 2.812 2.759 2.434 2.374How would you describe the facility’s supply of

supplies? 3.111 3.217 2.551 2.430 3.226 3.107 2.340 2.244How does supply availability compare to the

situation [in base year]? 2.808 2.781 2.483 2.520 2.712 2.667 2.085 1.946Do you think the facility needs physical

improvements? 0.679 0.669 0.352 0.453b 0.153 0.178 0.931 0.904Who makes repairs—fixes windows, furniture

and the roof—when they are needed? [don’t know=1; otherwise=0] 0.411 0.431 0.208 0.173 0.137 0.064a 0.328 0.430

Who pays for repairs? [don’t know=1; otherwise=0] 0.643 0.585 0.431 0.423 0.123 0.145 0.362 0.436When repairs are needed, how long does it take

for them to get done? 2.571 2.400 3.148 3.284 3.089 2.891 2.241 2.421Number 86 198 137 362 77 175 152 338a. 90% confidence level.b. 95% confidence level.c. 99% confidence level.

D i s a g g r e g a t i o n b y B o t t o m Q u a r t i l e w i t h i nS o c i a l F u n d – a s s i s t e d C o m m u n i t i e s

T a b l e M . 2 7

1 3 3

Albania: Clear Division of Roles andResponsibilities, Key to a SuccessfulPartnershipUnder a series of four Bank-supported projectssince 1993, the Albanian Development Fund(ADF) has developed a partnership with localgovernments to rehabilitate small-scale infra-structure, assigning roles and responsibilitiesaccording to each partner’s comparative advan-tage and capacity. At the national level, the ADFnegotiates and communicates with the govern-ment of Albania and the donors, coordinates theoverall program, and provides technical, finan-cial, and logistical support to local governments.At the local level, the local governments are inthe driver’s seat. With technical support from theADF, and in consultation with local communi-ties, they make the final decisions on subprojectswithin the financial envelope provided by theADF. Once a decision has been made, the ADFassists the local government in contracting outthe works through public tenders.

The partnership is working well because theADF has devolved to local governments thefunctions they perform best. In addition, the ADFhas built up their capacity to do so—throughstudy tours, short courses, and learning bydoing. The partners, in turn, have also maderoom for the private sector where appropriate.The system is not working perfectly—local com-munities are not always adequately consultedduring decisionmaking, and the infrastructureis not always sufficiently maintained—but theADF has succeeded in building effective part-nerships based on a shared vision among thepartners.

Bolivia: From Social Fund to Co-financingMechanismThe government of Bolivia introduced the Emer-gency Social Fund (ESF) in 1986. The ESF oper-ated between 1986 and 1991 with the primaryobjective of generating short-term employmentand alleviating the social costs of a macroeco-nomic adjustment program. By 1991, the needfor alleviating the social costs of adjustment waslessening. ESF changed its name and mandateto “Social Investment Fund” (SIF) and to invest-ments in social sector infrastructure projects. In1994, with the passage of the government’s Pop-ular Participation Law in Bolivia, the munici-palities assumed a much more important role inpublic investment vis-à-vis the centralgovernment. While the requirements of thePopular Participation Law created some chal-lenges for SIF, it adapted its role and beganworking more closely with municipalities. Allrequests to SIF had to be submitted through themunicipalities.

In the last half of 2000, further modificationsare being made to SIF in order to integrate bet-ter within a decentralized system of social serv-ice delivery. The SIF will be replaced by a newco-financing mechanism. Under the new mech-anism, municipal governments will continue tohave the main responsibility for investing. Cen-tral ministries are to have a normative role andwill be involved in quality assurance, but not inexecuting investments. The new mechanism, tobe implemented through an agency in the Min-istry of the Presidency, will become a vehicle tobring consistency between central and local pri-orities for poverty reduction.

ANNEX N: BEST PRACTICE EXAMPLES/ELEMENTS OF SOCIAL FUNDPROJECTS

Cambodia: Meeting Post-ConflictChallengesCambodia’s social fund faced the question ofwhether to focus on short-term results throughthe rapid delivery of infrastructure or to moveat a slower pace and emphasize long-term resultssuch as capacity-building, sustainability, andcommunity development. Although it wasacknowledged that addressing these problemsrapidly could make achieving sustainability dif-ficult, the decision was made to focus on therapid reconstruction of infrastructure because ofthe urgent needs of the population for basicsocial and economic infrastructure. Nearly con-tinuous civil war since 1970 had played havocon the infrastructure, requiring urgent action.Roads and bridges were destroyed throughoutCambodia, making access to many parts of thecountry impossible. Health centers, schools, andwater facilities were damaged and often aban-doned. The Cambodia Social Fund deliveredsignificant infrastructure on the ground quickly.

Egypt: Multi-Donor Partnership and CoordinationThe Egypt Social Fund for Development (SFD),one of the largest in the world, has had as manyas 17 donors, providing widely differing mixesof loans, grants, technical assistance, and train-ing with very different objectives, requirementsand procedures. Donor contributions haveamounted to more than $1.5 billion since thissocial fund was established in 1991. Over its nineyears of existence, the agency and its donorshave devised various practices to improve donorsupport and coordination: • High-level donor meetings with government

to help resolve key issues to get initial “lift-off” of the agency.

• Division of responsibilities among donors forproviding advice and for conducting super-vision on the basis of their comparative advan-tages and interests.

• A Cairo Donors Advisory Group dedicated tothe SFD, meeting monthly with a rotating sixmonthly chairmanship. This has provided aforum for catching policy and implementationissues early, and has facilitated efforts to com-bine/coordinate supervision, standardize

donor reporting and reduce the administra-tive cost to the SFD of so many donors.

• Comprehensive joint-donor review of theSFD, resulting in development of an ActionPlan by SFD, approved by its Board and bythe Prime Minister, to address the strategicissues identified.

Eritrea: Coordination with theGovernment’s Investment PlanningProcessEritrea’s social fund, Eritrean Community Reha-bilitation Fund (ECRF), started in 1993, has beencoordinated with the government’s investmentplanning process from the start. Project pro-posals for the coming fiscal year are put togetherby ECRF regional offices in collaboration withregional government staff. A list including thoseproposals presented by communities which meetthe selection criteria detailed in the operationalmanual and have passed desk and fieldappraisals is put together toward the end of thesummer and submitted to the ECRF managementunit.

ECRF’s management unit receives the listsfrom all regional offices, and consolidates themin an Annual Work Program and Budget (AWPB).In November, the AWPB is submitted to theMinister of Finance (MOF) and donors forapproval and financing, as well as to involvedline ministries to seek their approval on theproposed subproject. At that time, the MOF allo-cates to ECRF the whole amount of counterpartfunds (15 percent of the value of the activitiesfor that year), and the necessary recurrent coststo the line ministries.

Ethiopia, Moldova: IncorporatingMechanisms for Sustainable ServiceDeliveryEligibility criteria for the Ethiopia Social Reha-bilitation and Development Fund (ESRDF)included a minimum 10 percent communitycontribution for capital costs in cash or kind witha higher contribution to be encouraged wherefeasible. The project aimed to develop infra-structure to standards commensurate with thecommunity’s technical, managerial, and financialcapacity. Arrangements for cost recovery to


1 3 4

finance maintenance were in place: a clear con-tract in the financing agreement concerningmaintenance responsibilities of the sectorbureaus and communities; a specific mainte-nance schedule and standards; an organizationalstructure at the community level responsible formanaging the assets or services; ensuring avail-ability of spare parts and maintenance servicesand technical training in maintenance of com-munity specialists; and agreements for moni-toring maintenance and screening new proposalsbased on maintenance performance. The watersupply component of ESRDF was coordinatedwith the national programs for the sectors andinvolved Bank sector specialists.

The Moldova Social Fund (MSIF) project hasmade a strong effort to focus on sustainability ofsubproject investments. With every subproject,a Beneficiary Association is created in the formof a parent-teacher association, water users asso-ciation, or any other form chosen by the com-munity. Community participation is required atall stages, including a 15 percent contribution forinvestment costs. MSIF staff conduct social mon-itoring of subprojects through follow-up visits forone year after completion of subprojects and pro-vide any necessary advice. Recently, MSIF, as partof a two year performance contract, has startedrewarding Beneficiary Associations that suc-cessfully raise additional private funds with amatching grant of up to $500 a year. These fundscan be used to improve operations through sup-plemental (not routine operations and mainte-nance) activities, for example, buying books forthe library or educational trips for students.

Malawi: Effective Promotion andOutreachMalawi Social Action Fund (MASAF) promotionactivities, tailored for specific groups, are con-ducted through various channels, includingworkshops, radio, TV, posters, flipcharts, com-munity drama, and press conferences, as well asMASAF News, handbooks, and manuals. Thereis a focus on targeting under-served areas,women, and vulnerable groups. In a new ini-tiative, MASAF is collaborating with the MalawiInstitute of Education to develop low-cost andculturally sensitive information materials. The

success of these efforts is evident from thehousehold surveys conducted by OED: inMalawi, 98 percent of the households had heardof the social fund, compared to only 39 percentin neighboring Zambia where promotion effortshave been less intensive. A major promotioncampaign can also bring challenges: it may gen-erate demand and expectations that may be dif-ficult to meet. MASAF was aware of this tradeoff.It chose a policy of open information dissemi-nation and the payoffs came in terms of offset-ting rent-seeking behavior. Going beyondpromotion, MASAF is recently supporting shar-ing of subproject experience and lesson learn-ing between communities, and between serviceproviders and communities.

Moldova: Training “Prime Movers”In the two years since the Social Investment Fundwas established (one year since credit wasdeclared effective), community meetings to selectpriority micro-projects have been held in about650 villages. About 300 mayors and communityleaders received training for improving theircommunity organization and facilitation skills andmore than 50 community-based organizations(CBOs) received training in community strate-gic planning. Technical assistance has been pro-vided to about 400 villages and more than 200implementing agencies received training inmicro-project preparation and implementation,including fund raising.

Nicaragua: Broad-based Local PlanningSince 1998, the micro-planning process (MPP)has become the basis for subproject selection in60 out of the 147 municipalities in Nicaragua.Under MPP, municipalities organize workshopsin communities, sometimes with the help ofNGOs, in which the population can express itsdemand for subprojects and prioritize them. Onthe basis of these priorities, the municipalitiesmake a Municipal Investment Plan, and differ-ent groups including FISE finance different partsof this plan.

Peru: Achieving Success in TargetingThe Peruvian Social Fund, FONCODES, changedits targeting mechanism in 1996. Previously, a

A n n e x e s

1 3 5

province-level poverty map had been used withself-targeting through a positive list of subpro-jects. In 1996, a district-level map was intro-duced, combined with more disaggregatedpoverty information. Active promotion and out-reach were begun to help poorer communitiesprepare proposals. Some communities were alsoassisted by being allowed to submit a “pre-screening” proposal. The low wages offeredhelped self-target the employment programs.In later years, the Peru social fund focused onrural areas. These changes have resulted in largeimprovements in targeting performance overtime.

Thailand: Fostering PartnershipsIn Thailand, the Social Fund Office collaborateswith provincial and regional level members ofcivil society as well as different governmentagencies. Even civil servants have embraced theconcept of participatory community support inplace of top-down determination of commu-nity development. This concept encourages col-laboration between communities and civilservants because the communities are capableof determining needs, while the funds, the tech-nical expertise, and the necessary supportingmanpower remain in the hands of civil servants.

The Social Fund Office has managed to mobi-lize thousands of volunteers all over the coun-try, including NGO members, civil servants,academics, physicians, teachers, and religiousleaders, to help prepare, review, screen, andimplement projects. All of them put in longhours without receiving compensation for theirinvaluable support. The SIF is widely recog-nized in Thailand as a major element of thegovernment’s crisis alleviation strategy and apathway toward government collaboration withCSOs.

Yemen, Zambia: A Shift Toward“Software”: Changing Social FundActivities?The Second Yemen social fund will attemptthe transition from a focus on infrastructure pro-vision to a capacity building agency. It willadjust its project menu to include more “soft-ware” type interventions, such as informal edu-cation and training to help particularlydisadvantaged groups like the disabled or par-ticularly poor and women and children in diffi-cult circumstances. The Yemen Social Fund hasalso decided to reorganize its capacity buildingprogram to go beyond beneficiary communitiesand to target the lowest level of formal local gov-ernment as this level will have added develop-ment responsibilities under Yemen’s newdecentralization law.

In the Zambia Social Recovery Project II,findings of the Beneficiary Assessment prompteda significantly enhanced emphasis on trainingcompared to the first project, including a num-ber of initiatives to improve the content anddelivery of training and information, as well asexpansion of these programs. Under the recentZAMSIF, training and capacity building are fur-ther emphasized. Resources for capacity build-ing and skills training will be provided on ademand basis to community members and dis-trict officers. Training for community memberswill cover areas such as business and entrepre-neurship development, traditional birth attendantskills, and community health worker training.Training for district officers will cover areas suchas planning, financial management, design andcosting for projects, and training of trainers forcommunity capacity building. A capacity build-ing manager and two support staff will berecruited by ZAMSIF to organize and coordinatecapacity building under this component.


1 3 6

1 3 7

About 32 percent (21 of 66 projects) of theBank’s social fund projects mentioned commu-nity participation in the discussion of objec-tives.1 Participation has been expected tocontribute to: (a) the highest priority problemsbeing addressed; (b) effective subproject exe-cution (including subproject formulation, sub-mission, implementation); (c) communitycapacity building/skills development and socialcapital enhancement; and (d) sustainability ofbenefits (discussed in chapter 3). This Annexpresents the results of OED’s community-levelfield research comprising household surveysand qualitative focus group and key informantinterviews in four countries supplemented byOED’s portfolio assessment, literature review, andinstitutional analysis based on (qualitative) fieldresearch in an additional country (Argentina).The household surveys were conducted in 3 to5 randomly chosen social fund–assisted com-munities and 3 to 5 matched non-social fundcommunities in each of the four field researchcountries, Jamaica, Malawi, Nicaragua, and Zam-bia, and was administered on approximately 50randomly chosen households in each commu-nity for a total of 1,687 randomly-selected house-holds (845 social fund–assisted households and842 households in the matched communities).In the former 1,525 respondents were inter-viewed (roughly 2 per household, householdhead and where available another householdadult of opposite sex). This amounted to 284respondents in 4 communities in Jamaica, 499respondents in 5 communities in Malawi, 252respondents in 3 communities in Nicaragua, and490 respondents in 5 communities in Zambia).Among non-social fund households, 1,531respondents were interviewed (294 in Jamaica,493 in Malawi, 257 in Nicaragua, 487 in Zambia).

The total number of respondents, thus, amountedto 3,056. All results are reported by respondentunless indicated otherwise. The country-levelpercentages refer to responses from the sampledhouseholds in each of those countries rather thanto the population as a whole.

Subproject Identification Process

Project AssumptionsThe community model of social fund projects—even when other entities as well as communitiesare eligible to submit proposals—relies on somekey assumptions which are implicit in socialfund beneficiary assessments and a Bank reviewof them:2

• “Community” is a meaningful construct.• Community members are aware of the oppor-

tunities offered.• The terms of the offer are conveyed in an

unbiased way.• Community members understand the costs,

benefits, and obligations of the offeredoptions and their own potential roles andresponsibilities.

• The community makes an informed choicewith different groups in the community hav-ing a say.

The Practice The concept of community as envisaged in thesocial funds approach fits best in rural areaswhich are relatively homogenous and have higherdegrees of social cohesion. In urban areas, thecommunity may well be determined by thenature of the subproject rather than vice versa.The Zambian social fund has adopted a pragmaticand flexible approach to identifying “communi-ties,” financing subprojects such as water supply

ANNEX O: COMMUNITY PARTICIPATION AND SOCIAL CAPITAL

at a boys’ reformatory for which teachers and stu-dents constituted the “community.”

In terms of the participatory process, manyrural communities in Malawi and Zambia fit a dif-ferent pattern than implied by the social fundsmodel. In Zambia, where the majority of sub-projects are schools, a school headmaster usu-ally initiates the application. Occasionally ateacher, PTA member, or village leader takes theinitiative, although the headmaster tends to playan important role thereafter. The situation is dif-ferent in Malawi where an intensive outreachcampaign has reached both leaders and othermembers of the community. Even so, and unsur-prisingly, community leaders dominate the for-mulation of subprojects.3

The next step is to discuss with the wholePTA, which brings together important local fig-ures (such as larger farmers and retirees4) with“ordinary people.” The PTA will then seek theagreement of the village leaders.5 The villageleaders in turn seek the backing of the chief andthen call the community together. At this meet-ing, which may be addressed by the chief, thecommunity is told of the plan to apply for helprehabilitating the school,6 and that they need toprovide labor. All adults are expected to con-tribute and the leaders keep a register. Fines arealso imposed on those who do not contribute,usually additional workdays, though the finemay be money or livestock (a chicken).

The application can only be made to thesocial fund once a substantial part of the com-munity contribution has been made. So, thecommunity meeting with social fund staff orlocal officials may take place once consider-able work has already been done—there is lit-tle room at that stage for dissension even if thedynamic of a public meeting permits it. In fact,the focus group discussions seem to indicate thatmany would have preferred something differentthan that generated by the process (even thoughmost were satisfied with what was finallychosen).

The project committee is chosen throughpublic meetings, though evidence varies as tohow representative the process is. The leadersand the project committee members (key inform-ants) both say the committee was elected. How-

ever, the focus group participants did not men-tion taking part in an election.

The process has been working somewhatdifferently in Jamaica and Nicaragua althoughthere are many similarities as well. The differ-ences derive both from differences in socialstructures and from varying project designs. Forexample, in Malawi and Zambia communities arethe only entities eligible to sponsor subprojectswhile in Jamaica and Nicaragua other entities(government and NGOs) are also eligible. Thesimilarities lie in the role of local leadership inthe development process; in both Jamaica andNicaragua, mayors, engineers, and religiousleaders have typically acted as intermediaries forgetting social fund subprojects to communities.

In Jamaica, JSIF’s first in-person contact withthe community is usually at the field appraisalstage—after the subproject has already beenidentified and the application has been sent infrom the community. When required, JSIF holdscommunity meetings to agree on the terms ofthe implementation or provide training to com-munity leaders to strengthen the sponsoringagency’s capacity to manage the project. A sub-project steering committee, comprising a con-tractor and a community representative, followsup on subproject implementation. The com-mittee may also include other members of thecommunity as well as local representatives ofthe line ministry. Conflicts between the com-mittee and the community typically arise onlyin respect to employment of key personnel,especially contractors.

In Nicaragua, until 1998, subproject selec-tion did not follow a systematic pattern: FISEengineers traveled around with the FISE sub-project menu and followed the suggestions oflocal representatives or communities. In 1998,FISE aimed to give organized communities andmunicipalities a greater role in the selection ofsubprojects by supporting a participatory plan-ning process at the municipal level (MPP) in 60municipalities. The MPP can be expected tobring in the voice of a larger part of the com-munity into the local investment prioritizationprocess (“Nicaragua Broad-Based Local Plan-ning:” Annex N). While the MPP, as designed,has yet to be fully implemented in many com-


1 3 8

munities, early experience shows that it has ledto the expression of more varied communitydemand, including a higher demand for waterthan in past years.7

The above decisionmaking process based onOED’s institutional analysis conducted duringfield visits was broadly confirmed by OED’squantitative and qualitative survey data (boxO.1.).

The nature of participation in social fundprojects was also illuminated by the results ofthe OED Stakeholder Survey according to which

only 36 percent of the respondents thought thatthe choice of subprojects was made with theactive participation of diverse groups in thecommunity. Twenty-six percent thought thatthe choice was usually made with active par-ticipation of the community but that some groupshad little voice. Eleven percent thought that thechoice was inappropriately influenced by NGOs,private contractors, and community leaders.Among World Bank respondents, none thoughtthat the choice of subprojects was usually madewith the active participation of diverse groups

A n n e x e s

1 3 9

Role of “Prime Movers” in Subproject Selection Process. Accord-ing to the OED household survey data, a large percentage of therespondents who had heard of social funds noted that localleadersa (“prime movers”) played a major role in the subprojectselection (53 percent in Jamaica, 78 percent in Malawi, 71 per-cent in Nicaragua, and 32 percent in Zambia).b This role of theleader extended to encouraging the community to attend the sub-project selection meeting: of the respondents who knew of themeeting for subproject selection, the percentage of respon-dents encouraged by local leaders to attend the meeting was 39percent in Jamaica, 53 percent in Malawi, 36 percent inNicaragua, and 60 percent in Zambia. Many quotes from thefocus groups support this view, for example: “the chief calleda meeting with the village headmen and sent them to tell the peo-ple that there was a project to be done that involved communityparticipation”; “after hearing from the politicians that MASAFhas come in the country, the traditional authority called all theheadmen and group village headmen to decide what type ofprojects they could apply for”; and “we were not consulted onwhat project we wanted, the headman just imposed [it].” Thisview was also expressed in Nicaragua: “the teacher and mayorhave decided, but everyone supported the project” and “themayors are the ones who are really involved in the applicationprocess.” One of the subprojects was for assistance to a schoolrun by a religious group, and the qualitative data clearly showthat many community members felt that the process had notbeen participatory. Indeed, in Nicaragua, more respondents (10percent) felt they had had a larger role in another project in thesame community than they had had in the social fund subpro-

ject (6 percent). The OED household survey did not find evi-dence of contractors leading the subproject identificationprocess, but local political leaders such as mayors were foundto be important.

Role of “Prime Movers” in Deciding Community Contributionfor Construction: In Malawi and Zambia, the social fund sub-project requires up-front community contributions. This is not thecase in Jamaica and Nicaragua, where the contributions are tobe made during construction. The requirement in Malawi andZambia was taken seriously; the qualitative data in Malawiconfirms that the lack of up-front community contribution wasone reason why the matched community’s application for socialfund financing was turned down. “Prime movers” in Malawi andZambia assume an important role in mobilizing and enforcingcommunity contributions. Seventy-two percent of the respondentswho had heard of social funds in Malawi and 44 percent inZambia indicated that household contributions were decided bylocal leaders.

Role of Local Leaders (Headmen, Village Chief) in EnforcingCommunity Contributions: In addition, local leaders played arole in ensuring that the required contributions are actuallymade. Evidence is found of enforcement mechanisms from thequalitative data: “the village headmen fined individuals for lackof participation by asking them to pay a chicken” and “if some-one did not participate, the chief forced him to pay money.” TheOED household survey data support this view: 70 percent of theMalawian and 35 percent of the Zambian respondents who hadheard of social funds replied that there were some penalties fornon-contribution.

O E D Q u a n t i t a t i v e a n d Q u a l i t a t i v e S u r v e yR e s u l t s

B o x O . 1

a. Local leaders include, for example, village chiefs, headmen, community leaders, religious leaders, teachers, local elected officials, PTA members, or communityhealth workers.b. These figures use “respondents who have heard of the social fund” as the base. This is because respondents were not asked about the subproject selection processif they had not heard of the social fund.Source: OED quantitative and qualitative surveys.

in the community, and 42 percent thought thatchoice was usually made with the active par-ticipation of the community but that some groupshad little voice.

Participation and Addressing the HighestPriority ProblemsOED field research found that subproject selec-tion has been less a unified expression of com-munity will than a process in which “primemovers” usually determine project choice. Evenwhere virtually the whole community partici-pated in some aspect of the subproject, this didnot necessarily mean that the community as awhole drove project choice. While the vastmajority of beneficiaries have been satisfiedwith the subprojects financed,8 the OED house-hold surveys of 845 randomly selected socialfund-assisted households in four countries foundthat the subproject selection process could notbe counted on necessarily to meet the highestpriority problem of the majority of households,even if the relevant investments were on the sub-project menu. The OED household survey datafor 284 respondents in Jamaica, 499 in Malawi,252 in Nicaragua, and 490 in Zambia (for a totalof 1,525 respondents from 845 households)showed that (Annex M, table M.1):• The top priority was met for 27 percent of the

respondents in Jamaica, 34 percent in Malawi,23 percent in Nicaragua, and 22 percent inZambia.

• Even if only problems that were on the sub-project menu are considered the top prioritywas met for 31 percent of the respondents inJamaica, 47 percent in Malawi, 26 percent inNicaragua, and 28 percent in Zambia.

• One of the top three priorities was met for 42percent of the respondents in Jamaica, 52percent in Malawi, 35 percent in Nicaragua,and 38 percent in Zambia.

• OED household survey results by communityshowed that considering only the problemsthat could be addressed by an investment onthe subproject menu, in 9 of 17 communitiesthe top priority problem was addressed. In atleast 2 of these 9 communities, however, focusgroup interviews (which allowed for a morespecific definition of community priorities,

e.g., that distance to the nearest school wasa priority problem rather than improving theexisting building, or that a “technical school”was a priority not just simply a “school”) indi-cated that the top priority was not met. In 4of the remaining 8 communities the secondranked problem was addressed.

In order to avoid a bias in the householdresponse, the OED survey did not directly askbeneficiaries if the selected subproject was theirpriority (often the case in existing studies).Instead, the OED survey asked the question “inyour personal opinion what were the threebiggest problems facing your community (inthe year prior to the approval of the subproject).”This question was open ended and was askedimmediately following questions related to basicsocio-economic and demographic issues, wellbefore the community was asked about the par-ticular subproject financed by the social fund.Then, OED compared the household responsewith the subproject actually financed by thesocial fund in that community. This analysisprovided an indication of the extent to which thebiggest community problems were (or werenot) addressed by the social fund.

Several factors were found to influence thedemand-driven process, including the role of“prime movers,” who were critical to the mobi-lization of support and preparation of a suc-cessful subproject proposal and whose interestswere determined by the nature of their position.9

For example, when “prime movers” existed inthe form of headmasters or health workers, a biaswas found toward subproject investments inschools and health facilities. Typically, socialfund staff visited the community once they hadbeen informed of the community’s interest in aparticular type of subproject or had received anapplication. By this time, “prime movers” hadalready mobilized support for particular sub-projects. However, OED found that the vastmajority of the respondents still do express sat-isfaction with the chosen subproject indicatingthat the subproject was a community priority (butnot necessarily their highest priority). Overall,these findings are broadly consistent with theBank’s review of social fund Beneficiary Assess-


1 4 0

ments which noted: “Beneficiaries usually felt thatprojects reflected their needs and priorities evenwhen they had not been involved in identifica-tion…in Peru where 96 percent of respondentssaid the project was a community priority, [only]66 percent of the beneficiaries said the projectwas prioritized by the community itself…Onthe whole, the insertion of intermediaries, be theylocal governments, NGOs, promoters, or pri-vate contractors, was not found to have adverselyaffected the relevance of the project to the com-munity. This may be due, in part, to the basictypes of investments eligible for financing bysocial funds and the large range of unmet needsexisting in poor communities.”10 Satisfactionwith what was received may also result from lackof knowledge about what else might have beenavailable or because the required communitycontribution for that investment was low.

The Social Funds 2000 Impact Evaluationaddressed the issue of community priorities andpreferences primarily through qualitative bene-ficiary assessments asking about the importanceor priority of the subproject in reference to thealready-financed social fund subproject. Theoverall result was that communities consideredthe investments they had received to be meet-ing their most important needs (Annex E, tableE.4). When OED applied a similar methodology,the results were also positive, for example, whenasked, in reference to the actually chosen socialfund subproject “would you have preferred thatanother project was chosen instead,” a vastmajority of respondents in the four countriesanswered “no.” This response confirms the OEDconclusion that a vast majority of beneficiariesare satisfied with the financed subprojects. It isnot inconsistent with the results of the OEDhousehold surveys on the extent to which thebiggest community problems are met in thatsuch a response may have been influenced bybeneficiary perceptions of what their optionswere and what they assumed they could realis-tically have got. In its general conclusion, SocialFunds 2000 signals the need to ensure greaterrelevance to community needs and to movetoward more participatory local planning inorder to identify investment priorities within thebroader goals and activities of a community.

Knowledge of Social Fund Project: Whywere the community’s highest priority problemsnot addressed? Was it due to lack of knowledgeabout the social fund project on the part of house-holds or was it the nature of participation withinthese communities? The knowledge of the exis-tence of the social fund project in the social fund-assisted communities was low among surveyrespondents in Jamaica (65 percent had heardof JSIF) and Zambia (47 percent had heard ofMPU). Furthermore, only 15 percent of all therespondents in Jamaica and 37 percent in Zam-bia knew about at least one subproject typethat was eligible for social fund financing.11

While there could be other reasons, it is not sur-prising that the highest priority problems ofeach sampled community in the two countrieswere not addressed as the respondents had lit-tle knowledge about the social fund’s existence,or the nature of the offer.

In Malawi and Nicaragua, nearly all respon-dents in the social fund-assisted communities hadheard of the social fund (98 percent in Malawiand 97 percent in Nicaragua). The percentageof respondents who knew of at least one sub-project eligible for social fund financing was 90in Malawi and 71 in Nicaragua. The figures forknowledge about the requirement of communitycontributions were 50 percent and 23 percent,respectively. In these two countries, a large per-centage of respondents had knowledge of thesocial fund project, still the highest priority prob-lems of each community were not addressed.Thus, knowledge about the social fund does notnecessarily guarantee that the highest priorityproblems will be addressed—knowledge is anecessary but not sufficient condition. The natureof participation within the community (e.g.,level of input into the subproject selection meet-ing) also plays a role.

Nature of Participation in Selection Process.According to the OED household survey, inMalawi, of those who had heard about the socialfund project (MASAF), 81 percent knew aboutthe subproject selection meeting. Of theserespondents, 73 percent attended the meeting.Of those attending the meeting, only 28 percentactually spoke at the meeting.12 In Nicaragua, ofthose who had heard about the social fund proj-

A n n e x e s

1 4 1

ect (FISE), 62 percent knew about the subpro-ject selection meeting. Forty-five percent of theserespondents, actually attended the meeting.13

Forty-two percent of those who attended themeeting, spoke at the meeting. These numbersappear to over-estimate participation: for eachconsecutive item, the base gets reduced. If thebase is all respondents, then only 16 percent inMalawi and 12 percent in Nicaragua spoke at themeeting; this low proportion of respondentswho expressed their views at the meeting fitswith the picture of the participatory processdescribed above.14 This view is also supportedby qualitative data from focus groups and open-ended questions in the questionnaire, for whichrespondents indicated the subproject choicewas announced at the meeting.

Other aspects of participation also inhibitmeeting the highest priority problems of thepoor. For example, the MASAF II ProjectAppraisal Document notes “The criteria forupfront community contribution has met withproblems in those parts of the country where thenatural environment does not allow those peo-ple to produce fire-baked bricks or acquirematerials such as river sand and stones for con-struction purposes. In such places, it has becomedifficult for these communities to apply for theirfirst choice projects that require substantial con-tribution in local materials, instead they haveended-up with boreholes which do not requiremuch material contribution.” The Zambia socialfund faced a similar situation where counterpartcontribution requirements meant that commu-nities chose subprojects with a high unskilledlabor component than others for which theywould have to make a larger cash contribution.Another factor influencing community choice isthat communities request what they think theyare likely to get rather than what they truly wantand are willing to pay for—in many countries,social funds are known primarily as builders ofschools, so that a large proportion of commu-nity requests they receive are for schools.

Participation and Subproject ExecutionThe successful application for, and implemen-tation of, a social fund subproject requires threethings:

• The drive and initiative (usually in an indi-vidual or small group) to initiate the process(subproject selection, mobilizing the com-munity for community contribution, andenforcing the contribution)—and to carry itthrough to subproject completion.

• Sufficient community support to actually real-ize the community contribution before orduring construction.

• Community capacity to implement the sub-project, which means specific skills of mobi-lization, literacy, bookkeeping, and the abilityto interact with officials.

The existence of a PTA or other existing organ-ization to serve as an initial focal point helps, butit is usually a headmaster who meets the secondand third conditions. Respondents are well awareof this point. While the qualitative data fromsocial fund-assisted communities points to the roleof leaders, respondents from matched commu-nities lamented how lack of local leadershipimpeded making a social fund application: “wedon’t have any strong leader, that’s why we havea lot of problems, such as lack of assistance”(Jamaica), “the mayor here is the only problem”(Nicaragua), and “no leadership” (Malawi).

Four conclusions can be drawn:• Community participation in social fund sub-

projects is sufficient to see subprojects throughto completion. To the extent that the socialfund process uses existing community struc-tures, it is following good practice.15

• Depending on the particular social structureof the community, “prime movers” may ormay not represent the interests of the major-ity of community members. This puts a strongonus on the social fund to ensure dissemi-nation of relevant information and, wherenecessary, to undertake a longer-term andmore intensive participatory process.

• The role of facilitation is stressed. A similarpercentage of respondents in social fund-assisted communities and matched commu-nities had heard of the social fund from themedia (around one-quarter in Jamaica, justunder half in Malawi, 31 percent in Nicaragua,and around 4 percent in Zambia). However,a larger percent of respondents in social


1 4 2

fund–assisted communities compared tomatched communities had heard of the socialfund from local leaders, which seemed to beinstrumental in their getting the social fundsubproject (Annex M, table M.2).16 So, inaddition to pursuing promotion “to let themknow about it,” facilitation (through localleaders, for example) will also typically berequired to get communities to successfullyapply.17

• The results also show how facilitation shouldwork: that is, motivating local structures totake control of the process rather than creat-ing parallel ones not rooted in the existingsocial structure. A recent social fund aims atimproving facilitation skills among existinglocal leaders. (“Moldova Training “PrimeMovers”: Annex N).

These structures can be expected to workbetter for some sectors than for others. The con-ditions for successful subproject formulation aremet for schools and hospitals because they aremore likely to have “prime movers” and existingorganization that roads or water supply may not,and may lack the community structures to enforceoperations and maintenance. Water and roadsubprojects may require a different participatoryprocess—proactive information disseminationabout the social fund and its offer and/or exter-nal agents “igniting”—but not “leading”—thesubproject process and nurturing latent “primemover” capacity in local actors. Effective sub-project execution in these sectors will also befacilitated by the formation of community-basedorganizations dedicated to these sectors, forexample, road or water associations. Sufficienttime and resources are crucial for fostering thegrowth of such organizations and it is critical thatthey are rooted in existing social structures.

A number of sources confirmed the nature ofthe participatory process in social fund proj-ects, for example a review of beneficiary assess-ments (Annex K) and OED’s review of ICRs/PARs. The beneficiary assessment review found,for instance, that: “The characterization thatcommunities organize, enter into direct com-munication with the social fund, and preparetheir own projects was not fully borne out in the

beneficiary assessments. Beneficiaries perceiveda more complex web of actors inserted betweenthe social fund and the community in terms ofpromoting and preparing projects” (Owen andVan Domelen 1998). OED’s review of ICRs/PARsfound that the bottom-up process was not alwaysfollowed, and that project selection was notnecessarily based on any clear process and pro-cedure for considering and analyzing options—whether by community members, the socialfund, or other entity. According to a recentreview of social fund operational manuals “par-ticipatory planning is regarded as an essentialingredient in the work of most social funds, butnot all of the operations manuals discuss theexplicit procedures that are to be used to ensurethe participation of the community in the proj-ect cycle. Without explicit directions that explainhow the participatory process is expected towork, the actual participation of communitymembers is left to chance and for the sake ofexpediency, may never take place” (Weissman2001). A study of the design of social fund proj-ects also found weaknesses in terms of sup-porting participatory processes (Narayan andEbbe 1997).

In Jamaica, of those principal respondentswho had heard of the social fund, 37 percentresponded that they were asked to make con-tributions during construction (time, money, rawmaterial).18 These figures were 92 percent, 10 per-cent, and 84 percent for Malawi, Nicaragua andZambia, respectively. Of the households whowere asked to contribute during construction, 56percent in Jamaica, 100 percent in Malawi, 100percent in Nicaragua, and 98 percent in Zambiasaid they actually did contribute.19 The very highpercentages of community members makingcontributions in Malawi and Zambia are a resultof the way in which the participatory process isembedded in existing social structures (e.g., withsocial sanctions imposed by community leaders),and indicate that social funds in these countriescrowd-in local resources.

Participation, Capacity Building, andSocial CapitalThirty-two percent (21 of 66) of social fundprojects in the portfolio included community

A n n e x e s

1 4 3

capacity building among their objectives. Com-munity empowerment was mentioned amongsocial fund objectives in 12 percent of projects.Five percent of projects mentioned increasingsocial capital20 and social cohesion among theirobjectives.21 The emphasis on social capital isintended to be further strengthened in futuresocial fund projects (World Bank 2000f).

Social funds use a variety of mechanisms tosupport capacity building/skills developmentand social capital: project committee membersreceive training and gain experience on thejob;22 other community members may learnnew skills through participation in various sub-

project activities; and the whole communityhas the experience of working together. This lastelement is called building “bonding social cap-ital”: increasing community cohesion andthereby increasing the likelihood of future com-munity-based initiatives. Social funds can alsobuild “bridging social capital,” which is thecommunity’s links with outsiders. Higher bridg-ing social capital opens the channels throughwhich community members can hold local lead-ers accountable and increases access toresources. This section examines these link-ages. Box O.2 summarizes the results from theOED household survey.


1 4 4

Question Jamaica Malawi Nicaragua ZambiaCapacity Building

Have you become able to engage in [carpentry, masonry, MA — —brick making, or other skilled labour] or has your level of ability improved? TS — Negative —Have you become able to engage in [managing construction, MA — —managing maintenance/upkeep of community facilities, or bookkeeping/accounting] or has your level of ability improved? TS — — Negative

Bonding Social CapitalCompared to [base year], is it more difficult or easy to MA Positive Negative Positiveparticipate in groups and associations of people outside C2 Positive Negativeyour immediate household? TS Positive NegativeHow often last year did you participate in collective action, MAfor example attended a community meeting, etc? [compared C2 Negative Positiveto] How often did you participate in collective action TS Negative Positive[in base year]?Compared to [base year], is the level of trust and community MA Positive Negativecooperation between people from different backgrounds and C2 Positive Negative Negativeethnic groups in your community better or worse or the same? TS Positive Negative NegativeCompared to [base year], is it more difficult or easier to get MA Positivethe whole community to agree on a decision? C2 Positive Negative Negative Positive

TS Negative PositiveBridging Social Capital

Does the government respond more effectively to your needs MA Positive Negativenow than it did [in base year]? TS Positive Positive NegativeDo you feel your local leadership responds more effectively MA Negativeto your needs now than [in base year]? TS Positive Negative NegativeSource: This analysis is based on OED’s household surveys of 1, 687 households and 3,056 respondents. Only results that are statistically significant on multivariate analysis and are con-

firmed by the difference-in-difference approach (t-statistics and chi-square test) are reported here. MA: Statistically significant with multivariate analysis; C2: Statistically significant with

Chi-square test and TS: statistically significant using t-statistics (details are provided in Annex M, table M.7). Shaded area show confirmation of results by both multivariate analysis and

difference-in-difference approach (t-statistics and/or chi-square test).

I n d i c a t o r s f o r C a p a c i t y B u i l d i n g , B o n d i n g ,& B r i d g i n g S o c i a l C a p i t a l : M i x e d I m p a c t s

B o x O . 2

Community Capacity Building and SkillsDevelopmentCommunity participation is intended as an impor-tant means by which community capacity isenhanced. The quantitative data from the OEDhousehold surveys showed that in Malawi 36 per-cent and in Zambia 21 percent of all respondentsnoted improvements in blue collar skills (car-pentry, masonry, and/or brickmaking) and 25percent in Malawi and 23 percent in Zambianoted improvements in managerial skills (man-aging construction, bookkeeping, resolving dis-agreements).23 However, these did not representany statistically significant differences comparedwith the non-social fund communities (usingboth multivariate and difference-in-differenceapproaches; Annex M, table M.7).24

The limited capacity enhancing effect indi-cated by the household survey may reflect the factthat not many community members outside of theproject committee participate actively in mostsubproject activities—the project committee ofabout 10 people is only a small part of the com-munity and a community-wide survey may notpick it up. This argument fits with the qualitativefinding of new skills being mentioned in keyinformant interviews, particularly with projectcommittee members: “in the past we did notknow of any accounting but now we can at leastdo it here and there. We received training on lead-

ership, accounting, and this was done [in our vil-lage], only four members were trained” (Malawi).

Social Capital EffectsThe different kinds of possible social capitaleffects are demonstrated by Beneficiary Assess-ments conducted by the Participatory PovertyAssessment Group for the Zambia social fundproject (box O.3.).

Bonding Social Capital. Four aspects of bond-ing social capital were examined: (i) the respon-dent’s ease of participation in communityactivities; (ii) the number of times the respon-dent participated in collective action; (iii) the levelof trust and community cooperation betweenpeople of different ethnic groups; and (iv) easein getting the community to agree on a decision.

OED’s household survey data (Annex M,table M.7) showed that:• Fifty-five percent of the respondents in

Jamaica, 61 percent in Malawi, 43 percent inNicaragua, and 42 percent in Zambia reportedan increase in the ease of participation in com-munity activities compared to the base year.Using both multivariate and difference-in-dif-ference approaches, in one case (Jamaica),this increase was significantly higher than innon-social fund communities. In the otherthree cases (Malawi, Nicaragua, Zambia), nosignificant differences were found.25

A n n e x e s

1 4 5

Beneficiary Assessments show that the social capital effects ofprojects differ depending on factors of project design, leadership,and the social dynamics of a community. Three distinct possibil-ities of impact on social capital emerge from findings in Zambia.

Positive Social Capital Effects. Enhancement of social cap-ital through project participation is the most desired result ina participatory process. One water project exemplifies what canbe accomplished. The project was highly successful in bring-ing together the members of a recently established farm com-munity who hardly knew each other. It created a true sense ofcommunity, enhanced social capital, and even encouragedbeneficiaries to jointly plan activities beyond the scope of theoriginal project.

No Social Capital Effects. A complete absence of any posi-tive or negative social capital effects is also possible. This wasfound to have occurred in a community that was excluded fromproject implementation and did not contribute in any way. Instead,the church and the district council did all work. Consequently,the sense of ownership and responsibility among communityrespondents was nil and social capital remained unaffected.

Negative Social Capital Effects. Some interventions may actu-ally harm social capital. In one school project the PTA responsi-ble for construction work divided the community by hiring skilledlabor only from one village and discouraged free labor contribu-tions by beneficiaries, thus disrupting community participation andcompromising the sense of responsibility and ownership.

Z a m b i a : D i v e r s e S o c i a l C a p i t a l E f f e c t sB o x O . 3

Source: Milimo 1994.

• Fifteen percent of the respondents fromJamaica, 22 percent in Malawi, 7 percent inNicaragua, and 18 percent in Zambia per-ceived more participation in collective actionsince the base year. Using both multivariateand difference-in-difference approaches, thesedifferences were not found to be significantlydifferent from those in the matched non-social fund communities for any of thecountries.26

• Forty-five percent of the respondents fromJamaica, 7 percent in Malawi, 37 percent inNicaragua, and 11 percent in Zambia perceivethat the level of trust and community coop-eration among people of different ethnicgroups has increased since the base year.Using both multivariate and difference-in-dif-ference approaches, when compared withchanges in the matched non-social fundcommunities, only Jamaica indicated a sig-nificant positive social fund impact; in Zam-bia, the result was significantly negative forthe social fund community; and for Malawiand Nicaragua, no significant differences werefound.27

• Fifty percent of the respondents in Jamaica,61 percent in Malawi, 49 percent in Nicaragua,and 43 percent in Zambia perceived improvedease in getting the community to agree on adecision. Using both multivariate and differ-ence-in-difference approaches, a significantpositive social fund effect was found only inZambia; no significant differences were foundin the other three countries.28

The community-level qualitative data in theOED field research countries also provided a fewexamples of how the social funds have in somecases led to future collective action and raisedhopes and expectations for future developmentactivity. In one focus group in Malawi, all agreedthat they would work together in the futurehaving seen the fruits of MASAF, and the head-man confirmed that they were in the process ofapplying to build a bridge. In a Nicaraguanfocus group, it was said that “all people weremotivated and we worked together. Then, somepeople proposed we build a place like a gardenwhere the children play” (Nicaragua).

The OED qualitative studies for the socialfunds review pointed to the possibility of neg-atively affecting social capital through socialfund project activities, although these instanceswere not widespread. In Nicaragua, it was feltthat the conflict resolution method did notappropriately take into account the partiality ofleaders representing different groups within thecommunity. Respondents in Zambia, Jamaica,and Malawi emphasized access to informationand transparency of committee activities andproject processes as essential for building andpreserving trust.

Bridging Social Capital. Two aspects of bridg-ing social capital were examined: the effective-ness of the government in responding tocommunity needs and the effectiveness of thelocal leadership in responding to communityneeds.• OED’s household survey data showed that 25

percent of the respondents from Jamaica, 50percent in Malawi, 18 in Nicaragua, and 7 per-cent in Zambia perceived that the governmentresponded more effectively to their needsnow than it did in the base year. In compar-ison with the non-social fund communities,this represented a significant positive socialfund impact in Jamaica but a significantlynegative one in Nicaragua. No significant dif-ferences were found in the other two coun-tries (Annex M, table M.7).29

• As for the effectiveness of the local leadershipin responding to community needs, 29 percentof the respondents in Jamaica, 30 percent inMalawi, 26 percent in Nicaragua, and 23 inZambia perceived an improvement. In com-parison with non-social fund communities, asignificant negative social fund impact wasfound for Nicaragua—the only country wherethe difference was statistically significant.30

Key informant interviews were more posi-tive, for example, “the needs from the commu-nity to the ministries are more accessible now.”This difference is not surprising considering thatthe quantitative survey included a wide range ofcommunity members many of whom may havehad little involvement in social fund decision-making and management; however, for the hand-


1 4 6

ful of community leaders (key informants) whohave been involved in the contacts made withthe outside world, the social fund process wasreported by them to have had a positive impact.

Understanding the Findings on CapacityBuilding and Social CapitalOverall, the results with respect to the capacitybuilding and social capital effects of social fundprojects are mixed with the qualitative data pro-viding a more positive picture than the quanti-tative data. This may have been because thosemore involved in the social fund projects werelikely to both attend and then self-select forspeaking-up in a focus group discussion or qual-itative interview. The qualitative data show that,for the communities studied in each of the fourfield research countries, there has been capac-ity building and skill development impact amongthe community leaders who were active in deci-sion making. On the other hand, using both mul-tivariate and difference-in-difference approaches,the quantitative data showed no significant com-munity capacity building or skill developmentimpact of the social fund in the two countries(Malawi and Zambia) for which data were avail-able on this question.

With regard to social capital effects, using bothmultivariate and difference-in-difference ap-proaches, there are larger improvements in someelements of bonding social capital in socialfund-assisted communities in Jamaica and Zam-bia compared with matched communities, andno significant social fund impacts in Malawi andNicaragua. With regard to bridging social capi-tal, using both multivariate and difference-in-difference approaches, only Jamaica showspositive social fund effects, Nicaragua shows anegative social fund effect, and Malawi andZambia show no significant social fund impact.Possible reasons for the differences across coun-tries are: the predominance of “new” construc-tion in the sampled social fund-assistedcommunities in Jamaica versus rehabilitation inother cases; and the participatory research (Moserand Holland 1997) which highlighted the link-age between community centers/sports com-plexes and violence and influenced JSIF infinancing such investments (Rao and Ibanez

forthcoming). The findings in Jamaica suggestthat, in some social contexts, using in-depthparticipatory research may be a more effectivemethod of identifying community priorities andbuilding social capital than the invitation of sub-project proposals.

It is important to appreciate that social fundprojects have varied greatly in the priorityattached to capacity building and social capitalobjectives. For example, while some elementsof the Guatemala FIS strategy suggested anobjective of community empowerment, other ele-ments suggested that participation was a way toexpedite project implementation. The prioritieswere clear: in the first two years of operation,these contradictions were worked out in favorof subproject-oriented objectives. In ArgentinaFOPAR, priority was given from the outset tocommunity development. OED’s qualitative fieldresearch in Argentina (comprising semi-structuredinterviews with key stakeholders in governmentand non-government including community lead-ers, four community visits, focus groups withbeneficiaries, review of project files, literaturereview) comparing the social capital impacts oftwo projects in Argentina found that the socialfund project contributed to bonding within thecommunity and to linking individual communi-ties with formal organizations. It was less suc-cessful in generating changes in the way formalorganizations related to poor communities dueto the limited leverage that the poor can havewhen acting from the isolation of their individ-ual communities. The sectoral (agricultural) proj-ect, on the other hand, increased this leverageby organizing a large number of the rural pooraround a single identity shared across commu-nities—that is, being small farmers. This greaterleverage allowed the poor to pressure nationaland subnational governments to take them intoaccount in policy and budgetary issues. Thecreation of inter-community ties was facilitatednot only by organizing around a common iden-tity, but also by involving beneficiaries in state-level managerial structures, accompanyingbeneficiary groups for periods of three years ormore, (compared to an average of eight monthsfor the social fund), and adopting a flexible anddecentralized managerial strategy (box O.4).

A n n e x e s

1 4 7


1 4 8

“OED’s qualitative field research in Argentina (comprising semi-structured interviews with key stakeholders in government andnon-government including community leaders, four communityvisits, focus groups with beneficiaries, review of project files,literature reviewa) indicates that FOPAR (Fondo Participativo deInversion Social) contributed mainly to bonding social capital(within-community ties) while PSA (Programa Social Agro-pecuario) was effective mainly at strengthening bridging socialcapital (inter-community ties). This difference had implicationsfor the impact that each program had in terms of linking socialcapital (ties between the poor and formal organizations). Thebonding capital created by FOPAR, coupled with communitycontracting capacity-building, did allow individual communitiesto create new and better connections with formal institutionssuch as more and better articulated demands to local govern-ments, increased capacities to contract and oversee technicalassistance, and increased capacities to deal with banking insti-tutions. These connections remained a capital of individualcommunities isolated from each other, though, and they did notprovide the amount of leverage that the poor need as a collec-tive actor to demand formal organizations to be more pro-poor,demanding more participatory and transparent municipal plan-ning processes. The bridging capital created by PSA, on theother hand, did increase this leverage by organizing a largenumber of the rural poor around a single identity shared acrosscommunities—i.e., being small farmers. This greater leverageallowed the poor to pressure national and subnational govern-ments to take them into account in policy and budgetary issues.FOPAR is trying a new strategy to create ties across communi-ties (bridging social capital).

At least four elements differentiate PSA from FOPAR andother social fund projects that might explain why it has been eas-ier for the former to generate bridging social capital. First, PSAhad a sectoral (rural development) rather than a multi-sectoralapproach. PSA’s clients shared a concrete and vital identity as“small farmers” that facilitated collective-action endeavors

across communities. Additionally, there is a history of produc-ers organizing work which PSA could tap on and strengthen. Sec-ond, involving beneficiaries in PSA’s state-level managerialstructures created incentives for inter-community links anddemystified the notion of government as an inaccessible space.Beneficiary groups periodically gathered in each state to electand monitor the members representing them in the state public-private councils—these councils were responsible for issuessuch as targeting and subproject approval. In contrast, FOPAR andother social funds involve beneficiaries mainly at the individualsubproject level. Third, building social capital takes time andwhile PSA accompanies groups for prolonged periods (threeyears or more), FOPAR’s average time with a group is eightmonths, after which there is no other formal contact with thegroup. Finally, the more flexible and decentralized managerialstrategy of PSA is better suited for strengthening social capitalthan FOPAR’s more rigid approach. State-level coordinators andpublic-private councils in PSA enjoy significant discretion to pro-pose interventions that, although may not qualify as “projects,”are crucial for sustaining social capital—e.g., assisting a net-work of 36 small farmers’ town-markets to partner with provin-cial and national governments to solve a sanitary problem thatwas threatening the continuity of the network. This type of inter-vention (quick, “light,” customized) does not match well with theproject-cycle driven logic of social funds. Social funds aredesigned as “subproject processing machines,” with the frame-work for intervention being strictly predetermined by the sub-projects menu, and although changes in the menu are allowedthey are not common and take time. Usually, subprojects do notinclude interventions such as addressing bottlenecks threaten-ing the survival of organizations of the poor. The quick responseneeded for certain problems is often inconsistent with the projectcycle timing. The “subproject processing machine” design isappropriate for outputs where the input mix can be specified inadvance, such as building schools or clinics. It appears lessappropriate for strengthening collective action capacities.

A r g e n t i n a : B o n d i n g V e r s u s B r i d g i n g S o c i a lC a p i t a l

B o x O . 4

Note: This box compares FOPAR and PSA because both had community development/social capital creation as an explicit program goal (while TRABAJAR did not). a. Field research in Argentina comprised semi-structured interviews, focus groups, a short survey among FOPAR’s provincial managers, and a desk-review. Semi-structured interviews were conducted with a wide array of stakeholders: national- and provincial-level managers in government; key staff and fieldworkers of eachof FOPAR, TRABAJAR, and PSA; past and current managers from the Social Information Monitoring and Evaluation of Targeted Social Programs (SIEMPRO); mayorsand councillors; Ministers and key staff working in the areas of social development and rural development in the provincial government; and NGOs and communityleaders who had been assisted by these programs. Two focus groups were conducted with FOPAR beneficiaries from about 15 subprojects. Key staff from the WorldBank office in Argentina and TRABAJAR’s task manager were also interviewed. A short e-mail survey among the six FOPAR provincial managers aimed at gatheringdata on the extent to which: (i) FOPAR practices had been adopted by deconcentrated national agencies and provincial and local governments; and (ii) communitieshad increased their influence on government. The desk review comprised a review of Operational Manuals, World Bank documentation (such as PADs and ICRs), ex-post impact evaluation reports contracted by SIEMPRO, consultants’ reports, and academic literature.Source: Serrano 2000, OED portfolio assessment.

1 4 9

The members of the External Advisory Panel wel-come the OED review of World Bank experiencewith social funds as an important contributionto ongoing debate regarding the most appro-priate uses of this instrument. The review drawsextensively on evidence from earlier studies,and adds new survey-based evidence regardingthe character and impact of community partic-ipation in social fund projects and the implica-tions of social fund operations for local andnational government institution-building. Thereview also contributes to a more differentiatedand nuanced appreciation of the advantagesand drawbacks of this approach in differentsectors, for varied objectives, and in widelydivergent country contexts. Overall, the reviewprovides a carefully considered and balancedappraisal.

The panel finds that the conclusions of thereport do not build fully on—nor reflect thegist of—the evidence presented in the body ofthe report. The conclusions are more positive and“tender-hearted” than the evidence warrantswith respect to social funds as an effective instru-ment of service delivery. More specifically, thepanel believes the data and analysis in the reportcall for stronger conclusions with regard to theneed to rethink the rationale of the Bank’sexpanding support to social fund projects. Thepanel is concerned that social funds havebecome something of a “sacred cow,” not sub-ject to the same standards of objective assess-ment used for other kinds of projects.

Despite these reservations, the Panel endorsesthe general direction of the conclusions, in par-ticular the calls for:• Closer attention and analysis of how on-

going or proposed new social funds meshwith evolving institutions in client countriesand with World Bank comparative advantage;

• Fuller consideration of the relative merits ofsocial funds versus alternative approaches togoals of poverty reduction and improvedgovernance; and

• Fresh thinking regarding ways to addressweaknesses of conventional social fundapproaches in specific contexts, includingthe possibilities of modest or major changesin design of the fund itself, strengthenedcomplementary activities, transformation, orphase-out.

Samuel MorleyInternational Food Policy Research InstituteJoan NelsonWoodrow Wilson Center of the Smithsonian andSchool of International Science at AmericanUniversityElinor OstromIndiana UniversityChristina PaxsonPrinceton UniversityJudith TendlerMassachusetts Institute of TechnologyCarol H. WeissHarvard University

ANNEX P: STATEMENT OF THE EXTERNAL ADVISORY PANEL

1 5 1

I. IntroductionSocial funds represent an innovative approachto development assistance, and in recent yearshave become increasingly popular with theBank’s borrower countries. From the first socialfund project in Bolivia in FY87, the portfolio hasgrown to almost 100 projects in about 60 coun-tries as of end-FY01. In addition to the $3.5 bil-lion in Bank lending (45 percent of total socialfund resources), Bank-financed social fundshave leveraged $2.1 billion from bilateral sources(27 percent), $1.6 billion in local financing (20percent), and $630 million in multilateral assis-tance (8 percent).1 Social funds remain a smallshare of the Bank’s overall efforts—2 percent oftotal disbursements and 9 percent of poverty-targeted interventions. Nonetheless, given theinnovations and growing importance of thisapproach, this is an appropriate moment toevaluate experience to date.

Over the past ten years, there has been con-siderable adaptation and innovation in the basicsocial fund mechanism. Programs have beenmolded to country circumstance, and programobjectives have been broadened. Originally con-ceived of as emergency operations to address theshort-term social costs of economic crisis andadjustment, social funds have evolved to addresslonger-term poverty reduction objectives, includ-ing improving access to basic services, as wellas strengthening local governance and civilsociety.

Management welcomes this effort by theOperations Evaluation Department (OED) totake stock of what has been learned from thesocial fund experience. The OED review assessesthe development effectiveness of social fundprojects, and distills lessons and implicationsfor future Bank support. This Management

Response discusses the report’s main findingsand presents management’s views on the keyissues in supporting social funds. The OEDreview identifies constructive areas for futurework, as agreed by Management in the attachedManagement Action Record matrix; however,concerns remain regarding the database andmethodology used in some of the specific areasof the study, as noted below. The managementresponse was developed through consultationswith a broad range of stakeholders within theBank.2

II. General CommentsSocial Fund Performance. Social fund projectshave received some of the highest outcome rat-ings among projects in the Bank’s portfolio.OED evaluations of social fund projects closedas of the end of fiscal year 2000 rate 96 percentas “satisfactory” or better, compared to a Bankaverage of 71 percent and a poverty-targetedinterventions project average of 74 percent forthe same time period.3 OED’s 2000 AnnualReview of Development Effectiveness found thatamong the lowest-income countries, the Bank’ssocial protection projects—comprised largely ofsocial funds in this income group—had thehighest overall satisfactory ratings among sectoralareas (World Bank 2001). Empirical findings,OED’s own ratings of social fund performance,and Quality Assurance Group (QAG) ratings allverify that social funds have met their basicobjectives and have proven to be effective mech-anisms for delivering assistance in support ofpoverty reduction.

Social Funds 2000 Study and Methodol-ogy. Unlike most OED sectoral or thematicreports, this one comes on the heels of a majorimpact evaluation of social funds undertaken by

ANNEX Q: MANAGEMENT RESPONSE

Bank staff: Social Funds 2000 study (SF2000.)4

SF 2000 is a comprehensive and methodologi-cally advanced impact evaluation based on out-comes of social funds in six countries (Armenia,Bolivia, Honduras, Nicaragua, Peru, and Zam-bia). The study was a joint effort of the SocialProtection Unit of the HD Network, the PovertyAnalysis Unit of the PREM Network, and the LatinAmerica and Caribbean, Africa, and Europe andCentral Asia Regions. The evaluation uses someof the latest and most robust approaches avail-able, spanning both experimental and quasi-experimental designs, approaches rarely appliedto development projects. In all cases, controlgroups were constructed to identify the coun-terfactual “What would have been the status ofbeneficiaries without the social fund interven-tion?” Data were drawn from over 19,000 house-hold surveys in social fund communities and42,000 households from national household sur-veys, as well as facilities surveys of over 700schools, health centers, water and sanitationprojects, and 600 non–social fund facilities usedfor comparison. To ensure transparency andindependence, the evaluations in each of thecase study countries were carried out by indi-viduals or agencies external to the social fund.

The OED report draws on many of the find-ings of SF2000. This management response,therefore, takes note of how the SF2000 evidenceand findings are presented, used and interpretedin the OED study, as well making comparisonswith some of OED’s own findings. In both cases,the Management Response raises issues forconsideration.

OED Study and Methodology. Besidesdrawing on SF2000 results and desk reviews ofsocial fund projects, the OED report also pres-ents supplemental information on sustainabilityand social capital based on its fieldwork in a sam-ple of social fund project sites in four countries(three to five social fund communities per coun-try, or 17 of the total of more than 10,000 com-munities benefiting from social fund investmentsin these countries).5 The number of communi-ties sampled is not sufficient to build solid rec-ommendations for the social fund portfolio fromthis data alone, nor are they large enough to rep-resent all communities in the case study coun-

tries. OED acknowledged this limitation duringthe initial design of its study, stating that “theevaluation will not pretend that the case studyfindings are representative of the entire portfo-lio or that the communities being studied rep-resent all communities in the case studycountries.” The limitations of these data must beexplicitly recognized.6

III. Summary of OED Review Findings

A. Findings on Which There Is GeneralConsensus Social Fund Outputs. As reported in the OEDreview, OED project evaluations, and the SF2000report, expanded access to primary schools,health facilities, water supply and sanitation,and improved rural roads have benefited millionsof people. In almost all of the cases studied,social fund investments had a positive impact oninfrastructure, leading to an expansion in phys-ical capacity and higher quality service deliverycompared to control groups. The SF2000 impactevaluations indicate that this increased accesswas accompanied by increased utilization as aresult of social fund interventions.

Improvements in Household Welfare. Inaddition to the direct benefits of expandedaccess to basic services, increased utilizationusually translated into improvements in house-hold welfare. While the OED report presents amore qualified interpretation of the findings as“varied,” the impact results from the Social Funds2000 impact evaluations tell a clear and com-pelling story about changes at the householdlevel in communities that received social fundfinancing, namely:• Social fund investments in primary health cen-

ters resulted in a dramatic reduction in infantmortality in the one country (Bolivia) wherethis could be assessed. Infant mortality droppedfrom about 60 deaths per 1000 live births to30 in social fund households, but increasedfrom 60 to 67 in control group households.

• Substantial health gains were also observedfrom expanding access to water supply, par-ticularly in rural areas, including significantdrops in child mortality (all countries withdata) and reductions in diarrhea compared to


1 5 2

control groups (most countries with data).Where the investments studied were reha-bilitation of pre-existing urban water systems(Honduras), no net health impacts wereobserved. In all of the countries studied,households benefited from a decrease in thetime and/or distance to access water com-pared to nonbeneficiary households.

• Social fund interventions increased schoolsize in all of the countries studied and primaryenrollment rates in most of the countriesstudied. Typically, there was a positive impacton the number of years of schooling and/orage for grade, pointing to a direct link withfuture poverty outcomes. No significantimprovement in achievement test scores wasfound in the one country where this could beanalyzed.

• Impacts in the sanitation sector variedbetween sewerage and latrines. No net healthbenefits could be detected from social fundinvestments in sewerage systems, althoughinvestments in latrines tended to reduce theincidence of diarrhea.

Poverty Targeting. There has been a greatdeal of speculation about whether demand-driven mechanisms are capable of reaching poorareas and poor households. As reported in theOED review, the data from SF 2000 point to pro-poor targeting outcomes. At the geographiclevel, poorer areas received more social fundresources per capita than better-off areas. Thisresult attests to the strong demand for supportexpressed from poor areas, as well as improvedoutreach and targeting efforts by the socialfunds. At the household level, the majority ofbeneficiaries were poor, and the poorest of thepoor showed reasonable access to benefits—thepoorest ten percent of the people representedbetween 8 and 15 percent of social fund bene-ficiaries, depending on the country studied. Ascommunity infrastructure cannot exclude anycommunity member from access, better-offhouseholds also benefited. In all cases analyzed,social funds were at least as well targeted andusually better targeted than other social pro-grams, and typically much better targeted thangeneral public social and municipal spending.

The Social Protection Unit is currently workingwith Development Economics Research Groupto determine the extent to which householdtargeting can be improved using better povertytargeting maps, taking into consideration thelimits on targeting that come with the financingof community infrastructure to which all house-holds have access.

Efficiency. The OED review reports the SF2000 findings that the overhead expenses ofthe social funds studied ranged between 7 and14 percent. Both central and local governmentinvestments in similar infrastructure typicallyhad much higher overhead expenses. However,in terms of the unit costs of social fund sub-projects, no clear trend emerged between socialfunds, nongovernmental organizations (NGOs),local governments, or central agencies, andthere was great variability by country and sec-tor. Where community contributions were highand/or communities directly managed fundsand contracts, unit costs of investment werelower. These findings have broader implica-tions, pointing to potential efficiency gains incommunity-driven programs.

Sustainability. Drawing from the impactevaluation results as well as OED fieldwork,the review findings point to better sustainabil-ity than has generally been anticipated. Thequality of infrastructure was typically quite goodand generally better than comparators—animportant factor in long-term sustainability.7

More importantly, for social services such asschools and health centers, social fund facilitieswere found to be at least as well staffed andequipped as comparator facilities, and usuallybetter so. The meeting of recurrent cost obliga-tions appeared less strained in the education sec-tor than in the health sector, where systemicdifficulties were observed in both social fund andnon–social fund facilities in securing a reliablesupply of all essential drugs. Notwithstanding thispattern, in several countries social fund-supported health centers were better stockedwith key medicines than their non-social fundcounterparts, and in two of the countries stud-ied, average health center staffing was belowministry norms for social funds and controlgroup facilities alike.

A n n e x e s

1 5 3

Physical maintenance and repairs varied bycountry and sector. OED fieldwork found a highproportion of community members willing tocontribute toward maintenance activities of socialfund investments. From the impact evaluations,deficiencies in maintenance were more likely tooccur in water systems than in other types ofinfrastructure, although almost all systems werefound to be operating several years after con-struction. Similar maintenance issues may applyto rural roads, particularly for tertiary roads withthrough traffic. In water projects, cost recoverywas typically insufficient to ensure long-termoperations and maintenance, though this was trueof social fund and non–social fund water systemsalike. Management recognizes the challenges tosustainability, particularly in community-man-aged services. In response to previous concerns,social fund projects of more recent origin havepaid increased attention to training of water-user groups, and to establishing preventive main-tenance funds and maintenance contracts withcommunities, as the review notes. There is stilla great deal to do to improve communities’capacity to ensure proper maintenance in allsectors. Approaches need to be tailored to thespecific sector, taking into account evolving les-sons learned from the sectoral experience. Thisprocess is under way. For example, in the LatinAmerica and Caribbean and Africa Regions, therehave already been regional conferences bring-ing together social fund and sectoral ministriesin the water and sanitation sector to integrate sec-toral concerns into the social fund project cycle.

Against the backdrop of this empirical evi-dence, the report raises the general concernthat prevailing fiscal and sectoral constraintswithin a country may impact the provision ofrecurrent cost financing. The availability of recur-rent cost financing will condition the ultimate mixof rehabilitation versus expansion of infrastruc-ture and the sectoral mix of investments, and itmay ultimately determine the scale of socialfund operations. The availability of recurrentcost financing—whether to hire teachers andhealth workers, to operate water systems, or tomaintain rural roads—depends on the presenceof adequate sector policies and strategies. Man-agement is of the view that most general issues

of public finance are best addressed througheconomywide and sectoral reform programs.Such programs are long term in nature becausethey require fundamental institutional changes.These reform programs typically involve somedegree of decentralization of expenditure deci-sions. In this context, social funds that approvesmall-scale investments on a case-by-case basis,requiring evidence of recurrent cost financingand, where appropriate, prior ministry approval,have an important role to play in fostering theselonger-term reforms while minimizing risks ofescalating recurrent cost obligations.

B. Findings under QuestionCommunity Priorities. The review discussesthe extent to which social fund investments arerelevant to community priorities. OED bases itsconclusion on the small sample of communitiessubject to OED fieldwork. In 9 of the 17 com-munities studied, the top priority was financed, andin most of the others the second eligible prioritywas financed. There may be legitimate reasonswhy a community decides to opt for its secondor third priority, for example, one intervention maybe desired but technically or economically infea-sible, or one intervention may better fit the finan-cial and human capacity of the community. Thelarger body of evidence from impact evaluationsand beneficiary assessments confirms that com-munity members consistently report that socialfund investments reflect priority community needs.All studies, including the OED fieldwork, show thatcommunity priorities were met in the majority ofcases, leadership and citizens were activelyengaged in the process, and community membersas a whole express high levels of satisfaction.The fact that community leaders and existing localorganizations are involved in mobilizing citizensupport and participation is a positive sign thatthese investments are relevant to the communityand embedded in its social fabric.

C. Findings Where Evidence Is Unclear orFurther Research Is NecessaryOED’s project evaluation rating of the institutionaldevelopment impact of social funds is higherthan the average for all Bank projects—58 per-cent were rated “substantial” compared to 36 per-


1 5 4

cent for all Bank projects as of end-FY00. Theseimpacts extend beyond the social fund andinclude effects on central and local govern-ments, the private sector, NGOs, and commu-nities. Management agrees with OED that theseimpacts are difficult to capture, and that it is dif-ficult to generalize because of variations in socialfund design between countries and over time inthe same countries. As a consequence of limiteddata, the report is unable to fully evaluate thecorrelation between social funds and institu-tional development. The report raises appropriateflags of caution to indicate areas of possibleconflict between social fund design and institu-tional development. This useful contributiondirects country teams toward a more completeconsideration of a wide range of developmentissues related to social funds.

Effects on Central Government. The OEDreport recognizes social fund agencies’ achieve-ments in innovation, modern management tech-niques, transparency, and adaptation, which itattributes to their operational autonomy. As thereport states, views differ concerning the contin-uing justification for and ultimate effects of thisautonomy. To fill the empirical gap, the SocialFunds Thematic Group and the Civil Service andAdministrative Reform Thematic Group are assess-ing social funds in terms of institutional structure,budgeting, personnel, procurement, and account-ability procedures for coherence with broaderpublic sector management objectives. A first setof lessons learned and guidance for task teams,based on a checklist of key public sector concernsalready drafted by the Public Sector ManagementGroup, Poverty Reduction and Economic Man-agement Network (PRMPS), will be available totask teams in the second half of FY02.

The report notes that institutional impacts onother central agencies take many forms. Some ini-tial social funds had explicit objectives tostrengthen line ministries. This has either remained,appropriately, within the narrower confines ofthe social fund project cycle (involving ministrystaff in appraisal and supervision, for instance) orhas been largely superceded by the growing trendof local governments and community groups tak-ing on increased responsibilities for communityinfrastructure. Coordination and synergy are the

main issues with central agencies, not training lineministries to “do as social funds do.” Ministry rep-resentatives are usually part of the governancestructure of social funds. However, experienceshows that coordination is best achieved at thelocal level, among the community, local govern-ment, and local representatives of the line min-istries. As the OED report states, these mechanismshave been sufficient to secure central govern-ment acknowledgment, where relevant, of recur-rent cost obligations for the facilities financed bythe social fund, with follow-up by these centralagencies variable.

Effects on Local Governments. In responseto the growing importance of local governments,three years ago the Social Funds Thematic Groupand the Decentralization Thematic Group initi-ated joint research with the United Nations Cap-ital Development Fund in six countries to studythe interface between social funds and the evolv-ing decentralization framework. The OED reportdraws from the initial lessons generated from thiscross-Network cooperation. Sometimes, socialfunds have been drivers of greater local gov-ernment involvement; other times they havesimply responded to shifts in national decen-tralization policies, or have lagged behind them.As local governments increasingly assume respon-sibility for providing basic services, social fundsmay present an opportunity to support thisprocess—or they may undermine the develop-ment of local governments. To ensure that socialfunds fully support national decentralization poli-cies, best practice approaches and toolkits for taskteams are being developed. The Community-Driven Development and Decentralization the-matic groups are researching the interfacebetween local governments and community-based organizations to identify possible synergiesand tradeoffs between the two that will providevaluable insights into social fund design in termsof supporting improved local governance.

Community Capacity-Building and SocialCapital Effects. At the community level, evi-dence of building community capacity and socialcapital remains based on case study approaches,and appears to vary by community and by socialfund. Certainly, social funds have engaged com-munity members to an important degree in solv-

A n n e x e s

1 5 5

ing local development problems. The OED field-work found mixed effects from their 17 com-munity case studies in term of social capitalimpacts, with possible positive, negative, andneutral effects depending on the communityand the measure. Therefore, in management’sopinion, OED data confirm management’s con-clusion that more research and analysis is neededbefore we can say anything definitive on this sub-ject. Solid operational recommendations in thisarea will have to await both sufficient experiencefrom the more recent models that enter moreintensely into community facilitation, and moresystematic and robust evaluation of capacity-building and social capital effects. The Social Pro-tection Anchor is currently exploring options fordeveloping the analytic framework and empir-ical investigation necessary to begin to answerthese questions.

IV. Management’s Views: Issues GoingForwardManagement welcomes the acknowledgement inthe OED report that country circumstance shoulddictate social fund design and relevance. One ofthe strengths of social funds has been theircapacity to evolve and adapt to changing coun-try circumstance. As recognized, this evolutionis not always linear. For example, althoughsocial funds may have moved away from emer-gency objectives to longer-term poverty reduc-tion, their experience with swift delivery ofcommunity-level infrastructure has enabled themto respond effectively to natural disasters withemergency reconstruction. It is important to pre-serve this flexibility and responsiveness tonational circumstance.

The report rightly highlights the challenges ofmultisectoral programs. Multisectoral approachesafford communities choice and allow invest-ments to be tailored to local needs. They alsorequire a great deal of sector content and coor-dination to ensure that investments are not outof step with national policies. Social funds haveadopted many approaches to reconciling thesetensions. There have been shortcomings both onthe side of social funds in incorporating policiesfully into operating procedures, as well as on theside of sectoral ministries in clearly defining

these policies and providing technical input.However, there have also been substantialimprovements in incorporating sectoral experi-ence within social fund design and operation.The Cross-Network Community-Driven Devel-opment Group, comprised of staff working onboth sectoral and multisectoral operations, issupporting the development of Bankwide guide-lines and toolkits for improving sectoralapproaches within multisectoral projects, underthe technical guidance of the various infra-structure-related thematic groups. This experi-ence will be relevant beyond social funds, as theBank moves more into multisectoral instrumentsin community-driven development, decentral-ization support, and programmatic lending. Atthe country level, better definition of sectoralpolicies, including coherence among variousinvestment mechanisms in terms of technicalnorms, approaches to recurrent cost financing,and policies for community cost-sharing, willhelp to better integrate the various national pro-grams that are investing at the community level.

Management fully expects the evolution ofsocial funds to continue. In some countries,social funds are merging with the decentraliza-tion process, as in Bolivia, where the social fundoperates entirely through the municipal planningand budgeting process, and in Zambia, wheresocial funds are leading the experimentationwith shifting investment responsibilities to districtcouncils. In other countries, social funds arebringing communities more directly into thedevelopment process, improving local gover-nance from the bottom up, and strengthening civilsociety, particularly in post-conflict and transitioneconomies. Management is excited about recentadaptations and innovations to better address vul-nerable groups, support the decentralizationagenda, and further strengthen communities.Some of the innovations of the social fund modelhave been picked up in other Bank investmentprojects, encouraging an important diversificationin delivery mechanisms for poverty reduction.

Management does not believe that social fundsare the best instrument to address all poverty andinstitutional development concerns. Social fundshave always been viewed as complementary tothe broader public sector management, decen-


1 5 6

Management Response

1. Management agrees with this recommendation in principle. Specific

actions are outlined below.

Ensuring coherence with CASs. The CAS is the appropriate instrument to ensure

the consistency and coherence of the broad range of Bank support for a country, includ-

ing the relationship between macroeconomic, sectoral policy reform, and specific invest-

ment interventions. QAG QEA of social funds in the FY98 and FY99 samples gave highly

satisfactory ratings to project links with CAS objectives (average rating 1) and to ration-

ale for project approach, including choice of instrument (average 1.5). Management

will track QAG results to ensure continued consistency of social funds with CAS objec-

tives. Determining whether or not a social fund is appropriate given the unique insti-

tutional setting and how the design will complement the country’s institutional

development will be explicitly addressed in each Project Concept Document (PCD),

consistent with the draft OP10.06, Institutional Analysis in Bank-Financed Operations

(currently pending final approval). Task teams will refer to the Institutional Assess-

ment Tool developed by PRMPS for guidance during project preparation.

Social Funds and PRSPs. PRSPs are country documents. It is not the Bank’s role to

determine PRSP content. However it is worthwhile to review country experience in order

to provide information to client countries. HDNSP will conduct a review of PRSPs in

countries with social funds to assess the coherence and integration of social funds within

these national poverty alleviation strategies, to be completed in FY02.

tralization, and sectoral policy initiatives. One ofthe great dangers has been to heap multiple andcompeting objectives onto social funds becauseof their proven operational performance. OEDcautions about investing heavily in social fundsin the absence of progress on broader reformsof the state. However, the report also recognizesthat social funds have been extremely useful inemergency and post-conflict situations and inenvironments characterized by ineffective cen-tral agencies. This tension requires social fundsto operate in suboptimal policy and institutionalsettings, but to remain attentive to supportingthese broader reforms as they are developed.Management agrees with OED that the choice ofinstruments to meet Bank program objectives tai-lored to client demand and client circumstancesis an issue to be discussed in preparing CountryAssistance Strategies.

The OED report states that as social fundsbecome permanent, one measure of institutionaldevelopment should be a declining share ofexternal support, in order to show long-term gov-ernment commitment. This may be appropriatein certain cases. However, many of the poorestcountries still rely heavily on donor funding. Thedemonstration of government commitment mayneed to take other forms if the social fundremains the most viable option for attractingand channeling donor resources to poorcommunities.

V. Major OED Recommendations andManagement ResponseThe following Management Action Record matrixprovides management responses to the specificrecommendations highlighted in the report’sconclusions.

A n n e x e s

1 5 7

Major OED Recommendation

1. Strengthen integration of social funds into the

Bank’s country assistance and sectoral strate-

gies, and into clients’ PRSPs, where relevant.

• Rationale and objectives of Bank support need to

be clear and should drive the choice of instrument

rather than the other way around.

• Social fund projects should not displace policy

reform but should be designed as part of a pack-

age of Bank support rather than as isolated

interventions.

• In order to achieve compliance with Bank sectoral

policies and technical standards and consistency

with country sectoral and public sector management

reform strategies, the Bank should improve coun-

try-team coordination on social fund projects.

M a n a g e m e n t A c t i o n R e c o r d


1 5 8

Management Response

Social Funds and SSPs. Future SSPs will contain a menu of lending and nonlend-

ing services to guide country teams. HDNSP will work with staff preparing SSPs to

ensure that social funds are considered in this context.

Integration of sectoral expertise and country teams. The cross-Network CDD

group (including representation from HDNSP) and sectoral thematic groups are cur-

rently devising approaches for incorporating sectoral strategies in multisectoral

instruments. PRMPS is encouraging public sector thematic groups to contribute

toward a work program examining key areas of concern for public sector management

with regard to social funds (e.g., impact on civil service, flexibility and coordination

of budget, impact on local government incentives). Specific products under prepara-

tion will provide guidance to staff working on social funds, including:

• Rural Water TG will produce guidelines on appropriate water and sanitation

approaches for use by multisectoral instruments, to be available in FY02.

• Rural Transport TG has committed to produce guidelines on appropriate transport

approaches for use by multisectoral instruments, to be available in FY02.

• Rural and Microfinance TG and the Social Funds TG are currently finalizing

research into best practice case studies of microfinance approaches for social

funds, with guidelines to be developed in FY02.

• Joint research supported by HDNSP and PRMPS will review existing institu-

tional arrangements of social funds and identify best practice, in FY02.

In addition, increased efforts will be made to bring in sectoral staff as full-fledged

members of task teams. Efforts will be made to ensure that technical input to multi-

sectoral instruments be an explicit objective of the Bank’s sector unit strategy, and

recognized in sectoral staff work programs. Both the Bank’s Water and Sanitation Board

and the Transport Sector Board will facilitate engagement of sectoral expertise to ensure

consistency of sectoral practice in social funds. Increased norms for project prepa-

ration and supervision in the FY02 budget will help better resource these multisec-

toral operations.

2. Management agrees with general thrust of this recommendation.

Tradeoffs. At the margin, there may be tradeoffs between efficiency and impact, but

these should be determined in each specific case, consistent with the overall objec-

tives of that particular social fund. A social fund engaged in emergency reconstruc-

tion will face that tradeoff differently than one focused on longer-term, community

capacity building.

Demonstrated impact. Continued financing of any Bank project should be contin-

gent on demonstrated impact. In terms of longer-term impacts on household welfare,

the recent impact evaluations on social funds provide a richer basis for judgment than


2. Give more attention to long-term impacts.

• In design of social fund projects, the tradeoffs

between speed and efficiency of subproject pro-

cessing and long-term impacts need to be explic-

itly acknowledged and addressed, and reflected in

performance indicators.

• The continuation or extension of Bank financing to

a social fund should be based on evidence of that

project’s development impact.


A n n e x e s

1 5 9

Management Response

exists in most other Bank projects. HDNSP is currently developing a toolkit to assist

task teams in development of appropriate impact evaluation approaches based on the

experience of the Social Funds 2000 study. This product will be available in late FY02.

Role and Focus. In terms of the role and focus of social funds, design should be coher-

ent with the prevailing institutional framework and government objectives at the time,

and highly responsive to changes in that framework as they occur. Maintaining the

capacity for innovation and flexibility to changing circumstances will remain an

important design feature of social funds. HD’s Social Protection Board will monitor

this role and focus as part of its responsibility for project quality assurance, includ-

ing the use of QERs.

3. Management agrees with the general principle, with a caveat on some of

the specific implications.

Local Planning. Development of investment priorities must come from the community

level, using broad citizen consultations and/or established local planning processes

where available. In many countries, such processes do not yet exist or are very weak.

While social funds support these consultation and participatory planning processes,

having community groups and local governments able to perform systematic cost-ben-

efit analysis on all competing investment alternatives in that community should be

viewed as a long-term goal, taking into consideration that this is not yet achieved even

in industrial countries. In the short and medium terms, ensuring relevance and effi-

ciency of community investments is furthered by (a) the ex-ante economic analysis

social funds apply to each subproject proposal, (b) strengthening of local consulta-

tion processes to identify priority investments, and (c) requirements for local contri-

butions to upfront investment costs, where appropriate.

Informed Beneficiaries. Regarding beneficiary input to monitoring and evalua-

tion, social funds have been at the forefront in systematizing the use of beneficiary

assessments in Bank projects. Guidelines and best practice for beneficiary assess-

ments of social funds have been developed for staff and training modules on related

topics have been carried out since FY99. Such direct community member feedback

will continue to be used to monitor beneficiary awareness of their opportunities and

obligations within social fund programs, as well as to gauge citizen assessment of

benefits.

To promote improved information flows at the community level, the cross-Network

Community-Driven Development Group (including the Social Funds Thematic Group)

is supporting research into best practice approaches for information, education and

communications campaigns at the local level and the use of facilitation to identify

priorities and inform citizens. The results will be available to staff to improve the design


• The appropriate role and focus of the social fund

agency and its relationship with existing institutions

should be anticipated as far as possible from the

outset because of the difficulties experienced in

changing the orientation once the agency is well-

established.

3. Ensure efficiency of resource allocation.

• Social fund projects should ensure that investment

decisions include a systematic articulation of the

benefits as well as the costs of alternative invest-

ments by the community and/or local government

concerned, who should also be charged with mon-

itoring actual benefits in relation to their expecta-

tions at subproject appraisal.

• Stronger measures are needed to ensure that ben-

eficiaries are adequately informed and consulted

on investment options, costs, and benefits.

• The appropriate scale and scope of social fund

activities should be addressed at the project design

stage, and reassessed regularly during implemen-

tation, with reference to budgetary processes and

public expenditure analysis.



1 6 0

Management Response

the level of program knowledge by citizens and will provide a format to monitor

these improvements over time.

The scope and scale of social fund activities and their links with general public

expenditures are contained within the PAD. Recent QAG reviews have found these

issues to be addressed satisfactorily. Management will track performance in QAG

reviews and highlight these issues within QERs and other upstream quality enhance-

ment activities.

4. Management agrees with the need to provide clear criteria and guidance

for Bank support to social funds.

Criteria and advice are appropriately set out through guidance to staff, rather than

through the preparation of separate policy directives. Currently, the relevant policy

statements are provided as part of the Bank’s overall policy requirements for invest-

ment lending, which includes social funds. The overall policies cover the use of eco-

nomic, social, financial and institutional analysis.

The Bank’s Human Development Network already provides a considerable body of guid-

ance to Bank staff with respect to the use, design, and implementation of social funds

in areas such as project design and implementation, economic analysis, gender, envi-

ronmental guidelines, decentralization, emergency reconstruction/response, employ-

ment creation, impact evaluations and beneficiary assessments, management information

systems, procurement and community contracting, participation, working with NGOs,

poverty targeting, inclusion of vulnerable groups and the disabled, and delivery of social

services, with specific reference resources. These guidelines are available on the

HDNSP Social Funds Web Site.* Training programs in conjunction with WBI and rel-

evant thematic groups are conducted on an on-going basis to further disseminate these

guidelines and best practice experience. In consultation with PREM, PSI, and ESSD,

the HD Network will, as necessary, augment and update the guidance provided to staff.

Background work on several important inputs is already underway, including public sec-

tor management considerations and integration of sectoral strategies in multisectoral

instruments, as discussed in Recommendation 1 above. In addition, the Bank’s Safe-

guard Unit is currently preparing guidelines on the application of safeguard policies

to projects that use CDD approaches. This includes standard guidelines for environ-

mental classification to be applied consistently across Networks and Regions. HDNSP

is currently developing a discussion paper on future directions of social funds that will

review recent innovations and discuss institutional options for social funds that reflect

lessons learned from current practice, to be available to staff in FY02.

*http://ispace3.worldbank.org:7001/intranet/jsp/sectors_view.jsp?tab=2&gwitem=

474008


4. Develop policy requirements

Policy requirements on support to social fund projects

should be developed to provide clear criteria and pro-

cedures for Bank support to social funds. The policy note

should identify conditions that indicate appropriate-

ness or inappropriateness of using the social fund instru-

ment and its strategic justification within the CAS and

PRSP, provide guidance on the country-specific infor-

mation and analysis needed to ensure alignment of the

social fund with the institutional context, and identify

viable transformation or exit strategies. (See detailed

list of recommended items to include in the recom-

mendations section of the report.)


1 6 1

The Committee on Development Effectiveness(CODE) met on October 24, 2001 to discuss theOED report Social Funds: A Review of WorldBank Experience (CODE2001-0090) and thedraft Management Response (CODE2001-0091).The OED review is in response to the Bank’srapidly expanding social funds portfolio. A self-evaluation study of social funds projects under-taken by Management, Letting CommunitiesTake the Lead: A Cross Country Evaluation ofSocial Fund Performance (CODE 2001-0092),was circulated as background information. TheChair noted the serious effort made by OED,resulting in a comprehensive review. He alsocommended management for its self-evaluationreport and said both reports could be used toprovide future guidance to staff and to improvesocial funds. The Committee welcomed themanagement response and was pleased to notethe broad alignment between the OED recom-mendations, Management’s self-evaluation, andthe management response. The Committeeendorsed the OED report’s findings and itsrecommendations.

The OED Review. The review confirmedthat social funds had: (a) been highly effectivein delivering and improving access to small-scaleinfrastructure; (b) improved their performanceon poverty targeting over time; (c) been dis-bursed rapidly and had produced visible resultsquickly in difficult situations; and (d) had resultedin some positive institutional developmentimpacts, mostly at the local level. While the vastmajority of beneficiaries had been satisfied withthe resulting social infrastructure, the reviewnoted that social fund projects faced a numberof challenges. These included the: (i) variedresults on outcomes and welfare impacts insocial fund communities; (ii) need to improve

mechanisms to ensure the efficient allocation ofsocial fund investments and that the highest pri-ority community problems were addressed; (iii)significant number of non-poor beneficiariesparticipating and benefiting from the social fundprojects; (iv) insufficient complementary inputsfor ensuring adequate operations and mainte-nance; (v) negative impacts on public sectormanagement in large-scale operations; and (vi)the mixed impact of the social fund participa-tory process in building community capacity orsocial capital.

Members commented on a wide range ofissues raised in the report, including:

Broad Support for Report Findings. TheCommittee recognized the positive results ofsocial funds and appreciated the report’s atten-tion to poverty impact, sustainability and insti-tutional development of social funds. Manymembers believed that the report presented abalanced view of the social funds’ experienceand supported the report’s findings, although onemember felt the tone of the report was more crit-ical than warranted. Management emphasizedthe finding that 96 percent of the closed socialfund projects in the review had been rated “sat-isfactory” on outcome by OED, and both cau-tioned that a negative tone could lead to thereport being misinterpreted by other audiences.The DGO noted that according to the ExternalAdvisory Panel for the report, the conclusionsof the OED report could have been even moreforceful in its recommendation that the Bankrevisit its expanding support to social fundprojects.

Integration of Social Funds with CountryStrategies. The Committee stressed the impor-tance of improving the “alignment” of socialfunds with sector, CAS, and PRSP interventions,

ANNEX R: REPORT FROM CODE: COMMITTEE ON DEVELOPMENTEFFECTIVENESS

to become part of an integrated country strategy.Members underlined that the Bank should con-tinue to address issues of public participation,poverty targeting and budget setting in its policy-level engagements and some underlined thatsocial funds not become permanent, parallel,competing structures to other public sector insti-tutions, although it was noted by one memberthat industrial countries have institutions simi-lar to social funds that operate successfully in par-allel to other public sector institutions. The DGOexpressed his view that the ManagementResponse needed to address, in fuller detail,the relationship between social funds, country-assistance strategy, and country-institutionalsetup, particularly with regard to resource allo-cation. Management informed the Committee thatmany client governments did not view socialfunds as temporary parallel structures but ratherthey had become part of a more varied mix ofpublic institutions and that some funds werefinanced through the budget, and through Banklending. It added that social funds had beenmainstreamed into governments’ overall pro-grams and the Bank’s CASs in Eritrea andEthiopia and in the Latin American region.

Context and Design of Social Funds. TheCommittee discussed a number of issues relatedto the design of social funds and the challengesto using them to realize long-term developmentgoals. Members noted the need for manage-ment to: (a) clarify better what social fundscould and could not do; (b) a graduation pathfor social funds and an exit strategy; and (c) theneed to consider whether the Bank could haverealized similar or better results using alternative,less costly instruments. Members especiallyunderlined the importance of paying more atten-tion to the sustainability and institution-buildingdimensions of social funds and the need toaddress broader, contextual and systemic prob-lems that may impact on social fund perform-ance. They underlined the importance ofensuring “complementary” inputs for subprojectsand of explicitly addressing issues of recurrentcost and capacity and maintenance at the locallevel. They also noted the need to gain a fullerunderstanding of the life cycle of social fundsand how to address structural problems beyond

local-level control. One member asked thatfollow-up to OED’s recommendations extendbeyond a focus on fixing design issues to afocus on addressing broader, contextual prob-lems for which the social fund may be an inap-propriate response. Management cautioned thatsocial funds were limited in terms of what theycould be expected to accomplish, given theirinherent characteristics, and that this needed tobe better recognized. OED noted that the issueof recurrent costs extended beyond social fundsto the facilities in the comparator communities,and that it was an issue that often is not resolv-able at the community level. Speakers appreci-ated the best practice examples given in AnnexN and they suggested that this information bedisseminated more widely.

Guidance on Social Funds. The Committeesupported management’s proposal to updatethe operational guidelines for social funds basedon best practice as opposed to developing sep-arate policy requirements, as had been recom-mended by OED. They asked that the guidelinesbe kept simple so as to not tax the capacity ofclients. OED, however, did not believe that thecurrent policy requirements for investment lend-ing adequately covered all aspects of socialfunds, and it stressed that a formal good-prac-tice statement would need to be developed asa benchmark for evaluation. In its view, thiswould help in harmonizing sector policies andstandards between social funds and other Bankprograms. One speaker believed that it wouldbe more useful to undertake a strategic analy-sis of when and how the Bank should proceedwith respect to social funds, prior to develop-ing operational guidelines.

Community Priorities. The Committee dis-cussed the challenges to ensuring that projectsremained “demand-driven” and the scope andlimits of social funds in delivering public goodsacross communities. They noted, in particular,the tension between ensuring that projectsreflected community priorities and the needs ofthe poor and the role of “prime movers” indeveloping proposals. OED noted that the chal-lenge was to ensure that communities were wellinformed so that community priorities wouldreflect informed choice. One member asked


1 6 2

that in focusing on how to make social fundsmore responsive to the poor, fuller attention begiven to how to reduce vulnerability. OED notedthere was a limit to the extent to which socialfunds could provide public goods and distrib-ute risk across communities, given that socialfunds were community-driven. Managementstated that most community members were sat-isfied with the social-fund project choice and thatthe key was to continue to strengthen partici-patory planning processes.

Conclusion. The Chair noted the great enthu-siasm of staff and beneficiaries for social fundprojects and also noted the impressive per-formance ratings of OED’s evaluation. He notedthat the OED report had raised important ques-

tions about how social funds could be improvedto help address longer-term poverty-reductiongoals on a sustainable basis. He also noted thatthe benefits of social funds, as a tool to respondto local client needs, should not be lost in thisprocess. He concluded that by responding to theconcerns raised by OED in the report, socialfunds could be made even stronger and moresustainable, as well as better equipped todevelop institutions in countries. OED will issuethe report with the revised managementresponse. Management will update the opera-tional guidelines for social funds.

Pieter Stek, Chairman

A n n e x e s

1 6 3

Overview1. Management notes, “although OED char-

acterizes the Social Funds 2000 Impact Evalua-tion as a ‘self-evaluation,’ the impact analysis insix countries used experimental and quasi-exper-imental designs and in all cases data collectionand analysis was contracted to agencies outsideof the social fund, following best practice onevaluation design. The evaluation was a jointeffort between HD’s Social Protection Unit,PREM’s Poverty Analysis Unit, the AFR, LAC,and ECA Regions, with input from DEC.”

2. Management notes that the OED fieldworkshowed that in the majority of communities thetop priority was financed, and the broader bodyof evidence from impact evaluations and bene-ficiary assessments consistently report that thevast majority of social fund projects reflect com-munity priorities.

3. Management notes that OED’s evaluationof social fund experience is a comprehensivepiece of work in some very difficult territory, par-ticularly with respect to the impact of socialfunds on institutional development. The paperraises salient points in this regard, mainly not-ing where experience has suggested possibleconflicts between social fund design and insti-tutional development. However, due to verylimited data, the report is unable to answer out-standing questions that are necessary for eval-uating how social funds have affectedinstitutional development. Given the lack ofdata, Management believes that conclusionscannot be drawn; and the report could have beenmore explicit about the empirical gaps in ourunderstanding of the impact of social funds ongovernment.

Chapter 11. This review uses the Social Protection

Unit’s list as of May 2000 to define the socialfunds portfolio (Annex A), although this listdoes not include some projects with the sameor similar characteristics (such as AGETIPSand some rural development projects). Thefigures represent total Bank lending to theproject concerned, although in some cases thesocial fund was only one component in theproject.

2. In addition to Bank financing, social fundshave attracted substantial financing from otherdonors.

3. Social Funds Website, Social ProtectionUnit, November 2000.

4. The Social Protection Unit’s definition usesthe term “demand” in a general sense (what ben-eficiaries ask for) rather than in an economicsense (what they are willing and able to pay for).

5. “To make them more efficient social fundsusually enjoy special status for example: anindependent legal persona; control over theproject approval process; and/or exemptionfrom prevailing public sector rules and regula-tions relating to issues such as civil service salaryschedules, procurement and/or disbursements,”Social Funds Website, Social Protection Unit,November 2000.

6. At the far end of this spectrum, Argentina’sFOPAR is integrated in the Ministry of SocialDevelopment (Serrano 2000).

7. The 2001 social protection strategy setsocial funds in a risk-management framework. Tothe extent that social and economic infrastructureinvestments reduce risk, social funds contributeto risk management and these outcomes arecaptured as part of the general welfare impactof social fund projects. The present review couldnot assess social fund projects specifically for theirrisk-management impact: few projects includedexplicit risk-management objectives, incorpo-rated explicit mechanisms for risk-managementin their design, or used specific performanceindicators for monitoring risk management effects.

8. Management notes that “although OEDcharacterizes the Social Funds 2000 as a ‘self-evaluation,’ the impact analysis in six countriesused experimental and quasi-experimentaldesigns and in all cases data collection and analy-sis was contracted to agencies outside of thesocial fund, following best practice on evaluationdesign. The evaluation was a joint effort betweenHD’s Social Protection Unit, PREM’s PovertyAnalysis Unit, the AFR, LAC and ECA Regions,with input from DEC.”

Chapter 21. Based on task manager responses to the

basic information questionnaire and, where

1 6 5

ENDNOTES

these were unavailable, staff appraisal reports(SARs) or project appraisal documents (PADs)and operational manuals.

2. Based on SARs/PADs.3. Management notes that 8 out of 12 QAG

reviews of social fund projects rated monitoringand evaluation as “satisfactory,” with 4 “mar-ginally satisfactory.” While there is always scopefor improving monitoring and evaluation (M&E),none of the QAG reviews found fully unsatis-factory M&E; and there were examples of reviewfindings where monitoring indicators, both out-puts and outcomes in terms of developmentimpact, were judged as very good.

4. These ratings relate to the 23 projects thathad closed as of end-fiscal year 2000 and forwhich ICR reviews were conducted by OED. Forthe current review, OED examined all 66 socialfund projects in the social fund portfolio as ofend-fiscal 1999.

5. These numbers include all relevant projectsapproved between fiscal year 1987–99 and closedby the end of fiscal year 2000. Percentagesexclude projects not rated. There were 23 closedsocial fund projects, 103 closed PTI projects, and1,678 closed World Bank projects in the OEDproject database of February 28, 2001.

6. Social funds overwhelmingly support devel-opment of social infrastructure. Improving edu-cation and health outcomes are an important partof the broader conception of poverty; measur-ing the impact of social funds on income povertyis overall not relevant especially in the case ofthose social fund projects that do not supportincome transfers or microfinance.

7. Of nine QAG social fund quality at entryreviews, three were rated highly satisfactory,three satisfactory, and three marginal for appro-priateness of approach.

8. OED reviewed a random sample of eightCASs. Two were found to be moderately satis-factory with regard to a strategic discussion, therest were unsatisfactory.

9. Management notes that QAG has consis-tently rated as satisfactory or highly satisfactorythe links between social fund projects andachieving specific CAS objectives. Results are sim-ilar for QAG ratings of social funds in terms ofproject rationale, including choice of instru-

ment. In a sample of six quality at entry reviewsof social funds, 50 percent were rated as highlysatisfactory, 33 percent as satisfactory and 17 per-cent (one project) marginally satisfactory in thesearenas.

10. This result should be considered in lightof other possible explanations provided in theliterature, for example, that satisfaction may beinfluenced by the fact that communities have alarge range of unmet needs, or that communi-ties lack knowledge about options (confirmedoverall by the OED household survey in fourcountries) and assume they are confronted witha “this or nothing” choice.

11. Management notes that “Given the size ofthe community survey sample, the discussion onthe effectiveness of the demand-driven processcannot be generalized. Given the expected het-erogeneity of the population under review (ben-efiting from at least 66 different social funds, invarious Regions of 42 countries), this fieldresearch is not adequate to support the gener-alization across the entire portfolio of socialfund projects. While the results are an importantaddition to our knowledge of social funds inthese communities, the OED report should becareful in the presentation of its findings. Man-agement also notes that “In almost all caseswhere the second-ranked problem wasaddressed, the top problem not addressed waswater supply, where technical feasibility, cost,and greater organizational demands on thecommunity all come into play. More impor-tantly, more extensive data from impact evalu-ations, including cases with sample sizesrepresentative of the national level, found that“in every country surveyed, social fund invest-ments were consistent with the expressed pri-orities of the majority of community members”(World Bank forthcoming). This is consistentwith qualitative findings from 15 beneficiaryassessments covering 8 countries that concludedthat “social fund projects overwhelmingly reflectfelt needs of poor communities.” Most benefi-ciary assessments covered the issue of commu-nity priorities in some way and the responseswere consistently very high that projects indeedreflected pressing needs of the community”(Owen and Van Domelen 1998)”. OED points out


1 6 6

that it is not generalizing results across all 66social fund projects. Furthermore, OED hasfound no evidence to show that water was notfinanced for reasons of technical feasibility.Finally, OED’s methodology to get at this issuereduced the chances of biased responses.

12. In a few social funds, the social fundagency makes contact with the communitybefore the community sends in an applicationfor a subproject. For example, in Argentina,FOPAR itself selects the communities or neigh-borhoods eligible for applying to the social fundin contrast to other funds that allow all com-munities in the country (or within regions) toapply. Then, FOPAR’s staff directly assists com-munity decisionmaking processes through par-ticipatory techniques, again in contrast to mostsocial funds that enter in direct contact withcommunities only after the latter have selecteda subproject. A facilitation process such asFOPAR’s increases the likelihood of a broad-based participatory process, either by allowingnew “prime movers” to emerge or by encour-aging existing “prime movers” to take the leadin obtaining investments that the communitywants even if they are not of direct personal inter-est to them.

13. For example, in Malawi, 60 percent of thecommunity subprojects financed by MASAF werein the education sector as of October 1999; indollar terms, 65 percent of MASAF resourceswere spent on education. In Zambia, the per-centage of education subprojects approved byMPU exceeded 80 percent during 1995–98; indollar terms, MPU allocation was dominated byeducation, which absorbed 79 percent of the totalcommitments. In Yemen, schools accounted forabout 60 percent of the total portfolio as ofJune 1999.

14. Other reasons for the dominance of schoolprojects may include their localized benefits,which makes it easier for communities to mobi-lize around them or the lower levels (or differ-ent types) of required community contributioncompared with larger or more complex infra-structure subprojects.

15. Participatory research had indicated theimportance of communal facilities in addressingissues of violence: a JSIF-financed sports complex

involved limited community participation in iden-tification, but was perceived by community mem-bers to have directly addressed the violenceproblem by keeping youth off the streets.

16. The report contains a detailed discussionof participation and collective action issues in theJamaica Social Investment Fund.

17. Efficacy with respect to institutional devel-opment goals is discussed in Chapter 4.

18. According to the authors, the precisegroups that were directly hurt were not assistedsince they were assisted by other programs andminers received substantial severance packages.

19. Lustig 1997. Direct impact is limited toemployment effects in infrastructure, microen-terprise, or employment programs—all a smallshare of total social fund spending. Indirectimpacts through enhanced human capital trans-late into income gains only if other conditionsare in place (e.g., jobs).

20. Dijkstra 2000, Dijkstra and Green 2000,OED portfolio assessment.

21. In particular, the Social Funds 2000 find-ing that much lower stunting in FISE commu-nities in Nicaragua is the result of FISE waterfacilities may not be fully accurate since thefacilities are relatively recent while stunting is alonger-term indicator. A similar point applies tothe finding of reduced stunting in Zambia, espe-cially as wasting has increased.

22. Experience in the water and sanitation sec-tor has demonstrated that health benefits aremaximized by a three-pronged approach:providing safe water, sanitation, and healtheducation.

23. Further explanations provided by SocialFunds 2000 are presented in Annex E.

24. A holistic approach is based on puttingall the complementary components for achiev-ing a specific result in place simultaneously orin an appropriate sequence (for example, healthclinic together with safe water, access road, andnutrition education). This need not necessarilybe done through the social fund project itself.

25. The target groups are defined with vary-ing degrees of specificity. A majority of the proj-ects refer to broad and general categories of poorpopulations without further defining “poor,”“poorest,” “vulnerable,” or “low income.”

E n d n o t e s

1 6 7

26. Based on task manager responses to theBasic Information Questionnaire and, wherethese were unavailable, SARs/PADs and opera-tional manuals. N=59.

27. There are well-established tradeoffsbetween both Type I and Type II error andbetween the resources devoted to improving tar-geting and those available to spend in commu-nities. It is quite possible that there are cases inwhich the balance of these tradeoffs suggests thatfurther targeting efforts are not worthwhile.

28. Using data on the geographic distributionof poverty often combined with population data,proactive targeting allocates a specific quantityof resources up front to each area. By contrast,after applications have been received, reactivetargeting uses poverty data in prioritizing themto ensure that resources reach poor areas.

29. Management notes that QAG ratings onquality at entry for social funds have consistentlygiven technical and economic aspects a satis-factory rating. All social funds use ex-ante eco-nomic analysis criteria. The type of analysis isappropriate to the specific type of subproject andranges from unit cost analysis to rate of returnand ability to pay analysis on larger investmentsin water and sanitation. These approaches to eco-nomic analysis are presented in de Castillo 1998and de Castillo and Lema 1998.

30. Different methods have been used to cal-culate administrative costs. A number of socialfunds consider promotion and outreach activi-ties an investment rather than an administrativecost. For example, in Argentina, FOPAR decidedto count the field time spent by the promoters,evaluators, and supervisors as capacity building.With this definition the administrative expendi-tures of FOPAR I were 16 percent and in FOPARII are expected to be around 10 percent, simi-lar to the average of other social funds. If the fieldtime of FOPAR staff is factored in as adminis-trative cost, the figure increases from 16 percentto 30 percent in FOPAR I (Serrano 2000).

31. OED ratings relate to the 23 projects thathad closed as of end-fiscal year 2000 and forwhich ICR reviews were conducted by OED. Forthe current review, OED examined all 66 socialfund projects in the social fund portfolio as ofend-fiscal 1999.

32. These numbers include all relevant proj-ects approved between fiscal years 1987 and1999 and closed by the end of fiscal year 2000.Percentages exclude projects not rated. Therewere 23 closed social fund projects, 103 closedPTI projects, and 1,678 closed World Bank proj-ects in the OED project database of February 28,2001.

33. Management believes that this observationis inconsistent with recent QAG findings. A sam-ple of six recent quality at entry reviews (CY98-99) revealed all projects were rated either highlysatisfactory or satisfactory in their environmen-tal aspects, including adequate analysis andtreatment of environmental impacts, adequatearrangements for mitigating/managing environ-mental impacts, and compliance with Bank safe-guards on environmental assessment.

Chapter 31. OED ratings relate to the 23 projects that

had closed as of end-fiscal year 2000 and forwhich ICR reviews were conducted by OED. Forthe current review, OED examined all 66 socialfund projects in the social fund portfolio as ofend-fiscal 1999. The issue is complicated by thefact that the rating may refer to either or boththe sustainability of the social fund agency or thatof subprojects.

2. These numbers include all relevant projectsapproved between fiscal years 1987 and 1999and closed by the end of fiscal year 2000. Per-centages exclude projects not rated. There were23 closed social fund projects, 103 closed PTIprojects, and 1,678 closed World Bank projectsin the OED project database of February 28, 2001.

3. Three caveats need to be placed on the find-ings. First, social fund facilities and the com-parator facilities are not of the same age. Second,the social fund facilities surveyed were 1 to 3 yearsold, while sustainability problems typically showup in later years. Third, the coverage is mostly ofschools, health, and water and sanitation, not ofother social fund activities such as roads.

4. Tendler points out that not all projects areamenable to participatory design, a telling exam-ple being given by Richard Burghart, 1993, “HisLordship at the Cobblers’ Well,” in Mark Hobart(ed.) An Anthropological Critique of Develop-


1 6 8

ment: The Growth of Ignorance. In a recentpaper, Asim Ijaz Kwaja (2000) reviews 132 com-munity-maintained infrastructure projects (nosocial funds were included). He finds thatincreased community participation in projectdecisions has a positive effect on maintenancefor non-technical decisions, but a negative effectfor technical decisions. This finding is particu-larly interesting given the common perceptionthat all forms of community participation are uni-versally beneficial for sustainability.

5. In Zambia, the picture varies between posi-tion, in some cases staffing has fallen at bothsocial fund and non-social fund facilities but byless for the former, in some risen though by morein the former, and in some risen for the formerand fallen for the latter. However, the difference-in-differences are significant in only a few cases.

6. Of those who had heard of the social fundproject.

7. The analysis compared the perceptions ofrespondents at the time of the survey with thosebefore approval of the social fund project (baseyear). The base year was the year before the startof the subproject. In Jamaica and Nicaragua,the base year was 1995 across all communities.In Malawi and Zambia, the base year varied bycommunity depending on the year of subpro-ject approval.

8. The multivariate analysis confirms positivesocial fund impact in Malawi, Nicaragua, andZambia for adequacy of staff and in Jamaica,Nicaragua, and Zambia for improvement instaffing. For improvement in staffing it indicatespositive correlation with women respondentsfrom the social fund–assisted communities inJamaica (Annex M, tables M.17 and M.19).

9. The multivariate analysis confirms a posi-tive social fund impact for all countries andimprovements in supplies in Jamaica, Nicaraguaand Zambia. For adequacy of supplies it alsoindicates a positive correlation with the poorest-strata respondents in Nicaragua (Annex M, tablesM.18 and M.20).

10. The multivariate analysis indicates that thelikelihood of the facility needing physicalimprovement is positively correlated with theeconomic status in Malawi and negatively cor-related in Zambia (Annex M, table M.21).

11. Social Funds 2000 background researchon Honduras.

12. The multivariate analysis confirms thepositive social fund impact for perception thatrepairs are done quickly in Nicaragua.

13. These conditions are drawn from a com-prehensive review of the literature includingsystematic quantitative analysis, for example,Ostrom, Scgriederm and Wynne 1993 and OEDanalyses of sustainability.

14. OED sustainability matrix.15. Social Funds 2000 background research

for Honduras and Nicaragua.16. In Argentina, there was the confusion

over the legal ownership of the FOPAR infra-structure. If the NUB (Nucleo of Beneficiaries)was a subgroup of a legal community organi-zation, and each of these two groups (viz., theNUB and the community organization) washeaded by a different leader, it sometimes hap-pened that the two leaders entered into con-flict about which faction was the real owner ofthe facility. Having observed this situation in aconsiderable number of cases, FOPAR decidedthat wherever there is a community organiza-tion with legal status, the existing leadership willconform the NUB and the facility will belongto existing organization. NUBs will be createdonly if there is no legal community organiza-tion, and no interest in creating one (Serrano2000).

17. The multivariate analysis confirms thatthe likelihood of knowing who is responsiblefor making repairs is (i) positively correlatedwith being in social fund–assisted communityin Nicaragua and Jamaica, (ii) negatively cor-related with women from the social fund–assisted communities in Malawi and Zambia,and (iii) negatively correlated with the poor-est-strata respondents from the social fund–assisted communities in Nicaragua (Annex M,table M.23).

18. Data for these indicators used “respon-dents who have heard of the social fund” as thebase. This was because respondents were notasked about upkeep and maintenance of the sub-project if they had not heard of the social fund.Where the base is different, this is specified inthe text.

E n d n o t e s

1 6 9

19. The multivariate analysis confirms that thelikelihood of knowing who is responsible forpaying for repairs is (i) positively correlatedwith being in social fund–assisted communitiesin Malawi and Nicaragua, (ii) negatively corre-lated with women from the social fund–assistedcommunities in Malawi and (iii) positively cor-related with the poorest-strata respondents fromthe social fund–assisted communities in Zambia(Annex M, table M.24)

20. Social Fund 2000 background research onHonduras.

21. Federal Ministry for Economic Cooperationand Development (BMZ), Germany, Division411 (Poverty Reduction), Bonn November 1999.

22. Based on task manager responses to thebasic information questionnaire and, wherethese were unavailable, SARs/PADs and opera-tional manuals.

Chapter 41. Management notes that OED’s evaluation

of social fund experience is a comprehensivepiece of work in some very difficult territory, par-ticularly with respect to the impact of socialfunds on institutional development. The paperraises salient points in this regard, mainly not-ing where experience has suggested possibleconflicts between social fund design and insti-tutional development. However, due to verylimited data, the report is unable to answer out-standing questions that are necessary for eval-uating how social funds have affectedinstitutional development. Given the lack ofdata, Management believes that conclusionscannot be drawn; and the report could have beenmore explicit about the empirical gaps in ourunderstanding of the impact of social funds ongovernment.

2. OED ratings relate to the 23 projects thathad closed as of end-fiscal year 2000 and forwhich ICR reviews were conducted by OED. Forthe current review, OED examined all 66 socialfund projects in the social fund portfolio as ofend-fiscal 1999.

3. These numbers include all relevant projectsapproved between fiscal years 1987 and 1999and closed by the end of fiscal year 2000. Per-centages exclude projects not rated. There were

23 closed social fund projects, 103 closed PTIprojects, and 1,678 closed World Bank projectsin the OED project database of February 28, 2001.

4. OED portfolio assessment. 5. OED portfolio assessment.6. Based on task manager responses to the

Basic Information Questionnaire and, wherethese were unavailable, SARs/PADs and opera-tional manuals.

7. OED portfolio assessment; Dijkstra 2000;Dijkstra and Green 2000.

8. OED 2000b. See also, OED 2001: “Forexample, most social protection projects in lowincome countries are social funds which oper-ate outside line agencies to finance and imple-ment projects, often through local communities.Yet, while “enclave” approaches can improveservice delivery to the poor, they may limit thecatalytic influence of projects and underminegovernment capacity if not properly designed.”

9. Based on task manager responses to theBasic Information Questionnaire and, wherethese were unavailable, SARs/PADs and Oper-ational Manuals.

10. AGETIP stands for Executing Agency forPublic Works for Employment. The MauritaniaUrban Infrastructure and Pilot Decentralizationproject, Benin Urban Rehabilitation and Man-agement project, and the Togo (Lomé) UrbanDevelopment project are AGETIPs. AGETIPmanagement does not have investment pro-gramming powers but simply executes subpro-jects selected by the municipalities and presentedto it for execution (principal-agent relationship).

11. Management notes that OED’s review often years of social fund activities in Bolivia(OED Precis No. 147, May 1997) found that “FIShas become an efficient organization capable offinancing investments in basic social sectors rel-atively quickly. It has successfully translatedpolicies into actions and clarified the roles andresponsibilities of the central government andlocal authorities in the delivery of social servicesto the poor.”

12. Of the 160 respondents, 35 percent werefrom social fund agencies, 17 percent fromNGOs, 14 percent from government, 12 per-cent from the World Bank, and 8 percent fromother donor agencies.


1 7 0

13. Based on SARs/PADs.14. OED Institutional Development Matrix.15. OED portfolio assessment.16. As reported in Daniel Owen and Julie Van

Domelen, 1998, when central government agen-cies have filled this role, their methods werefound to be less participatory than other agencies.

17. Guatemala Ministry of Education, OEDportfolio assessment.

18. Nicaragua FISE, for example, financeshealth posts only if the ministry of healthapproves and it also asks for approval at depart-mental and municipal levels. Nevertheless, healthposts have been found closed due to the non-availability of health personnel. During OEDfield research, FISE personnel confirmed that 6FISE health posts were without personnel in1999 (Dijkstra 2000).

19. Education Sector Strategy, World Bank,May 18, 1999; Sector Strategy, Health, Nutritionand Population, World Bank, 1997.

20. In the context of tracking poverty-reducingpublic spending in heavily indebted poor coun-tries, IMF and World Bank staff concluded that cre-ating separate institutional poverty funds would,in many cases, undermine the significant progressalready achieved in most HIPCs in providing com-prehensive budgets, making it more difficult totrack aggregate performance in increasing publicresources allocated to poverty reduction.

21. Based on SARs/PADs.22. Based on task managers’ responses to

the Basic Information Questionnaire and, wherethese were unavailable, SARs/PADs and Oper-ational Manuals.

23. In Honduras, until recently individualcommunities could request intercommunitypiped water systems, but only municipalitiesare now eligible to sponsor these investments.This change resulted from, among other things,the realization that the community was not theappropriate level for decisionmaking for suchinvestments.

24. From beneficiary assessments, interviewswith task managers, Serrano 2000.

25. Parker and Serrano 2000; Dijkstra andGreen 2000.

26. OED’s qualitative review of 4 JSIF road andwater projects in Jamaica.

27. The Bank financed only one of thesefunds.

28. Social capital is “a community’s capacityfor collective action” (Woolcock and Narayan2000). This definition encompasses: the natureand density of social networks in a communitythat build bonds within the community andbridges across communities, civil society, andgovernment; and the community’s ability tointernalize information and experience andenhance its (present and future) ability to man-age projects. Social capital effects may be pos-itive or negative: social funds may decrease orincrease divisiveness and stratification in thecommunity; they may include or exclude thepoor/poorest members; and they may or may notreplace one set of elite networks with another(equally or more exclusionary) one.

29. Based on SARs/PADs.30. Malawi and Zambia. Data relating to

capacity building and skill development wasnot available for Jamaica and Nicaragua.

31. This is despite the fact that all sampledprojects in Nicaragua were 1998 or post 1998-approved projects by when FISE had givenexplicit attention to community participation.

32. Field research in Argentina comprisedsemi-structured interviews, focus groups, a shortsurvey among FOPAR’s provincial managers,and a desk-review. Semi-structured interviewswere conducted with a wide array of stake-holders: national- and provincial-level managersin government; key staff and fieldworkers of eachof FOPAR, TRABAJAR, and PSA; past and cur-rent managers from the Social Information Mon-itoring and Evaluation of Targeted SocialPrograms (SIEMPRO); mayors and councilors;Ministers and key staff working in social devel-opment and rural development in the provincialgovernment; and NGOs and community leaderswho had been assisted by these programs. Fourcommunity visits were undertaken. Two focusgroups were conducted with FOPAR beneficiariesfrom about 15 subprojects. Key staff from theWorld Bank office in Argentina and TRABA-JAR’s task manager were also interviewed. Ashort e-mail survey among the six FOPAR Provin-cial Managers aimed at gathering data on theextent to which: (i) FOPAR practices had been

E n d n o t e s

1 7 1

adopted by deconcentrated national agencies andprovincial and local governments; and (ii) com-munities had increased their influence on gov-ernment. The desk review comprised a reviewof operational manuals, World Bank documen-tation (such as PADs and ICRs), ex-post impactevaluation reports contracted by SIEMPRO, con-sultants’ reports, and academic literature.

33. Participatory research had indicated theimportance of communal facilities in addressingissues of violence: a JSIF-financed sports complexinvolved limited community participation in iden-tification, but was perceived by community mem-bers to have directly addressed the violenceproblem by keeping youth off the streets.

34. The report contains a detailed discussionof participation and collective action issues in theJamaica Social Investment Fund.

35. This information is available for 53 of the66 social fund projects in the portfolio. Of the53 projects, 13 included microcredit activities.

36. OED portfolio assessment, Dijkstra 2000.37. Based on SARs/PADs.38. Based on task manager responses to the

Basic Information Questionnaire and, wherethese were unavailable, SARs/PADs and Oper-ational Manuals.

Chapter 51. For example, in Jamaica, the identification

of the sports complex (financed by JSIF) involvedlittle community participation, but reflected thefindings of participatory research which hadshown the linkages between communal facili-ties and violence: the sports complex addressedthe top priority problem (lack of security) of thecommunity.

2. Education Sector Strategy, World Bank,May 18, 1999; Sector Strategy, Health, Nutritionand Population, World Bank, 1997.

3. For example, OED field research inArgentina found that the social fund (FOPAR) hadless impact in building bridging social capitalthan a sectoral rural development project whichaccompanied communities for three years ormore, compared to an average of 8 months forFOPAR (Serrano 2000; Annex O, box O.3).

4. For example, in Albania, Armenia, or Hon-duras (after Hurricane Mitch).

5. This was found to be the problem in Boliviaunder the SIFs where the activities of the socialfund were impeding the intergovernmental fis-cal framework.

Chapter 61. OD 4.15 Poverty Reduction, December

1991, paragraph 26 noted that “Because socialaction programs in many cases have a multi-sectoral focus and are administered by separateproject units, special efforts are required toensure the consistency of the program’s sub-projects with sectoral strategies.”

Annex B1. A random sample (stratified by region) of

30 social fund projects was selected out of theuniverse of 66 social fund projects. Of the 30projects, 21 were active. PSRs for these activeprojects were reviewed. In addition, PSRs for allactive social fund projects in the four fieldresearch countries (i.e., Jamaica, Malawi,Nicaragua, and Zambia) were also examined;there were a total of 8 social fund projects in thefour field research countries of which 5 wereactive. In total, PSRs for 26 projects werereviewed.

2. Included six consultants, one internationalNGO, one social development partnership organ-ization, one networking institute, one UN spe-cialized agency, and one university.

Annex E1. While utilization of both FISE and non-FISE

health centers increased significantly, a slightlygreater and significant increase was observed inthe FISE health posts.

Annex F1. For example, “the development objective

of the project is to increase on a sustainable basisthe incomes of the poor and improve theiraccess to services” (SAR, Tajikistan Pilot PovertyAlleviation Project), or “the project aims atpoverty alleviation by addressing the demand ofthe poor for priority infrastructure and services”(PAD, Panama Social Investment Fund Project).Alternatively, reaching the poor is listed as oneof several objectives; for example, the Philippines


1 7 2

SZOPAD Social Fund Project has the overallobjective of facilitating the implementation ofdevelopment activities in areas formerly affectedby the civil war, but the specific objective is “toincrease the access of the population in thepoor and most conflict-affected areas to basiceconomic and social infrastructure, services andemployment opportunities” (PAD, p. 2).

2. Project documents mention television (10projects), radio (7 projects), newspapers (9 proj-ects), and leaflets, brochures, and posters (13projects).

3. Project documents mention public meetings(16 projects) but also using NGOs (8 projects),government (10 projects), local leaders (2 proj-ects), and politicians (1 project).

4. This result reflects the finding that 30 per-cent of beneficiaries of health and education facil-ities are from the bottom decile, reflecting bothgood geographic targeting and, for education,targeting within communities.

5. For school infrastructure, “the incidence ofinvestments in educational infrastructure is morepro-poor when these are made by FONCODESthan when they are made by Parent’s Commit-tees or INFES.” (Paxson and Schady 1999.) [OEDnotes that FONCODES has a more progressivedistribution than that of the INFES program, butthis is to be expected given that the latter focuseson secondary schools in urban areas].

6. Management notes that the authors of theSocial Funds 2000 study have stressed that cau-tion is needed in directly comparing geographictargeting (Figure F.1) with household targeting(Figure F.2) due to differences in time span,sectoral mix and types of measurement used.

7. For results on other countries, see SocialFunds 2000.

8. Walker and others 2000, table 6.5, p. 22.9. Data from Social Protection Unit, 12/4/00.10. Although education did well in reaching

the bottom 10 percent.11. There are well-established trade-offs

between both Type I and Type II error andbetween the resources devoted to improving tar-geting and those available to spend in commu-nities. It is quite possible that there are cases inwhich the balance of these trade-offs suggest thatfurther targeting efforts are not worthwhile.

12. Expenditure incidence is from Bank data.Poverty headcount is from same source for Zam-bia and World Development Indicators forNicaragua; these data use national poverty linesand so are not comparable between countries.

13. World Bank data.14. Under FHIS 1, the target for resource dis-

tribution to municipalities with “deficient,” “reg-ular,” and “acceptable” levels of poverty (i.e.,non-poor municipalities) was 48 percent, andunder FHIS 2, the target was 50 percent.

15. Fumo and others 2000. Phrase in italicsadded.

16. The bias is the result of the populationweighting rather than the indicators chosen.

17. Dijkstra 2000. 18. For the other quintiles there is not too

great a deviation from their “expected” share of20 percent.

19. Under this rule, 45 percent of resourcesare planned to be spent in the poorest districtquintile (OED field visit Zambia 2000). The allo-cation for district i (Ai) is given by:

20. Social Funds 2000 research for Nicaragua,page 16.

21. E-mail from J. Van Domelen 12/4/00.22. Schady 1999.23. World Bank communication.24. World Bank communication.

Annex I1. These numbers include all relevant projects

approved between fiscal year 1987–99 and closedby the end of fiscal year 2000. Percentagesexclude projects not rated. The data are fromOED’s project database and includes closedprojects as of end-fiscal 2000. There were 23closed social fund projects, 103 closed PTI proj-ects, and 1,678 closed World Bank projects.

2. QAG defines “at risk” projects as thosethat are at risk of not meeting their developmentobjectives and including both actual and poten-tial problem projects. Actual problem projects arethose for which Implementation Progress is

AP P

P P

ii i

j , j

j

n=

−

−=

∑/ ( )

/ ( )

1

11

0,

0

E n d n o t e s

1 7 3

unsatisfactory and/or the Development Objec-tives are not likely to be achieved. Potentialproblem projects are those that are rated satis-factory on IP and DO but have other risk fac-tors historically associated with unsatisfactoryoutcomes.

3. A Category C rating is given to projectslikely to have minimal or no environmentalimpact. No environmental assessment is requiredfor Category C projects. (World Bank OP 4.01Environmental Assessment.)

4. Management notes that on safeguard andenvironmental issues, on average QAG assess-ments, both quality at entry and quality of super-vision, consistently found social fundperformance on environmental issues to berated as satisfactory. The cases cited here do notreflect the general findings. A sample of sixrecent quality at entry reviews (CY98–99) showedthat all projects were rated either ‘highly satis-factory’ or ‘satisfactory’ in their environmentalaspects, including adequate analysis and treat-ment of environmental impacts, adequatearrangements for mitigating/managing environ-mental impacts, and compliance with Bank safe-guards on Environmental Assessment.

Annex O1. Based on World Bank data.2. Owen and Van Domelen 1998. 3. As argued later, the issue is not that leaders

should not bring project ideas to the community(leaders play a crucial role in a demand-drivenprocess), but whether there are mechanisms toensure that the ideas of the leaders are also themost important ones for the community.

4. Retired officials, police, or army occupy animportant position in the community.

5. A single school will have a catchment areaof several villages, typically around six for aprimary school.

6. In practice, rehabilitation usually includesthe construction of new classroom blocks, oftenreplacing wood and thatch structures.

7. Wider local planning processes (as opposedto exclusively community-centered ones) appearto emphasize a shift away from schools in othersocial fund projects as well, for example, inAlbania or Bulgaria where roads or other sub-

projects have dominated. 8. Commentators have pointed out that sat-

isfaction may be influenced by the fact thatcommunities have a large range of unmet needs,or that communities lack knowledge aboutoptions and assume they are confronted with a“this or nothing” choice.

9. The issue is not that “prime movers” shouldnot bring project ideas to the community (lead-ers play a crucial role in any demand-drivenprocess), but whether there are mechanisms toensure that the ideas of the leaders are alsoimportant ones for the community.

10. Owen and Van Domelen 1998, p. 20,para. 3.6.

11. The OED stakeholder survey found that28 percent of respondents thought that at thetime they make their proposal communities arefully aware of the full range of eligible subpro-ject types. Forty-three percent thought they werepartially aware, and 6 percent that they wereunaware.

12. Youths, particularly males, saw them-selves as excluded from the participatoryprocesses. As a consequence, they were alsoapart from social capital accumulation. In onecommunity in Malawi, youths reportedly wouldnot work without being paid.

13. Multivariate analysis indicates that thelikelihood of attending the meeting is (i) nega-tively correlated with economic status in Jamaicaand Nicaragua and (ii) positively correlated withwomen from social fund-assisted communitiesin Nicaragua. And the likelihood of speaking atthe meeting is (i) negatively correlated witheconomic status in Nicaragua, and (ii) nega-tively correlated with women from the socialfund communities in Jamaica, Malawi, and Zam-bia (Annex M, table M.3).

14. The participatory process for Malawi andNicaragua discussed earlier showed that the com-munity often comes together in a meeting onlyafter the “prime mover” has already selected andproposed a particular subproject. The meeting ismore concerned about discussing next steps,rather than re-opening questions of subprojectchoice. While speaking up at such a meeting maybe futile in influencing the choice of the sub-project, this may not necessarily be the case—it


1 7 4

is conceivable that if a large number of com-munity members attend the meeting and speakup, the “prime mover” might feel pressured toreconsider the chosen subproject in favor ofother subprojects that may better meet commu-nity priorities (though this is unlikely if the “primemover” is associated with their preferred project).

15. The case for participatory interventionsrooting themselves in existing social structuresis well-established in the literature; see, forexample, Wade 1988.

16. Multivariate analysis confirms the socialfund impact for all four countries. Furthermore,the likelihood of having heard of the socialfunds from their local leader is (i) positivelycorrelated to women from the social fund-assisted communities in Jamaica and (ii) nega-tively correlated to the poorest strata respondentsfrom the social fund-assisted communities inNicaragua, and (iii) negatively correlated towomen from social fund-assisted communitiesin Zambia (Annex M, table M.6).

17. FOPAR in Argentina itself selects the com-munities or neighborhoods eligible for applyingto the social fund in contrast to other funds thatallow all communities in the country (or withinregions) to apply. Then, FOPAR directly assistscommunity decision-making processes, againin contrast to most social funds that enter in directcontact with communities only after the latterhave selected a subproject. A facilitation processsuch as FOPAR’s can allow new “prime movers”to emerge or encourage existing “prime movers”to take the lead in obtaining investments that thecommunity wants even if they are not of directpersonal interest to them.

18. The question asked was: “were you or yourhousehold asked to make contributions of time,money, or materials during the construction?”.

19. The multivariate analysis indicates that thelikelihood to be asked to contribute not relatedto gender or economic status (Annex M, tableM.5).

20. Social capital is “a community’s capacityfor collective action.” (Woolcock and Narayan2000). This definition encompasses: the natureand density of social networks in a communitythat build bonds within the community andbridges across communities, civil society, and

government; and the community’s ability tointernalize information and experience andenhance its (present and future) ability to man-age projects. Social capital effects may be pos-itive or negative: social funds may decrease orincrease divisiveness and stratification in thecommunity; they may include or exclude thepoor/poorest members; and they may or may notreplace one set of elite networks with another(equally or more exclusionary) one.

21. Based on Bank data.22. This is true in three of the four countries

studied. JSIF works through intermediaries, han-dling contracting itself, and does not providetraining to community members in projectmanagement.

23. The multivariate analysis indicates that theimprovements in the blue collar skills were neg-atively correlated with the poorest strata respon-dents of the social fund-assisted communities inMalawi and Zambia. And the improvements in themanagerial skills were positively correlated to thewomen from the social fund-assisted communi-ties in Malawi. (Annex M, table M.8 and table M.9).

24. In all cases, the change over time beforeand after the introduction of the social fundwas examined. A positive difference between thesocial fund-assisted community and the matchedcommunity implies that the variable under con-sideration improved more in the socialfund–assisted community. Two techniques wereused to compare the social fund communitywith the matched community: difference-in-dif-ference and multivariate regression. The analy-sis used all respondents in the OED survey ineach country and compared the perceptions ofrespondents at the time of the survey with per-ceptions preceding approval of the social fundproject in the country (base year). The baseyear was 1995 for Jamaica and Nicaragua, 1996for Malawi, and 1992 for Zambia. Data forJamaica and Nicaragua relating to capacity build-ing and skill development was not available.

25. For an increase in the ease of participa-tion in community activities, the multivariateanalysis confirm a positive social fund impact forJamaica and Zambia. The difference-in-differenceconfirms the social fund impact for Jamaica. ForMalawi, the multivariate analysis also indicates

E n d n o t e s

1 7 5

a negative correlation with (i) being a respon-dent from social fund-assisted communities; and(ii) the poorest strata respondents from thesocial fund-assisted communities (Annex M,table M.10).

26. For increased participation in collectiveaction, the multivariate analysis indicates (i) apositive correlation with women from the socialfund-assisted communities in Zambia, and (ii)negative correlation with poorest strata respon-dents from the social fund-assisted communitiesin Nicaragua and Zambia (Annex M, table M.11).

27. For increase in level of trust and com-munity cooperation among people of differentethnic groups, the difference-in-difference andmultivariate analysis confirm a positive socialfund impact for Jamaica and a negative one forZambia. The multivariate analysis also indicatesa positive correlation with being a respondentfrom the poorest strata from the socialfund–assisted communities in Malawi (Annex M,table M.12).

28. For improved ease in getting the com-munity to agree on a decision, the difference-in-difference and multivariate analysis confirma positive social fund impact in Zambia. The mul-tivariate analysis also indicates a negative cor-relation with the poorest strata respondents insocial fund-assisted communities in Malawi(Annex M, table M.13).

29. For a more effective government response,the difference-in-difference and multivariateanalysis confirm a positive social fund impact inJamaica and negative social fund impact inNicaragua. (Annex M, table M.14).

30. The multivariate analysis and thedifference-in-difference approach indicate neg-ative social fund impact for Nicaragua. And indi-cates a negative correlation with poorest stratarespondents from the social fund-assisted com-munities in Malawi (Annex M, table M.15).

Annex Q1. Data from the Social Funds Database, Social

Protection Unit, Human Development Network(HDNSP), accessible at the World Bank website(click Sectors, Social Protection, Social Funds).

2. Comments and inputs were received fromthe Social Protection Board of the Human Devel-

opment (HD) Network, the Private Sector andInfrastructure (PSI) Network, the Poverty Reduc-tion and Economic Management (PREM) Net-work, the Environment and SustainableDevelopment (ESSD) Network, and countrydepartments of the various Regions active insocial funds.

3. OED notes that these ratings relate to the26 projects (roughly a third of the social fundportfolio) that have closed and for which imple-mentation completion report reviews were con-ducted by OED. For the current review, OEDexamined all 66 social fund projects in the socialfund portfolio as of end-fiscal 1999.

4. See HDNSP’s Social Funds website, acces-sible at the World Bank website (click Sectors,Social Protection, Social Funds). OED notes thatas with any research, SF2000 has both strengthsand weaknesses. For example, there are diver-gent views on methodological issues relating tothe SF2000 study. The counterfactual used by theSF2000 study (comparing social fund house-holds with households who have not receiveda comparable investment) sets a low bar forassessing social fund performance. Managementnotes that the counterfactual applied is consis-tent with international best practice for impactevaluations.

5. OED notes that its social funds review isbased on a portfolio assessment of all 66 socialfund projects approved by the Bank as of end-fiscal 1999, a literature review, task managerinterviews, stakeholder surveys, household sur-veys of 1,687 randomly-selected households andcommunity-level key informant interviews andfocus groups in four countries, and institutionalanalysis based on field research in seven coun-tries. Conclusions relating to the portfolio in allcases are based not just on its household surveyresults but on triangulation of information fromall the sources listed here as well as SF2000.

6. Management notes that Bank Manage-ment/staff and client-country officials have hadlimited access to the OED country-level casestudies quoted in the report. While these coun-try cases studies will now be available on request,it would have been preferable for the Manage-ment Response to have been based on a fullreview of this work.


1 7 6

7. These findings were drawn from impactevaluations of subprojects in education, health,water supply, sewerage and latrines—the mainsectors financed by social funds. Roads and

other sectors were not included as they accountfor a small share of the portfolio of the socialfunds studied.

E n d n o t e s

1 7 7

Background papers to this review are identifiedwith an asterisk*.

Abbott, T., and Jane Covey. 1996. Social Invest-ment Funds: An Expanded Critique. IDRReport, 12(4). Institute for DevelopmentResearch, Boston.

Alcazar, L., and E. Wachtenheim. 2000.”Deter-minants of Success of Social Fund Projects:The Case of FONCODES.” Instituto Apyao,Lima. Photocopy.

Alemayehu, A. 1997. “Appraisal and EnvironmentAssessment of Social Fund Sub-Projects.”Paper presented for the International Work-shop on Social Funds and Regional Consul-tations, World Bank.

Barrientos, J. 1999. Coordinating Poverty Alle-viation Programs with Regional and LocalGovernments: The Experience of the ChileanSocial Fund-FOSIS. World Bank Social Pro-tection Discussion Paper No 9933. Washing-ton, D.C.

Bigio, A. (ed.). 1998. Social Funds and Reach-ing the Poor: Experiences and Future Direc-tions. EDI Learning Resources Series.Washington, D.C.: World Bank.

Burghart, R. 1993, “His Lordship at the Cob-blers’ Well.” In Mark Hobart, ed., An Anthro-pological Critique of Development: The Growthof Ignorance. London: Routledge.

Carvalho, S. 1999. “Social Funds.” Paper pre-sented at Center for Urban DevelopmentStudies, Harvard University, 12 October, Cam-bridge, Massachusetts.

——. 1994. “Social Funds: Guidelines for Designand Implementation.” World Bank HRO Work-ing Paper 34, Washington, D.C.

Carvalho, S., and H. White. 1997. Combining theQuantitative and Qualitative Approaches toPoverty Measurement and Analysis: The Prac-tice and the Potential. World Bank TechnicalPaper 366. Washington, D.C.

Cornia, G. 1999. Social Funds in Stabilizationand Adjustment Programmes. Research forAction 48. Helsinki: UNU World Institute forDevelopment Economics Research.

Covey, J. 1997. “Role of NGOs in the Design,Management and Implementation of SocialFunds.” Paper presented for the International

Workshop on Social Funds and Regional Con-sultations. Washington, D.C., World Bank.

Das Gupta, M., H. Grandvoinnet, and M. Romani.2000.” State Synergies in Development: Lay-ing the Basis for Collective Action.” WorldBank Policy Research Working Paper 2439.Washington, D.C.

De Castillo, C. P. 1998. Economic Analysis ofSocial Investment Fund Projects. Washing-ton, D.C.: The World Bank.

De Castillo, C. P., and R. Lema. 1998. EconomicAnalysis of Social Investment Fund Projects:Case Studies and Minimum RequirementsProposal. Washington, D. C.: The World Bank.

De Silva, S. 2000. “Community-based Contract-ing: A Review of Stakeholder Experience”World Bank Working Paper 20414. Washing-ton, D.C.

Demery, L. 2000. “Benefit Incidence: A Practi-tioner’s Guide. World Bank Poverty and SocialDevelopment Group, Africa Region. Wash-ington, D.C. Photocopy.

Desmarchelier, A. 2000. “Rural Transport Infra-structures: Guidelines for Community-BasedDevelopment Projects.” Washington, D.C.,World Bank. Photocopy.

*Dijkstra, G. 2000. “Nicaragua Fondo de Inver-sión Social de Emergencia (FISE).” Back-ground paper for OED Social Funds Review.Washington, D.C., World Bank. Photocopy.

*Dijkstra, G., and J. Green. 2000. “Jamaica SocialInvestment Fund (JSIF).” Background paperfor OED Social Funds Review. Washington,D.C., World Bank. Photocopy.

DiMaggio, P., and W. Powell. 1991. “The IronCage Revisited: Institutional Isomorphism andCollective Rationality.” In P. DiMaggio and W.Powell, eds., The New Institutionalism inOrganizational Analysis. Chicago: Univer-sity of Chicago Press.

El Gamal, H. M. 1997. “Financial ResourcesMobilization for Social Funds.” Paper pre-pared for the International Workshop onSocial Funds and Regional Consultations,World Bank.

Ellerman, D. 2001. “Helping People Help Them-selves: Towards A Theory of Autonomy-Com-patible Help.” World Bank Policy ResearchWorking Paper 2693. Washington, D.C.

1 7 9

BIBLIOGRAPHY

Federal Ministry for Economic Cooperation andDevelopment, Poverty Reduction Division.1999. Support for Social Funds under Ger-man Official Bilateral Development Cooper-ation. Bonn.

Feinstein, O., and R. Picciotto. (eds.) 2000. Eval-uation and Poverty Reduction: Proceedingsfrom a World Bank Conference. Washington,D.C.: World Bank.

Fledderjohn, D. 1995. An Assessment of TwoPilot Programs of FONCODES in Peru in theContext of Future Regional Economic Devel-opment. Washington, D.C.: Inter-AmericanDevelopment Bank.

Frigenti, L., and A. Harth, with R. Huque. 1998.Local Solutions to Regional Problems, TheGrowth of Social Funds and Public Worksand Employment Projects in Sub-SaharanAfrica. Washington, D.C.: World Bank.

Fumo, C., A. de Haan, J. Holland, and N. Kanji.2000. Social Funds: “An Effective Instrumentto Support Local Action for Poverty Reduc-tion.” Social Development Department, DFID,Working Paper No. 5. London.

Glaessner, P., K. W. Lee, A. Sant’Anna, and J. deSt. Antione. 1994. Poverty Alleviation andSocial Investment Funds: The Latin Ameri-can Experience. World Bank DiscussionPapers 261. Washington, D.C.

Goodman, M., S. Morley, G. Siri, and E. Zuck-erman. 1997. Social Investment in Latin Amer-ica: Past Performance and Future Role.Evaluation Office, Social Programs and Sus-tainable Development Department. Washing-ton, D.C.: Inter-American Development Bank.

Graham, C., and C. Kane. 1998. “OpportunisticGovernment or Sustaining Reform? ElectoralTrends and Public-Expenditure Patterns inPeru, 1990–1995.” Latin America ResearchReview 33 (1).

Grootaert, C., G-T. Oh, and A. Swamy. 1999. “TheLocal Level Institutions Study: Social Capitaland Development Outcomes in Burkina Faso.”Local Level Institutions Series, No. 7, WorldBank Social Development Family, Environ-mentally and Socially Sustainable Develop-ment Network. Washington, D.C.

Grosh, M. 1990. “What Should Social FundsFinance?: Portfolio Mix, Targeting, and Effi-

ciency Criteria.” View from LATHR 3, WorldBank Human Resources Division. Washington,D.C.

Heiser, T. 1990. “Socio-Economic DevelopmentFunds: An Operational Guideline.” SocialDimensions of Adjustment Project Unit, WorldBank, Washington, D.C.

Holdcroft, L. 1978. The Rise And Fall Of Com-munity Development In Developing Coun-tries, 1950–1965: A Critical Analysis And AnAnnotated Bibliography. Department of Agri-cultural Economics, Michigan State University.

IDB (Inter-American Development Bank). 1998.The Use of Social Investment Funds as ainstrument for Combating Poverty. Washing-ton, D.C.

——. 1997. Social Investment Funds in LatinAmerica, Past Performance and Future Role.Washington, D.C.

ILO (International Labor Organization). 1997.Social Funds Employment and Gender Dimen-sions: Report on the Technical BrainstormingWorkshop. Women in Development and SocialGroups Unit. Geneva.

IMF (International Monetary Fund). 2001. Track-ing of Poverty-Reducing Public Spending inHeavily Indebted Poor Countries (HIPCs).IMF’s Fiscal Affairs Department and the WorldBank’s Poverty Reduction and Economic Man-agement Network. Washington, D.C.: TheWorld Bank.

Jorgensen, S., and J. Van Domelen. 1999. “Help-ing the Poor Manage Risk Better: The Role ofSocial Funds.” Paper presented at IADB Con-ference on Social Protection and Poverty,February 1999.

Jorgensen, S., M. Grosh, and M. Schacter (eds.).1992. Bolivia’s Answer to Poverty, EconomicCrisis, and Adjustment. The Emergency SocialFund. World Bank Regional and SectoralStudies. Washington, D.C.: The World Bank.

Kakhobwe, S. 1997. “Information Systems, Out-reach and Communication of Social Funds:The Case of Malawi Social Action Fund.”Paper prepared for the International Work-shop on Social Funds and Regional Consul-tations, Washington, D.C., World Bank.

Kariuki, M. 1998. “Social Fund Projects in the Eastand Southern Africa Region.” UNDP-World


1 8 0

Bank Water Sanitation Program, paper pre-sented at the Community Water and Sanita-tion Conference, Washington, D.C.

Katz, T., and J. Sara. 2000. Making Rural WaterSupply Sustainable: Recommendations froma Global Study. UNDP/World Bank Waterand Sanitation Program. Washington, D.C.:World Bank.

Kessides, C. 2000. “Decentralization and Com-munity Driven Development.” Paper pre-sented at the World Bank Africa RegionSeminar, Washington, D.C.

——. 1997. World Bank Experience with theProvision of Infrastructure Services for theUrban Poor: Preliminary Identification andReview of Best Practices. TWU-OR8. Wash-ington, D.C.: World Bank.

Khadiagala, L. 1995. Social Funds: Strengths,Weaknesses, and Conditions for Success. WorldBank ESP Discussion Paper Series No. 52.Washington, D.C.

Kwaja, A. I. 2000. Can Good Projects Succeed inBad Communities? Collective Action in theHimalayas. Cambridge, MA: Harvard University.

Klees, R., J. Godinho, M. Lawson-Doe. 1999.Health, Sanitation and Hygiene in RuralWater Supply and Sanitation Projects andother World Bank-Financed Projects. ECARegional Studies Program. Washington, D.C.:World Bank.

Lebo, J., and D. Schelling. 2001. Design andAppraisal of Rural Transport Infrastructure:Ensuring Basic Access for Rural Communities.World Bank Technical Paper 496. Washington,D.C..

Lecuit, L., J. Elder, C. Hurtado, F. Rantrua, K. Sib-lini, and M. Tovo. 1999. DeMIStifying MIS:Guidelines for Management Information Sys-tems in Social Funds. World Bank TechnicalPaper 443. Washington, D.C.

Levine, A. (ed.). 2000. Social Funds: Accom-plishments and Aspirations. Washington, D.C.:World Bank.

Lustig, N. 1997. The Safety Nets which Are NotSafety Nets: Social Investment Funds in LatinAmerica. Note prepared for the Conferenceon Governance, Poverty Eradication, andSocial Policy, Harvard University, November12–13.

Malena, C. 1997. NGO Involvement in World-Bank Financed Social Funds. World BankEnvironment Department Paper No. 052.Washington, D.C.

Marc, A., C. Graham, M. Schacter, and M.Schmidt. 1995. Social Action Programs andSocial Funds, A Review of Design and Imple-mentation in Sub-Saharan Africa. WorldBank Discussion Paper 274. Washington, D.C.

Mesa-Lago, C. 1993a. Recent Developments in SocialFunds and Safety Nets. UNCTAD. Genevea.

Mesa-Lago, C. 1993b. Safety Nets and SocialFunds to Alleviate Poverty: Performance, Prob-lems and Policy Options. Issues note pre-pared for UNCTAD Standing Committee onPoverty Alleviation, Geneva.

Milimo, J. 1994. “Report on Beneficiary Assess-ment Phase III.” Lusaka, Zambia.

Moser, C., and J. Holland. 1997. Urban Povertyand Violence in Jamaica. World Bank LatinAmerican and Caribbean Studies, Viewpoints.Washington, D.C.

Mosse, D. 1995. Peoples Knowledge in ProjectPlanning: The Limits and Social Conditionsof Participation in Planning Agricultural.London: Overseas Development Institute,Agricultural Research and Extension Network.

Narayan, D. 1995a. The Contribution of People’sParticipation: Evidence from 121 Rural WaterSupply Projects. Environmentally SustainableDevelopment Occasional Papers Series, No.1. Washington, D.C.: The World Bank.

Narayan, D. 1995b. Designing Community BasedDevelopment. World Bank Social Develop-ment Papers 7. Washington, D.C.

Narayan, D. and K. Ebbe. 1997. Design of SocialFunds: Participation, Demand Orientation,and Local Organizational Capacity. WorldBank Discussion Paper 375. Washington, D.C.

Nelson, J. 1999. Reforming Social Sector Gov-ernance: A Political Perspective. In J. Tulchinand A. Garland (eds.) The Politics and Ad-ministration of Social Development in LatinAmerica.

Newman, J., and others 2000. “An Impact Eval-uation of Education, Health and Water Sup-ply Investments of the Bolivian SocialInvestment Fund.” Washington, D.C., WorldBank. Photocopy.

B i b l i o g r a p h y

1 8 1

OED. 2001. 2000 Annual Review of Develop-ment Effectiveness: From Strategy to Results.Washington, D.C.: The World Bank.

——. 2000a. Poverty Reduction in the 1990s.An Evaluation of Strategy and Performance.Washington, D.C.: The World Bank

——. 2000b. “Utilization of Project Implemen-tation Units.” OED Evaluation Lessons Num-ber 1. Washington, D.C.

——. 1999. The Drive to Partnership: Aid Coor-dination and the World Bank. Washington,D.C.: The World Bank.

Ostrom, E. 2000. “Collective Action and the Evo-lution of Social Norms.” Journal of EconomicPerspectives 14(3): 1–21.

Ostrom, E., L. Schroeder, and S. Wynne. 1993.Institutional Incentives and Sustainable Devel-opment: Infrastructure Policies in Perspec-tive. Boulder: Westview.

Owen, D., and J. Van Domelen. 1998. Getting anEarful: A Review of Beneficiary Assessments ofSocial Funds. World Bank Social ProtectionDiscussion Paper 9816. Washington, D.C

Parker, A., and R. Serrano. 2000. PromotingGood Local Governance through Social Fundsand Decentralization. Washington, D.C.: TheWorld Bank.

Parker, R. and T. Skytta. 2000. “Rural Water Proj-ects. Lessons from OED Evaluations.” OEDWorking Paper 3. Washington, D.C.

Paxson, C., and N. Shady. 1999. “Do SchoolFacilities Matter? The Case of the PeruvianSocial Fund (FONCODES).” World Bank Pol-icy Research Working Paper 2229. Washing-ton, D.C.

Perkins, G., and R. Yemtsov. 2001. Armenia:Restructuring to Sustain Universal GeneralEducation. World Bank Technical Paper No.498. Washington, D.C.

Picciotto, R. 1995. Putting Institutional Eco-nomics to Work: From Participation to Gov-ernance. World Bank Discussion Paper 304.Washington, D.C.

——. 1992. “Participatory Development: Mythsand Dilemmas.” Policy Research WorkingPaper 930. Washington, D.C.: The World Bank.

Piriou-Sall, S. 1998. Decentralization and RuralDevelopment: A Review of Evidence. Wash-ington, D.C.: World Bank.

Platteau, J-P. 2000. “The Dilemma of Participa-tion with Endogenous Community Imper-fections, and Community Imperfections.”Paper for the Annual Bank Conference onDevelopment Economics., Washington, D.C.

Pouliquen, L. 1999. Rural Infrastructure from aWorld Bank Perspective: A Knowledge Man-agement Framework. Environmentally andSocially Development Network. Washington,D.C.: The World Bank.

Rao, V., and A. Ibanez. Forthcoming. “The SocialImpact of Social Funds in Jamaica: A Mixed-Methods Analysis of Participation, Targetingand Collective Action in Community DrivenDevelopment. Washington, D.C.: The WorldBank.

Ravallion, M., and G. Datt. 1994. “Transfer Ben-efits from Public Works Employment: Evi-dence from Rural India.” Economic Journal:The Journal of the Royal Economic Society.104: 1346–1469.

Ravallion, M., and G. Datt. 1995. Growth andPoverty in Rural India. Washington, D.C.:The World Bank.

Reddy, S. 1998. Social Funds in DevelopingCountries: Recent Experiences and Lessons.UNICEF Staff Working Paper EPP-EVL-98-002. New York.

Salmen, L. 1987. Listen to the People: Participant-Observer Evaluation of Development Projects.Oxford, U.K.: Oxford University Press.

Sara, J., and T. Katz. 1999. Making Rural WaterSupply Sustainable: Report on the Impact ofProject Rules. UNDP-World Bank Water andSanitation Program. Washington, D.C.: WorldBank.

Schady, N. 1999. “Seeking Votes: The PoliticalEconomy of Expenditures by the PeruvianSocial Fund (FONCODES), 1991–1995.” Pol-icy Research Working Paper 2166. Washing-ton, D.C.: World Bank.

Schmidt, M., and A. Marc. 1995. “Participation inSocial Funds.” Environment Department Paper004. Washington, D.C.: World Bank.

Schmidt, M., and A. Marc. 1998. “Social Invest-ment Funds in Transition Economies: Oppor-tunities and Risks.” Human DevelopmentDepartment, World Bank, Washington, D.C.Photocopy.


1 8 2

*Serrano, R. 2000. “The Institutional DevelopmentEffects of Social Funds: The Case of Argentina’sFOPAR.” Background paper for OED SocialFunds Review. World Bank Operations Eval-uation Department, Washington, D.C.

Stewart, F., and W. van der Geest. 1994. Adjust-ment and Social Funds: Political Panacea orEffective Poverty Reduction? EmploymentPapers 2. Geneva: ILO.

Subbarao, K., and others. 1997. Safety Net Pro-grams and Poverty Reduction: Lessons fromCross-Country Experience. Washington, D.C.:World Bank.

Tendler, J. 2000a. “Why Are Social Funds SoPopular?” In Local Dynamics in the Era ofGlobalization, S. Yusuf and others, eds. NewYork: Oxford University Press for the WorldBank.

——. 2000b. “Why Social Policy Is CondemnedTo a Residual Category Of Safety Nets, andWhat To Do About It: Thoughts on A ResearchAgenda for UNRISD.” Photocopy.

——. 1993. “New Lessons from Old Projects: TheWorkings of Rural Development in North-east Brazil.” World Bank Operations Evalua-tion Department, Washington, D.C.

Tendler, J., with R. Serrano. 1999. “The Rise ofSocial Funds: What Are They A Model of?”Paper presented at the World Bank Confer-ence on Evaluation and Poverty Reduction,June 14–15, Washington, D.C.

UNCTAD (United Nations Conference on Tradeand Development). 1993. Recent Develop-ments in Social Funds and Safety Nets.UNCTAD/PA/2. Standing Committee onPoverty Alleviation. Geneva.

Uribe, E. D. 1997. “Social Investment Funds andtheir Role in Reducing Poverty.” Paper pre-pared for the International Workshop onSocial Funds and Regional Consultations,World Bank, Washington, D.C.

Van Hoffman, A. 1997. Issues in Assessing theImpact of Social Investment: The Local Ini-tiatives Support Corporation. Cambridge, MA:Kennedy School of Government Case Pro-gram, Harvard University.

Wade, Robert. 1988. Village Republics: EconomicConditions for Collective Action in South India.Cambridge, U.K.: Cambridge University Press.

Weiss, C. 1998a. Evaluation, 2d ed. Upper Sad-dle River, NJ: Prentice Hall.

——. 1998b. Evaluation. Methods for StudyingPrograms and Policies. Upper Saddle River,NJ: Prentice Hall.

Weissman, J. 2001. Operating Instructions Included:A Review of Social Investment Fund OperationsManuals. World Bank Social Protection Dis-cussion Paper 0110. Washington, D.C.

Wiens, T., and M. Guadagni. 1998. Design Rulesfor Demand-Driven Rural Investment Funds:The Latin American Experience. World BankTechnical Paper No. 407. Washington, D.C.

Woolcock, M., and D. Narayan. 2000. “Social Cap-ital: Implications for Development Theory,Research, and Policy.” World Bank ResearchObserver 15(2): 225–49.

World Bank. Forthcoming. Letting CommunitiesTake the Lead: A Cross-Country Evaluation ofSocial Fund Performance (Social Funds 2000Impact Evaluation). Washington, D.C.

——. 2000a. World Development Report 2000/2001: Attacking Poverty. Washington, D.C.

——. 2000b. Promoting Good Local Governancethrough Social Funds and Decentralization.Social Protection Discussion Paper 0022.Washington, D.C.

——. 2000c. Including the Most Vulnerable:Social Funds and People with Disabilities.Social Protection Discussion Paper 0023.Washington, D.C.

——. 2000d. Social Protection Strategy: FromSafety Net to Spring Board. Washington, D.C.

——. 2000e. Social Funds: Accomplishmentsand Aspirations. Proceedings of the SecondInternational Conference on Social Funds,June 5–7, 2000. Washington, D.C.

——. 2000f. “Scaling Up Community Driven Devel-opment for Poverty Reduction.” CDD Group,World Bank, Washington, D.C. Photocopy.

——. 2000g. “Sourcebook on Community DrivenDevelopment in the Africa Region: Commu-nity Action Programs.” World Bank AfricaRegion, Washington, D.C. Photocopy.

——. 2000h. Spectrum: summer/fall 2000. WorldBank Social Protection Unit, Washington, D.C.

——.1999a. Casualties Between Social Capitaland Social Funds. Social Protection Discus-sion Paper 9908. Washington, D.C.

B i b l i o g r a p h y

1 8 3

——. 1999b. Beneficiary Assessment Manual forSocial Funds. Social Protection DiscussionPaper 9930. Washington, D.C.

——. 1999c. Coordinating Poverty AlleviationPrograms with Regional and Local Govern-ments: The Experience of the Chilean SocialFund. Social Protection Discussion Paper9933. Washington, D.C.

——. 1999d. Helping the Poor Manage Risk Bet-ter: The Role of Social Funds. Social ProtectionDiscussion Paper 9934. Washington, D.C.

——. 1998. The Initiative on Defining, Monitor-ing and Measuring Social Capital. Overviewand Program Description. Social DevelopmentFamily, Environmentally and SociallySustainable Development Network, Social Cap-ital Initiative Series, No. 1. Washington, D.C.


1 8 4

1 8 5

The Operations Evaluation Department (OED), anindependent evaluation unit reporting to the WorldBank’s Executive Directors, rates the developmentimpact and performance of all the Bank’s completedlending operations. Results and recommendations arereported to the Executive Directors and fed back intothe design and implementation of new policies andprojects. In addition to the individual operations andcountry assistance programs, OED evaluates the Bank’spolicies and processes.

Summaries of studies and the full text of the Précisand Lessons & Practices can be read on the Internet athttp://www.worldbank.org/html/oed

How To Order OED PublicationsOperations evaluation studies, World Bank discussionpapers, and all other documents are available from theWorld Bank InfoShop.

Documents listed with a stock number and price codemay be obtained through the World Bank’s mail orderservice or from its InfoShop in downtown Washington,D.C. For information on all other documents, contactthe World Bank InfoShop.

For more information about this study or OED’sother evaluation work, please contact ElizabethCampbell-Pagé or the OED Help Desk.

Operations Evaluation DepartmentPartnerships & Knowledge Programs (OEDPK)E-mail: [email protected]: [email protected]: (202) 458-4497Facsimile: (202) 522-3200

Ordering World Bank PublicationsCustomers in the United States and in territories notserved by any of the Bank’s publication distributors maysend publication orders to:

The World BankP.O. Box 960Herndon, VA 20172-0960Fax: (703) 661-1501Telephone: (703) 661-1580The address for the World Bank publication databaseon the Internet is: http://www.worldbank.org (selectpublications/project info).E-mail: [email protected] number: (202) 522-1500Telephone number: (202) 458-5454

The World Bank InfoShop serves walk-in customersonly. The InfoShop is located at:

701 18th Street, NWWashington, DC 20433, USA

All other customers must place their orders through theirlocal distributors.

Ordering by e-mailIf you have an established account with the World Bank,you may transmit your order by electronic mail on theInternet to: [email protected]. Please include youraccount number, billing and shipping addresses, the titleand order number, quantity, and unit price for eachitem.

OPERATIONS EVALUATION DEPARTMENT PUBLICATIONS

Study Series2001 Annual Review of Development Effectiveness: Making ChoicesAgricultural Extension: The Kenya ExperienceAgricultural Extension and Research: Achievements and Problems in National SystemsBangladesh: Progress Through PartnershipDeveloping Towns and Cities: Lessons from Brazil and the PhilippinesThe Drive to Partnership: Aid Coordination and the World BankFinancial Sector Reform: A Review of World Bank AssistanceFinancing the Global Benefits of Forests: The Bank’s GEF Portfolio and the 1991 Forest Strategy and Its ImplementationFiscal Management in Adjustment LendingIDA’s Partnership for Poverty ReductionIndia: The Challenges of DevelopmentIndia: The Dairy RevolutionInformation Infrastructure: The World Bank Group’s ExperienceInvesting in Health: Development Effectiveness in the Health, Nutrition, and Population SectorMainstreaming Gender in World Bank Lending: An UpdateNongovernmental Organizations in World Bank–Supported Projects: A ReviewPaddy Irrigation and Water Management in Southeast AsiaPoland Country Assistance Review: Partnership in a Transition EconomyPoverty Reduction in the 1990s: An Evaluation of Strategy and PerformancePromoting Environmental Sustainability in DevelopmentReforming Agriculture: The World Bank Goes to MarketSocial Funds: Assessing EffectivenessUganda: Policy, Participation, PeopleThe World Bank’s Experience with Post-Conflict ReconstructionThe World Bank’s Forest Strategy: Striking the Right BalanceZambia Country Assistance Review: Turning an Economy Around

Evaluation Country Case SeriesBosnia and Herzegovina: Post-Conflict ReconstructionBrazil: Forests in the Balance: Challenges of Conservation with DevelopmentCameroon: Forest Sector Development in a Difficult Political EconomyChina: From Afforestation to Poverty Alleviation and Natural Forest ManagementCosta Rica: Forest Strategy and the Evolution of Land UseEl Salvador: Post-Conflict ReconstructionIndia: Alleviating Poverty through Forest DevelopmentIndonesia: The Challenges of World Bank Involvement in ForestsUganda: Post-Conflict Reconstruction

ProceedingsGlobal Public Policies and Programs: Implications for Financing and EvaluationLessons of Fiscal AdjustmentLesson from Urban TransportEvaluating the Gender Impact of World Bank AssistanceEvaluation and Development: The Institutional Dimension (Transaction Publishers)Evaluation and Poverty ReductionMonitoring & Evaluation Capacity Development in AfricaPublic Sector Performance—The Critical Role of Evaluation

Multilingual EditionsAssessing Development Effectiveness: Evaluation in the World Bank and the International Finance CorporationAppréciation de l’efficacité du développement :L’évaluation à la Banque mondiale et à la Société financière internationaleDeterminar la eficacia de las actividades de desarrollo :La evaluación en el Banco Mundial y la Corporación Financiera InternacionalCôte d’Ivoire : Revue de l’aide de la Banque mondiale au paysPhilippines: From Crisis to OpportunityFilipinas: Crisis y oportunidadesRebuilding the Mozambique Economy: Assessment of a Development PartnershipReconstruir a Economia de Moçambique

http://www.worldbank.org/oed

OED PUBLICATIONS