The fectivenesofWorl B The Effectiveness S of World Bank ... · Indonesia: The Challenges of World...

THE WORLD BANK

THE WORLD BANK 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

The Effectiveness of World Bank Supportfor Community-Basedand -Driven Development

Th

eEffectiveness

ofWorld

Bank

Supportfor

Com

munity-B

asedand

-Driven

Developm

ent

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 drawnfrom evaluation findings.

OPERATIONS EVALUATION DEPARTMENT

Study SeriesAddressing the Challenges of Globalization: An Independent Evaluation of the World Bank's Approach to Global Programs2003 Annual Review of Development Effectiveness: The Effectiveness of Bank Support for Policy Reform2004 Annual Review of Development Effectiveness: The Bank's Contributions to Poverty ReductionAgricultural Extension: The Kenya ExperienceAssisting Russia’s Transition: An Unprecedented ChallengeBangladesh: Progress Through PartnershipBrazil: Forging a Strategic Partnership for Results—An OED Evaluation of World Bank AssistanceBridging Troubled Waters: Assessing the World Bank Water Resources StrategyCapacity Building in Africa: An OED Evaluation of World Bank Support The CGIAR: An Independent Meta-Evaluation of the Consultative Group on International Agricultural ResearchThe CIGAR at 31: An Independent Meta-Evaluation of the Consultative Group on International Agricultural ResearchCountry Assistance Evaluation Retrospective: OED Self-Evaluation Debt Relief for the Poorest: An OED Review of the HIPC InitiativeDeveloping Towns and Cities: Lessons from Brazil and the PhilippinesThe Drive to Partnership: Aid Coordination and the World BankEconomies in Transition: An OED Evaluation of World Bank Assistance Evaluating a Decade of World Bank Gender Policy: 1990-99The Effectiveness of World Bank Support for Community-Based and –Driven Development: An OED EvaluationEvaluation of World Bank Assistance to Pacific Member Countries, 1992-2002 Financial 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 ReductionImproving the Lives of the Poor Through Investment in CitiesIndia: The Dairy RevolutionInformation Infrastructure: The World Bank Group’s ExperienceInvesting in Health: Development Effectiveness in the Health, Nutrition, and Population SectorJordan: Supporting Stable Development in a Challenging RegionLesotho: Development in a Challenging EnvironmentMainstreaming Gender in World Bank Lending: An UpdateMaintaining Momentum to 2015? An Impact Evaluation of Interventions to Improve Maternal and Child Health and Nutrition Outcomes in Bangladesh The Next Ascent: An Evaluation of the Aga Khan Rural Support Program, PakistanNongovernmental Organizations in World Bank–Supported Projects: A ReviewPoland Country Assistance Review: Partnership in a Transition EconomyPoverty Reduction in the 1990s: An Evaluation of Strategy and PerformanceThe Poverty Reduction Strategy Initiative: An Independent Evaluation of the World Bank’s Support Through 2003Power for Development: A Review of the World Bank Group’s Experience with Private Participation in the Electricity SectorPromoting Environmental Sustainability in DevelopmentPutting Social Development to Work for the Poor: An OED Review of World Bank ActivitiesReforming Agriculture: The World Bank Goes to MarketSharing Knowledge: Innovations and Remaining ChallengesSocial Funds: Assessing EffectivenessTunisia: Understanding Successful Socioeconomic Development Uganda: 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 ForestsThe Poverty Reduction Strategy Initiative: Findings from 10 Country Case Studies of World Bank and IMF SupportUganda: 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 EditionsAllègement de la dette pour les plus pauvres : Examen OED de l’initiative PPTEAppréciation 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 paysFilipinas: Crisis y oportunidadesReconstruir a Economia de MoçambiqueСодействие России в переходе к рыночной экономике: беспрецедентная задача

All OED evaluations are available, in whole or in part, in languages other than English. For our multilingual selection, please visithttp://www.worldbank.org/oed

OED PUBLICATIONS

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 drawnfrom evaluation findings.

OPERATIONS EVALUATION DEPARTMENT

Study SeriesAddressing the Challenges of Globalization: An Independent Evaluation of the World Bank's Approach to Global Programs2003 Annual Review of Development Effectiveness: The Effectiveness of Bank Support for Policy Reform2004 Annual Review of Development Effectiveness: The Bank's Contributions to Poverty ReductionAgricultural Extension: The Kenya ExperienceAssisting Russia’s Transition: An Unprecedented ChallengeBangladesh: Progress Through PartnershipBrazil: Forging a Strategic Partnership for Results—An OED Evaluation of World Bank AssistanceBridging Troubled Waters: Assessing the World Bank Water Resources StrategyCapacity Building in Africa: An OED Evaluation of World Bank Support The CGIAR: An Independent Meta-Evaluation of the Consultative Group on International Agricultural ResearchThe CIGAR at 31: An Independent Meta-Evaluation of the Consultative Group on International Agricultural ResearchCountry Assistance Evaluation Retrospective: OED Self-Evaluation Debt Relief for the Poorest: An OED Review of the HIPC InitiativeDeveloping Towns and Cities: Lessons from Brazil and the PhilippinesThe Drive to Partnership: Aid Coordination and the World BankEconomies in Transition: An OED Evaluation of World Bank Assistance Evaluating a Decade of World Bank Gender Policy: 1990-99The Effectiveness of World Bank Support for Community-Based and –Driven Development: An OED EvaluationEvaluation of World Bank Assistance to Pacific Member Countries, 1992-2002 Financial 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 ReductionImproving the Lives of the Poor Through Investment in CitiesIndia: The Dairy RevolutionInformation Infrastructure: The World Bank Group’s ExperienceInvesting in Health: Development Effectiveness in the Health, Nutrition, and Population SectorJordan: Supporting Stable Development in a Challenging RegionLesotho: Development in a Challenging EnvironmentMainstreaming Gender in World Bank Lending: An UpdateMaintaining Momentum to 2015? An Impact Evaluation of Interventions to Improve Maternal and Child Health and Nutrition Outcomes in Bangladesh The Next Ascent: An Evaluation of the Aga Khan Rural Support Program, PakistanNongovernmental Organizations in World Bank–Supported Projects: A ReviewPoland Country Assistance Review: Partnership in a Transition EconomyPoverty Reduction in the 1990s: An Evaluation of Strategy and PerformanceThe Poverty Reduction Strategy Initiative: An Independent Evaluation of the World Bank’s Support Through 2003Power for Development: A Review of the World Bank Group’s Experience with Private Participation in the Electricity SectorPromoting Environmental Sustainability in DevelopmentPutting Social Development to Work for the Poor: An OED Review of World Bank ActivitiesReforming Agriculture: The World Bank Goes to MarketSharing Knowledge: Innovations and Remaining ChallengesSocial Funds: Assessing EffectivenessTunisia: Understanding Successful Socioeconomic Development Uganda: 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 ForestsThe Poverty Reduction Strategy Initiative: Findings from 10 Country Case Studies of World Bank and IMF SupportUganda: 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 EditionsAllègement de la dette pour les plus pauvres : Examen OED de l’initiative PPTEAppréciation 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 paysFilipinas: Crisis y oportunidadesReconstruir a Economia de MoçambiqueСодействие России в переходе к рыночной экономике: беспрецедентная задача

All OED evaluations are available, in whole or in part, in languages other than English. For our multilingual selection, please visithttp://www.worldbank.org/oed

OED PUBLICATIONS

The Effectiveness of World Bank Supportfor Community-Basedand -Driven Development

An OED Evaluation

2005The 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/cbdcdd

i i

Printed on Recycled Paper

World Bank InfoShop

E-mail: [email protected]

Telephone: 202-458-5454

Facsimile: 202-522-1500

Operations Evaluation Department

Knowledge Programs and Evaluation Capacity

Development (OEDKE)

E-mail: [email protected]

Telephone: 202-458-4497

Facsimile: 202-522-3125

© 2005 The International Bank for Reconstruction and Development / The World Bank1818 H Street, NWWashington, DC 20433Telephone 202-473-1000Internet www.worldbank.orgE-mail [email protected]

1 2 3 4 09 08 07 06 05

All rights reservedManufactured in the United States of America

The findings, interpretations, and conclusions expressed here are those of the author(s) and do not necessarily reflect theviews 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,denominations, and other information shown on any map in the work do not imply on the part of the World Bank anyjudgment of the legal status of any territory or the endorsement or acceptance of such boundaries.

Rights and PermissionsThe material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permissionmay be a violation of applicable law. The World Bank encourages dissemination of its work and will normally grantpermission promptly.

For permission to photocopy or reprint any part of this work, please send a request with complete information to theCopyright 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: © Giacomo Pirozzi/Panos Pictures.

ISBN-10: 0-8213-6390-5 ISBN-13: 978-0-8213-6390-4e-ISBN: 0-8213-6391-3 DOI: 10.1596/978-0-8213-6390-4

i i i

vii Acknowledgments

ix Foreword

xi Executive Summary

xvii Acronyms and Abbreviations

1 1 Origin, Scope, Design, and Methodology of Evaluation2 Scope2 Design4 Methodology4 Some Limitations of the Study Design

7 2 The CBD/CDD Portfolio7 Temporal, Regional, and Sectoral Distribution of the Portfolio7 Ways That CBD/CDD Projects Differ from Those in the

Non-CBD/CDD Portfolio

11 3 Outcome of Bank-Supported CBD/CDD Projects11 Outcome Ratings11 Relevance15 Efficacy22 Targeting the Poor Is Not Enough to Reach the Poor25 Efficiency

29 4 Institutional Enhancement and Sustainability29 Institutional Enhancement36 Sustainability

41 5 Bank Operational Policy Requirements, Processes, and CBD/CDD Interventions41 The Bank Has Attempted to Adapt Its Policies to Design and

Implement CBD/CDD Projects45 Bank Capacity to Ensure Effective Implementation of

CBD/CDD Interventions

Contents

49 6 Conclusions

51 7 Recommendations

53 Annexes55 Annex A: Definition of “Community” 57 Annex B: Results Chain for World Bank CDD Projects59 Annex C: World Bank Guidance on Key Design Principles for CDD63 Annex D: Study Framework, Methods, and Instruments67 Annex E: The Universe of CBD/CDD Projects and Its Distribution71 Annex F: Sample of CBD/CDD and CDD Projects77 Annex G: OED Ratings of Completed Projects81 Annex H: Focus on CBD/CDD and Related Aspects in Bank

and Borrower Strategy 85 Annex I: Central Government and Local Government Surveys89 Annex J: Efficiency95 Annex K: Nkayi District Formal and Informal Systems97 Annex L: Results of Bank Staff Survey103 Annex M: Methodology for Community-Level Data Collection and Analysis119 Annex N: Enhancing Community Capacities141 Annex O: Poverty Targeting143 Annex P: Evidence from Fieldwork on Sustainability147 Annex Q: Safeguard Thematic Study: A Summary151 Annex R: Advisory Committee Comments159 Annex S: Management Response173 Annex T: Chairman’s Summary: Committee on Development Effectiveness

177 Endnotes

191 References

Boxes3 1.1: Primary Data Collection for the Evaluation of CBD/CDD Projects15 3.1: Change in Social Capital and Empowerment as a Means of

Assessing the Community Capacity–Enhancing Impact of BankCBD/CDD Interventions

17 3.2: Highly Satisfactory Aspects of Design and Implementation in Selected CBD/CDD Projects

18 3.3: Local Champions Can Effectively Use Bank CBD/CDD Funds:A Case from Ghana

24 3.4: Example of Sophisticated Poverty Targeting in Two Recent Vietnam CDD Projects

32 4.1: Inconsistent Strategies: The Road Sector Experience in Nepal33 4.2: Bank CDD Projects Have Added to the Proliferation of

Municipal Councils in Rio Grande do Norte (Brazil)40 4.3: Why Formal Groups Do Not Last Long43 5.1: The Meaning of the Environmental Categories44 5.2: The Fiduciary Challenge: The Case of Vietnam48 5.3: Some Monitoring and Evaluation Features to Emulate:

The Case of the Indonesia Kecamatan Development Project

THE EFFECT IVENESS OF WORLD BANK SUPPORT FOR COMMUNITY-BASED AND -DRIVEN DEVELOPMENT

i v

Figures2 1.1: The Universe of CBD/CDD Projects8 2.1a: Bank Commitments to Projects with CBD/CDD

Approaches Have Grown8 2.1b: Africa Has the Largest Share of Projects with CBD/CDD Approaches9 2.2: Communities Have More Responsibility in CDD Projects12 3.1: Outcome Ratings of CBD/CDD Projects Have Been Better

Than Those of Non-CBD/CDD Projects13 3.2a: CBD/CDD Projects in the Latin American and Caribbean Region

Have Done Better Than Projects in Other Regions13 3.2b: CBD/CDD Projects in Education and Social Protection Do Better

Than Projects in Health and Rural Development14 3.3: CBD/CDD Is Increasingly Important in Country Programs16 3.4: Community Priorities Are Better Met When There Is Greater Choice20 3.5: Focus Groups Report Significant Decision Making by Local Leaders21 3.6: Subproject Cycle Is Too Short for Meaningful Enhancement

of Community Capacity26 3.7: CBD/CDD Projects Cost the Bank More Than Non-CBD/CDD Projects34 4.1: Low-Income Countries Work with More Partners36 4.2: Sustainability Has Been Consistently Lower for CBD/CDD

Projects but Is Improving

Tablexii ES.1: Overview of CBD/CDD Strengths and Weaknesses

C O N T E N T S

v

v i i

Acknowledgments

The report was prepared by a core team con-sisting of Nalini Kumar (Task Manager), BarbaraPozzoni, Anju Vajja, and George Garner Woodall.Additional team members were Ewa Gradzka,Kavita Mathur, Sonali Moonesinghe, Ridley Nel-son, Floribert Ngaruko, Camilo Gomez Osorio,and Peter Whitford. In addition to the core team,data collection was undertaken by Dr. AlbertoCoelho Gomes Costa and his team of surveyorsin Brazil, the staff and students of the Centre forDevelopment Economics, Delhi School of Eco-nomics in India, and Dr. Roch Mongbo and histeam of surveyors in Benin. William Hurlbut ed-ited the original report and Caroline McEuenedited the manuscript for publication. TomasCaspellan, Soon-Won Pak, and Helen Phillip pro-vided administrative support. Julius Gwyer pro-vided technical support for the maintenance ofthe Web site. Jacintha Wijesinghe and EtsegenetNigussie advised on the budget. The reportbenefited considerably from the guidance pro-vided by Alain Barbu, Ajay Chhibber, Nils Fos-tvedt, Patrick Grasso, Gregory Ingram, and KylePeters.

The External Advisory Panel for the reviewconsisted of Dr. Norman Uphoff, Director of theCornell International Institute for Food, Agri-culture and Development and Professor of Gov-ernment at Cornell University; Mr. PaiboonWattanasiritham, Chairman of the CommunityOrganizations Development Institute (CODI) in

Thailand; and Professor Robert Chambers, Re-search Associate at the Institute of DevelopmentStudies at the University of Sussex (UK).

Colleagues inside the Bank provided com-ments at various stages. They include: RuthAlsop, Hans Binswanger, Victor Botini, Ed Bres-nyan, Mark Cackler, Robert Chase, Luis Coirolo,Monica Das Gupta, Gershon Feder, ScottGuggenheim, Jeffrey Gutman, Steen Jorgensen,Christine Kessides, Ghazala Mansuri, AlexanderMarc, Robin Mearns, Deepa Narayan, DanielOwen, Vijayendra Rao, Martin Ravallion, JohnRedwood, Louise Scura, Susan Wong, MarkWoodward, and Michael Woolcock.

We would like to thank the staff of the WorldBank Country Offices in Benin, Brazil, Egypt,India, Nepal, and Vietnam for all their support.OED also gratefully acknowledges the input ofnumerous community members and other stake-holders across various countries who patientlyparticipated in interviews and surveys and pro-vided valuable perspectives.

We are grateful for comments from the fol-lowing OED staff: Martha Ainsworth, Bill Bat-taile, Soniya Carvalho, Victoria Elliott, FareedHassan, Edoardo Masset, Janardan Prasad Singh,Denise Vaillancourt, and Howard White.

This review was greatly enhanced by the gen-erous support and input of several donors. Sub-stantial feedback and financial support from theSwiss Agency for Development and Coopera-

tion—in particular Laurent Ruedin, Joerg Frieden,Robert Groeli, Walter Meyer, and Ursula Funk—as well as contributions from the Norway MFA

Partnership and the Norwegian ConsultancyTrust Fund, are gratefully acknowledged.


v i i i

Acting Director-General, Operations Evaluation: Ajay ChhibberActing Director, Operations Evaluation Department: R. Kyle Peters

Manager, Sector, Thematic, and Global Evaluation: Alain BarbuTask Manager: Nalini Kumar

i x

Foreword

This report analyzes the effectiveness of the World Bank’s lending sup-port for the growing area of community-based development (CBD) andcommunity-driven development (CDD). The latter supports the em-

powerment of the poor by giving communities control over subproject re-sources and decisions, while CBD gives communities less responsibility andemphasizes collaboration, consultation, or sharing information with them onproject activities. Since the late 1990s, the focus of Bank-supported CBD/CDDprojects has shifted toward CDD, though many CDD projects also include CBDcomponents.

The share of projects that include a CBD/CDDcomponent has grown from about 2 percent infiscal 1989 to 25 percent in 2003. Over the1994–2003 period, the outcome ratings ofCBD/CDD projects have been better than thosefor non-CBD/CDD projects. Much more successhas been achieved in CBD/CDD projects onquantitative goals such as construction of infra-structure than on qualitative goals such as ca-pacity enhancement. Sustainability ratings forthe projects have improved over time, but thereis considerable room for improvement. Bank in-terventions have often failed to provide the con-sistent, long-term support needed for an activityto become sustainable (for example, in a forestryproject, support should be provided until the for-est starts yielding adequate returns from tim-ber and non-timber products).

The study finds that Bank-supportedCBD/CDD projects have typically performed bet-ter on capacity enhancement—an important ob-jective of the projects—when they havesupported indigenously matured participatory ef-forts or when the Bank has provided sustained,long-term support to communities beyond thelength of a single subproject.

The distribution of costs and benefits—to theinstitution, the borrower, or the communities—related to undertaking CBD/CDD projects hasnot been systematically assessed by the Bank. TheOED study finds that CBD/CDD projects aremore expensive to prepare and supervise thanother Bank projects. Governments also spendconsiderable resources to put a participatoryapproach in place. While a participatory ap-proach lowers the cost to the government for ser-

vice-delivery infrastructure, the communitiesbear a significant part of this cost burden.

CBD/CDD projects have increased access toservice delivery infrastructure such as schoolsand health centers for remote communities. Inaddition, several CBD/CDD projects in conflictand post-conflict countries have helped in re-habilitating infrastructure and have providedsignificant employment benefits to the localpopulation. However, as with other projects, in-creased access to infrastructure does not alwaystranslate into effective service delivery. Further,the poorest may not always benefit from theseprojects. There is little hard evidence to date onthe poverty-reducing and community capac-ity–enhancing impact of the projects.

Bank-supported CBD/CDD projects have en-hanced the capacity of government institutions toimplement participatory interventions, but fewborrower governments have adopted the ap-proach more widely in their development pro-grams.

Adaptation of operational policies and decen-tralization of the Bank to field offices have en-hanced the Bank’s capacity to implementCBD/CDD interventions, but additional changesare needed. Weaknesses in monitoring and eval-uation, the need for development of adequateguidelines for staff on safeguards for CDD projects,and the short time span of the Bank’s subprojectcycle constrain the Bank’s capacity to implementCBD/CDD projects. These projects, most partic-ularly CDD, also continue to pose a challenge forboth safeguard and fiduciary compliance.

Going forward, the Bank could do a numberof things to improve its effectiveness for com-munity-based and community-driven develop-ment. The study makes three recommendations:

• In undertaking new CBD/CDD projects, theBank should analyze whether it is building onexisting local initiatives or starting a new pro-gram in a country and give priority to the for-mer. In the latter case, the Bank should tailorthe initiative to the country and communitycontext and undertake selective, rigorous im-pact assessments of its projects before scalingup.

• The Bank needs to strengthen operationalguidance for the application of safeguardpolicies and fiduciary oversight of CBD/CDDprojects and for cost-benefit analysis and mon-itoring and evaluation (M&E) systems, and itshould commission an audit of the fiduciaryaspects of a representative sample of CDDprojects to be submitted to the Board withina year.

• CBD/CDD projects need to be integrated witha country’s assistance strategy. Future CountryAssistance Strategies (CASs) should show howthey have analyzed and addressed linkagesnot only between CBD/CDD projects, but alsobetween CBD/CDD and relevant non-CBD/CDD projects. Such analysis should alsoaddress whether arrangements for CBD/CDDproject implementation are made at the ex-pense of local government capacity develop-ment.


x

Ajay ChhibberActing Director-General, Operations Evaluation

x i

Executive Summary

Participatory approaches that involve local communities in their owndevelopment have gained substantial support among internationaldonors over the past quarter-century and have become increasingly

important in the work of the World Bank. Community participation is an ap-proach to development that can be used with any Bank lending instrumentand across sectors. Projects can involve communities in different ways—by shar-ing information, consulting, collaborating, or empowering them. The processof involving communities in project activities is also expected to contributein most cases to community capacity enhancement.

The World Bank’s support for communityparticipation has been manifested in the de-sign and implementation of either com-munity-based development (CBD) orcommunity-driven development (CDD)projects. Although the literature does not clearlydistinguish CBD from CDD,1 there is increasingconsensus that CDD projects give communitiescontrol over resources and decisions in the de-sign and implementation of subprojects. CBDprojects, however, give communities compara-tively less responsibility and emphasize collab-oration, consultation, and information sharingwith them. Since the late 1990s, the focus ofsuch Bank-supported projects has shifted to-ward CDD, although many CDD projects also in-clude CBD components.

Interest in community empowermentemerged in large part because donors, in-

cluding the World Bank, were impressed bythe poverty-reducing effects of local ini-tiatives that developed independently inseveral countries. In these islands of success,local communities had taken control of theirlives as a result of independent forms of socialaction. In its ongoing effort to reduce poverty,the Bank has emulated these local initiatives byattempting to enhance community capacity bybuilding social capital and fostering empower-ment in communities through its projects. Re-search into the multidimensional nature ofpoverty has further reinforced the importanceof empowerment. Today, the Bank’s StrategicFramework identifies empowering poor peopleto participate in development as one of the twopriorities in the fight against poverty.

Interest in a community-based approachis also predicated on a belief that not only

would it lead to better allocation of re-sources to help communities, but wouldalso lead to reduced corruption and mis-use of resources, and thereby more devel-opment assistance would reach the poor.Community involvement is expected to increasetransparency and accountability by working di-rectly with the ultimate beneficiaries, especially

where state capacity is weak or has been weak-ened by conflict and other factors.

Evaluation FindingsThe Bank has not, until recently, system-atically identified and tracked its portfolioof CBD/CDD projects,2 and therefore haslacked a comprehensive understanding of


x i i

Outcome ratings are generally better for CBD/CDD than for non-

CBD/CDD projects. Much more success has been achieved in

CBD/CDD projects on quantitative goals, such as the construction of

infrastructure, than on qualitative goals, such as capacity enhance-

ment or quality of training.

Borrower officials believe that a participatory approach can con-

tribute to poverty alleviation.

CBD/CDD projects help lower the cost to government of delivering

infrastructure.

They have increased access of remote communities to service-delivery

infrastructure such as schools, health centers, and the like.

Capacity-enhancement effort in a CBD/CDD project has been more

successful when a Bank project supports indigenously matured ef-

forts or provides sustained, long-term support to communities beyond

a Bank subproject cycle.

Sustainability ratings have improved over time.

CBD/CDD projects have enhanced government capacity to implement

participatory interventions.

Adaptation of Bank policies and decentralization to field offices have

enhanced Bank capacity to implement CBD/CDD projects.

The Bank’s project M&E systems do not allow systematic assessment

of the capacity-enhancing impact of CBD/CDD interventions. It is often

assumed that meeting the quantitative goals will automatically fulfill

the qualitative goal—for example, holding a certain number of train-

ing courses is expected to enhance capacity.

Borrower officials do not necessarily believe that community control over

decisions and resources in a subproject is the best means of engaging

communities.

Communities bear an increased share of the burden for service deliv-

ery infrastructure.

But the poorest may not always benefit from CBD/CDD projects.

The individual subproject cycle is too short to sustainably enhance

community capacity where it is weak or does not exist.

Infrastructure and services have been difficult to sustain beyond the Bank

presence because of a lack of resources from the government and com-

munities to ensure their operation and maintenance.

Few governments appear to have adopted the CBD/CDD approach more

widely in their own development programs.

More changes are needed to improve fiduciary and safeguard compli-

ance in CBD/CDD projects.

Strengths of Bank CBD/CDD projects Weaknesses of Bank CBD/CDD projects

Table ES.1: Overview of CBD/CDD Strengths and Weaknesses3

the evolution and scope of its work in thisfield. Although the Bank has been supportingsuch projects for more than a quarter-century, ithas no database that has tracked these projectssince their inception. It is only recently (for proj-ects approved from fiscal year 2000 onward) thatthe Bank has started maintaining a database.

Despite a rapid increase in lending, mostBank-supported CBD/CDD projects havenot yet been subjected to rigorous evalua-tion.4 The share of projects in the Bank’s port-folio that include a CBD/CDD component grewfrom about 2 percent in fiscal 1989 to 25 percentin 2003. However, as yet there is very little hardevidence on the impact of these projects in re-ducing poverty or enhancing community capac-ity. This is because of: (i) the Bank’s failure toinclude baseline surveys in most of the projectscompleted thus far and (ii) inadequacies in thedesign of the Bank’s project monitoring and eval-uation systems to systematically assess changesin community capacity resulting from the Bankintervention. Some recent CBD/CDD projectsare setting up baselines, but the results will notbe available for several years. To fill these gaps,in the absence of baseline data, this evaluationused a pragmatic approach based on householdsurveys to assess the association betweenCBD/CDD projects and change in communities’capacity and drew on a variety of sources for in-sights into the projects’ poverty impact.

OutcomeOutcome ratings of Bank-supportedCBD/CDD projects were, on average, bet-ter than those for non-CBD/CDD projectsbetween 1994 and 2003. The education sec-tor has had the highest percentage of projectsrated satisfactory on outcome, followed by proj-ects in the transport, urban development, and so-cial protection sectors. The rural developmentsector, with the largest CBD/CDD portfolio, is abelow-average performer on outcome in the ag-gregate, as are projects under the water supply,health, and environment sectors. Between 1999and 2003, the outcome rating for CBD/CDD proj-ects in post-conflict countries was better than theoutcome rating for CBD/CDD projects in non-conflict countries.

The borrower government officials in-terviewed for this review were convincedthat a participatory approach can con-tribute to poverty alleviation in their coun-tries. However, they did not necessarily believethat allowing community control over invest-ment decisions and resources in a Bank sub-project is the best means of engagingcommunities. This appears to be partly becausegovernment officials have concerns about the ca-pacity of communities to manage resources, butalso partly because they feel threatened by de-volution of complete control over decisions andresources to communities.

The Bank has not systematically andrealistically assessed the costs and benefitsof undertaking CBD/CDD projects to theinstitution, the borrower, or the commu-nities.5 CBD/CDD projects are more expensivethan non-CBD/CDD projects for the Bank toprepare and supervise, and there are substantialcosts in time spent by the borrower in puttinga participatory approach in place. WhileCBD/CDD projects have helped lower the costto governments for delivering service delivery in-frastructure, the communities now bear an in-creased part of the cost of that infrastructure. Theinsufficient focus on costs and benefits inCBD/CDD projects, especially on measures ofpoverty impact, has prevented convincing com-parisons with non-CBD/CDD investments andpolicy and institutional reform programs.

Much more success has been achievedin CBD/CDD projects on quantitative goals,such as the construction of infrastructure,than on qualitative goals, such as capacityenhancement or quality of training. How-ever, without baseline data, and without ap-propriate indicators, it is often assumed thatmeeting the quantitative goals will automati-cally fulfill the qualitative goal—for example,holding a certain number of training courses isexpected to enhance capacity. That said, sinceCBD/CDD projects have supported construc-tion of infrastructure in scattered communities,they have also increased access to infrastruc-ture for schools, health centers, and the like forthe communities in which they intervene. Sev-eral of the CBD/CDD projects in conflict and

E X E C U T I V E S U M M A R Y

x i i i

post-conflict countries have met quantitativetargets for infrastructure rehabilitation and havealso provided substantial employment benefitsto the local population. However, as with otherBank projects, increased access to infrastruc-ture does not always translate into effectiveservice delivery because of the inadequacy ofcomplementary inputs such as teachers, doctors,and medicines.

Bank-supported CBD/CDD projects havehad much more success with communitycapacity enhancement when they have sup-ported indigenously matured participa-tory efforts or when they have providedsustained, long-term support to commu-nities. The one year of a typical subproject cycleis generally too short a time to bring about thekind of enhancement of community capacitythat is envisioned in Bank-supported CBD/CDD,particularly CDD projects; it is sufficient to allowsuccessful subproject execution, but not to con-sistently enhance community capacity.

SustainabilitySustainability ratings have improved overtime for both CBD/CDD and non-CBD/CDDprojects, although several concerns re-main. Infrastructure and other activities sup-ported by the Bank’s CBD/CDD projects havebeen difficult to sustain beyond the Bank pres-ence because of a lack of the needed resourcesfrom the government and communities to ensuretheir operation and maintenance. More broadly,Bank projects have often failed to provide con-sistent long-term support for an activity to be-come sustainable (for example, in a forestryproject, support should be provided until the for-est starts yielding adequate returns from tim-ber and non-timber products).

Institutional Development Impact Bank-supported CBD/CDD projects haveenhanced the capacity of governmentinstitutions to implement participatoryinterventions, but so far few borrower gov-ernments appear to have adopted the ap-proach more widely in their owndevelopment programs. At the country levelthe Bank’s approach to promoting government

decentralization under various CBD/CDD proj-ects has not always been consistent. In othercases, the ad hoc parallel arrangements set upto implement Bank projects have hindered thelong-run enhancement of local government ca-pacity.

The Bank’s Operational PolicyRequirements and ProcessesIt is difficult to ensure fiduciary and safe-guard compliance in CBD/CDD projects.It is easier for the Bank to monitor resource useand comply with safeguards in non-CBD/CDD in-vestments, such as large bridges or a powerplant, than where small subprojects are being im-plemented by hundreds of remote communi-ties in scattered locations. Over time, adaptationof operational policies and decentralization of theBank to field offices have enhanced the Bank’scapacity to implement CBD/CDD projects, butadditional changes are needed. Among the issuesthat need to be addressed are: weaknesses inmonitoring and evaluation (particularly with re-gard to monitoring progress on community ca-pacity enhancement), development of adequateguidelines for staff on safeguards for CDD proj-ects, and the short time span of the Bank’s sub-project cycle.

In the end, the Bank should be aware that itis largely trying to use a single financial channel—project financing—to bring about changes inempowerment and social capital, which are af-fected by a long history of social, cultural, andpolitical forces embedded in the societies inwhich the Bank is trying to support develop-ment. Sometimes, the Bank support works well,especially when it supports existing grass rootsinitiatives. But when the Bank tries to initiate em-powerment and enhance social capital throughCBD/CDD projects, it is often not enough—orcan even be counterproductive if the better-offsections of the community gain more than theless-well-off. Where the Bank supports the cre-ation of “temporary” arrangements for the im-plementation of CBD/CDD projects at the locallevel, they could further undermine long-termcapacity building efforts and should be carefullyconsidered, especially if financing is provided foronly a few years.


x i v

RecommendationsGiven the mixed and limited evidence on the im-pacts of CBD/CDD projects—particularly in termsof poverty reduction and empowerment—andquestions about sustainability and safeguard andfiduciary compliance, the Bank should approachfuture CBD/CDD projects, particularly CDD,with greater care. In countries where the Bankis already supporting a CDD program, the insti-tution needs to rigorously assess the povertyand institutional development impact of its proj-ects before scaling them up. A cautious approachwould be especially important in countries orareas where the Bank is just beginning to sup-port CDD. In its future assistance to CBD/CDD,the Bank should:

At the corporate level, strengthen opera-tional guidance and management over-sight. • The Bank should provide operational guid-

ance for the application of Bank safeguardpolicies and fiduciary oversight of CBD/CDDprojects and for the strengthening of cost-ben-efit analysis and M&E systems and should com-mission an audit of the fiduciary aspects of arepresentative sample of CDD projects for sub-mission to the Board within a year.

At the country level, design the CBD/CDDprogram as an integral part of the overallassistance strategy and carry out periodic

assessment of ongoing CBD/CDD projectsto ensure relevance and effectiveness of theprogram to the country context. • Future CASs should show how they have ana-

lyzed and addressed linkages, not only amongvarious CBD/CDD projects to be undertakenin the country, but also among CBD/CDD andrelevant non-CBD/CDD projects. In particu-lar, the analysis should address whetherarrangements for CBD/CDD project imple-mentation come at the expense of local gov-ernment capacity development.

At the project level, the Bank should givepriority to helping countries build up ex-isting indigenously matured initiatives;where there are no such existing initia-tives, the Bank should tailor its project tothe country and community context, whileundertaking selective, rigorous impact as-sessments to ensure learning.• For any new CBD/CDD project, the Bank

should analyze (using existing processes, suchas social assessments) whether it is building onindigenously matured initiatives or attemptingto begin a CDD program in a country, andthen tailor the intervention to local capacity;and the Bank should also selectively undertakerigorous impact assessments upon comple-tion of its ongoing CBD/CDD projects to learnfor the future.

E X E C U T I V E S U M M A R Y

x v

x v i i

ACRONYMS AND ABBREVIATIONS

AFR Africa RegionAIDS Acquired immunodeficiency syndrome APL Adaptable Program LendingASIP Agriculture Sector Investment Project (Ghana)CA Community associationCAS Country Assistance StrategyCAE Country Assistance EvaluationCBD Community-based developmentCDD Community-driven developmentCBRIP Community-Based Rural Infrastructure Project (Vietnam)CFAA Country Financial Accountability Assessment ReportCIDSS Comprehensive and Integrated Delivery of Social Services (Philippines)CPAR Country Procurement Assessment ReportDFID Department for International Development (U.K.)DOLIDAR Department of Local Infrastructure Development and Agricultural Roads

(Nepal)EA Environmental assessmentEAP East Asia and the Pacific RegionECA Europe and Central Asia RegionERR Economic rate of returnESSD Environmentally and Socially Sustainable Network (World Bank)FI Financial intermediaryFUMAC Fundo Municipal de Apoio Comunitario (municipal-community scheme, Brazil)FUMAC-P Conselho Municipal de Apoio Comunitario – Piloto (Pilot Municipal

Community Schemes, Brazil)FY Fiscal yearHIV Human immunodeficiency virusHNP Health, Nutrition, and Population (sector)IAD Internal Auditing DepartmentICR Implementation Completion ReportID Institutional developmentIDB Inter-American Development BankIFAD International Fund for Agricultural DevelopmentJFM Joint forest managementKDP Kecamatan Development Project (Indonesia)LAC Latin America and the Caribbean RegionLIL Learning and Innovation LoanM&E Monitoring & evaluationMAP Multi-Country HIV/AIDS ProgramMDG Millennium Development Goal

MIS Monitoring and information systemMNA Middle East and North Africa RegionNGO Nongovernmental organizationNMPRP Northern Mountains Poverty Reduction Project (Vietnam)O&M Operation and maintenanceOED Operations Evaluation DepartmentOD Operational DirectiveOP Operational PolicyOPCS Operational Policy and Country ServicesPAC Programa de Apoio Comunitario (state-community schemes)PAD Project Appraisal DocumentPPAR Project Performance Assessment ReportPRSP Poverty Reduction Strategy PaperPSR Project Status ReportQAG Quality Assurance GroupQSA Quality of Supervision AssessmentRPAP Rural Poverty Alleviation Program (Brazil)RPRP Rural Poverty Reduction ProjectSAR South Asia RegionSEWA Self-Employed Women’s AssociationVLPA Village-Level Participatory ApproachZAMSIF Zambia Social Investment Fund


x v i i i

1

Origin, Scope, Design,and Methodologyof the Evaluation

Participatory approaches that involve communities in their own devel-opment have gained substantial support among international donors overthe past quarter-century and have become increasingly important in

the work of the World Bank.1 Community participation is an approach todevelopment that can be used with any Bank lending instrument and acrosssectors. Projects can involve communities in different ways—by sharinginformation, consulting, collaborating, or empowering them.2

The World Bank’s support for these participatoryapproaches has been largely manifested in thedesign and implementation of either community-based development (CBD) or community-drivendevelopment (CDD) projects. Although the lit-erature makes no clear distinction between theCBD and CDD approaches, there is increasingconsensus that CDD projects further the Bank’ssupport for empowerment by, inter alia, givingcommunities control over resources and deci-sions in the context of a Bank project. These proj-ects are distinguished from CBD, where thecommunities have less responsibility. Instead,the emphasis is on collaboration, consultation,and sharing information with the communitiesabout the project. Since the late 1990s, the focusof Bank-supported CBD/CDD projects has shiftedtoward CDD, though many CDD projects alsoinclude CBD components.

Interest in community empowermentemerged in large part because donors, includ-ing the World Bank, were impressed by the

poverty-reducing effects of local initiatives thatdeveloped independently in several countries.In these islands of success, local communitieshad taken control of their lives as a result of in-dependent forms of social action. In its ongoingeffort to reduce poverty, the Bank has emulatedthese local initiatives by attempting to enhancecommunity capacity by building social capital andfostering empowerment in communities throughits projects. Research into the multidimensionalnature of poverty has further reinforced the im-portance of empower-ment. Today, the Bank’sStrategic Frameworkidentifies empoweringpoor people to partici-pate in development byinvesting in them as oneof the two basic priori-ties in the fight againstpoverty (World Bank2001b).

11

The Bank’s Strategic

Framework identifies

empowering poor

people to participate in

development by investing

in them as one of the two

basic priorities in the

fight against poverty.

2


Interest in a community-based approach isalso predicated on a feeling that not only wouldit lead to better allocation of resources to helpcommunities, but also to reduced corruptionand less misuse of resources, and thereby moredevelopment assistance would reach the poor.Community involvement would increase trans-parency and accountability by working directlywith the ultimate beneficiaries, especially wherestate capacity is weak or has been weakened byconflict and other factors.

While Bank lendingfor CBD and CDD ap-proaches has increasedsignificantly over theyears, recent reports bythe Bank’s research andsocial development de-partments (Mansuri andRao 2004; Wassenich andWhiteside 2004) note thelack of rigorous evalua-tive evidence on the de-

velopment effectiveness of these approaches.

ScopeAs noted in the Approach Paper approved by theBank’s Committee on Development Effectiveness(CODE) in July 2003, the goal of this evaluationis to assess the development effectiveness of theBank’s CBD/CDD projects, not that of participa-tory development projects in general. Even though

the Bank has been giving more emphasis to CDDin recent years, a large percentage of these proj-ects continue to include both CBD and CDDcomponents, which makes it impossible to carryout an evaluation of CDD projects only.3 Socialfunds, in which funds are channeled through anautonomous agency, are also a subset of CBD/CDDapproaches. Figure 1.1 illustrates the relationshipbetween CBD, CDD, and social funds.

Because CBD/CDD projects can involve a com-munity in different ways, depending on the na-ture of interaction required with that community,there is no typical CBD/CDD project, nor can asingle causal chain—from inputs to outputs, out-comes, and impacts—be identified for all theseinterventions. However, a causal chain can beidentified for CDD projects. (See figure B.1 inAnnex B.) Support for a CDD project typically in-cludes: (a) strengthening community groupswith training support and providing them with anopportunity to control decisions and resourcesin a project context for building small infra-structure and (b) creating an enabling environ-ment for these activities through appropriatepolicy and institutional reform. The underlyinghypothesis in this approach is that such com-munity control in the preparation and imple-mentation of their donor-supported developmentplans will make interventions more suited tolocal needs and enhance community capacityfor self-development. This will allow the com-munity to hold accountable the institutions thataffect their lives. More recently, emphasis is alsobeing given to decentralization reform and pro-moting partnerships between local governmentinstitutions and communities. Strengthened localgovernment institutions are also expected tobuild an enabling environment for CDD.

DesignThe Bank has no explicit benchmark, such as apolicy or strategy paper, against which the designand performance of CBD/CDD approaches ingeneral, and CDD projects in particular, can beevaluated.4 The CDD chapter of the PRSP Source-book (World Bank 2003b) and a Web site man-aged by the CDD Anchor are the principalsources of guidance for Bank staff on CDD(Annex C).

Bank lending for CBD

and CDD approaches has

increased significantly,

but there is a lack of

rigorous evaluative

evidence on the develop-

ment effectiveness of

these approaches.

Figure 1.1: The Universe of CBD/CDD Projects

Community-baseddevelopment

(CBD)

Social funds

Community-drivendevelopment

(CDD)

Source: Study research.

ORIGIN , SCOPE, DES IGN, AND METHODOLOGY OF THE EVALUAT ION

3

Two kinds of primary data provide perspectives from a range ofstakeholders on the appropriateness of the CBD/CDD approachesto development: from the field and from Bank staff. Primary datawere collected because (i) “projects lack careful evaluations withgood treatment and control groups and baseline and follow updata” (Mansuri and Rao 2004); and (ii) though most BankCBD/CDD projects are expected to contribute to community ca-pacity enhancement, project monitoring and evaluation sys-tems as designed are unable to systematically assess changesin that capacity because of the Bank intervention. Chapter 3 (box3.1) explains how this evaluation assessed the capacity-build-ing impact of Bank CBD/CDD interventions.

From the Field: Primary data were collected in the five case studycountries and in India to assess project outcomes. All five coun-try studies included interviews or surveys with central or stategovernment officials and bilateral and multilateral donor repre-sentatives. Focus groups or interviews were also held with localand international nongovernmental organizations (NGOs) ineach of the countries. Local government officials were inter-viewed in the State of Rio Grande do Norte in Brazil and the Bor-gou Region of Benin.a Extensive fieldwork was carried out in therespective areas of Benin and Brazil and in the context of twoOED project assessments (Uttar Pradesh Sodic Land Reclama-tion Project and Madhya Pradesh Forestry Project) in India.These four project areas also involved focus group sessions andkey informant interviews. In the absence of baseline data, OED’sfieldwork adopted a non-experimental evaluation design that

compared randomly selected CBD/CDD communities with com-parator communities in the four projects (Annex M).

Comparators for the Fieldwork: The selected comparators var-ied according to project and country context. In two projectareas (Benin and Brazil), the comparison group communitieshad benefited from similar subprojects as the CBD/CDD com-munities, but through a non-participatory approach adopted ei-ther by the government or by a religious organization. These twocases allowed the evaluation to assess whether a program thatinvolves communities is more effective than one that does not.In another project area (Madhya Pradesh) the comparison-groupcommunities benefited from a similar activity carried out througha participatory approach, but supported by the government. Herethe evaluation assessed whether there is any difference in out-comes because of the participatory approach pursued by the Bankcompared with the participatory approach pursued by the gov-ernment. Finally, in the fourth project area (Uttar Pradesh), com-parison communities did not benefit from a similar activity asproject communities. Here the evaluation assessed the overalloutcomes of the Bank CBD/CDD project, not only that of its par-ticipatory approach.

The Egypt, Nepal, and Vietnam studies also involved a limitednumber of focus group sessions with communities.

Within the Bank: An electronic survey was administered to 400Bank staff and managers familiar with CBD/CDD issues. A responserate of 38 percent resulted in 152 completed surveys (Annex L).

Box 1.1: Primary Data Collection for the Evaluation of CBD/CDD Projects

Uttar Madhya Benin Brazil Pradesh Pradesh

Number of household surveys 1,376 1,097 1,197 1,147

Number of focus groups with community members 32 56 60 58

Number of local leader interviews 32 33 29 30

Number of interviews with committee/community association members 32 28 30 30

Number of interviews with municipal council members 32

a. Henceforth, references to primary data collected in Benin and Brazil refer only to the areas studied.

In assessing the relevance, efficacy, efficiency, in-stitutional development impact, and sustainability ofthe Bank’s CBD/CDD projects, the evaluation addressedsix questions:

1. Are Bank-supported CBD/CDD projects relevant tothe achievement of the Bank’s poverty reduction mis-sion and borrower and community priorities? (Rele-vance)

2. To what extent have CBD/CDD projects mettheir objectives? (Efficacy)

3. How efficient have CBD/CDD projects been rel-ative to alternatives? (Efficiency)

4. To what extent have these projects enhancedthe capacity of the communities and of cen-tral/state and local government institutions?(Institutional Enhancement)

5. What are the challenges for ensuring sustain-ability of the benefits from a CBD/CDD proj-ect? (Sustainability)

6. Do internal policies and processes position theBank to adequately support implementationof CBD/CDD approaches (and CDD projectsin particular)? (Bank Policy Requirements)

MethodologyThe greatest challenge this evaluation faced wasin identifying the portfolio of CBD/CDD projects

to be reviewed.5 Thoughthe Bank has been sup-porting CBD/CDD ap-proaches for more than aquarter-century, the in-stitution has no databasethat has tracked theseprojects since their in-ception. It is only very re-cently that the CDDAnchor in the Social De-velopment Department

has started maintaining a database. Records arebeing maintained on the basis of self-reportingfrom the Regions, but only for projects approvedfrom fiscal year 2000 forward. Hence, OED hadto develop a methodology for identifying theuniverse of projects approved during fiscal years1989–2003 (see Chapter 2).

Both qualitative and quantitative evaluationtools were used to address the six evaluationquestions (Annex D). The study drew on thefollowing:

• A Portfolio Review included a desk review ofproject documents for a proportionate andrepresentative random sample of 84 projects,stratified by time and sector board, drawn fromthe universe of 847 projects identified by OED.The Portfolio Review also reviewed relevant

economic and sector work, as well as 73 Coun-try Assistance Strategies (CASs), 29 PovertyReduction Strategy Papers (PRSPs), and povertysector work for a large number of countries cov-ered by the sample.

• Five country case studies—Benin, Brazil, Egypt,Nepal, and Vietnam—involved desk reviewsof the literature and project documents plus vis-its to the countries. Two of the country stud-ies included extensive fieldwork (box 1.1).

• A Bank Staff Survey and interviews.• Two thematic studies, one a review of portfo-

lio projects for their compliance with safe-guard policies, the other an assessment of howBank capacity to undertake CBD/CDD, partic-ularly CDD, projects in client countries hasevolved over time.

• A Literature Review. • Nineteen project assessments, with extensive

fieldwork in the context of two (box 1.1).• A desk review of documents for six (one in each

Region) ongoing CDD projects. • OED reviews of 33 Implementation Completion

Reports.• Past OED studies.

An external Advisory Panel of three expertsprovided guidance to the evaluation.

Since OED’s rating methodology is objective-based, the Portfolio Review was able to comparethe ratings for outcome, institutional develop-ment impact, and sustainability of CBD/CDD proj-ects with those of non-CBD/CDD projects. For theprojects covered by household surveys, country-specific comparators were selected as describedin box 1.1 and Annex M. The compliance of theCBD/CDD portfolio with the Bank’s fiduciary andsafeguard policies was examined to assess whetherinternal policies and processes position the Bankto adequately support implementation ofCBD/CDD projects, with particular attention paidto the cumulative impact of small subprojectsand community control over resources in CDD.Results of this analysis were assessed against theBank’s benchmark of full compliance.

Some Limitations of the Study DesignBecause few completed CBD/CDD projects hadcommissioned baseline surveys at the outset

4


The Bank has no explicit

benchmark against

which the design and

performance of CBD/CDD

approaches in general,

and CDD projects in

particular, can be

evaluated.

and project monitoring and evaluation systemsare not systematically designed to assess changesin community capacity, OED had to adopt apragmatic methodology based on a non-exper-imental evaluation design to assess the possibleimpact of CBD/CDD projects on changes incommunity capacity (boxes 1.1 and 3.1).6 Sucha methodology has its limitations, because thereare several complexities in identifying com-parators, as highlighted in Annex M, and thevariables for measuring change in social capitaland empowerment, as highlighted in Annex N.Nevertheless, the findings of the fieldwork aresuggestive of the community capacity–enhanc-

ing impact of these proj-ects and provide usefulinsights until sufficientbaseline data are com-piled under Bank proj-ects to permit morecomprehensive and rig-orous (longitudinal) im-pact evaluations. Theevidence from the fieldwork and the desk reviewof Bank-supported participatory developmentapproaches was also corroborated with evi-dence from the literature (both Bank and non-Bank).

ORIGIN , SCOPE, DES IGN, AND METHODOLOGY OF THE EVALUAT ION

5

Both qualitative and

quantitative evaluation

tools were used to

address the evaluation

questions.

7

The CBD/CDD Portfolio

As noted in Chapter 1, the Bank has only recently developed a systemfor tracking its CBD/CDD projects.1 Because the Anchor’s databasewas incomplete and could not be directly used for the evaluation’s

purposes, OED developed a methodology to identify all CBD/CDD projectsapproved between 1989 and 2003.

A universe of 847 projects was identified (AnnexE) this way, and a proportionate and represen-tative random sample of 84 projects, stratified bytime and sector board, was drawn for intensivereview.2 OED then categorized the sample of 84projects into CBD, CDD, and mixed CDD/CBDinterventions. Annex F explains how the largelyCDD interventions were separated from theCBD/CDD ones.

Temporal, Regional, and SectoralDistribution of the Portfolio The total number of Bank projects that includea CBD/CDD component has increased substan-tially over time (figure 2.1a). The largest per-centage of these projects is in the Sub-SaharanAfrica Region (AFR), with the Latin America andCaribbean Region (LAC) in second place (fig-ure 2.1b). Among sectors, the rural sector has thelargest percentage of CBD/CDD projects (27percent), with health, nutrition, and populationin second place (16 percent) and social protec-tion close behind (15 percent) (Annex E).

Analysis of the random sample of 84 projects(Annex F) found that over the period 1989–2003,

overall CBD/CDD projects have grown at about14 percent per year, whereas CDD projects havegrown at about 19 percent per year. The analy-sis also found that “pure” CDD projects formabout 23 percent of the CBD/CDD portfolio.The Latin America and Caribbean Region hadthe highest, and the Middle East and North AfricaRegion the lowest, number of CDD projects(Annex F). Further, more than 75 percent of theCBD/CDD portfolio consists of projects for whichthe majority of investment funding is for a largenumber of small and scattered subprojects.

Ways That CBD/CDD Projects Differ fromThose in the Non-CBD/CDD Portfolio

A Greater Focus onLearning by Doingand Multisectorality The literature on par-ticipatory developmenthighlights the impor-tance of a learning-by-doing, flexible approach,

22

The total number of Bank

projects that include a

CBD/CDD component has

increased substantially.

The largest percentage

of these projects is in

the Sub-Saharan Africa

Region.

a concept developed from the learning processideas of Korten (1980). Flexibility is beingbrought to Bank CBD/CDD projects throughgreater use of flexible lending instruments such

as Adaptable Program Loans (APLs) and Learn-ing and Innovation Loans (LILs); by allowingcommunities greater choice in the selection ofactivities, as in a multisectoral project; and byproviding them the opportunity to control in-vestment decisions and resources during proj-ect implementation in the case of CDD. Otherissues that are relevant for flexibility are dis-cussed in Chapter 5.

The percentage of APLs and LILs in theCBD/CDD universe is more than double that inthe non-CBD/CDD universe (32 percent against13 percent for the period 1999–2003). Over time,research into the multidimensional nature ofpoverty has led to an increase in multisectoral proj-ects in the Bank’s portfolio, and the percentageof these projects in the CBD/CDD universe ismuch higher than that in the non-CBD/CDD uni-verse (51 percent versus 31 percent between fis-cal years 1989 and 2003; see Annex E, table E.1).Analysis of the sample of 84 CBD/CDD projectsalso revealed that the CDD portfolio has a higherpercentage of multisectoral projects (74 percent)than the overall CBD/CDD portfolio (48 percent).However, a large number of CDD projects con-tinue to be sectoral interventions.

8


Figure 2.1a: Bank Commitments to Projects with CBD/CDD Approaches Have Grown

Num

ber o

f pro

ject

s

0

10

20

30

40

50

60

70

80

90

100

Commitment Projects

1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003

In 2

001

mill

ion

US$

0

1,000

2,000

3,000

4,000

5,000

6,000

Source: World Bank database.

Figure 2.1b: Africa Has the Largest Share of Projectswith CBD/CDD Approaches

MNA7%

LAC23%

ECA11%

SAR11%

EAP17%

AFR31%

Source: World Bank database.Note: MNA = Middle East & North Africa; SAR = South Asia; AFR = Africa; EAP = East Asia & the Pacific; ECA = Europe and Central Asia;LAC = Latin America & the Caribbean.

A Greater Focus on Giving MoreResponsibility to CommunitiesCBD/CDD projects, particularly CDD, empha-size process issues more than the non-CBD/CDDprojects, because they attempt to enhance socialcapital and further the Bank’s support for com-munity empowerment. Within the CBD/CDDportfolio, the CDD projects differ from the oth-ers in that they are designed to put greater re-sponsibility on communities for each phase of thesubproject cycle (figure 2.2). Although there islittle discussion in Bank documents (at appraisal,supervision, or completion) about the partici-patory process itself,3 the importance given toempowerment in the Bank’s Strategic Frame-work, the evidence from the Egypt and Nepalcountry studies, and the Portfolio Review indi-cate that the projects involving community par-ticipation have evolved toward allowing for acommunity role in decision making, rather thanmerely “informing” them of decisions. (See fig-ure 3.3 and Annex H.)

CBD/CDD ProjectsHave Been Usedto Promote a Varietyof Objectives CBD/CDD projects havesought to achieve severalkinds of objectives—poverty reduction, decen-tralization, employmentgeneration, basic infrastructure development, ac-cess to health care and education, nutrition, nat-ural resource management, private sectordevelopment, urban upgrading, mitigation of thesocioeconomic impact of HIV/AIDS, and eco-nomic recovery. It is not uncommon to have sev-eral objectives combined in a single project,without a clear hierarchy. However, more recentprojects generally have fewer objectives thanthose approved in the earlier part of the period.

The Portfolio Review found that most proj-ects have sought to achieve their objectivesthrough two broad kinds of activities: (i) ma-

THE CBD/CDD PORTFOLIO

9

Figure 2.2: Communities Have More Responsibility in CDD Projects

Perc

enta

ge o

f pro

ject

s

Responsible for subproject design

Responsible for subproject implementation

Responsible for subproject M&E

Responsible for subproject O&M

0

25

50

75

100

Community responsibility in subproject cycle

CBD/CDD CDD

Source: Review of 84 project appraisal documents.

Note: M&E = monitoring and evaluation, O&M = operation and maintenance.

Projects have sought to

achieve their objectives

through (i) material

development and

(ii) capacity-building

support.

terial development, such as infrastructure con-struction and (ii) and capacity-building supportfor government (central, provincial, and local),communities, and other stakeholders, such asnongovernmental organizations (NGOs).Though the word “empower” is present in thestated objectives of only 2 of the 84 sampleprojects, other projects de facto seek to em-power communities by building their capac-ity—for example, the Nepal Community SchoolSupport Project (2003) proposes to providetechnical assistance and other financial supportto build the capacity of communities to takeover the management of government-fundedschools.

CBD/CDD is Being Used in New ActivitiesSuch as AIDS and Post-Conflict WorkThe recent Africa Multi-Country HIV/AIDS Pro-jects, aimed primarily at assisting national gov-ernments in their strategies to cope with thedisease, are also considered community-driven.These projects are the subject of another inde-pendent evaluation in OED. Community-drivenreconstruction has also been used recently as anapproach in the transition from war to peace. Theportfolio of 847 projects includes 198 in conflictand post-conflict countries, several of which aredesigned to provide speedy delivery of recon-struction assistance and support for infrastruc-ture rehabilitation.

1 0


1 1

Outcome ofBank-SupportedCBD/CDD Projects

This chapter first reviews the overall outcome ratings of CBD/CDDprojects before going on to examine their relevance, efficacy, and effi-ciency.

Outcome Ratings

Outcome Ratings of CBD/CDD ProjectsHave Been Better than Those ofNon-CBD/CDD ProjectsThe outcome ratings (investment lending only)of closed CBD/CDD projects (when compared bynumber of projects), on average, have been bet-ter than those of non-CBD/CDD projects over theperiod covered by the evaluation, although thegap has narrowed over time (figure 3.1). Out-come ratings of CBD/CDD projects are also bet-ter when disbursement-weighted, with a fewlarge projects—mostly concentrated in middle-income countries—doing much better than alarge number of smaller projects.

The Latin America and Caribbean Region,with the second-largest portfolio (Chapter 2), hasa significantly higher percentage of CBD/CDDprojects rated satisfactory on outcome than allother Regions. Although the Africa, Middle Eastand North Africa, and Europe and Central AsiaRegions have a lower percentage of CBD/CDDprojects rated satisfactory on outcome, theseRegions also show an increase in satisfactoryoutcome ratings in the period 1999–2003 over

1994–98 (figure 3.2a). The education sector hashad the highest percentage of projects ratedsatisfactory on outcome, followed by projects inthe transport, urban development, and socialprotection sectors. The rural development sec-tor, with the largest CBD/CDD portfolio, is abelow-average performer on outcome in the ag-gregate, as are projects under the water supply,health, and environment sectors (figure 3.2b).1

OED’s data also shows that between 1999 and2003 the outcome rating for CBD/CDD projectsin post-conflict countries was 4 percentage pointshigher than the outcome rating for CBD/CDDprojects in non-conflict countries (Annex G,table G.2).

Relevance

Increasing Importance Is Being Given toCBD/CDD, Particularly CDD, in Recent CASs A review of 62 CAS doc-uments (two each from31 countries covered bythe sample of 84 proj-ects) found that theCBD/CDD approach isconsidered an important

33

The CBD/CDD approach is

considered an important

element of the Bank’s

strategy in over 74 per-

cent of countries.

element of the Bank’s strategy in over 74 percentof countries, and that the emphasis given togreater community participation in decisionmaking and resource allocation in Bank inter-ventions has increased over time (See Annex Hand figure 3.3).

Borrower Government Officials InterviewedWere Convinced that a ParticipatoryApproach Is BeneficialSurveys of government officials in the 5 casestudy countries, past OED studies, and a reviewof 29 PRSPs (Annex H) indicate that borrowergovernment officials are convinced that a par-ticipatory approach can contribute to poverty re-duction in their countries. For example, inVietnam, 86 percent of central government of-ficials interviewed reported that projects withbeneficiary participation can address poverty re-duction better than those without (Annex I).One reason that governments in countries suchas Nepal and Vietnam have accepted and adopted

participatory approaches is because these coun-tries have a relatively long history of communityparticipation in their own development. Anotherreason why governments are convinced of thebenefits of the approach is that in an age ofshrinking budgets, it allows them to do more withless. For example, the OED India country studyon forestry (Kumar and others 2000) reportedthat the forest department staff valued commu-nity participation in forest protection. BeforeJoint Forest Management (JFM), the forest de-partment was fighting a losing battle, and one for-est guard could not effectively patrol the largearea under his control without the participationof the communities.

But They Didn’t Necessarily Believe in GivingCommunities Control over Decisions and Resources Surveys of government officials in Benin, Nepal,and Vietnam also indicated that they did notnecessarily believe that allowing communitiescontrol over investment decisions and resources

1 2


Figure 3.1: Outcome Ratings of CBD/CDD Projects Have Been BetterThan Those of Non-CBD/CDD Projects

Satis

fact

ory

(%)

1994–98 1999–20030

40

20

60

80

100

Exit fiscal yearCBD/CDD Non-CBD/CDD

74%

66%73% 72%


Note: OED ratings are based on OED reviews of Implementation Completion Reports (ICRs), 25 percent of which are subsequently revisited through OED field assessments.

The outcome ratings of the closed investment projects reveal insignificant differences between CBD/CDD and non-CBD/CDD projects in the two time periods con-

sidered (1994–98 and 1999–2003). The differences between CBD/CDD and non-CBD/CDD projects were also insignificant for each exit year between fiscal years 1999

and 2003.

OUTCOME OF BANK-SUPPORTED CBD/CDD PROJECTS

1 3

Figure 3.2a: CBD/CDD Projects in the Latin America and Caribbean Region Have Done Betterthan Projects in Other Regions

Satis

fact

ory

(%)

AFR EAP ECA LAC MNA SAR0

40

20

60

80

100

Region1994–98 1999–2003


Note: OED ratings are based on OED reviews of ICRs, 25 percent of which are subsequently revisited through OED field assessments. AFR = Africa; EAP = East Asia

and the Pacific; ECA = Europe and Central Asia; LAC = Latin America & the Caribbean; MNA = Middle East & North Africa; SAR = South Asia.

Figure 3.2b: CBD/CDD Projects in Education and Social Protection Do Better than Projectsin Health and Rural Development

Ruraldevelopment

Socialprotection

Health Education Other*0

40

20

60

80

100

1994–98 1999–2003

Satis

fact

ory

(%)


*Urban, Environment, Water Supply & Sanitation, Transport.

Note: OED ratings are based on OED reviews of ICRs, 25 percent of which are subsequently revisited through OED field assessments.

in a Bank project context was the best means ofengaging them (Annex I). Over 90 percent of thePRSPs reviewed are also silent on such commu-nity control (Annex H). Central government of-ficials in Benin were asked what form ofparticipation is effective and efficient for devel-opment approaches; nearly 80 percent of the re-spondents said that it is where the communityis informed and consulted on the governmentdevelopment plan for the area. The local gov-ernment surveys in that country also suggestthat local officials feel threatened by devolutionof complete control over decisions and resourcesto communities (Annex I).

Central and local gov-ernment officials sur-veyed in case studycountries also appear tobe unconvinced of theability of communities tohandle resources. For ex-ample, among the Viet-namese officials, only 21

percent said that more than 75 percent of the com-munities had the ability to identify their needs andprioritize them (Annex I). None of the local gov-ernment officials in Benin said that the commu-nities had the ability to manage and mobilizeexternal or internal resources. Even in Brazil, amiddle-income country that has had a CDD pro-gram for a decade, while over two-thirds of themunicipal government officials interviewed inthe state of Rio Grande do Norte said that mostcommunities are capable of identifying and pri-oritizing their needs, the majority stated that mostcommunities are not capable of preparing de-velopment plans, implementing and maintainingcommunity projects, or mobilizing resources ei-ther within or from outside the community. Halfof the municipal government officials interviewedalso said that most communities are not capableof managing financial resources.2

Nor Did They Have the Same Understandingas the Bank about How Community EmpowermentIs To Be Promoted OED’s field research, desk review of project doc-uments, and interviews with Bank staff found that(i) within the Bank, there is a mismatch betweenthe understanding of empowerment and the pri-mary means of promoting it; and (ii) betweenthe Bank and its clients, there is a consequent dif-ference in the understanding of how to promoteempowerment. The Bank’s Web site defines em-powerment as the process of increasing the ca-pacity of individuals or groups to make choicesand to transform those choices into desired ac-tions and outcomes (box 3.1). However, theBank’s primary channel for promoting communityempowerment is through design and imple-mentation of CDD projects that have a definite yetnarrower interpretation of how empowerment isto be brought about in a community—that is,through giving communities control over deci-sions and resources in a Bank project context. In-terviews with borrower government officials incase study countries found that they see activitiesthat help build the capacity of the communitiesto participate in the development process as em-powering activities. In Uttar Pradesh, the imple-menting agency whose operational capacity wasconsiderably enhanced with support from the

1 4


Figure 3.3: CBD/CDD Is Increasingly Importantin Country Programs

Perc

enta

ge o

f CA

Ss

Previous CAS Latest CAS0

25

50

75

100

Community participation in decision making and resource allocation,especially of the poor

Partnership (community rolein decision making and planning responsibilities)

Information and coming closer to communities

Top-down approach (no participation in theformal decision making)

28%14%

66%

59%

0%

21%

7%

7%

Source: Review of CASs (Annex H).

There appears to

be a difference in

understanding between

the Bank and its clients

on how to promote

empowerment.

Sodic Land Reclamation Project did not see “em-powering” communities as building toward astrategy of putting them in control. The OED as-sessment of the project notes:

From the implementing agency perspec-tive, the most important aspect of com-munity participation was ensuring farmercommitment to the reclamation processand sharing of reclamation costs. Hence,for them, even if they ‘directed’ and ‘con-trolled’ the discussion in the village-level im-plementing bodies, it was not a concern, aslong as farmers participated and con-tributed and the reclamation was under-taken as per the technical standards.

The way the government officials view em-powerment is in harmony with the definition ofempowerment displayed on the Bank’s Web site,and not with the way it is interpreted in the im-plementation of the CDD projects. The needfor clarity on how empowerment is to be pro-moted by the Bank becomes even more urgentwhen seen in the context of the discussion of theissue in the literature.3

Priority Needs Are More Likely to BeAddressed When Communities Are GivenMultiple Options from which to Choose Household survey data collected by OED inBenin and Brazil show that where communitieswere given a wide menu to choose from, Bankprojects were more likely to meet one of thetop-priority needs of the communities. TheMadhya Pradesh Forestry and Uttar PradeshSodic Land Reclamation projects in India, bycontrast, did not meet the priority needs of thecommunities (figure 3.4). These sectoralinterventions were not designed to givecommunities a choice of activities.

Efficacy

So Far, More Success Has Been Achievedon Quantitative than on Qualitative GoalsMaterial development and capacity building ac-tivities (see p. 9) have both quantitative andqualitative dimensions. CBD/CDD interventionshave generally been successful in achieving quan-titative goals such as infrastructure built, em-ployment generated, and number of trainingcourses held. Since these interventions are sup-


1 5

This evaluation assessed changes in community capacity toundertake development through respondents’ perception ofchanges in social capital and empowerment (as defined below)in CBD/CDD and comparator communities as captured throughhousehold surveys. The capacity assessed did not includechanges in a community’s technical capacity.

Social capital refers to the norms and networks that enable col-lective activity in a community. By drawing people in a communitytogether to collectively decide and manage project activities andoutputs, Bank CBD/CDD projects expect to expand the depth andrange of communities’ social networks. To assess the extent towhich Bank-funded interventions have succeeded in enhancingsocial capital at the community level, the household surveys col-lected information on respondent perception of change in trust, as-

sociational life, participation in traditional events and in non-tra-ditional/political events, and circle of friends.

The Bank’s Web site defines empowerment as the process ofincreasing the capacity of individuals or groups to make choicesand to transform those choices into desired actions and out-comes. The Sourcebook on Empowerment and Poverty Reduc-tion (World Bank 2002b) identifies four key elements of successfulempowerment approaches: access to information, inclusion/par-ticipation of poor people, accountability, and local organizationalcapacity. This understanding of empowerment has informed datacollection for this study, which explores both the levels of em-powerment at the time of fieldwork and respondents’ perceptionsof changes in empowerment before and after subproject imple-mentation.

Box 3.1: Change in Social Capital and Empowerment as a Means of Assessing the CommunityCapacity–Enhancing Impact of Bank CBD/CDD Interventions

Source: See Annex N for details and results from household surveys on these variables.

porting subprojects in scattered communities,they are also likely to help in increasing accessto schools, health facilities, and the like for thesecommunities. On the basis of evidence from alimited number of evaluations, a recent Bankreview of CDD projects (Wassenich and White-side 2004) also presents a favorable picture ofCDD impacts on access to infrastructure. Find-ings from 4 of the 5 case study countries and 13of the 19 related OED project assessments sup-port the conclusion that there has been rela-

tively more success inachieving quantitativegoals, such as infrastruc-ture built and number oftraining courses held,than on qualitative goals,such as community ca-pacity enhancement andtraining quality.4

The Portfolio Reviewfound that several of theCBD/CDD interventionsin conflict and post-con-flict areas have success-

fully met quantitative targets for infrastructurerehabilitation and reconstruction and have alsoprovided substantial employment benefits tothe local population (see box 3.2 for the EritreaCommunity Development Fund). For example,378,805 persons are reported to have benefiteddirectly or indirectly from infrastructure activi-ties under the Kosovo Community DevelopmentFund Project (2000). In post-conflict situations,where a large part of the basic infrastructurehas been destroyed by war or civil strife, this isa considerable achievement. The experience ofthe Nepal country study also found that well-planned participatory interventions can con-tribute to the mitigation of the social dissentthat fuels conflict. However, in conflict and post-conflict situations, where the focus is on get-ting things done quickly, it is even more difficultto achieve the qualitative goals. While commu-nities can play a major role in ensuring ac-countability and proper use of donor resourcesin these situations, what is often lacking in post-conflict communities is the ability to act to-gether. As noted by an OED report on The WorldBank’s Experience with Post-Conflict Recon-

1 6


Figure 3.4: Community Priorities Are Better Met When There Is Greater Choice

Com

mun

ities

(%)

Brazil [N = 27] Benin [N = 13] Uttar Pradesh [N = 19] Madhya Pradesh [N = 21]

Multisector Single sector

0

25

50

75

100

Top two priorities Top priority

None None

Source: Household survey data.

Note: Priorities are aggregated by community.

Several of the CBD/CDD

interventions in conflict

and post-conflict areas

have successfully met

quantitative targets

for infrastructure

rehabilitation and

reconstruction and have

provided substantial

employment benefits.

struction (OED 1998) “Inherent in violent civilconflict is the destruction of social capital, par-ticularly institutions of governance and civil so-ciety and such basic attitudes and behaviors astrust and participation.”

Inadequate tracking of progress on qualitativeaspects has been a shortcoming in most Bank in-terventions. However, it becomes an even big-ger issue with CBD/CDD projects, particularly inCDD projects, where enhancing the capacity ofcommunities and local governments is a cen-tral objective. However, without baseline data,and without appropriate indicators, it is often as-sumed that meeting the quantitative goals will

automatically fulfill the qualitative goal—for ex-ample, holding a certain number of trainingcourses is expected to enhance capacity.

The Bank’s CBD/CDD Projects Appear toHave Enhanced the Social Capital ofCommunities But Have Had VariableSuccess in Empowering ThemWassenich and Whiteside (2004) found thatonly two Bank impact studies for CDD projectshave findings on social capital that are reliablyrepresentative of all CDD projects, and those twostudies showed mixed results regarding thecontribution of CDD projects to the enhance-


1 7

The Peru Rural Roads Rehabilitation and Maintenance Project(1996) achieved its key project objective of providing a well-integrated and reliable rural road system through rehabilitationand maintenance of rural roads and key links connecting to theprimary road system. In this project community participation wasan important part of project preparation. A logical framework de-signed by the participants in the first meetings was used tostructure a community’s involvement in the project. (PPAR July2001)

The experience of the Nepal Hill Community Forestry Project(1990) shows that clear policies, laws, and procedures are cru-cial to building trust between the communities and the Forest De-partment. Even though the concept of User Group Managementwas adopted in 1988, implementation was slow in the early yearsas a supportive framework was missing. However, things changedwith the passing of the Forest Act of 1993. The procedural clarifi-cations on the implementation of the act followed with the Rulesin 1995. As a result, identification of user groups and handing overof forests accelerated after 1995. (PPAR June 2001)

The Uttar Pradesh Sodic Land Reclamation Project (1993) ex-ceeded expectations in carrying out large-scale reclamation ofsodic soils. Over 68,400 hectares were reclaimed with farmer par-ticipation. The project also contributed to poverty reduction by help-ing increase returns to many small and marginal farmers. Theproject effort in sequencing activities helped harness farmercommitment for agricultural development on reclaimed lands. Forexample, the transparent land titling was done before undertak-ing of technical solutions to sodicity. (PPAR June 2004)

The Turkey Eastern Anatolia Watershed Rehabilitation Project(1993) largely delivered on its objective of restoring sustainable man-agement of forest and farm activities in the upper watersheds ofthe three project provinces, reducing soil degradation, erosion, andsedimentation in reservoirs, as well as increasing productivityand incomes in this impoverished region of Turkey. With respectto processes, important experience was gained by the public sec-tor with community-based participation. There was very good—unprecedented in Turkey—coordination between ministries anddepartments in this project. (PPAR March 2004)

The Benin Borgou Pilot Project (1998) interventions took placein 229 villages where the communities successfully completed atotal of 296 infrastructure subprojects. A substantial number of lit-eracy and training courses were held to improve technical capacityin the communities. The project built on the Bank’s experience inBenin with the Village-Level Participatory Approach (VLPA) in the1990s. Under the VLPA, extension staff and other government andprivate agencies encouraged village communities to analyze theirsituation, identify priority problems, prepare action plans to dealwith them, and implement those action plans. (PPAR June 2003)

The Eritrea Community Development Fund (1996) financed therehabilitation or creation of a significant amount of social andeconomic infrastructure in the rural and war-devastated areas ofEritrea. The Eritrea Community Development Fund made an im-portant contribution to population needs in a post-conflict situa-tion. The project adopted a short-term problem-solving approachand delivered outputs under extremely difficult conditions. (PPARJune 2002)

Box 3.2: Highly Satisfactory Aspects of Design and Implementationin Selected CBD/CDD Projects

ment of social capital.Similar findings emergedfrom the fieldwork car-ried out for OED’s SocialFund Evaluation (OED2002b). OED’s analysisof the household datafrom Benin, Brazil, andIndia found the associa-tion between CBD/CDDprojects and respon-dents’ perceptions ofchanges in social capitalto be positive in Benin,

Madhya Pradesh, and Uttar Pradesh; results inBrazil were mixed. Household data also showthat the association between CBD/CDD proj-ects and respondents’ perceptions of changes inempowerment was variable across the four proj-ect areas. Four results patterns emerged from theinformation gathered for this evaluation thatare suggestive of the differing impacts that BankCBD/CDD projects can produce at the com-munity level. This section also attempts to ex-plain these results.

First, the Bank has had the most success inareas where it has either supported indigenouslymatured participatory initiatives, as it did withthe India dairy program Operation Flood, or

where it has provided consistent, long-term ca-pacity building support to communities. Evi-dence from other Bank studies supports thisfinding.5 The Bank provided support to India’sOperation Flood through five projects during theperiod 1974–87. The program is dominated byfarmer-controlled, village-level dairy coopera-tive societies; the Bank’s financial support addedvalue to this indigenously matured CBD/CDD ef-fort, which had already experienced a criticalamount of learning and institution building.6

The OED impact study for the program notedthat “Operation Flood is an Indian pro-gram.…when the program was massively ex-panded under Operation Flood II, there was analready existing indigenous institution ready toimplement the project. Bank institutional sup-port thus involved genuine institution building,as distinguished from the institution creationcharacteristic of many Bank projects” (Candlerand Kumar 1998). There are several other ex-amples (such as the Self-Employed Women’s As-sociation [SEWA] in India, and the Orangi PilotProject in Pakistan7) of well-known developmentinitiatives where the initial idea and effort startedindigenously, long before the Bank or anothersource of external support was provided. Bysimilar logic, communities that have effectiveleadership and pre-existing ability to organize for

1 8


Although the outcome of the Ghana Agriculture Sector Invest-ment Project (ASIP, 1994) as a whole was rated unsatisfactoryby OED, a few established groups were able to turn the finan-cial opportunities offered by the ASIP to their advantage. TheNangbanyini Nyagsa Women’s Group, Savelugu Nanton Dis-trict, Northern Region, is an example. The group of 24 members(2 men and 22 women) had come into existence 10 years beforethe project and was then known as the Tiyum Taba Women’sGroup. It had started its activity with an agro-forestry unit forwhich the village chief had allotted the group five acres of land.The group is now considered a success story under the project,

which provided the group with an agro-processing unit. How-ever, the success of this subproject has less to do with the proj-ect than with the existing group capacity and the dynamicpersonality of the group leader, who is also helping other, less-effective groups in nearby villages. The agro-processing unit runby the group now consists of a grain mill, a shea nut crusher, anda rice huller. When the group heard that funds were being pro-vided under the ASIP for income-generating activities, it con-tacted the local District Assemblies with a request for anagro-processing unit.

Box 3.3: Local Champions Can Effectively Use Bank CBD/CDD Funds: A Case from Ghana

Source: OED Assessment of the Ghana Agriculture Sector Investment Project (OED 2001).

The Bank has had the

most success in areas

where it has either

supported indigenously

matured participatory

initiatives or where it

has provided consistent,

long-term capacity

building support to

communities.

collective action have been able to turn the fi-nancial opportunities offered by a Bank-sup-ported project to the advantage of the group, asin Ghana (box 3.3). Field visits for OED’s proj-ect assessments of the Mali Natural ResourceManagement Project, the West Bengal ForestryProject, and the Pakistan Northern ResourceManagement Project support this finding.

Extending this logic further, when a Bank in-tervention has been built on past experiencewith a similar capacity-building approach andwith a focus on the process of bringing com-munities together to organize for collective ac-tion, reasonable success has also been achieved.In these cases, the same communities have beenthe focus of the capacity- building effort for sev-eral years. As shown by OED’s evaluation of theAga Khan Rural Support Program in Pakistan(OED 2002c),8 one of the basic reasons for suc-cess in this program has been the sustainedsupport for processes in the same communi-ties over 20 years. In Benin, the findings fromOED’s household survey indicate that the Bank’sstrategy for community participation had moresuccess in enhancing social capital and em-powering communities in the Borgou Pilot proj-ect than in the Social Fund or the Food Securityprojects. The Borgou Pilot built on the Bank’sexperience with the Village-Level ParticipatoryApproach exercise in Benin, which included par-ticipatory rural appraisal and other partnershipefforts that enabled communities and villages tocoordinate and execute their own rural devel-opment, with assistance from extension agentsand financial resources from a variety of pro-grams. Bank staff are increasingly aware of theneed for long-term support to build commu-nity capacity. One of the criteria for selection ofcommunities under the Indonesia KecamatanDevelopment Fund Project (KDP) 3, which be-came effective in January 2005, is that they re-ceived support under KDP 1 or 2. About 52percent of Bank staff surveyed said that com-munity groups initially formed under a projectneed support for at least 6–10 years to reach alevel of sustainability in community processes(Annex L).

Second, when the borrower has used theopportunity provided by Bank financing to sup-

plement its own effortsto organize for collectiveaction, and the commu-nities believe in the long-term benefits offollowing the approach,interventions have metwith reasonable success,as with forestry projectsin India, where Banklending helped the coun-try bridge the financialresource gap it faced inimplementing its foreststrategy directed towardsupporting joint forestmanagement (Kumar and others 2000). TheGovernment of Uttar Pradesh also placed a highpriority on reclamation of sodic lands and lookedto the Bank for financial support. In MadhyaPradesh, analysis of the household survey dataindicated that the Bank’s strategy of communityinvolvement was positively associated withchange in social capital, but much less so withempowerment. In Uttar Pradesh, the house-hold data analysis indicates that the Bank’s proj-ect was positively associated with change insocial capital, and to a greater extent than inMadhya Pradesh with empowerment.9 Indeed,had the Bank taken steps to ensure sustain-ability in both cases (see section on Sustain-ability in Chapter 4), even more positive resultscould have been achieved.

Third, when undertaken in circumstanceswhere political and social settings have not his-torically favored participation and collectiveaction, as in the state of Rio Grande do Norte inBrazil, the Bank’s strategy of community partic-ipation may have little influence on communitysocial capital and empowerment.10 Analysis of thehousehold survey data from that state indicatethat, at best, there is no difference in respon-dents’ perceptions of change in empowerment,while results for social capital have been minimal(Annex N). The literature (Costa and others1997; Tendler 1997) reveals that the states ofNortheast Brazil are known for their clientelis-tic ways of governing. The patron-client rela-tionships that pervade the Northeast create a


1 9

When the borrower has

used the opportunity

provided by Bank

financing to supplement

its own efforts to

organize for collective

action, and the

communities believe in

the long-term benefits of

following the approach,

interventions have met

with reasonable success.

social system in which vertical ties of mutualdependence hinder development of strong hor-izontal links of solidarity within communities.11

Although the Brazilian government’s decentral-ization policies appear to have weakened thesetraditional forces, they continue to exercise astrong influence. An issue raised in some focusgroup sessions with communities in Rio Grandedo Norte is that communities that receive sub-project funds do so through political relation-ships, and not because of need. It is thus notalways a case of a community needing a moti-vated individual, but a “connected” one.

Fourth, when a Bank-supported interventionattempts to build social capital and empowercommunities, the capacity-building benefits

may be cornered by the“better-off ” communitymembers. Various stud-ies in the literature onparticipatory develop-ment point out that thebetter-educated mem-bers of the communityand the relatively better-off are often the oneswho represent the com-munity in participatory

interventions (Desai 1996; Gibson and Marks1995; van der Linden 1997; Ribot 1998). The2004 World Development Report, Making Ser-vices Work for Poor People, also notes that “elitescan mobilize more quickly, master the rules ofsubmitting applications (if they can read andthe majority of the community cannot), andpresent themselves to the community as aneffective conduit for receiving funds” (WorldBank 2004d, p. 73). Therefore, the report cau-tions that “Rushing to create social capital whereit does not exist can do more harm than good.”OED’s focus groups in Brazil found that manycommunities indicated dependence on an indi-vidual or a small group of community leaders tobring donor-funded projects to the community.In Benin also, focus group sessions revealedthat decision-making procedures surroundingthe selection of subprojects lacked the activeparticipation of community members. Villageleaders, whether traditional, administrative, orthe Groupement Villageois (which in the Borgouregion are mainly structured around the cottonsector), took the lead in identifying subprojectactivities, and only later brought them to thecommunities for approval. In Madhya Pradeshand Uttar Pradesh also, a majority of villages inOED’s focus groups indicated that most decisionstaken by the Panchayat were largely devoid ofany broad community participation. OED’shousehold data shows that in Madhya Pradesh,Uttar Pradesh, and Brazil, respondents who weremembers of community organizations set up bythe Bank projects had a higher socioeconomicprofile, including greater mobilization skills anda more extensive social network, than non-mem-bers before the Bank intervention. Further, inMadhya Pradesh and Uttar Pradesh, these mem-bers of project organizations also reported agreater increase in social capital than did thenon-members. In Madhya Pradesh, Uttar Pradesh,and Brazil they also reported a greater increasein empowerment (Annex N, endnote 7).12

These results are not difficult to explain. Threefactors appear to be responsible.

Projects generally do not tailor capacity build-ing to community capacity: The literature showsthat communities are at different stages in the evo-

2 0


Figure 3.5: Focus Groups Report Significant DecisionMaking by Local Leaders

Perc

enta

ge o

f foc

us g

roup

s

Benin[N = 26]

Brazil[N = 44]

MadhyaPradesh[N = 40]

UttarPradesh[N = 38]

0

40

20

60

80

100

Source: Focus group interviews.

When a Bank-supported

intervention attempts to

build social capital and

empower communities,

the capacity-building

benefits may be cornered

by the “better-off ”

community members.

lution of social capital and empowerment, andhence have different capacities (Greiner 1972;Handy 1985; Pretty and Ward 2001). Communitiesin Brazil, Benin, and India (and within these coun-tries) likely have different capacity levels; there-fore, the change that can be expected as a resultof exposure to a Bank-supported project is alsolikely to be different. Yet project documents in-dicate that Bank-supported projects do not di-agnose community capacity or tailor capacitybuilding to existing community capacity.

The Bank subproject cycle is too short: The lit-erature also shows that building capacity is atime-consuming process (Ostrom 1999; Prettyand Ward 2001).13 However, Bank interventionshave not been designed to provide long-termsupport. Village societies traditionally have beenhierarchical, with the local leaders making the de-cisions. It is difficult for a Bank intervention tochange this with the limited money and short pe-riod of support in a subproject (figure 3.5).14

The flexibility that many of the CBD/CDD proj-ects appear to have because of the use of anAPL or LIL instrument (Chapter 2) has not

changed the subproject cycle at the communitylevel. The one year of a typical subproject cycle(figure 3.6) is enough time to implement a sub-project and, in several cases, to start a processof change in a group, or move a group furtheralong the evolutionary process if it is at a higherlevel of development, but in most cases is in-sufficient to take it far enough for sustainabilityto be ensured. OED’s Social Fund Evaluationalso found that the nature and extent of infor-mation sharing and participation by communitymembers in social fund projects was sufficient toallow successful subpro-ject execution, but not toconsistently have a sig-nificant positive impacton community capacity.15

Further, a group with lit-tle capacity may evenregress at the end of thesubproject cycle.16 If agroup’s capacity is at ahigher level, as was theone in Ghana (box 3.3),then it is able to use the


2 1

Figure 3.6: Subproject Cycle Is Too Short for Meaningful Enhancement of Community Capacity

Elect Project Management Committee Prepare/submit subproject application and budget

Mobilize community contribution Plan launch ceremony

Sign Financial Agreement

ImplementationProcure materials and services

Supervise construction Manage labor contribution Prepare progress reports

Manage funds

Monitoring and EvaluationDevelop own indicators/M&E plan

Provide day-to-day oversight Conduct audits

± 12

mon

ths

CompletionTake over operation and management of asset

Preparation

The one year of a typical

subproject cycle is enough

time to implement a

subproject and to start a

process of change in a

group, but in most cases is

insufficient to take it far

enough for sustainability

to be ensured.

Bank financing to its ad-vantage.

Also, Bank projectstypically have not goneback to the same com-munity with the same ap-proach to enhancingcapacity to follow upwhere the initial subpro-jects left off. Even if an-other Bank project does

go back to the same community, in most casesit does not build on the capacity-building effortof the earlier interventions. Where it does, as inthe Borgou Pilot in Benin, positive results can beexpected. The change a year of effort can bringabout in social capital and empowerment in agiven community cannot be expected to be verydramatic.

Communities have a different understandingof the role they are expected to play in CDDprojects than does the Bank: Focus group datain all three countries indicate that the commu-nities’ understanding of participation in a sub-project cycle is different from that of the Bank.17

This does not appear surprising given that thereis also a difference in the understanding of par-ticipation between the Bank and the borrowinggovernment, which is technically “in charge” ofimplementation and sees it primarily as an op-portunity to do more with less (see p. 12). TheBank CDD projects visualize communities takingthe lead in the choice and implementation of thesubproject. However, analysis of qualitative datafrom Benin and Uttar Pradesh found that vil-lagers see participation in a Bank project pri-marily as a requirement for them to meet partof the subproject cost, and they see the advan-tage of meeting the 10 to 15 percent communitycontribution requirement, if that amount canleverage a much larger sum of money.

With this understanding of participation, andgiven that a large num-ber of communities aretrying to “attract” thelimited amount of donorresources, the existingsocial capital and the en-

ergy of the communities and their leaders ismarshaled toward ensuring the maximum re-source inflow to their village. Hence, as notedby OED’s Social Fund Evaluation, it appears thatthe participatory interventions are “users” ofexisting social capital rather than “producers” ofit. “Using” social capital may ultimately con-tribute to its increase, but this increase doesnot seem to be of the kind envisioned in a Bankintervention. The social capital that appears tobe strengthened is that which can ensure thateach community has the best opportunity toattract the maximum external resources. Whilethe literature confirms that social capital repre-sents a potential—a propensity for collectiveaction (Narayan 1995; Narayan and Pritchett1997)—and provides examples where investingin it has led to desirable returns through in-creased benefit flows (Uphoff and Wijayaratna2000), it also notes that whether or not its po-tential is activated and for what purpose de-pends on several factors (Krishna 2001).18

In conclusion, this study finds that CBD/CDDprojects can enhance social capital and foster em-powerment at the community level, but the linkbetween CBD/CDD and social capital and com-munity empowerment is weak. It also finds thatthe extent to which a Bank project is able to en-hance the capacity of whole communities is de-termined by various local social, cultural, andpolitical factors and by the very approach of theBank’s capacity building effort. The short timethat the Bank allows for implementation of asubproject can lead to the benefits of the ca-pacity-building efforts being cornered by thebetter-off in the community.

Targeting the Poor Is Not Enoughto Reach the Poor Even strong NGO interventions, such as the Pak-istan Aga Khan Rural Support Program, most re-cently evaluated in 2001 and operating for nearly20 years, have found it difficult to reach thepoorest. The reason it is so difficult is that it in-volves not just economic change, but also socialand cultural changes. Effecting such fundamen-tal changes requires considerable time and sus-tained effort of a sort that is unusual in aBank-supported project of any kind. 19

2 2


CBD/CDD projects can

enhance social capital

and foster empowerment

at the community level,

but the link between

CBD/CDD and social

capital and community

empowerment is weak.

Even strong NGO

interventions have

found it difficult to

reach the poorest.

CBD/CDD Projects Have Yet to OvercomeMajor Obstacles to Reaching the Poor The Bank-financed CBD/CDD projects have triedto reach the poor through targeting, but thereis limited evidence to show that they have donethis more successfully than any other Bank in-vestment. It is not surprising, therefore, that arecent literature review (Mansuri and Rao 2004)found that projects that rely on community par-ticipation have not been particularly effective attargeting the poor. A recent study on community-driven rural development projects carried out bythe Inter-American Development Bank notesthat the poorest and the most vulnerable gen-erally are not reached (Dahl-Ostergaard and oth-ers 2003). OED’s evaluation of the Aga KhanRural Support Program in Pakistan came to asimilar conclusion (OED 2002c).

OED project assessments have found in-stances of improved living standards, but theimprovement was greater for the better-offamong the communities than for the poor.20

This suggests that the project may have had lit-tle effect on socioeconomic factors. Specifically,the OED assessment for the Egypt Matrouh Re-source Management Project (approved in 1993)noted that although genuine attempts to reachthe poorer farmers were made, the gains to thelarge and medium-size farmers were more thanthose to small farmers, since many benefits werebased on land ownership. The literature alsosupports this finding.21

The thoroughness of the Bank’s effort alsofell short in some cases. OED project assess-ments and studies found that even when sup-porting an activity such as forestry, which canbenefit the poor, issues critical to their liveli-hood have not received adequate attention. Forexample, in Bank-supported community forestryinterventions in Nepal and India, marketing ofnon-timber forest products has been neglected(Kumar and others 2000; Kumar 2002). One rea-son for this is found in the literature, which notesthat the poor remain largely excluded from par-ticipatory “spaces” created by donor-supportedCBD/CDD interventions (Kumar and Corbridge2002; Turton and Farrington 1998).22, 23 Evenwhere they are “formally” included in a partici-patory “space” because projects may require that

there be representativesfrom the poorest in meet-ings, their views, as wellas their priorities, arelikely to remain excludedfrom collective decision-making processes.24

There are even caseswhere the position ofthe poor has actuallyworsened in the contextof a Bank project. For example, in the EasternAnatolia Watershed Rehabilitation Project (1993)in Turkey, OED’s assessment found “that therehad been some short-term losers, in particularlandless livestock owners.” While the assess-ment does acknowledge that village leaders gen-erally have attempted to allocate compensatorybenefits to losers, there have been challenges indoing this effectively. In India, the OED assess-ment of the Andhra Pradesh Forestry Projectfound that livestock herders, fuelwood head-loaders, shifting cultivators, and a dispropor-tionate number of women, all within the poorestgroups, may have been losers. In Nepal, theOED Review of Community Forestry (Kumar2002) noted several reasons why the poorestmay be the losers. The OED assessment of theBorgou Pilot Project in Benin noted that thecommunity contribution typically required inBank interventions created hardships for thepoor. It is very difficult for the poorest to maketheir cash contribution, so they usually have tocontribute time and labor, which takes themaway from income-earning activities. In situa-tions where the rich contribute on behalf of thecommunity, the position of the elite is strength-ened relative to that of the poor.25

Sophisticated Targeting Strategies MayHelp, but They Are Too New to Assess Some more recent CDD projects incorpo-rate quite sophisticatedpoverty targeting strate-gies (box 3.4), and theirability to achieve greatersuccess in reaching thepoor will need to be care-fully assessed upon com-


2 3

The poorest are likely to

remain excluded from

the collective decision-

making processes.

OED project assessments

have found instances

of improved living

standards, but the

improvement was

greater for the better-off

among the communities

than for the poor.

pletion. But findings from project assessments,ICR Reviews, country studies, and a desk reviewof project documents for six ongoing CDD proj-ects indicate several reasons why it may be dif-ficult to realistically implement these strategies.

First, political pressures in the country maymake things difficult. For example, in Indonesia,fieldwork done for the OED assessment of theKecamatan Development Fund Project foundthat while poverty-related survey data were usedfor ranking communities, provinces with lowerpercentages of poor still had to be included forpolitical reasons. The early findings of the OEDassessment for the Pakistan North West FrontierCommunity Infrastructure Project also reported

political interference inthe selection of commu-nities.

Second, projectscould find it difficult toreach the poor commu-nities because of datalimitations. For example,in Vietnam, governmentdata are being used toidentify poor communesin the latest CDD proj-ects (box 3.4), but thereis skepticism about the

reliability of the methodology used by the gov-ernment to identify poor people (Minot andBaulch 2004; Yukio 2001). In India, the lack of re-liable information on who the poor are has beena major constraint in the Andhra Pradesh DistrictPoverty Initiatives Project (2000).26

Third, in some countries, Bank efforts to reachpoorer communities have been constrained bytheir lack of easy accessibility. For example, thehousehold data for the forestry project in Mad-hya Pradesh indicate that, on average, commu-nity members in dispersed villages are less wealthythan those adjacent to towns that are less dis-persed. However, it appears that a relatively largerpart of the project resources were directed towardvillages of the latter type.

Substantive Participation of Women in theDevelopment Process Has Remained Elusive The Bank considers promoting gender inclusiona key design principle for CBD/CDD interventionsbecause women in most parts of the world ex-perience significant socio-cultural constraints ontheir participation in development. There aretwo aspects to assessing progress on gender in-clusion for CBD/CDD interventions: how muchhave women benefited from the participationprocess, and has the infrastructure or activitythat was supported specifically benefited them?

2 4


The two CDD projects in Vietnam—the Northern MountainsPoverty Reduction Project (NMPRP; fiscal 2002) and the Com-munity Based Rural Infrastructure Project (CBRIP; fiscal 2001)—include sophisticated targeting mechanisms to reach the poor.

The NMPRP targets 368 of the poorest communes in 44 districtsin the 6 provinces of the Northern Mountains Region using es-tablished government criteria (geographic targeting). Using par-ticipatory planning processes, the project attempts to involve thepoor within the targeted communes. Subproject selection criteriafavor small-scale subprojects implemented in or giving access tothe poorest and most remote villages within each project commune(self-targeting). The commune development budget component,

which puts small budgets under the control of the communes,will target the needs of the very poor households and groups (so-cial targeting).

The CBRIP targets 3,600 poor people in the 540 poorest com-munes in 13 provinces (geographic targeting). Poor communeswere selected based on the criteria established by the governmentfor its own Program 135. The project will support small-scale pub-lic infrastructure works intended to increase household incomes,improve living standards, and reduce poverty and vulnerability(self-targeting).

(See Annex O for definitions for different targeting mecha-nisms.)

Box 3.4: Example of Sophisticated Poverty Targeting in Two Recent Vietnam CDD Projects

Source: Vietnam Country Study.

CBD/CDD projects have

contributed to greater

“formal” inclusion of

women in participatory

“spaces” than was

possible in the past.

However, they have had

limited success in

promoting women’s

“substantive” inclusion.

The literature, OED assessments, and countrystudies show that because of specific gender-tar-geting strategies and the support these projectshave given to laws that promote gender inclusion,CBD/CDD projects have contributed to greater“formal” inclusion of women in participatory“spaces” than was possible in the past.27 However,they have had limited success in promotingwomen’s “substantive” inclusion, which, ac-cording to the literature, can be largely attributedto social norms that define gender roles (Agar-wal 2000a, b, 2001; Cornwall 2003). OED’s the-matic review of the community forestry programin Nepal noted that “the extent of women’s par-ticipation and involvement in user group activi-ties is a reflection of their position in Nepalesesociety. Community forestry, in so far as it em-powers women, can help improve their social sta-tus, but it is unreasonable to expect that it wouldradically transform gender relations” (Kumar2002). OED’s assessment of the Benin BorgouPilot Project noted the need to give careful at-tention to local social and cultural factors in fram-ing realistic gender-related project objectives. Inthe Matrouh Project in Egypt, where there hadbeen a substantial focus on women, focus groupmeetings with communities revealed that thepercentage of women who believed that theyhad benefited from the project was highly vari-able. The seven focus groups conducted foundthat there were no elected women leaders orwomen’s associations, and there were substan-tial concerns about marketing of products pro-duced by women.

Efficiency

Efficiency Has Not Been a PrimaryConsideration in the Bank’s Support forCBD/CDD ProjectsIdeally, efficiency should be addressed at twolevels in CBD/CDD projects: first, absoluteefficiency at the level of the individual project,as is normally expected in all Bank projects,through rate of return or other measures; and,second, comparative efficiency as an alterna-tive approach to development, comparing thecosts and benefits of attempting to undertakedevelopment through a CBD/CDD approach

versus achieving the same results through anon-CBD/CDD approach (Annex J).

Less than a Fifth of Closed CBD/CDD ProjectsHave Calculated an Economic Rate of ReturnAt the individual project level, an economic rateof return (ERR) has not been calculated eitherex-ante or ex-post for the majority of CBD/CDDinterventions. Of the closed projects in the pop-ulation of 847, only 24 percent attempted anERR at appraisal and 17 percent reported anERR at completion. This is significantly differentfrom 39 percent of the non-CBD/CDD projectsat appraisal and 34 percent at completion. Whileit may be reasonable to argue that an ERR can-not be calculated in CBD/CDD projects ex-antebecause the subproject investments are not iden-tified at appraisal, what reasonably can be cal-culated is expected ERRs for typical subprojectsthat the intervention is proposing to support. Cal-culation of ERRs at completion for at least a sam-ple of subprojects should be a requirement forall projects, because by then all costs and manybenefits are known or can be estimated. In manyICRs that do not calculate an ERR at comple-tion, the only reason given for not doing so is thatit was not done at appraisal.

In recent years, much more attention alsohas been given to impact studies. In several in-terventions, impact studies have been attemptedat midterm and completion. This study reviewedsome of these impact studies for their quality andfound them lacking in methodological sound-ness.28 A major reason for this seems to be thepoor data and lack of a reliable baseline for mak-ing comparisons. In the absence of baseline data,the “with-without” approach is attempted, butoften with limited success.

Bank Costs for CBD/CDD Have BeenSignificantly Higher than forNon-CBD/CDD Projects Other than limited at-tempts as a part of cost-effectiveness analysis,there has been nosystematic analysis ofcosts and benefits ofCBD/CDD projects in


2 5

OED found the

operational costs to

the Bank to be higher

for CBD/CDD than for

non-CBD/CDD projects.

2 6


Figure 3.7: CBD/CDD Projects Cost the Bank More Than Non-CBD/CDD Projects

Non-CBD/CDD CBD/CDD

Cost

s (0

00 U

S$)

0

100

200

300

400

500

600

IBRD/IDA commitment (million US$)

0–20 20–40 40–60 60–80 80–100 100–120 120–140 140–160 160–180 > 180

R2 = 0.6743N = 1493

R2 = 0.7894N = 374

Average Supervision Costs (closed investment lending)

Source: World Bank data.

Note: *Significant difference is noted between CBD/CDD and non-CBD/CDD projects for investment lending of less than $60 million for costs prior

to Board approval. The mean investment lending for CBD/CDD projects is $54 million. Results on costs were aggregated across projects and rep-

resent an average. Hence, individual project and country experience could vary. CBD/CDD costs are higher than non-CBD/CDD, even when costs

are normalized by loan size.

Non-CBD/CDD CBD/CDD

Cost

s (0

00 U

S$)

0

100

200

300

400

500

600

IBRD/IDA commitment (million US$)

0–20 20–40 40–60 60–80 80–100 100–120 120–140 140–160 160–180 > 180

R2 = 0.9274N = 2361

R2 = 0.7142N = 839

Average Costs Prior to Board Approval (investment lending)

comparison with non-CBD/CDD projects fromthe Bank’s perspective. The only relevantsource that OED could locate is a workingpaper (Hentschel 1994) that undertook a

comparative study of costs for preparation ofBank participatory versus non- participatoryprojects. It found those for the former to behigher. For this study, OED compared

operational costs to the Bank of CBD/CDDversus non-CBD/CDD projects and found themto be higher for CBD/CDD by about 10 percent(figure 3.7 and Annex J). Bank staff perceptions(Annex L), revealed through a staff survey, arein line with these findings. No study has yettaken this issue further to explore what thishigher cost to the Bank means in comparisonwith benefits from a CBD/CDD approach.

For the Bank, these higher costs could bejustified under two conditions: first, if the highercosts are fixed, and ultimately—as the Bank doesmore CBD/CDD—its average costs for under-taking these interventions would potentially fall;and, second, if the benefits to the client coun-tries are going to be so large from the new ap-proach that the Bank is willing to bear theadditional costs in the interest of achievingpoverty reduction. There is no evidence thatthe first condition is going to hold. If Bank proj-ects adhere to a true learning process approach(Korten 1980), each community interventionmust be made specific to the needs of the par-ticular community involved. Under these cir-cumstances, the likelihood of a decline in costsresulting from scaling up is low.

The Cost to the Government of Introducing aCBD/CDD Approach Has Been Substantial...In dealing with the second condition, the surveysof government officials in case study countries(Annex I) revealed that there is a substantialcost in time spent by government officials inputting a participatory approach in place, thoughit is reasonable to expect that these costs will de-cline over time. About 75 percent of central gov-ernment officials in Benin, Nepal, and Vietnamand 75 percent of local government officials inBenin and 80 percent in Brazil recognized the in-creased time and, hence, cost implications for thegovernment in initially putting a participatoryprocess in place.

Though the Actual Costs to the Governmentfor Infrastructure Have Been Lower…The cost to the government for infrastructurethrough CBD/CDD projects was found to belower than that for non-CBD/CDD projects. Stud-ies carried out in Nepal (SAPROS & IFAD 2002;

SAPROS and World Bank2000) and work done forthe Egypt and Brazil casestudies and an OED as-sessment in Indonesiaconfirm this finding.

…Communities Bear a Part of the Cost ofService Delivery Infrastructure inCBD/CDD Interventions Lower infrastructure unit costs to the govern-ment are often the result of communities’ shar-ing in the cost of construction and contributing(or providing) for operation and maintenance.The Bank’s recent self-evaluation of social fundsin six countries (World Bank 2003d) also notedthat community management of investmentsprovides a significant opportunity for cost savingsof as much as 25–50 percent.29 However, themerit (or its lack) of shifting part of these coststo the communities has not been a factor in thedecision to increase support for CBD/CDD proj-ects. As the literature shows, the cost of com-munity participation (in cash, kind, or labor) canbe substantial, particularly under the more in-tensive participatory approaches.30 If the op-portunity cost of the time that communitymembers spend in meetings with donor and gov-ernment officials is also considered, the costsare higher still. The lack of discussion of theissue of beneficiary time in any appraisal reportssuggests that the opportunity cost of time is as-sumed to be low for households in poor com-munities. However, farm management studiesover the years have generally shown this not tobe so, and have shown high costs at peak agri-cultural labor periods (Schultz 1964; Collinson1982; Renata and Houston 2002). 31

Further, considering costs only in terms of in-frastructure construction and flow of services isnot enough (even if it takes into account com-munity costs) because CBD/CDD projects, andparticularly CDD projects, are also trying to en-hance the capacity of the communities. It is thusnecessary to consider the cost for the capacity-enhancing exercises, both in resources and intime spent by government officials and by com-munities. By estimating community time forboth subproject implementation and capacity


2 7

There is a substantial

cost in time spent by

government officials in

putting a participatory

approach in place.

enhancement, it may be possible to approxi-mate overall costs for a CBD/CDD intervention.(See Annex J, box J.1, for an example.)

Undertaking a Comparative Assessmentof All Benefits Constitutes an EquallyChallenging ExerciseBenefits from capacity-enhancing activities areparticularly difficult to assess—for instance, whatshould be done if the gains from the capacity-enhancing exercise are concentrated in a smallsection of the population? Is that a positive or anegative? The only sure way to assess benefits ap-pears to be in terms of the poverty impact. IfCBD/CDD projects are delivering poverty-re-duction benefits, and non-CBD/CDD projectsare failing to do so, or doing so inadequately, thenspending more to get the desired results may beworthwhile. However, if poverty-reduction im-

pact is not noticeably better for CBD/CDD proj-ects, then the extra costs of the CBD/CDD ap-proach may not be worthwhile. Even thoughpoverty reduction is the major objective of manyof these projects, a full-fledged comparative as-sessment of the poverty impact of different ap-proaches would require further research. Thelimited evidence available is in the form of thebeliefs of the government officials in the casestudy countries and some limited references inthe literature.32 Government officials in four casestudy countries reported that projects with par-ticipation have better outcomes than those with-out participation (Annex I). It is also commonlyargued that if communities are willing to con-tribute for subprojects, they must perceive thebenefits of the CBD/CDD interventions to be atleast equal to the costs.

2 8


2 9

Institutional Enhancement and Sustainability

This chapter draws on country studies, project assessments, and the lit-erature to discuss the extent to which CBD/CDD interventions haveenhanced the borrowers’ capacity at the central, state, and local

government levels to allow them to make effective use of their human, financial,and natural resources.

Recent CBD/CDD projects have given greateremphasis to building partnerships betweencommunity groups and local governmentorganizations, and decentralization reform.While it is beyond the scope of this evaluationto delve into the merits and demerits ofdecentralization, it does attempt to assesswhether the design and implementation ofBank-supported CBD/CDD projects has helpedenhance the capacity of local governments inclient countries. Capacity at the communitylevel, a critical issue for CBD/CDD projects, iscovered in Chapter 3.

This chapter also assesses the sustainability ofCBD/CDD projects. Interviews with Bank staff re-vealed that in the context of CBD/CDD inter-ventions, the understanding of sustainabilityitself varies among Bank staff. For some it impliespredominantly sustainability of communityprocesses, for others CBD/CDD is simply a meansto an end, and sustainability is related to infra-structure investments. For still others, it is relatedto overall resource allocation, including supportfor decentralization.

Institutional Enhancement

Institutional Development Impact HasImproved over Time

Projects Have Not Brought about theRadical Reorientation in InstitutionsRequired to Undertake CBD/CDD, but TheyHave Helped Enhance Government Institutions The literature shows that the institutionalizationof a CBD/CDD approach requires a radical re-orientation of the way governments and bu-reaucracies operate (Shepard 1998; Thompson1995). This requires changes in managementand organizational procedures, as well as in theattitudes and behaviors of personnel, that taketime to consolidate (Pimbert and others 2000).This study found that such a radical reorientationhas not yet come about in most of the Bank’sclient countries, as seen by the experience inOED case study and project assessment countries,although CBD/CDD interventions have helpedchange the attitude of government officials andenabled supportive policy and legal reform.1

44

Projects Have Supported Changes in theAttitude of Government Officials towardWorking with Communities Surveys of government officials in all five casestudy countries and findings of project assess-ments indicate greater acceptance at all levels ofgovernment of the value of involving commu-nities in service delivery/activities and greaterunderstanding and ability to implement a par-ticipatory approach to development. For exam-ple, the OED assessment of the Eastern AnatoliaWatershed Rehabilitation Project in Turkey notesthat at the central, provincial, and community lev-els it was widely acknowledged that the Bank hadintroduced new ideas related to community par-ticipation processes. The OED assessment for theCommunity Development Fund Project in Eritreaalso notes that the project’s emphasis on com-munity contribution was perceived as a usefullesson by line agencies in the country.

Projects Have Supported Policyand Legal Reforms The presence of the Bank has often provided theincentive and opportunity for the governmentto “push” for legal and policy changes on polit-ically sensitive issues. OED studies indicate andproject assessments confirm that the Bank’spresence in sectors such as forestry encour-aged the governments of India and Nepal totake action on several critical policy issues thathelped generate momentum for change in thesector. However, important overarching policyissues have occasionally been shortchanged, asin Egypt, probably because of the burden oftrying to get the community participation inplace.2

It appears that mostcountries today subscribeto an approach to partic-ipation that is some-where between a fully“bottom-up” and a fully“top-down” way of doingdevelopment. As notedin Chapter 3, surveys andinterviews of govern-ment officials in casestudy countries suggest

that officials at the central and local levels in-terpret community participation a little differentlythan does the Bank—more as a means of doingmore with less, rather than actually putting com-munities in control. As also noted, however, at-titudes are changing.

But neither the policy and legal reforms northe change in attitude guarantee that theCBD/CDD approach will be implemented effec-tively. Several factors could challenge the trans-lation of reform initiatives into effective workingarrangements. The following sections exploresome of the issues around policy reform, in-cluding the connection with decentralization.

CBD/CDD Projects are Hampered by WeakCoordination across Government Departmentsand Government Levels Although the majority of central governmentstaff in case study countries report an increase inthe number of interministry meetings held(Annex I), progress in actual coordination hasbeen limited, at least in the case study countries.Interdepartmental coordination problems ariseprimarily because government ministries con-tinue to be organized sectorally, and the sectoralculture is so firmly ingrained that it is difficult fordepartments to work together in the context ofa Bank intervention. While weak coordinationbetween government departments negatively af-fects the implementation of both CBD/CDD andnon-CBD/CDD interventions, the negative im-plications are greater for CBD/CDD interven-tions, which also require interdepartmentalcoordination at the local level. For example, inEgypt, most government staff interviewed be-lieved that interdepartmental coordination atthe higher levels across sectors had improvedsomewhat over the past few years, but coordi-nation problems among government staff at thecommunity level persist. Each CBD/CDD projectis implemented by a different department and hasits own unique mechanism for community con-sultation and subproject phasing.

Further, despite progress on decentralization,the Benin, Nepal, and Vietnam country studiesfound that relations between different layers ofgovernment remain difficult. Capacity at thelower levels of government in all three countries

3 0


The presence of the Bank

has often provided

the incentive and

opportunity for the

government to “push”

for legal and policy

changes on politically

sensitive issues.

continues to be weak, and the roles of officialsat various levels are not clearly defined. As in thecase of coordination among government de-partments, weak capacity at the local govern-ment level is more problematic in the context ofCBD/CDD interventions, where local officialsare often directly involved in project imple-mentation. Only about 20 percent of Bank staffsurveyed reported being satisfied or very satis-fied with coordination within the government ofcountries borrowing for CBD/CDD interventions(Annex L).

There Is No Evidence to Indicate That theParticipatory Approach Has Been WidelyAdopted beyond Bank CBD/CDD ProjectsThe evidence from country studies and projectassessments also indicates that while a Bankproject may succeed in getting a country to ex-periment with (or adopt) a demand-driven ap-proach in a project context in a particular sector,other government departments do not neces-sarily support the approach. For example, inVietnam, evidence from an internal Bank reviewindicates that while the Coastal Wetlands Pro-tection and Development Project (2000) is un-dertaking mangrove plantations with beneficiaryparticipation, the government has its own sep-arate programmatic approach to replanting man-grove coastlines. The report notes that it isunclear how government ownership will beachieved for the Bank project when there is a rivalgovernment project under way, or why the Bankis not financing the government program, withwhatever cost-effective improvements couldhave been devised and negotiated. In Indonesia,while the field mission for the OED assessmentfound quite strong evidence that the processesintroduced by the Kecamatan Development FundProject had been accepted by the governmentat the district government level and below as theright approach for development, sustainability re-mained difficult to predict and depended to alarge extent on the attitude of the new govern-ment toward decentralization.

Finally, the Bank itself has not had a consis-tent policy across projects in the same country,a point that is evident in its approach to decen-tralization reform.

CBD/CDD Projects Have Increasingly SupportedGovernment Decentralization ReformSupport for decentralization reform and en-hancing local government capacity underCBD/CDD projects has increased significantlyin recent years. Of the 84 sample projects, 57 sup-ported some form of government decentraliza-tion, and the largest percentage of these was inthe later years. Analysis carried out for this studyas a part of the Portfolio Review and countrystudies revealed two fundamental factors thatcould constrain progress in decentralization, nomatter how well-designed the CBD/CDD project.First, success in promoting decentralization in acountry depends on borrower commitment tothe reform process. For example, in Egypt, thecountry study found that progress on decen-tralization has been limited primarily because oflack of government ownership. In this context,the CBD/CDD projects reviewed in that countrycould not accelerate decentralization and re-sponsiveness at the local level. Second, successcan also be constrained if there is lack of har-monization between the legal framework for de-centralization and the level at which the Bank’sCBD/CDD project is implemented. For exam-ple, in Benin, the OED assessment of the BorgouPilot Project found that the project supported ac-tivities at the village level, while decentraliza-tion stopped at the commune level.3, 4

Individual CBD/CDD Projects AdoptVarying StrategiesThe evidence from four of the five country stud-ies indicates that the Bank has not followed aconsistent decentralization strategy under itsCBD/CDD projects within each country. Forexample, in Nepal, recognition of weak capacityat the local level has ledBank-supported CBD/CDD projects to providetechnical assistance tostrengthen local govern-ment institutions. Yet theBank has also supportedcreation of “temporary”arrangements for the im-plementation of CBD/CDD projects at the local

INST ITUT IONAL ENHANCEMENT AND SUSTAINABIL ITY

3 1

Support for decen-

tralization reform

and enhancing local

government capacity

under CBD/CDD

projects has increased

significantly in

recent years.

level. These arrangements have had a negative ef-fect on the progress of decentralization. The im-plications are particularly worrying when formaltraining and other capacity-enhancing activitiesfor district-level organizations fall short of plans,as in the Rural Infrastructure Project in Nepal. Thisperpetuates the need to continue the “tempo-rary” arrangements for a longer period, imped-ing further decentralization progress (box 4.1).Perhaps it is this inconsistency in Bank strategythat is picked up in the Central Government Sur-vey in Nepal, where only 19 percent of officialsinterviewed said that the Bank has the capacityto enhance local government capacity to sup-port participatory interventions (Annex I). Sim-ilarly, in Vietnam, while projects such asCommunity-Based Rural Infrastructure (2001)have provided technical support to local gov-ernments, their implementation has proceededthrough establishment of temporary parallelstructures that manage implementation at thelocal level under central ministries.

Brazil, where the Bank has been supportingCDD projects in the Northeast for more than adecade, is another case where the Bank has pro-

vided support for parallelstructures. The CDD proj-ects in the Northeast areessentially building a struc-ture that is parallel to the

planning process of the municipal government(box 4.2). By instituting ad hoc municipal councilsfor implementation, the project has contributed tothe proliferation of municipal councils, with littlecoordination between them. In Rio Grande doNorte, OED found that municipal councils hadweak capacity, in part because they had receivedlittle training.5 Although the project Technical Unitprovides assistance to the councils, it is not enough.Most of the councilors interviewed said theyneeded more assistance, as they often need clar-ifications on issues, without which they cannotmake progress. However, it is important to put thelow level of institutional development of the proj-ects’ municipal councils into context. Many of theother municipal councils share similar weakness,and some are less effective than the Bank-sup-ported councils. Indeed, it is worth asking whetherit is even beneficial for both the municipality andproject or program to have so many councils.Were the disparate efforts for capacity building tobe focused on a single integrated municipal coun-cil, the result would likely be a stronger, more ef-fective, and efficient council.6

This lack of consistency in Bank support to de-centralization under CBD/CDD projects withinthe same country is particularly visible in coun-tries where the Bank has been supporting proj-ects under different institutional arrangements.This has the potential to send a conflicting mes-

3 2


To assist with the limited capacity of District Development Com-mittees, the government established the Department of Local In-frastructure Development and Agricultural Roads (DOLIDAR)under the Ministry of Local Development, with branches in thedistricts. DOLIDAR’s mandate, which is to provide guidance andtechnical support to district committees in development activi-ties, has often stretched to interference with the committees’ au-tonomy. While the ICR for the Rural Infrastructure Project claimsthat a balance between the roles of DOLIDAR and the districtcommittees in implementing the project was achieved, the verypresence of a temporary solution and a branch of a central gov-

ernment organization at the district level has two implications.First, a significant amount of resources get spent on strength-ening the temporary structures. A part of the resources from theRural Infrastructure Project also went toward strengtheningDOLIDAR. These resources could have been spent on strength-ening the district-level bodies. The proposed Rural Access Im-provement Project, which is still under preparation, is expectedto continue to provide capacity-building support to DOLIDAR.Second, the presence of the parallel structures sends a confusingmessage about the authority of the district-level bodies to takecharge of development activities.a

Box 4.1: Inconsistent Strategies: The Road Sector Experience in Nepal

Source: Nepal Country Study.a. Management notes that most CDD operations in Nepal have performed well.

The Bank may not have

supported a consistent

decentralization strategy.

sage to the borrower and district officials, espe-cially when both projects are expected to resultin similar poverty outcomes. For example, in thePhilippines, the Bank is supporting more than twocompeting institutional arrangements.7 Pilotingwith different institutional arrangements makessense if the purpose is to study which wouldwork best before scaling up. But supporting dif-ferent arrangements side by side over large areason a long-term basis does not send the right sig-nals to the borrower and does not augur well forlong-term institutional development.

Recent Social Funds Have Given Much MoreAttention to Decentralization IssuesOn the positive side, the question of whether,how, and in what circumstances social funds cansupport decentralization is receiving increasingattention (OED 2002b). This is a notable im-provement, since older social funds frequentlyresulted in structures outside local governmentthat have had limited (perhaps even negative) im-pact on enhancing local government capacity.

Some projects, such asthe Zambia Social In-vestment Fund (ZAM-SIF), approved in fiscalyear 2000, are designedto integrate into thelarger decentralization ef-fort in the country. Dis-trict authorities are to bedevolved increasingresponsibility in the project cycle for community-level subprojects. But while supervision docu-ments report on the considerable amount ofcapacity enhancement of district officials beingundertaken, they also note that the actual trans-fer of social fund responsibilities to the districtsis taking much longer and is a much more diffi-cult exercise than was visualized.

NGOs Have Been DevelopmentPartners in CBD/CDD According to Bank data, 36 percent of theCBD/CDD portfolio, versus 8 percent of the non-


3 3

Brazilian municipalities have large numbers of municipal coun-cils (IBGE 2003). Many federal programs require the creation ofan ad-hoc council to implement them at the municipal level.These councils typically have representatives from both gov-ernment and civil society organizations. Field research in RioGrande do Norte found that many of these councils are weakin capacity and that there is a substantial overlap in their mem-bership. In rural agricultural development, the issue of councilproliferation is particularly evident. Municipalities receive fundsfrom two main sources, the World Bank (through the RuralPoverty Alleviation Program, RPAP, and RPRP [Rural Poverty Re-duction Project]) and the Ministry of Agriculture (through itsPRONAF [Brazilian Federal Program to Support Family Agricul-ture] program). Both funding bodies require that municipalcouncils be set up to implement their programs.a These two typesof councils differ in two main respects: membership structureand funding modality. While the Bank requires civil society tohold the majority of the seats in FUMAC (Municipal Community

Scheme) councils, in PRONAF councils, representation of civilsociety and the government is equal. Under PRONAF, funds aretransferred to the municipal government, which is responsiblefor allocation. Communities do not manage funds directly, butreceive equipment and infrastructure from the municipal gov-ernment. Under the Bank’s program, the municipal governmentnever manages project funds, which are transferred directly, orthrough the FUMAC-P (Pilot Municipal Community Scheme)council, to the communities. These differences (especially thesecond) render it difficult for municipalities to argue for the fu-sion of the two councils, even if the overlap in membership isoften significant. Only one of the 13 municipalities surveyed inRio Grande do Norte was able to persuade the 2 funding bod-ies that a single council for rural development constitutes a bet-ter institutional arrangement, and that having two parallelmunicipal councils that work in an uncoordinated fashion onrural development is likely lead to a suboptimal allocation of re-sources.

Box 4.2: Bank CDD Projects Have Added to the Proliferation of Municipal Councilsin Rio Grande do Norte (Brazil)

Source: Brazil Country Study.a. Follow-on Bank projects use the municipal councils established under earlier projects.

The question of whether,

how, and in what

circumstances social

funds can support

decentralization is

receiving increasing

attention.

CBD/CDD portfolio, hadsome form of NGO in-volvement. NGOs havetraditionally providedcommunity developmentsupport, including par-ticipatory diagnosis andthe preparation of localdevelopment plans. Insome projects, such as

the Uttar Pradesh Rural Water Supply and Envi-ronmental Sanitation Project (1996), however,NGOs have also provided engineering supportto communities. NGOs have also played a rolein design and implementation of communitysubprojects, as in the HIV/AIDS projects in Africa.

CBD/CDD Projects Have Contributedto the Development of NGO Capacity The evidence from several project assessmentsand all five country studies shows that local NGOcapacity varies widely across—and even within—countries, and it is difficult to generalize. How-ever, several Bank CDD projects, such as theUttar Pradesh Sodic Land Reclamation Project,

have also provided support for development ofNGO capacity.

Evidence from the Bank’s database (figure4.1) and the case study countries shows thatmiddle-income countries (Brazil, Egypt, andTurkey among the cases) have a limited numberof NGOs, and consequently less participation ofNGOs in CBD/CDD projects compared with low-income countries. Among other reasons, this isbecause of the presence of other private andpublic technical assistance providers in the mid-dle-income countries. Among the low-incomecountries, Vietnam has active international NGOsbut few active local NGOs. In contrast, Beninand Nepal have numerous active local NGOs.In Benin, particularly, the country study foundthat momentum created by CBD/CDD inter-ventions was largely responsible for the recentmultiplication of local NGOs. Elites in the coun-try that traditionally were a part of the govern-ment began forming NGOs and increasinglyshifted their focus toward donors, where re-sources, and hence opportunities, were avail-able.8, 9 What this means for the future is stillunclear. However, surveys show that a largely

3 4


Figure 4.1: Low-Income Countries Work with More Partners

Perc

enta

ge o

f pro

ject

s

Donor involvement NGO partnership0

10

20

30

40

50

Lowincome

Low-middleincome

Upper-middleincome


Middle-income countries

have a limited number of

NGOs, and consequently

less participation

of NGOs in CBD/CDD

projects compared with

low-income countries.

shared opinion among both donor agency rep-resentatives and government officials in thecountry is that these intermediaries have little ac-countability toward either the communities orthe government.

Partly as a result of the difference in capacityamong NGOs, the degree of success of the part-nership between the Bank and NGOs has alsovaried. This was also the finding of an OED studyof NGO involvement in all Bank projects (OED1999a). Bank staff, when asked in the staff sur-vey whether NGO-supported interventions gen-erally achieve a better outcome than Bankinterventions, were fairly evenly divided: 31 per-cent disagreed or strongly disagreed, 26 per-cent agreed or strongly agreed, and 24 percentfell in between. Such responses may not be par-ticularly revealing because this is a complicatedquestion, and the answer is likely to depend onthe respondent’s experience. NGO focus groupsin country studies, however, revealed dissatis-faction among NGOs with the unequal relation-ship that they have with the Bank in a CDDintervention and with Bank procedures.

NGO Partnership Can Be Importantin Reaching the Poor The Portfolio Review, project assessment find-ings, and country studies found that local NGOs,because of their familiarity with local conditions,have been important in helping some CBD/CDDprojects to reach the poor and disadvantagedpopulations and as catalysts in mobilizing com-munities. For example, NGOs helped mobilizefarmers and disseminate technology, and assistedvillage-level institutions in developing links withgovernment agencies in the Uttar Pradesh SodicLand Reclamation Project. OED’s project assess-ments also found that NGOs played an importantrole in enabling the Peru Rural Roads Rehabili-tation and Maintenance Project to exceed itstarget number of villages benefited and in im-plementing the Andhra Pradesh Forestry Pro-ject. That said, the Benin country study found thatNGOs with poor qualifications handicapped proj-ect implementation in that country. In both theSocial Fund and the Food Security Projects inBenin, a large number of NGOs had to be sus-pended for unacceptable performance. Many

stakeholders, including Bank staff and otherdonor agency representatives, expressed con-cern about the potential role of NGOs in a num-ber of CBD/CDD projects in Benin, noting thatwhen they are paid a fixed proportion of thetotal cost of the project, NGOs tend to push thechoice of the subprojects that are the most costlybut have the minimum operating cost. WhenNGOs behave this way, they may compromisethe participatory process. The OED assessmentof the Northern Resource Management Projectin Pakistan found that there are both pros andcons to handing over all or most community mo-bilization to NGOs. The assessment notes, “Infavor of such an approach is the generally ac-knowledged skills of NGOs. Against it, however,is that public technology transfer will continue tobe needed and, in order to play a key role, pub-lic services can benefit from the first-hand expe-rience of participatory approaches. The aimshould be to work out effective and efficient ser-vice delivery arrangements involving all actors:central government, NGOs, private sector, localgovernments, and local communities.”

Most Donors Support Community Participationas a Strategy, but There Is Lack of Agreementon Implementation Procedures Bank data show that 40 percent of bothCBD/CDD and non-CBD/CDD projects have mul-tidonor involvement. Several evaluations of Bankprojects (CBD/CDD and non-CBD/CDD) havehighlighted problems that arise when donor ef-forts are not well coordinated. However, thecase study and project assessment evidenceshows that the challenge of donor coordinationis greater in a CBD/CDD project than in a non-CBD/CDD project. The country studies in Benin,Nepal, and Vietnam andfieldwork for project as-sessments in Benin,Ghana, and Mali confirmthat several donors areoften present in the samecommunity, with differ-ent CBD/CDD strategies,providing infrastructureor activity support in aseemingly uncoordinated


3 5

Local NGOs have been

important in helping

some CBD/CDD projects

to reach the poor and

disadvantaged

populations and as

catalysts in mobilizing

communities.

manner. Each donor in-tervention at the com-munity level may requirea separate committee tomeet the implementationrequirements. The sub-stantial confusion that

this creates at the community level is a muchmore serious issue than the strain on institu-tional capacity at the government level. Thirty-nine percent of the Bank staff surveyed agreedor strongly agreed with the observation that in-adequate donor coordination in a cofinancedproject is likely to have a greater negative impacton outcomes in a CBD/CDD project than in anon-CBD/CDD project.

Interviews with Bank and other donor staff forthree of the five country studies found that al-though most donors endorse community par-ticipation as a strategy, there is less agreementon implementation plans and procedures. For ex-ample, in Nepal, the coordination efforts of in-dividual donor representatives are hindered bya feeling of competition among donors. Mostdonor officials interviewed in Benin said that

the coordination issue could be resolved by al-locating different sectors among donors basedon their comparative advantage, with regularmeetings among them to keep the group in-formed. In Vietnam, in contrast, nearly all donorrepresentatives said that stronger coordinationby the government could resolve the issue.

Coordination among Donors Is MoreDifficult for Small, Low-Income CountriesThan for Middle-Income CountriesThe experience of Benin, Nepal, and Vietnam alsoindicates that the presence of numerous bilateraland multilateral donors, each implementingCBD/CDD projects, but with different institu-tional arrangements and procedures, has put atremendous coordination burden on these gov-ernments and stretched their limited institu-tional capacity. This finding is supported byevidence from the literature.10

In middle-income countries, however, theBank’s data reveal (see figure 4.1), and the ex-perience of the Brazil and Egypt country studiesconfirms, that fewer donors are involved inCBD/CDD projects in those countries. In Brazil,in the state of Rio Grande do Norte, for exam-ple, no other donor is supporting CBD/CDDprojects. In addition, countries such as Egypthave better administrative and institutional ca-pacity to handle coordination.

Few of the Poverty Reduction Strategy Papers(PRSPs) reviewed—but most Country AssistanceStrategies (CASs)—note donor coordination as anissue (Annex H). Since PRSPs are meant to bethe outcome of a country-driven process of dis-cussion on priorities and challenges, this findingindicates that for the large majority of the bor-rowers, this issue may not be high on their agenda.

Sustainability

Infrastructure and Activities Have BeenDifficult to Sustain beyond the ProjectsThe sustainability of the CBD/CDD portfolio hasbeen improving, but shows considerable scopefor further improvement (figure 4.2). The sus-tainability ratings have varied by Region, with thehighest rating in the Middle East and NorthAfrica, followed closely by Latin America and the

3 6


Figure 4.2: Sustainability Has Been ConsistentlyLower for CBD/CDD Projects but Is Improving

Like

ly o

r bet

ter (

%)

1994–98 1999–20030

40

20

60

80

100

Exit fiscal year

CBD/CDD Non-CBD/CDD

37%

48%55%

63%


The challenge of

donor coordination is

greater in a CBD/CDD

project than in a

non-CBD/CDD project.

Caribbean11 (Annex G, table G.3). As with the out-come ratings, some of the larger CBD/CDD proj-ects show higher sustainability compared with alarge number of smaller ones.

About 37 percent of the OED project assess-ments that inform this study rate sustainabilityas unlikely, and more than 30 percent rate it ei-ther nonevaluable or uncertain. Even those thatrate sustainability likely raise concerns eitherabout maintenance of supported infrastructureor activities, as in the OED assessment for theEgypt Matrouh Development Project, or lack ofadequate analysis of the capacity of the govern-ment with respect to government contributionduring the project and post-project phase, as inthe OED assessment of the Pakistan Northern Re-source Management Project.12

Scarcity of Resources for Operations andMaintenance Has Been a ConstraintCBD/CDD projects have supported a largeamount of social infrastructure—such as schoolsand heath centers—at the community level inmany countries. These have been difficult tosustain beyond the Bank intervention. Mansuriand Rao (2004, p. 32), quoting studies on waterprojects in Sub-Saharan Africa and South Asia,found that even if communities are initially suc-cessful in creating the project, they may lackthe material resources to sustain their efforts.OED project assessments also show that poorcommunities find it difficult to raise resources in-ternally to provide for continuous operation andmaintenance (O&M).13 Governments, too, are fis-cally constrained.

A previous OED study (Kumar 2003) notesthat Bank projects typically do not plan for si-multaneous investment in social and produc-tive sectors. Doing so could ensure long-termsustainability, because village-level capacity toprovide for O&M of social investments ultimatelydepends on increased capacity to generate rev-enue at the local level. Field research in Brazilfound that most of the communities in the stateof Rio Grande do Norte had only one subprojectapproved, largely because project municipalcouncils could not justify a second investmentin any community before all communities hadbeen covered with at least a single investment.14

While the Portfolio Re-view shows that CDDprojects provide for agreater role for commu-nities in O&M of sub-projects than CBD/CDDprojects (figure 2.2), the resource constraint re-mains an issue even in these projects.15

Several Bank projects require the setting upof maintenance funds.16 The Benin country studyfound that commitment to maintenance fundsis significantly reduced if communities knowthat they can go to another donor or to the gov-ernment when earlier investments cease to befunctional.17, 18 The uncoordinated presence ofseveral donors and NGOs in the same villages insupport of different activities can foster de-pendence and soften the commitment of com-munities toward maintenance activities.19

Fieldwork carried out for the OED assessmentof the Indonesia Kecamatan Development Pro-ject found that the community approach is gen-erally to postpone maintenance until it isunavoidable—for example, the road is about tobecome impassable, or the bridge is about to col-lapse, and then to do the minimum that will getit back into usable shape. OED’s project assess-ment of the Mali Natural Resource ManagementProject also found that a large amount of infra-structure was constructed at the village level,but there were few arrangements made at thecommunity level for its maintenance. Hence, al-though community members were aware of theimportance of maintenance and a significantnumber of them had been trained, the lack ofavailable resources on a continuous basis toallow for O&M was a significant constraint.

Maintaining a Quality Flow of Servicesor Income Has Proven Even More Difficult The World DevelopmentReport Making ServicesWork for Poor People(World Bank 2004d) ac-knowledges the chal-lenge of ensuring servicedelivery.20 While it is ar-gued that services can beimproved by changing


3 7

Social infrastructure has

been difficult to sustain

beyond the Bank

intervention.

Bank projects typically do

not plan for simultaneous

investment in social and

productive sectors. Doing

so could ensure long-term

sustainability.

the relationships of accountability as with CDD,there is little evidence yet. A recent Bank re-view (Wassenich and Whiteside 2004) found thatthere is little evidence regarding CDD impactson quality of service delivery. OED project as-sessments and a review of ICRs show that mostCBD/CDD projects, including CDD projects, givelittle thought at the design stage to the issue ofmaintaining a quality flow of services or income.21

Hence, even though infrastructure may be stand-ing in a village or community, it is often under-used or not being used for the purpose that wasoriginally intended. Further, the flow of servicesor income from an infrastructure requires con-sideration of issues related to coordination withdifferent government departments. It also impliesincreased financial responsibility for the gov-ernment in the form of supplemental resources,such as to pay the salaries of teachers and doc-tors. Annex P draws on the experience of theprojects in Benin to illustrate this point in thecase of school subprojects.

And Some Types of Services Have BeenMuch More Difficult to Maintain Than OthersOED project assessments and country studyfindings show that sustaining a quality serviceflow from some types of infrastructure dependson the scale and complexity of the service. Forexample, quality flow of education services froma school will require coordination with the ed-ucation department to ensure the availability ofcertified teachers and books, as well as adherenceto a centrally planned curriculum, among otherthings. Key informant interviews in Benin re-vealed that 50 percent of the teachers in schoolssupported under the Borgou Pilot Project and 80percent of teachers in schools supported under

the Social Fund were notstate certified; in thecomparator schools,fewer than a third werenot certified (Annex P).In Eritrea, the OED as-sessment of the Com-munity DevelopmentFund Project found thathealth centers supportedunder the project have

been negatively affected by the severe overallshortage of doctors in the country. In contrast,flow of services from a village water supplyscheme can be maintained with minimum tech-nical support from outside the village. Ironically,in several of the poorest countries where theinstitutional environment is the weakest, andcoordination among various government de-partments the biggest problem, Bank projects aresupporting social infrastructure such as schoolsand health centers. In contrast, in Brazil, In-donesia, and the Philippines, where the level ofinstitutional development is higher, BankCBD/CDD projects have provided support for ac-tivities that often do not require the same kindof coordination and support from higher levelsof government, such as water supply schemesand roads.22 Since the focus is on the number ofschools and health centers built, the actual issueof how the flow of services from these facilitieswill be maintained receives less attention.

Villagers May Also Not Have the NecessaryInformation or Technical Knowledge Project assessment findings show that commu-nities often may not have the information andtechnical expertise they need to allow for main-tenance. The OED assessment of the CommunityDevelopment Fund Project in Eritrea notes thatin addition to ownership and willingness of dif-ferent actors to sustain investments, sustain-ability requires that the relevant actors have thefinancial, managerial, and technical capacity tooperate and maintain the infrastructure. Focusgroup sessions with villagers in Benin found thatseveral communities had not received adequatetraining to be able to maintain their subprojects.In Uttar Pradesh, where the Bank supportedsodic land reclamation, villagers are interestedin keeping the reclaimed land from reverting toits former state, but the OED fieldwork foundthat they may not have the technical knowledgeor capacity to do so.23 Maintenance of maindrains critical to the sustainability of reclaimedlands could only be done by the Irrigation De-partment. The implementing agency staff be-lieved that the political pressure from the farmerswould be sufficient to ensure that the govern-ment provided adequate resources to the Irri-

3 8


Since the focus is on the

number of schools and

health centers built, the

actual issue of how the

flow of services from

these facilities will

be maintained receives

less attention.

gation Department to maintain the main drains.However, household survey data analysis foundthat 96 percent of the farmers in the CDD com-munities were unaware that the Irrigation De-partment had the responsibility for maintainingthe main drains. Further, the household surveyanalysis also indicated that most villagers werenot even aware of the critical importance ofdrainage for containing sodicity (Annex P).

Lack of Clear Communication about theRole of Communities in a Bank ProjectHas Added to the ProblemAs already indicated, villagers mainly see partic-ipation in a Bank project as a requirement forthem to meet part of the subproject cost. Theproject assessment of the Uttar Pradesh SodicLand Reclamation Project also found that thecommunity members’ understanding of theircontribution to the participation process hasled them to expect support from the governmentfor maintenance of drains that were critical to sus-tainability. In a situation where the implement-ing agency expects communities to take the leadand the communities are not clear about this, sus-tainability issues are generally neglected. Theassessment noted, “The farmers continued tothink of the reclamation activity as an ‘outside’effort brought to them by UPBSN [implement-ing agency] rather than something that they hadto carry out on their own. It is understandablethen that they would think that the drainageproblem too would be ‘taken care’ of by UPBSNand the Gram Pradhan.”

And Sometimes Bank Interventions Have Failedto Provide Consistent Support Long Enough for aSustained Income Flow To Be EstablishedProject assessments and OED country studiesfound that where the issue is a sustained flow ofincome from a particular activity, such as forestry,project support is often unavailable until the re-turns from the forest could allow for a sustainedflow of income from non-timber forest prod-ucts and timber (Annex P). For example, inIndia’s forestry projects, popular support forjoint forestry management (JFM) was contin-gent on forging a link in villagers’ minds be-tween protection of the forest and improvements

in livelihoods. Hence,the projects providedfor complementary in-vestments in communalinfrastructure to give vil-lagers an incentive to co-operate. The OED studyon forestry in Indianotes: “The regeneratedforest area can be keptunder tree cover only ifthe FPC [forest protec-tion committee] members get enough returns tocompensate for the income forgone. This wouldmean that JFM and the Economic DevelopmentProgram have to be part of one strategy of en-suring returns in the future. Currently this isnot the case” (Kumar and others 2000). Althoughin several states, the projects, when appraised,were presented as the first phase of a long-termoperation to consolidate the JFM strategy, inseveral states (including Madhya Pradesh) theBank did not commit to funding a follow-on op-eration, which considerably jeopardized sus-tainability of the effort already made.

Formal and Informal Organizations AreBoth Important in Determining CollectiveActivity at the Community LevelThe literature shows that, at the community level,both formal and informal organizational systemsinfluence collective activities. Community Asso-ciations set up in Brazil under the Rural PovertyAlleviation Program (RPAP) are a good exampleof formal organizations, because these groupsneed to be legally constituted before they can par-ticipate in the Bank-supported project. Village so-cieties also typically haveinformal arrangementsthat determine howgroups collectively man-age resources such aswater. These informalarrangements are not ex-plicit rules or regulations,but are based on customsand conventions or whatpeople consider “the gen-erally accepted way of


3 9

But customs and

conventions that could

be specific to a particular

community and are

important in determining

collective activities have

received inadequate

attention.

The recent emphasis on

empowerment and social

capital has focused

much more attention

on the importance of

understanding the rules

and regulations that

govern behavior in

village society.

doing things” (Cleaver 1997, 1998; Tripp 2001)(see Annex K).

Yet Bank Projects Have Primarily Focusedon Formal OrganizationsA review of project documents and evidencefrom country studies shows that in Bank projects,the focus is primarily on formal organizations andmanifestations of collective action, such as thecreation of a user group or committees and theholding of their meetings. The recent emphasison empowerment and social capital has focusedmuch more attention on the importance of un-derstanding the rules and regulations that gov-ern behavior in village society, but customs andconventions that could be specific to a particu-lar community and are important in determin-ing collective activities have received inadequateattention. As a result, little thought appears tohave been given to how the formal structures that

are created under a Bank intervention will affectthe informal organizations, customs, and con-ventions of a village society and how the inter-action of the formal and informal rules couldinfluence empowerment.

Since the formal and informal systems influ-ence each other, it could be argued that the for-mal arrangements created under a Bankintervention will influence and bring about ad-equate changes in the informal arrangementsthat are in keeping with the formal systems.However, as will be seen in Chapter 5, Bank sup-port is rarely provided long enough to allow thenew structures to become an effective part of theway the village operates. Project assessmentsand evidence from focus groups (box 4.3) showthat new structures established to implementBank projects tend to fade away once the proj-ect implementation period is over.

4 0


“Only 7 out of 12 [committee] members participated in meetings.Today, this number has been reduced to three. The other mem-bers of the committee said that there is no profit in their beingon the committee and for this reason they prefer to go deal withtheir own affairs [fieldwork].” Benin Borgou Region (AgeFIB)

“No committee was set in the village to monitor or managethe project. Only the [village committee] secretary had playedsome role. There has been no discussion as to community con-tribution and participation. Financial contribution was paid on[committee] revenue while households were requested to con-tribute with free labor, especially for fetching water for the build-ing.” Benin Borgou Region (PAMR)

“People do not even mention collective work anymore. Noteven meetings are any longer held. There is only a meeting whenmoney is concerned.” Rio Grande do Norte, Brazil

“[Maintenance of the infrastructure] is the job of the districtgovernment and the project people. No one has come to repair

the field drains and connecting drains. On being asked why theydon’t repair the connecting drains the reply is, that we werepaid to make them, no one has paid us to repair them.” UttarPradesh, India

“Why has our community association become inactive? …ourmain target was to get the water system; we got it and the peo-ple stopped mobilizing so we were benefited with no other proj-ects. Our target was to have water and then people stoppedmobilizing.” Rio Grande do Norte, Brazil

“When the Samiti was formed, it had promised to start 7 vil-lage-based organizations. They could remember just 4…Oncethese Samitis were started, they operated for 4 to 5 years but laterthey were dissolved… None of them exist today….One man saidthat there had been no meetings for the last 2 years.” MadhyaPradesh, India

Box 4.3: Why Formal Groups Do Not Last Long

Source: Focus groups.

4 1

Bank Operational PolicyRequirements, Processes,and CBD/CDD Interventions

This chapter assesses the extent to which internal policy requirementsand processes position the Bank to support implementation ofCBD/CDD, in particular CDD, interventions. Although the Bank’s mis-

sion is to fight poverty and improve the living standards of people in thedeveloping world, it is also a lending institution, and its shareholders wantassurance that funds provided by their taxpayers achieve expected resultsand that operations are economically, socially, and environmentally sound.

Consequently, in addition to meeting efficiencyconditions (Chapter 3), all Bank projects need tomeet two basic policy requirements—fiduciary,which govern the use of project-related funds, andsafeguards, to prevent unintended adverse ef-fects on third parties and the environment.

The Bank Has Attempted to AdaptIts Policies to Design and ImplementCBD/CDD ProjectsBoth fiduciary and safeguard policies were orig-inally developed for non-CBD/CDD projects thatgenerally involved large-scale “lumpy” invest-ments at specific locations, typically implementedby a central government department or agencythat monitored and reported on resource use.Bank missions supervised the investment site pe-riodically and reported on resource use sys-tematically. A typical CBD/CDD project is verydifferent. Each project includes numerous smallsubprojects that are heterogeneous and scat-

tered, sometimes in remote locations with poorcommunication. They involve multiple actors,and many communities with varying socioeco-nomic, cultural, and political backgrounds. Un-like the more traditional investments, thesubprojects under CBD/CDD are often not evenknown in advance. Moreover, in the case ofCDD, communities are also expected to controlresources and decisions and be in charge of con-tracting for their implementation. These signif-icant differences have made it difficult to ensurethe compliance of CBD/CDD projects with fi-duciary and safeguard policies that were devel-oped for non-CBD/CDD investments.

As a result, the Bank attempted to adapt its poli-cies to be able to design and implement CBD/CDDinterventions, while meeting the institution’s fi-duciary and safeguard obligations.1 This chapterexamines three issues that are pertinent to as-sessing Bank capacity to implement CBD/CDDprojects: first, whether CBD/CDD projects can

55

pose a challenge for safe-guard and fiduciary com-pliance; second, whetheradequate changes havebeen made in Bankpolicies to effectively sup-port implementation ofCBD/CDD projects; third,whether the Bank has the

capacity to ensure effective implementation ofCBD/CDD projects.

CBD/CDD Projects and the Challenge forSafeguard and Fiduciary Compliance

Compliance with Safeguard at EntryHas Improved over the YearsA thematic study that reviewed the sample of 84projects for their compliance with safeguards(Annex Q) found that such compliance at entryhas improved over the years. In terms of Re-gions, all projects in Europe and Central Asiawere found to be satisfactory. The Middle East andNorth Africa recorded the next strongest record,followed by Africa, and East Asia and the Pacific.The study also found that despite format changesin the Project Status Reports that encourage de-tailed reporting on the implementation of safe-guard measures, such reporting remains sparseand inadequate. Further, the study found that theoverall quality of implementation was rated sat-isfactory for fewer than 40 percent of the cases.In terms of Regions, South Asia and East Asiaand the Pacific score highest for quality of im-plementation of safeguard issues. These findingsat entry and implementation are disturbing giventhe Bank’s current emphasis on full compliancewith safeguard requirements. The thematic studyalso found that 6 of the 11 projects rated unsat-isfactory on overall quality were in the Bank’slargest borrower countries.

However, Cumulative Impact of Subprojects HasBeen an Issue for Safeguard ComplianceSome have argued that individual subprojects inCBD/CDD interventions are so small that theycannot have a substantial negative social or eco-nomic impact. However, the thematic study foundthat while the environmental and social impact

of individual subprojects may be insignificant,their cumulative impact can be substantial. Thestudy also notes that too little attention is beingpaid in CBD/CDD projects to the environmentaland social consequences of changes in land use,especially for livestock, irrigation, and reforesta-tion projects. Moreover, subprojects are oftennot small and may include investments—such aswastewater treatment plants in Poland or damsin China and Brazil—with the potential for majornegative environmental and social impacts.2 Fur-ther, as decision making is decentralized, thereis some danger that potential impacts, particularlyin resettlement cases, may not be recognizedand suitably mitigated. For example, it was onlywhen OED carried out an assessment of theAndhra Pradesh Forestry Project (1994) that casesof uncompensated land were discovered.

Wrong Environmental Category AssignmentCan Have Serious ImplicationsTypically, projects in areas such as health, edu-cation, nutrition, institutional development, tech-nical assistance, and human resources are placedin environmental Category C (see box 5.1) be-cause they are considered to be unlikely to haveadverse environmental impacts, or that any suchimpact would be minimal. However, health proj-ects supporting immunization programs, basicpackages of drugs and syringes, and laboratoryservices for infectious diseases (including AIDS)raise concerns about the safe collection, storage,and disposal of medical waste. The thematic re-view found that 9 percent of the projects classi-fied as Category B and 38 percent classified asCategory C had been misclassified—that is, proj-ects had been classified as C or B when theyshould have been B or A. Assignment of CategoryB induces a B mindset, which implies concen-trating attention on documentation to be pro-duced before Board approval, rather than onappraising the capacity of the project agenciesto screen subprojects, analyze their potentialimpacts, and design and implement mitigationmeasures, and on specifying the needed insti-tutional strengthening and monitoring systems.Assignment of Category C generally means thatno further work is done to identify and mitigateimpacts and there is no further review by safe-

4 2


While the environmental

and social impact of

individual subprojects

may be insignificant,

their cumulative impact

can be substantial.

guard specialists. For example, OED’s projectassessment of the Mali Natural Resource Man-agement Project also found that since the proj-ect was wrongly categorized as a C, importantenvironmental issues were not given attention.

Attention to Safeguards during ImplementationHas Also Been Inadequate While quality at entry needs improvement, safe-guards compliance during implementation war-rants much greater attention by the Bank andborrowers, and may indicate the need for greaterallocation of supervision resources. The low per-centage of moderately satisfactory or better Cat-egory A projects appears to conflict with theexpectation that A’s would receive much moreintensive scrutiny during supervision. The in-adequacy of funding to address safeguard issuesis strongly endorsed by Bank staff in the staff sur-vey. If resources for safeguard supervision wereincreased, it would add to the already highercosts of supervision for CBD/CDD projects incomparison with non-CBD/CDD projects, withefficiency implications. Some of the more re-cent projects (such as Nigeria Community-BasedPoverty Reduction) have had special assessmentsof safeguards implementation, usually by spe-

cialist consultants, a practice that would be valu-able to adopt more widely, especially wherethere are several CBD/CDD projects in the samecountry. The Bank is also currently exploringways to streamline the application of safeguardpolicies by delegation, both within the institutionand to national authorities.

Although Fiduciary Compliance IsMore an Issue for CDD ProjectsWhere the handling of resources in scattered sub-projects remains in the hands of a central imple-menting unit, the fiduciary challenge is not verydifferent from that for a non-CBD/CDD project—that is, the need to set in place within the imple-menting unit a system that can adequately monitorand report on resource use. The challenge occurswhen control over resources, and often pro-curement responsibility as well, is transferred tocommunities (often remote ones), as happens inCDD projects. The coun-try studies show that thisfiduciary challenge islikely to be greater incountries where institu-tional capacity is weaker,such as Benin, Nepal, and

BANK OPERAT IONAL POL ICY REQUIREMENTS, PROCESSES, AND CBD/CDD INTERVENTIONS

4 3

The Bank classifies proposed projects into one of four categories,depending on the type, location, sensitivity, and scale of theproject and the nature and magnitude of its potential environ-mental impacts.

Category A: A project is classified as Category A if it is likely tohave significant adverse environmental impacts that are sensi-tive (a potential impact is considered sensitive if it may be irre-versible), diverse, or unprecedented. These impacts may affectan area broader than the sites or facilities subject to physicalworks.

Category B: A project is classified as Category B if its poten-tial adverse environmental impacts on human populations or en-

vironmentally important areas—including wetlands, forests,grasslands, and other natural habitats—are less adverse thanthose of Category A projects. These impacts are site-specific;few if any of them are irreversible; and in most cases mitiga-tory measures can be designed more readily than for CategoryA projects.

Category C: A project is classified as Category C if it is likely tohave minimal or no adverse environmental impacts.

Category FI: A proposed project is classified as Category FI if itinvolves investment of Bank funds through a financial interme-diary, in subprojects that may result in adverse environmentalimpacts.

Box 5.1: The Meaning of the Environmental Categories

Source: Bank Operational Policy 4.01—Environmental Assessment.

Safeguards compliance

during implementation

warrants much greater

attention by the Bank

and borrowers.

Vietnam (box 5.2), than in middle-income coun-tries such as Brazil, Egypt, and Indonesia, wherecapacity to monitor resource use at the local levelmay be greater. The country studies also show thatthe understanding and interpretation of fiduciaryresponsibility and accountability can be very dif-ferent in local communities in several countries.In Benin’s rural communities, for example, prop-erty rights are poorly defined and enforced. Com-mercial exchanges are based on trust rather thanon enforceable contracts, and fiduciary rules aretypically informal. In this kind of environment, ap-plying fiduciary management and accountabilitythrough the Bank’s rules may be more difficult andcostly than is envisaged in CDD projects.

Most Bank Documents Have Not Reportedon Community Capacity to Undertake FiduciaryManagement Responsibility

Fiduciary sector work, in-ternal audit reports, andproject documents for 12CDD projects were re-viewed to assess fiduci-ary compliance. Thereview found that themajority of appraisal andsupervision documents

still do not report on community capacity to un-dertake fiduciary responsibility, even though theproject envisages communities being responsi-ble for managing resources. It is thus unclear howextensively capacity at the community level is ac-tually assessed before a Bank CDD project is in-troduced in a particular setting. The majority ofCountry Financial Accountability Assessments(CFAAs) and Country Procurement AssessmentReports (CPARs) reviewed also do not reporton community capacity.

The Readiness of Country Financial Proceduresand Internal Control Systems to Support CDDIs Not Given Adequate Attention While CFAA and CPAR documents can indicatewhether the existing processes in the country atthe central and regional levels are strong enoughto monitor resource use, this information doesnot seem to be used to assess country readi-ness for CDD. For example, the documents onNepal, Nigeria, Pakistan, and Senegal have raisedconcerns about the extent of capacity in thesecountries to monitor resource use and to re-port on poor compliance with financial proce-dures and internal control systems.3 However, allthese countries have several CDD projects eitherongoing or in the planning stages. While some

4 4


The Country Financial Accountability Assessment for Vietnamnotes the challenge for financial management at the subna-tional government levels created by weak capacity, especiallyin budgeting, accounting, and financial reporting.a With defi-ciencies and irregularities in procurement as well (VietnamCPAR 2002), it is unclear how the Bank will be able to managethe fiduciary risk in its ongoing portfolio of CDD interventions. Aninternal Bank review of the Quality of Supervision for a CBD/CDDproject, the Coastal Wetlands Protection and Development proj-ect (2000), identified the serious challenge that the Bank teamfaces.b It reports that the Bank’s procurement procedures werealmost completely rejected by the client in favor of directing

works under the International Development Association (IDA)credit to monopoly state-owned enterprises. In the view of thereviewers, this represented a possible failure of the Bank’s en-tire project implementation system in Vietnam. The report ac-knowledges that Bank staff must be attentive to situations wherethe Bank’s policies may not apply or where there may be moreeconomical, efficient, or transparent ways of doing things, but,in the end, Bank management is responsible for the environmentin which projects are conceived, prepared, negotiated, and im-plemented. This should be an environment in which both partiesare focused on getting results they value.

Box 5.2: The Fiduciary Challenge: The Case of Vietnam

a. Concerns are already being noted by Bank staff on financial management in the supervision reports for the Community-Based Rural Infrastructure Project (2001) thatstarted disbursement to communes in fiscal 2003.b. Management notes that the operational issues highlighted are not related to the CDD aspects of the project.

This fiduciary challenge

is likely to be greater

in countries where

institutional capacity is

weaker than in middle-

income countries.

have argued that it is much more likely that re-sources provided through CBD/CDD interven-tions will reach the poor than those providedthrough non-CBD/CDD work, the concern isthat, given the level of transparency and ac-countability in several poor countries, unlessadequate processes are put in place, it is quitepossible that despite the best intentions, theseresources may not actually benefit the poor.4 Inaddition to fiduciary sector work, the weaknessin these aspects of public sector governance hasalso been noted in other Bank-IMF documents.5

The importance of adequate follow up to suchdiagnostic sector work in the design and imple-mentation of CDD projects cannot be empha-sized enough.6 It is worrying that several internalaudit reports have picked up lack of fiduciarycompliance as an issue in several CDD projects.It is worth mentioning that some project su-pervision reports have also raised concerns aboutfiduciary issues.7

Changes in Bank Policies to EffectivelySupport CBD/CDD Projects

More Progress Has Been Made on RefiningFiduciary than Safeguard PoliciesWhile it is to the credit of the institution that theneed to adapt its policies has been recognized,policies have not yet been completely refined, al-though CBD/CDD lending is growing rapidly.More progress seems to have been made on re-fining the fiduciary policies (as a fiduciary refer-ence guide at least exists and was issued in May2002) than on safeguard policies. But it is im-portant to recognize that the procedure for de-veloping adequate guidance on safeguards iscomplicated. The Andhra Pradesh forestry expe-rience, and more recently that of the IndonesiaKecamatan Development project, shows the dif-ficulty of applying the Bank’s safeguard policies,particularly resettlement policies, to situationswhere it is not easy to determine what is a “loss”for which a displaced person needs to be com-pensated.8 The challenge for the Bank today is toensure that safeguard violations are minimizedwhile the new guidelines are being developed.

Although the thematic study found that guid-ance to staff on safeguard issues is still being de-

veloped, with fiduciaryissues, application of theguidelines is a challenge.The guidance note statesthat the Bank rules andguidelines apply to CDDprojects in the same waythat they do for any other Bank project, but thattheir application needs to be adapted to the ca-pacity of the project and the community. Sincethere are capacity differences among commu-nities, fiduciary requirements need to be adaptedto each project—if not to each community’s ca-pacity. The Bank’s policy on fiduciary manage-ment for CDD projects leaves the decision aboutwhat communities are required to do for eachCDD project’s appraisal team. As a result,whether and how this policy translates into sim-pler procedures at local level is less clear, and thesimplification, if any, may be quite variable acrossprojects.

In response to the questionnaire on the sub-ject in the staff survey—that task managers ofCBD/CDD projects can monitor fiscal account-ability as satisfactorily as can managers of moretraditional non-CBD/CDD projects—36 percentagreed or strongly agreed, 28 percent were in themiddle, and 23 percent disagreed or stronglydisagreed (Annex L). While on other questionsthis might be considered a reasonably positiveresponse, on matters related to accountability,calling for some necessary minimum standard,having approximately one-quarter of staffexpressing concern about the ability of taskmanagers to monitor suggests a significantproblem.

Bank Capacity to Ensure EffectiveImplementation of CBD/CDD Interventions Two issues are central to assessing Bank capac-ity to ensure effective implementation ofCBD/CDD interventions. These are: the Bank’smode of operation and institutional organizationand the Bank’s capacityto undertake adequatemonitoring and evalua-tion of its operations.The following sectionsdeal with them in turn.


4 5

Policies have still not

been completely refined,

although CBD/CDD

lending is growing

rapidly.

With fiduciary issues,

application of the

guidelines is a

challenge.

Mode of Operation andInstitutional Organization

The Information Access Chain is Much Longerand Has Greater Gaps in CBD/CDD Interventionsthan in Non-CBD/CDD Interventions

Regular “supervision”missions were largelyadequate to ensure thatimplementation was pro-ceeding as planned innon-CBD/CDD opera-tions. In CBD/CDD inter-

ventions, and particularly CDD, where theimplementation is being undertaken by commu-nities, the information access chain for the Bankis much longer and has larger gaps:

• First, while the Bank relies on the borrowerto provide reports, several layers of governmentare involved. The extent to which the bor-rower is able to maintain accurate records de-pends on its ability to coordinate amongdifferent layers and get information from thecommunity level.

• Second, the reliability and accuracy of the in-formation from the community level dependson the capacity of the communities to main-tain the records and of the local governmentagencies or facilitators to monitor communityrecord keeping.

• Third, it is almost impossible to expect the bor-rower to “supervise” and cross-check for ac-curacy and consistency the information that iscoming from thousands of remote communi-ties. The government has to rely for this ac-curacy on facilitators that may have a vestedinterest in reporting that implementation isgoing well, since the facilitator’s survival oftendepends on the availability of donor resources.

• Fourth, given that there is no systematic wayof checking for cumula-tive impact, and that base-line data are often notavailable, it is almost im-possible to tell how, say,the simultaneous diggingof 500 wells will affect thewater table in a given area.

• Fifth, the Bank’s own supervision resourcesallow missions to “inspect” only a very limitednumber of subprojects, and internal audit re-ports show that these are mostly in sites thatcan be easily visited.

Bank Decentralization Has Increased the Capacityof the Institution to Track CBD/CDD InterventionsNearly 3,000 of the 10,000 Bank staff now liveand work in client countries.9 With this decen-tralization to the field, the Bank is today better con-nected with its borrowers than it in the past.During OED project assessment field missions,borrowers almost universally said that supervisionby field offices has helped them. Sixty-five percentof Bank staff were somewhat satisfied or betterwith the impact of Bank decentralization to fieldoffices on the efficacy of Bank support for par-ticipatory projects. Only 13 percent were not sat-isfied, suggesting fairly widespread satisfactionwith Bank decentralization. While decentralizationhas undoubtedly brought the Bank closer to theborrower, it can do little about the governmentand community side of the information chainand about the Bank’s own ability to monitor whatis happening in thousands of remote communi-ties. The staff survey raised concerns about theavailability of adequate supervision resources tobe able to monitor compliance with safeguards.As Chapter 3 indicates, the Bank’s preparation andsupervision costs for CBD/CDD projects are al-ready higher than for non-CBD/CDD project, andthere are no additional incentives for country di-rectors to provide the additional resources re-quired to prepare and supervise these operations.

But the Sectoral Organization of the BankContinues to Handicap the Design andImplementation of CBD/CDD InterventionsFurther, despite the attempt by Regions tocreate multisectoral teams to coordinateCBD/CDD, the Bank itself is compartmental-ized, so integrated approaches across severalsectors have remained limited.10 Only 9percent of Bank staff surveyed reported beingsatisfied or very satisfied with coordinationwithin the Bank across sectors in CBD/CDDinterventions. It is striking that this concernabout Bank coordination was actually greater

4 6


Borrowers almost

universally said that

supervision by field

offices has helped them.

The Bank itself is

compartmentalized, so

integrated approaches

across several sectors

have remained limited.

than staff concern about borrower coordina-tion. Only about 14 percent reported beingfully satisfied with the support from the currentmatrix-management organizational structurefor CBD/CDD projects and about 27 percentwere not satisfied (Annex L). A key finding of arecent Regional review of Bank supervision ofCDD projects in the East Asia Region was thatthere was little collaboration between theRegion’s sector units in supervising CDDprojects despite the multisectoral scope ofthese operations (World Bank 2003c).

Monitoring and Evaluation

Monitoring and Evaluation Have ImprovedOver the YearsSince an Operational Memo in January 1996from the Operations Policy and Country Ser-vices (OPCS) unit provided guidance to staff onpreparing indicators, most projects, includingCBD/CDD, have given more attention to M&E ca-pacity and indicators, though there is variationin quality across projects. A review of the sam-ple of 84 projects found that M&E was veryweak in most of the early projects (pre-1994),and most of the indicators were output-relatedrather than outcome-related. Some projects hadvery little monitoring of any kind. For example,the OED assessment for the Turkey Eastern Ana-tolia Project, a 1993 intervention, notes the weak-ness of M&E and the fact that the first baselinewas not done until 1998. In Egypt, the OED as-sessment of the Matrouh Resource ManagementProject (1993) also noted that M&E started toolate and analysis focused largely on number ofadoptions rather than quantified impacts, so notmuch could be said about gains in productivity.

But Little Has Been Done aboutMonitoring Capacity EnhancementThere has been progressive improvement, andmore projects approved in later years (after fis-cal 1995) have outcome and impact indicators.The Portfolio Review found that 95 percent ofthese projects have indicators to monitor progressand impact, compared with 50 percent in theearlier period. However, most indicators con-tinue to focus on quantity rather than quality.

For example, the OED as-sessment of the BorgouPilot Project, a fiscal 1998intervention, shows thatmonitoring indicators stilltrack quantitative inputand output achievements(such as days of trainingprovided, number of vil-lages covered, and thelike) rather than qualita-tive progress toward achieving the primary proj-ect objective of improving the capacity of villagecommunities to better manage their socioeco-nomic environment. Similarly, the Yemen ThirdSocial Fund and the recently approved CameroonCommunity Development Program APL havelarge capacity building components and progresstoward the objectives should thus be measuredby qualitative and process-oriented indicators.Yet most of the outcome and impact indicatorsnoted in the report continue to be quantitativeand will be able to say little about the qualityand impact of the capacity-building effort. How-ever, on the positive side, there are examples ofsome projects, such as the India Andhra PradeshDistrict Poverty Initiatives Project (2000), whichmake some provision for process monitoring indesign and may be able to indicate improvementin capacity assuming adequate follow up.

Several projects, such as the CWPII in Albania,Kalahi CIDSS in the Philippines, Kecamatan De-velopment Project in Indonesia, and the ThirdSocial Fund for Development in Yemen, are em-phasizing participatory M&E to involve commu-nities in tracking progress on activities. This isnot only likely to support the Bank capacity-en-hancing effort at the community level, but is alsolikely to help build greater ownership of Bank ac-tivities in communities, with positive implicationsfor sustainability (Estrella and Gaventa 1998).Some projects have attempted to combine par-ticipatory M&E with other measures in a pluralis-tic approach that could prove invaluable in trackingprogress (box 5.3). But, given the difference in theunderstanding of participation between the Bankand the borrower/community noted earlier, it re-mains to be seen how effective participatory M&Ewill be in tracking progress on process issues.


4 7

Most projects, including

CBD/CDD, have given

more attention to M&E

capacity and indicators,

but most indicators

continue to focus on

quantity rather than

quality.

Further, the preparation for most Bank projectsdoes not include establishing a baseline. The Port-folio Review found that fewer than 10 percent ofthe projects involve establishing a baseline againstwhich to assess the impact of Bank intervention,instead relying largely on with-without compari-son with weak counterfactuals. Hence, even ifstudies are carried out, it is very difficult to tellwhether there are any achievements.

Furthermore, well-designed M&E is pointlessif it is not effectively used. For example, the OEDassessment of the Uttar Pradesh Sodic LandReclamation Project notes that the latest moni-toring technology is available to the RemoteSensing Application Center in Uttar Pradesh, butthe large amount of data being generated is notbeing used effectively. This is also because proj-ect-related M&E procedures contribute little to

systematically buildingevaluation capacity in thecountry. Most project-re-lated M&E effort comesto an end when projectsclose. There appears to

be little, if any, systematic relationship betweenevaluation capacity development activities andindividual project-level M&E.

An ideal M&E system for CBD/CDD should beable to do at least five things:

• Tell whether adequate qualitative and quanti-tative progress is being made toward meetingthe project objectives.

• Tell whether the Bank resources are beingused effectively and efficiently.

• Give some indication of whether progress isbeing made in reaching the poor and the poor-est.

• Provide information on safeguard and fiduci-ary compliance.

• Give an indication of whether sustainabilitycan be ensured.

• If the first five are not happening, it should pro-vide flags for mid-course corrections.

Existing M&E systems in Bank CBD/CDD in-terventions are a long way from meeting thesecriteria.

4 8


The KDP project illustrates some features that an M&E systemfor CDD should emulate:

• Involvement of beneficiaries in monitoring performance (forexample, measuring quantities of materials delivered by con-tractors).

• Public display of financial data at the village level on noticeboards so that all beneficiaries could see, monitor, and ques-tion.

• The establishment of relatively strong record-keeping sys-tems and bookkeeping skills at the village level (althoughskills in the wrong hands can make corruption easier tohide also).

• A quite strong central government monitoring system.• Baseline studies, impact studies, and other studies on par-

ticular issues that were contracted out (though not all thesereports have been assessed for their quality).

• Qualitative and quantitative indicators measure physicalachievement, corruption, and conflict.

• Contracted, but independent, journalists to be another eye inmonitoring, which enabled the publicizing of corruption cases.

• Contracted NGOs to monitor performance at the communityand local government levels (in most CDD projects NGOs arethe facilitators, in KDP they were only monitors).

• A project-run public grievance system that generally re-sponded quite quickly.

• A series of related studies, such as an innovative quantitativeassessment of corruption in infrastructure (for example,through core sampling of roads), a study of microfinance per-formance, and studies of conflict problems.

• A general readiness to respond with new, quick studies as newperformance issues arose.

This said, the KDP system also has some weaknesses (for ex-ample, related to community process measurement, insufficientmeasurement of poverty and gender impact, and methodologicalproblems with measuring impacts), which are planned to be ad-dressed under KDP3, which became effective in January 2005.

Box 5.3: Some Monitoring and Evaluation Features to Emulate: The Case of the IndonesiaKecamatan Development Project (KDP)

The preparation for

most Bank projects

does not include

establishing a baseline.

Source: Portfolio Review.

4 9

Conclusions

This evaluation of the Bank’s support for CBD/CDD interventions in clientcountries supports four broad findings.

The Bank has not, until recently, system-atically identified and tracked its portfolioof CBD/CDD projects, and therefore haslacked a comprehensive understanding ofthe evolution and scope of its work in com-munity development. It also has not beensufficiently clear about the objectives ofusing CBD/CDD approaches, criteria forchoosing among different community de-velopment approaches, or about how tomeasure the results.

Although the Bank has been involved inCBD/CDD for a long time, a database of projectsusing community development approaches wasonly established for projects approved from 2000onward. Overall development effectiveness canbe assessed only on a clearly defined portfolio.Furthermore, effectiveness can only be assessedagainst clear objectives, preferably with clear in-dicators against which to judge success. Differ-ent community development approaches havevarying community capacity requirements, so itis important to ensure that Bank staff, as well asthe Bank’s clients and the ultimate beneficiaries,understand the expectations under the project.For example, the concept of empowerment, amajor justification for the most recent generation

of CDD projects, is not uniformly understoodwithin the Bank or, even more important, be-tween the Bank and its clients. Regarding com-munity participation, the surveyed beneficiariesappear to have a very limited, pragmatic under-standing of the concept that differs significantlyfrom the Bank’s intent.

The Bank’s structure and mode of opera-tion limit its ability to ensure sustainableoutcomes from CBD/CDD projects. Thislimitation has become much more appar-ent since the institution began emphasiz-ing CDD in the late 1990s.

It is easier for the Bank to monitor resourceuse and be in compliance with safeguards innon-CBD/CDD investments such as bridges or apower plant than where small subprojects arebeing implemented by hundreds of remote com-munities in scattered locations. In CBD/CDDprojects, and more so in CDD ones, the criticalchallenge that the Bank faces is that the processmust be managed “close to the ground,” butnormally without direct Bank involvement atthe local level. As a result, with its mode of op-eration, distance from implementation, and itscurrent monitoring and evaluation system, the

66

Bank has found it difficult to ensure safeguardcompliance and sustainability of developmentoutcomes from its CBD/CDD projects.

The Bank’s support for CBD/CDD has pro-duced different, though systematic, resultpatterns depending on local political andsocial conditions, government commit-ment, and community capacity.

Bank-supported CBD/CDD projects have hadmuch more success, particularly regarding ca-pacity enhancement, in supporting indigenouslymatured participatory efforts or where it hasprovided consistent long-term capacity-build-ing support to communities over time. How-ever, most projects make little effort to tailorcapacity building to community capacity or to goback to the same communities with a consis-tent capacity-building strategy; the one year ofa typical subproject cycle is sufficient to allow suc-cessful subproject execution, but not to consis-tently have a significant positive impact oncommunity capacity; and communities do not ap-pear to have understood that their participationis meant to drive the development process, andsee participation in a Bank project primarily asa requirement for them to meet part of the sub-project cost.

To effectively support CBD/CDD projects,and especially CDD ones, the Bank willnot only need to carefully consider its owncapacity but also to assess borrower com-mitment, community capacity, and thecosts and benefits of the alternatives avail-able. Four issues need special attentionwhen future CBD/CDD projects are con-sidered:

• Clear articulation of expected achievementsof CBD/CDD interventions. While the designof CBD/CDD projects has emphasized bothmaterial development and capacity building

activities, during project implementation rel-atively greater importance has tended to begiven to achievement of material developmentgoals. This raises concerns about whether theBank is using CBD/CDD as a means for facili-tating an investment program rather than forsustainably improving community decisionprocesses.

• Calculation of the costs and benefits, includingthe long-term poverty impact, of undertakingthe CBD/CDD approach as a basis for com-parison with alternatives. The Bank has notsystematically and realistically assessed the dis-tribution of costs and benefits of undertakingCBD/CDD projects to the institution, the bor-rower, and the communities. The insufficientfocus on costs and benefits, especially meas-ures of poverty impact, in CBD/CDD projectshas prevented convincing comparisons withmore traditional investments and policy and in-stitutional reform programs.

• Focus on sustainability and long-term devel-opment. Project experience indicates that ina number of cases there has been a lack of ad-equate follow through of activities supportedby Bank projects in order to address and min-imize risks to long-term outcomes. In othercases, the ad hoc parallel arrangements madeto implement Bank projects have hinderedthe long-run enhancement of local govern-ment capacity.

• Addressing constraints related to the Bank’smode of operation, its operational policies,and its monitoring and evaluation systems.For individual communities, the Bank’s sub-project cycle is generally too short to bringabout the kind of enhancement of commu-nity capacity that is visualized in Bank-sup-ported CBD/CDD, particularly CDD projects.Further, Bank processes and systems have notbeen geared toward supporting long-termprocesses such as empowerment and socialcapital enhancement.

5 0


5 1

Recommendations

Given the mixed and limited evidence on the impacts of CBD/CDD proj-ects—particularly in poverty reduction and empowerment—andquestions about sustainability and safeguard and fiduciary compliance,

the Bank should approach future CBD/CDD projects, particularly CDD ini-tiatives, with greater care. In countries where the Bank is already supportinga CDD program, the institution needs to rigorously assess the poverty and in-stitutional development impact of its projects before scaling them up. A cau-tious approach would be especially important in countries or areas where theBank is just beginning to support CDD. In its future assistance to CBD/CDD,the Bank should:

At the corporate level, strengthen opera-tional guidance and management over-sight. • The Bank should provide operational guid-

ance for the application of Bank safeguardpolicies and fiduciary oversight of CBD/CDDprojects and for the strengthening of cost-ben-efit analysis and M&E systems; and shouldcommission an audit of the fiduciary aspectsof a representative sample of CDD projectsfor submission to the Board within a year.

At the country level, design the CBD/CDDprogram as an integral part of the overallassistance strategy and carry out periodicassessment of its ongoing CBD/CDD proj-ects to ensure relevance and effectivenessof the program to the country context.

• Future CASs should show how they have ana-lyzed and addressed linkages not only betweenvarious CBD/CDD projects to be undertakenin the country but also between CBD/CDDand relevant non-CBD/CDD projects. In par-ticular, the analysis should address whetherarrangements for CBD/CDD project imple-mentation come at the expense of local gov-ernment capacity development.

At the project level, the Bank should givepriority to helping countries build up ex-isting indigenously matured initiatives;where there are no such existing initia-tives, the Bank should tailor its project tothe country and community context, whileundertaking selective rigorous impact as-sessments to ensure learning.

77

• For any new CBD/CDD project, the Bankshould analyze (using existing processes, suchas social assessments) whether it is building onindigenously matured initiatives or attemptingto begin a CDD program in a country and then

tailor the intervention to local capacity; and theBank should also selectively undertake rigor-ous impact assessments upon completion of itsongoing CBD/CDD projects to learn for the fu-ture.

5 2


ANNEXES

The Community

All Bank participatory projects exhibit threebasic assumptions about communities (fromOED 2003):

• They comprise a group of people who sharebroad development goals.

• Their social behavior and relationships are gov-erned by social norms that are expected toprovide solidarity.

• By extension, those who do not belong to thatcommunity are “excluded.”

The “community” in this approach is oftenconsidered a “unified, organic whole” (Agrawaland Gibson 1999). Since the group of people ina “community” live in a particular area, share acommon interest (water users associations,herders, and the like), and are governed by a setof norms, its members are assumed to be in thebest position to identify their most pressingneeds and problems.

The latter idea suggests that there are com-mon problems that can be solved through com-munity consensus. While this may be true, itneglects community members’ differences andpower relationships, the conflicts, and the di-versity of interests that determine day-to-daybehavior and that have an impact on the effec-

tiveness of participatory approaches. The poorthemselves are rarely a homogenous group; theylive in different geographic areas and face dif-ferent kinds of deprivations, and each seeks apersonalized way of reducing poverty (Schnei-der 1999).

The shared norms that are expected to unifythe community can themselves hinder commu-nity action (Western and Wright in Agarwal andGibson 1999). Such norms may dictate patternsof behavior, such as deference to the elite, whichdo not allow the poorest and the marginal to ef-fectively demonstrate their choice. Moreover,participation may lead to significant psycholog-ical and even physical duress for the most sociallyand economically disadvantaged, typically theprime potential beneficiaries of CDD projects,since genuine participation may require them totake positions that are contrary to the interestsof more powerful groups (Mansuri and Rao2004).

Participation starts a process of institutionalchange in communities. A distinction betweenformal and informal “rules of the game” and or-ganizations is essential to understanding thischange process. Many crucial decisions in a vil-lage community are made not through formalcommittees and groups, but through informal or-ganizations that vary from community to com-munity.

ANNEX A: DEFINITION OF “COMMUNITY”

5 5

5 7

How are CDD interventions expected to workin a Bank project? Since the 1992 Wapenhansreport, the World Bank has tried to increase theresults focus of its operations to track theprogress of Bank interventions, including CDD.The relationship among Bank inputs, outputs,expected outcomes, and impacts of CDD oper-ations are shown in figure B.1. The arrows in-dicate the direction of the results-based chain

that links inputs to impacts through outputsand outcomes.

Within this framework, the principal impactof a CDD approach is expected to be sustaineddevelopment and positive impact on the lives ofthe poor. Underlying this is a hypothesis that em-powered communities (outcome) can partici-pate in decision making, create and implementtheir own development plans, and hold ac-

ANNEX B: RESULTS CHAIN FOR WORLD BANK CDD PROJECTS1

Figure B.1: The Results Chain in a Bank-Supported CDD Intervention

Increased access of communities to basic

infrastructure, services, and income-generating activities

Favorable policy and legal environment and strong local

government institutions

Output

Empowered communities

Community control and management of decisions

and resources

Outcome

Resources for financing capacity building at the

community level, for preparing and implementing

development plans according to community priorities

Support for strengthening legal and policy environment

and local government institutions

Input

Sustained development �and positive impact on �

lives of poor

Impact


5 8

countable the institutions that affect their lives.This is expected to allow for improved effec-tiveness and targeting of development inter-ventions, which in turn is expected to promotesustainable development. However, this can hap-pen only if benefits are not captured by the eliteand donor support is available over a defined pe-riod to allow elements of sustainability to bebuilt. The major outputs would be increased ac-cess of communities to basic infrastructure and

services and income-generating activities, a fa-vorable policy and legal environment, andstronger local government institutions. Bank in-terventions—through resources for financingcapacity-enhancing efforts at the communitylevel, resources for preparation and implemen-tation of development plans, support to thecountry for improving the legal and policy en-vironment and for strengthening local govern-ment institutions are the inputs.

5 9

The CDD Anchor has identified 10 principles toguide policy formulation and program designto enhance the effectiveness and sustainabilityof support to CDD.1

1. Establish an enabling environmentthrough relevant institutional and policyreform. CDD involves more than strengtheningcommunity-based organizations (CBOs) and fund-ing their projects—it also requires active measuresto establish an appropriate enabling environment.Large programs of support to CDD will not be sus-tainable without the policies, laws, systems, andgovernance processes that encourage effectivecollaboration among local governments, centralgovernments, civil society, service providers, andCBOs. Specifically, such an environment shouldinclude: (a) elected local governments that are re-sponsive to constituents and are empowered toserve them; (b) intergovernmental arrangementsfor fiscal flows to local governments and CBOs;(c) a conducive legal and regulatory frameworkthat supports community action; and (d) clear sec-tor policies with well-defined financing rules anddefined roles and responsibilities of key playersin each sector.

2. Make investments responsive to informeddemand. Enabling communities to be involvedin decision making is not sufficient to achievesustainable outcomes. Decisions need to be basedon accurate information about the costs and ben-efits of various options, and communities need tohave some of their own resources invested.

• Informed, meaningful choice. Communitiesand stakeholders should have access to suffi-cient information to weigh tradeoffs and makerealistic choices from a range of options that

meet their needs and fit local conditions, cul-ture, values, and available operation and main-tenance capacity.

• Community contributions to investment andrecurrent costs. Community co-financing hasbeen shown to be an important factor in build-ing ownership and in helping to ensure that ap-propriate choices are made and thatinvestments are sustainable. People seem tomake better choices when they have their ownresources at stake and when opting for a moreexpensive option implies a proportionallyhigher cost.

3. Build participatory mechanisms for com-munity control and stakeholder involve-ment. Communities that have ownership of aproject or program are more likely to sustainoutcomes. This implies providing inclusive com-munity groups with knowledge, control, and au-thority over decisions and resources throughoutall phases from program inception. Programsshould be designed to engage relevant stake-holders (government, local leaders, NGOs, civilsociety, the community) at the earliest opportu-nity and dynamically over time. Political will—gar-nered through broad-based support and/or“political champions” to drive necessary re-forms—have played critical roles in the scaling upof many existing CDD programs. Broad stake-holder participation helps tap into local techni-cal and financial resources in support ofcommunity initiatives. It also ensures that localknowledge and preferences are incorporatedinto the project design.

4. Ensure social and gender inclusion. Com-munity-driven development has the potential toincrease the power of poor communities to ne-

ANNEX C: WORLD BANK GUIDANCE ON KEY DESIGN PRINCIPLES FOR CDD

gotiate with government, the private sector, andcivil society. But to fulfill this potential, CDDneeds to be responsive to the priorities of allpoor groups. Communities are not homoge-neous; thus CDD needs to be designed to besocially inclusive, giving voice and decision-mak-ing responsibility to women, the elderly, youth,religious and cultural minorities, indigenous andother ethnic groups, those with HIV/AIDS, andthe disabled. When community-driven develop-ment does not pay attention to issues of socialinclusion, groups of poor people may be ex-cluded, investment choices may not reflect thetrue needs of the poor, and impacts may be sig-nificantly compromised.

5. Invest in capacity-building of CBOs. Thelasting impact of CDD programs depends on thecapacity of CBOs to provide services and goodson a sustainable basis, often in partnership withresponsive formal institutions. Capacity buildingof CBOs, and strengthening linkages with formalinstitutions, is a critical area for investment. Theimpact of CDD programs is directly related tothe strength of the CBOs driving the process. Ex-perience and studies have shown that those CBOswith clear lines of responsibility, open decision-making processes, and direct accountability tothe community improve service provision, makemore effective use of resources, and are more sus-tainable.

6. Facilitate community access to informa-tion. Support to CDD is as much about facilitat-ing flows of information among all groups in acommunity as it is about facilitating flows of funds.The lack of information is often the most signif-icant limitation on CBOs’ capacity to play a partin the development enterprise—community or-ganizations need information on market oppor-tunities, on what support resources are available,and on how to use these resources productivelyand efficiently. A variety of media may be used tofacilitate access to and stimulate flows of infor-mation. Information technology and the inter-net, adapted to community needs, are playing agrowing role in this process and can dramaticallyaccelerate local learning and connections with awide range of opportunities.

7. Develop simple rules and strong incen-tives, supported by monitoring and evalu-ation. Experience indicates that sustainabilityand effectiveness of CDD is enhanced whenprocesses are simple and transparent and whenactors have strong and consistent incentives forperformance. Regular monitoring and evaluationthen provides the necessary information to ensurethat the integrity of the system is maintained.

• Simple rules. Community access to resourcesneeds to be governed by simple rules that areeasy for participating communities to inter-pret and apply. To maintain the credibility of thesystem, these rules should be monitored andtransparently enforced.

• Strong performance incentives. Key ac-tors at all levels should be rewarded for per-formance through objective evaluation basedon clear criteria.

• Regular monitoring and evaluation. Sys-tematic monitoring and evaluation of programprocesses and outcomes is critical for ensuringthat programs continue to grow and adapt tochanging conditions.

8. Maintain flexibility in design of arrange-ments. Flexibility in design, often through pi-loting, is essential to allow systems to evolve andbetter adapt to local demand and capabilities.Flexible program planning and decentralized de-cision-making mechanisms, situated as close to thecommunity as possible, facilitate quick responseto change. For example, in Zambia, the Social Re-covery Program is experimenting with more directcapacity building and integration of local gov-ernments into the project cycle. In both theMoldova and Albania Social Investment Funds,the initial pilot phase was extremely important towork out operational procedures before the pro-gram was offered nationwide. As part of this learn-ing process, direct feedback from the communityon program performance is essential.

9. Design for scaling-up. Despite the manyislands of success in community-driven devel-opment, most countries still have significant op-portunities for scaling up CDD. To have a materialimpact on macro indicators of poverty, CDD


6 0

6 1

needs to take place in many communities si-multaneously. It is no longer acceptable to designCDD as small, non-replicable, isolated interven-tions. However, the challenge of scaling up isnot about bigger projects or bigger organiza-tions, but rather about achieving sustainable re-sults in a large number of communities. Thesection entitled “Scaling Up” provides more de-tail and links to documents and sites with moreinformation.

10. Invest in an exit strategy. An exit strategyfor external support is a critical component of all

CDD interventions. A clear distinction must bemade between support services that are recurrentor permanent in nature and those that are tem-porary. For recurrent services, sustainability re-quires putting in place permanent institutionaland financing arrangements at a cost that canbe supported over the medium- and long-term.Temporary services, such as initial intensive ca-pacity-building support to community-based or-ganizations, may, however, not require sustainablefinancing or permanent institutional structures.For such temporary services, explicit exit strate-gies need to be designed and implemented.

A N N E X C

6 3

Study Methods and InstrumentsThe study had four components: a Portfolio Re-view, country case studies, a Literature Review,and thematic studies.

Portfolio ReviewThe Portfolio Review was a desk study ofCBD/CDD projects (and project components)supported by the Bank between fiscal 1989 andfiscal 2004. The study:

• Identified all International Bank for Recon-struction and Development (IBRD), IDA, andSpecial Financing CBD/CDD and non-CBD/CDD lending approved Bankwide be-tween fiscal 1989 and fiscal 2003 using theBank Business Warehouse database. (Annex Enotes the methodology used to identify the 847CBD/CDD projects.)

• Examined CBD/CDD-related nonlending ac-tivities—economic and sector work in eachcase study country, relevant participatorypoverty assessments, beneficiary assessments,local-level studies of institutions, and otherformal and informal sector work.

• Reviewed project documents—appraisal doc-uments (PAD, SAR), Project Status Reports(PSRs), Aide Memoires, Implementation Com-pletion Reports (ICRs), and Operational Man-uals—for a sample of 84 CBD/CDDprojects.

• Reviewed 73 CASs and 29 PRSPs and povertysector work for a number of countries cov-ered by the sample of 84 projects. (See AnnexH for details.)

• Reviewed six recent CDD projects (one in eachBank Region) to better assess the attributes ofthe most current CDD projects under imple-mentation in the Bank.1

• Reviewed 33 OED ICR Reviews.2

• Reviewed 19 OED Assessments.

Country Case StudiesThe country case studies were undertaken tocomplement the portfolio review. The five coun-try studies include two middle-income coun-tries, Brazil and Egypt, and three low-incomecountries, Benin, Nepal, and Vietnam. The se-lection of case study countries was done to pro-vide an opportunity for pairing of middle- andlow-income countries. This provided a basis forcomparing the performance of CBD/CDD in-terventions in countries where institutions arerelatively more developed, where literacy levelsare relatively higher, and where the policy andlegal environment is stronger with countriesthat have less developed institutions, lower lev-els of literacy, and a weaker enabling environ-ment for CBD/CDD interventions. These countrycase studies also provided an opportunity togain in-depth understanding of participatoryapproaches supported by the Bank in clientcountries and to provide national perspectivesfrom a range of stakeholders on the appropri-ateness of the approach to development. Allfive case studies—Benin, Brazil, Egypt, Nepal,and Vietnam—involved desk reviews plus visitsto the country, interviews and surveys of centralgovernment officials and other internationaldonors, and focus group sessions with NGOs.The Egypt, Nepal, and Vietnam case studies in-volved limited visits to relevant project sites tomeet with communities and hold focus groupsessions. In Benin and Brazil, extensive house-hold-level fieldwork was undertaken in approx-imately 30 communities involving 1,200household surveys, 60 focus group sessions,

ANNEX D: STUDY FRAMEWORK, METHODS, AND INSTRUMENTS


6 4

Figure D.1: Study Framework

Resources for capacity building,

basic infrastructure, and services

Input

Community-Level Key Evaluation

Concerns

Bank Project-Level Key Evaluation

Concerns

Basic infrastructure and services created, more

income-generating activities, favorable enabling environment

and local institutions strengthened

Output

Empowered communities, community control and

management of decisions and resources

Outcome

Sustainable development and poverty alleviation

Impact

Have Bank-supported subproject interventions been relevant to

community priorities?

How well are objectives of CBD/CDD interventions derived

from the country's overall development priorities?

Is the CBD/CDD approach the most efficient way to create this

infrastructure for the country?

Have adequate forward and backward linkages been

established for income-generating activities?

Is there sufficient ownership of the CBD/CDD intervention in

the government?

Is there harmonization of donor approaches?

Have sustainability issues been addressed during project design

and implementation?

Is CBD/CDD seen as an important poverty alleviation tool

by the borrower?

How far have the stated goals of CBD/CDD projects been directed

at poverty alleviation?

Have the poor been specifically targeted?

To what extent are CBD/CDD interventions relevant to borrower

priorities?

Does the participatory process ensure the representation of the

whole community?

To what extent are communities satisfied with the nature of the

subprojects financed?

Are the subprojects being maintained and are they being

adequately utilized?

Do communities have greater access to information and are they aware of development activities?

Is there sufficient ownership of the CBD/CDD intervention in

the community?

Have Bank-supported interventions had an impact on

the living standards of the poor?

How far have CBD/CDD interventions improved the institutional capacity of the communities to take charge

of their own development?

Has CBD/CDD built capacity & social capital at the community level?

Source: From the “CDD Evaluation Design Paper”: http://www.worldbank.org/oed/cbdcdd/documents/discussion_paper.pdf.

and 60 key informant interviews with local gov-ernment officials. (See box 1.1 of the main re-port and Annex M for details.)

Similar household-level fieldwork was un-dertaken in two states in India—Uttar Pradeshand Madhya Pradesh. (See box 1.1 in the mainreport and Annex M for details.) The two proj-ects covered were Uttar Pradesh Sodic LandReclamation project and Madhya PradeshForestry.

Thematic StudiesTwo thematic studies were undertaken to in-vestigate issues that could not be adequatelyaddressed in either the Portfolio Review or thecase studies:

• The Bank Capacity study reviewed Bank doc-uments and interviewed and surveyed Bankstaff (152 completed surveys analyzed of 400mailed to the relevant group of staff). AnnexL presents the main results of the survey.

• The Safeguards study reviewed project ap-praisal documents, ICRs, and a limited numberof supervision reports for the 84 sample proj-ects to gather information related to safeguardcompliance. Relevant safeguard literature wasalso reviewed, as were a limited number ofOED assessments. (Annex Q).

In addition, a small number of projects fromthe portfolio were reviewed specifically for theirfiduciary compliance. A limited number of Coun-try Financial Accountability Assessments (CFAAs),Country Procurement Assessments (CPARs), andInternal Audit Department reports were also re-viewed.

Literature Review and Associated Eventson CBD/CDD around the BankThe Literature Review had four objectives: (i) togather qualitative, quantitative, and anecdotalevidence on participatory approaches to localdevelopment; (ii) to draw on the evidence in theliterature to understand the different kinds of“participatory spaces” that the Bank’s CBD/CDDinterventions have fostered at the local level;3

(iii) to explore the evidence on factors thathave a bearing on the development effectivenessof CBD/CDD–type interventions; (iv) to pro-vide a means for “testing” the validity of findingsemerging from other study components, par-ticularly case study countries and the PortfolioReview.

OED also participated in or attended brownbags, seminars, and other training events aroundthe Bank on issues relevant to CBD/CDD anddrew on relevant information disseminated atthese events.

Project Assessments of ParticipatoryAssessmentsNineteen project assessments informed thestudy: Uttar Pradesh Sodic Lands ReclamationProject (India); Borgou Pilot Project (Benin);Household Energy Project (Mali); Natural Re-source Management Project (Mali); Eastern Ana-tolia Watershed Rehabilitation Project (Turkey);Matrouh Resource Management Project (Egypt);Nepal Hill Community Forestry Project; NepalSecond Forestry Project; West Bengal ForestryProject (India); Kerala Social Forestry Project(India); Ghana Agricultural Sector InvestmentProject; Second Village Infrastructure Project(Indonesia); Kecamatan Development Project(Indonesia), Andhra Pradesh Forestry (India),Northwest Frontier Province Community Infra-structure Project (Pakistan), Community Devel-opment Fund (Eriteria), and Rural RoadsRehabilitation and Maintenance Project (Peru),Northern Resource Management Project (Pak-istan).

Review of All Relevant OED WorkAll relevant Country Assistance Evaluations, Im-pact Evaluations, and studies were reviewed, in-cluding: The Next Ascent: An Evaluation of theAga Khan Rural Support Program, Pakistan; So-cial Funds: A Review of World Bank Experience;India’s Dairy Revolution; An OED Review of So-cial Development in Bank Activities; Non-governmental Organizations in Bank-SupportedProjects: An OED Review; a participation processreview; Books, Buildings, and Learning Out-

6 5

A N N E X D


6 6

comes: An Impact Evaluation of World BankSupport to Basic Education in Ghana, WorldBank Forestry Strategy, Striking the Right Balance,and associated country studies.

Interviews with Bank StaffSupplemental interviews were conducted withBank staff working on CBD/CDD and related is-sues to get their views on various aspects of theBank’s work.

6 7

UniverseThe universe of CBD/CDD projects was identi-fied using a key word search on a textbase of ap-praisal documents (Project Appraisal Documents,or PADs, and Staff Appraisal Reports, or SARs) forall Bank projects approved between fiscal 1989and fiscal 2003.1 A total sampling frame of 847projects was identified. The population of 847 (asof September 2004) includes projects that arelargely CBD/CDD and others with CBD/CDDcomponents (a complete list of the CBD/CDDportfolio as identified is available upon request).Since the portfolio was identified using a word

search, it is likely that some projects with verysmall CBD/CDD component were missed.2

The total number (and commitment) of Bankprojects that include a CBD/CDD componenthas increased substantially overtime (figure E.1).

Distribution of the CBD/CDD Portfolio Regional. Africa had the largest number ofCBD/CDD projects approved between fiscal1989 and fiscal 2003 (266 projects, 31 percent),followed by Latin America and the Caribbean(193 projects, 23 percent). South Asia, East Asia

ANNEX E: THE UNIVERSE OF CBD/CDD PROJECTS AND ITS DISTRIBUTION

Figure E.1: Commitment and Number of CBD/CDD Projects Have Increased from Lessthan 5 Percent to 25 Percent of Bank Totals

Num

bero

fpro

ject

s

0

10

20

30

40

50

60

70

80

90

100

Commitment

1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003

2001

mill

ion

US$

0

1,000

2,000

3,000

4,000

5,000

6,000

Projects



6 8

and the Pacific, and Europe and Central Asiahad 110 projects (13 percent), 118 projects (14percent), and 94 projects (11 percent), respec-tively. The Middle East and North Africa Regionhad the smallest portfolio with 66 projects (8percent).

Sector Board. The Rural Development (RDV)Sector Board had the largest number ofCBD/CDD projects approved between fiscal 1989and fiscal 2003 (226 projects, 27 percent). TheHealth Sector Board (HE) followed with 135projects (16 percent) over the same period. So-cial Protection (SP) was a close third with 131projects (15 percent). The other important sec-tor boards for CBD/CDD projects were Education(111 projects, 13 percent), Urban Development(61 projects, 7 percent), Water Supply And San-itation (53 projects, 6 percent), and Environ-ment (43 projects, 5 percent).

Sector. While the sector board under which aproject is categorized manages the project, eachproject is also assigned, at most, five subsec-tors. The number of projects assigned two ormore sectors (multisectoral operations) has beenincreasing over time (table E.1). The percentageof multisectoral projects has been rising for thenon-CBD/CDD portfolio as well. However, thepercentage of multisectoral projects is much

Fiscal years

1989– 1994– 1999– 1989–93 98 2003 2003

CBD/CDD 43 51 53 51

Non-CBD/CDD 28 31 33 31 Source: World Bank data and calculations.

Table E.1: A Majority of CBD/CDDProjects Are Multisectoral (percent)

Figure E.2: Distribution by Lending Instrument

Perc

enta

ge

SIL APL SIM LIL ERL TAL FIL SAD PRC SAL SSL0

40

20

10

30

50

70

60

80

Lending Instrument

71.43%

10.74%6.97% 6.38%

1.53% 1.18% 1.06% 0.35% 0.12% 0.12% 0.12%

Source: World Bank data and calculations.

Note: APL=Adaptable Program Loan; ERL=Emergency Recovery Loan; FIL=Financial Intermediary Loan; LIL=Learning and Innovation Loan; PRC=Poverty Reduction

Support Credit; SAD=Sector Adjustment Loan; SAL=Structural Adjustment Loan; SIM=Sector Investment and Maintenance Loan; SIL=Specific Investment Loan;

SSL=Special Structural Adjustment Loan; TAL=Technical Assistance Loan.

6 9

higher for the CBD/CDD portfolio compared tothe non-CBD/CDD portfolio for the period (fis-cal 1989–2003).

Lending Instrument (figure E.2). Of the 847CBD/CDD projects, 841 are investment lending.The instrument chosen for 605 Bank CBD/CDDprojects was the Specific Investment Loan (SIL).

Lending instruments deemed to allow forgreater flexibility, APLs and LILs, were adoptedin 91 projects and 54 projects, respectively. Thepercentage of lending channeled through thetwo instruments has increased over time (the in-struments were introduced in 1997). Since fis-cal 1999, 32 percent of CBD/CDD projects

approved have used either an APL or a LIL, com-pared to 13 percent of the non-CBD/CDD proj-ects.

Income Category. The portfolio of 847 projectswas distributed among 119 countries: 60 in thelower-income category, 38 in the middle-incomecategory, and 18 in the upper-middle-incomecategory (figure E.3A). Three countries/territo-ries in the portfolio, Barbados, Kosovo, and WestBank, had unspecified poverty categories. Ofthe 847 projects, 503 were in low-income coun-tries, 230 were in middle-income countries, 101were in upper-middle-income countries, and 13were in the unspecified category (figure E.3B).

A N N E X E

Figure E.3: Distribution of the Portfolio Countries/Projects by World Development Indicator(WDI) Poverty Category

Unspecified poverty category

2.52%

Upper-middleincome 15.13%

Middle income31.93Lower income

50.42%

Source: World Bank data, SIMA, and calculations.

Unspecified poverty category

1.53%

Upper-middleincome 11.92%

Middle income27.15%

Lower income59.39%

A: Portfolio countries (119 countries) B: Portfolio projects (847 projects)

According to accepted sampling methodology, a10 percent proportionate random sample of 84projects, stratified by time and sector board, wasdrawn from the universe of 847 projects for in-tensive review. The stratification was done toensure that important characteristics in the uni-verse of 847 projects were adequately repre-sented in the sample. (See table F.3 for the listof CBD/CDD sample projects.)1

The CDD Anchor classifies CDD projects infiscal year 2000 and beyond into four categoriesthat are not mutually exclusive. It was difficultto apply this classification to projects approvedin the early 1990s, when the four kinds of CDDhad not been identified. Hence, OED devel-oped a simple methodology to categorize theprojects into two broad groups: CBD/CDDand CDD based on percentage of project cost(box F.1).

In the sample of 84 projects, 19 (23 percent)were identified as CDD. Extrapolating this pro-portion to the entire universe (since the sample

was randomly selected) implies that the WorldBank has approved close to 192 CDD projects inthe fiscal years 1989–2003 period (table F.1).

Some characteristics of the sample ofCBD/CDD and CDD projects are presented intable F.2. CDD projects have grown at an an-nual rate of 19.6 compared with 11.7 for theCBD/CDD projects (excluding CDD). Nearlythree-quarters of the CDD are multisectoral,compared to 40 percent of the CBD/CDD proj-ects (excluding CDD). Over 80 percent of theCDD projects are under two sector boards—So-cial Protection and Rural Development.

7 1

ANNEX F: SAMPLE OF CBD/CDD AND CDD PROJECTS

Step 1: For each project in the sample of 84, project cost was di-vided among different aspects of community participation, start-ing from information sharing and ranging to community controlover decisions and resources based on information in the WorldBank appraisal documents (PAD/SARs).a

Step 2: All costs devoted to (i) community control over deci-

sions and/or resources, (ii) creating an enabling environment, and(iii) capacity enhancing for the community were combined andthis cost was divided by total project cost.

Step 3: A project was classified as CDD if the percentage cal-culated in step 2 was 85 percent or more (85 percent being anarbitrary cutoff).

Box F.1: Methodology for Identifying CDD Projects

a. The level of information contained in appraisal documents varies widely, so some value judgments were required. For example, if the project was participatory, but only in a consulta-

tive sense, it was designated CBD, but if it involved communities in a more holistic sense it was designated CDD. Within the latter, it was a matter of determining whether or not com-

munities would be in charge of the funds allocated to them or if an outside group would manage their funds. Consequently, the amount of CDD in a project as classified by OED may be

different from the amount reported by the CDD Anchor. However, an attempt was made to follow a clearly defined strategy.

Sample Universe

CBD/CDD projects 84 847

CDD projects 19 192 [= 19/84*847]

Table F.1: CBD/CDD Projects Approved by the Bank


7 2

The CDD operations are designed to providecommunities with greater responsibility for eachaspect of the subproject cycle. Based on infor-mation from the appraisal documents, the com-munity’s role in different aspects of thesubproject cycle was categorized (figure 2.2 inChapter 2). The analysis revealed that commu-nities were responsible for subproject design innearly all CDD projects (compared with a thirdin other CBD/CDD projects, excluding the CDDprojects). Communities were also responsible forsubproject operation and maintenance for over75 percent of the CDD projects, and for sub-project implementation, for nearly two-thirdsof the CDD projects.

While the community’s role in subprojectM&E for CDD projects was higher than that ofother CBD/CDD projects, the absolute numberof projects identifying community responsibili-ties was low.

Table F.2: Composition of the Sample

CDD CBD/CDD

Number of projects 19 84

Annual growth rate (%) 19.6 14.2

Multisectoral 74 48

Sectoral (single sector) 26 44

Sectoral (single subsector) 0 8

Sector Board

Social Protection 47 18

Rural Development 37 29

Education 5 13

Urban Development 5 6

Water Supply and Sanitation 5 2

Health, Nutrition and Population 0 18

Transportation 0 7

Region

Africa 26 32

East Asia and the Pacific 21 17

Europe and Central Asia 5 11

Latin America and Caribbean 32 23

Middle East and North Africa 0 7

South Asia 16 11Source: World Bank data and calculations.

Perc

enta

ge o

f pro

ject

s

Table F.3: The List of 84 Projects, Their Regions and Sector Boards

Lending Fiscal Date, rev Commit-Project name Country Project ID Sector Board instrument year closing ment ($m)

Africa Region

Borgou Pilot Benin P057345 Rural Sector LIL 1998 6/30/02 4.00

Community-Based Rural Burkina Faso P035673 Rural Sector APL 2001 6/30/06 66.70

Development

Urban II Burundi P000205 Urban Development SIL 1989 12/31/96 21.00

SDA/Human Resources Cameroon P000405 Social Protection SIL 1990 21.50

National Livestock Central African P000474 Rural Sector SIL 1995 6/30/00 16.60

Development Republic

Public Works and Chad P000533 Social Protection SIL 1994 6/30/99 17.40

Capacity Building

Third Education Comoros P000603 Education SIL 1997 12/31/03 7.00

Emergency Recovery/ Congo, P081924 Poverty Reduction ERL 2003 12/31/07 41.00

Community Project Republic of

A N N E X F


Pastoral Community Ethiopia P075915 Rural Sector APL 2003 12/31/08 30.00

Development

AG Services Gambia, The P000818 Rural Sector SIL 1993 3/1/99 12.30

Secondary Schools Ghana P000954 Education SIL 1991 6/30/95 14.70

National Health Guinea-Bissau P035688 Health, Nutrition SIL 1998 6/30/05 11.70

Development Program and Population

Sexually Transmitted Kenya P001333 Health, Nutrition SIL 1995 6/30/01 40.00

Infections Project and Population

Rural Transport Project Madagascar P073689 Transport APL 2003 6/30/09 80.00

Urban Development Program Mauritania P069095 Urban Development APL 2002 12/31/06 70.00

HIV/AIDS Response Project Mozambique P078053 Health, Nutrition APL 2003 12/31/08 55.00

and Population

Community Based Nigeria P069086 Social Protection SIL 2001 2/28/06 60.00

Poverty Reduction

Human Resources Rwanda P045091 Education SIL 2000 6/30/06 35.00

Development

Rural Water Supply Rwanda P045182 Water Supply SIL 2000 12/31/06 20.00

& Sanitation and Sanitation

Quality Education For All Senegal P047319 Education APL 2000 12/31/04 50.00

Social Development Fund Senegal P041566 Social Protection APL 2001 12/31/05 30.00

HIV/AIDS Prevention Senegal P074059 Health, Nutrition APL 2002 9/30/07 30.00

& Control and Population

Health Sector Sierra Leone P074128 Health, Nutrition SIL 2003 2/28/08 20.00

Reconstruction & and Population

Development

Poverty & Social Costs Uganda P002966 Health, Nutrition SIL 1990 9/30/95 28.00

and Population

Small Towns Water Uganda P002957 Water Supply and SIL 1994 6/30/03 42.30

Sanitation

EMCBP Uganda P002978 Environment SIL 1996 6/30/01 11.80

Pilot RDC Zimbabwe P045029 Urban Development SIL 1997 6/30/00 12.30

Total number of African projects 27 Total from 22 countries 848.30

East Asia & Pacific Region

Social Fund Cambodia P037088 Social Protection SIL 1995 6/30/00 20.00

Shanxi Poverty Alleviation China P003649 Rural Sector SIL 1996 12/31/03 100.00

Disease Prevention (Hlth7) China P003589 Health, Nutrition SIL 1996 6/30/04 100.00

and Population

Anning Valley Agricultural China P049665 Rural Sector SIL 1999 12/31/04 120.00

Development

7 3

(continued on following page)

Table F.3: The List of 84 Projects, Their Regions and Sector Boards (continued)


7 4



Sustainable Forestry China P064729 Rural Sector SIL 2002 8/31/09 93.90

Development

Irrigation Subsector II Indonesia P003953 Rural Sector SIL 1992 7/31/95 225.00

Third Community Health Indonesia P003914 Health, Nutrition SIL 1993 3/31/01 93.50

and Nutrition and Population

WSSLIC II Indonesia P059477 Health, Nutrition SIL 2000 6/30/09 77.40

and Population

Second Kecamatan Indonesia P073025 Social Development SIL 2001 12/31/06 320.20

Development Project

Community Based Philippines P004595 Rural Sector SIL 1998 6/30/06 50.00

Resource Management

Kalahi-CIDSS Project Philippines P077012 Social Development SIL 2003 6/30/09 100.00

Agriculture Rehabilitation Timor-Leste P070533 Rural Sector SIL 2000 3/15/03 6.80

Project

Small Enterprises Project II Timor-Leste P072654 Private Sector SIL 2002 12/31/06 7.50

Development

Second Education Project Vanuatu P004823 Education LIL 2001 6/30/05 3.50

Total number of East Asia and Pacific projects 14 Total from 6 countries 1,317.80

Europe & Central Asia Region

Natural Resource Armenia P057847 Rural Sector SIL 2002 7/31/08 8.30

Management

Highway Azerbaijan P040716 Transport SIL 2001 6/30/05 40.00

Farmer Support Services Croatia P008335 Rural Sector SIL 1996 12/31/02 17.00

Social Safety Net Kyrgyz Republic P008515 Social Protection SIL 1995 4/30/00 17.00

Rural Development Poland P058202 Rural Sector SIL 2000 6/30/05 120.00

SDF 2 (APL 2) Romania P068808 Social Protection APL 2002 8/31/06 20.00

Rural Education Romania P073967 Education SIL 2003 9/15/09 60.00

Rural Infrastructure Tajikistan P058898 Rural Sector SIL 2000 3/31/06 20.00

Rehabilitation

Health I Uzbekistan P009125 Health, Nutrition SIL 1999 12/31/04 30.00

and Population

Total number of Europe and Central Asia projects 9 Total from 8 countries 332.30

Latin America & Caribbean Region

Renewable Energy in Argentina P006043 Energy and Mining SIL 1999 9/30/05 30.00

Rural Markets

Health Sector Reform Bolivia P074212 Health, Nutrition APL 2001 6/30/06 35.00

and Population

Land Management 3 Brazil P006474 Rural Sector SIL 1998 12/31/05 55.00

(Sao Paulo)

7 5

A N N E X F

Basic Education Costa Rica P006938 Education SIL 1992 9/30/00 23.00

Provincial Health Dominican P007015 Health, Nutrition SIM 1998 6/30/04 30.00

Services Project Republic and Population

Reconstruction & Local Guatemala P049386 Social Protection SIL 1999 6/30/05 30.00

Development

Nutrition/Health Honduras P007392 Health, Nutrition SIL 1993 6/30/01 25.00

and Population

PROFUTURO Honduras P057350 Environment SIL 1999 10/31/04 8.30

Fifth Social Investment Honduras P064895 Social Protection SIL 2001 6/30/05 60.00

Fund Project

On-Farm & Minor Irrigation Mexico P007701 Rural Sector SIL 1994 3/31/02 200.00

Second Decentralization Mexico P007702 Private Sector SIL 1995 6/30/00 500.00

Development

Basic Education Mexico P040199 Education APL 1998 12/31/01 115.00

Development APL I

Rural Development in Mexico P007711 Rural Sector APL 1998 6/30/03 47.00

Marginal Areas

Rural Development in Mexico P057530 Rural Sector APL 2000 6/30/05 55.00

Marginal Areas II

Social Investment Fund Nicaragua P007786 Social Protection SIM 1993 9/30/96 25.00

Basic Education II Panama P052021 Education SIL 2001 6/30/05 35.00

Social Development Fund Peru P008062 Social Protection SIM 1994 6/30/97 100.00

Second Rural Roads Project Peru P044601 Transport SIM 2001 6/30/05 50.00

Caracas Slum Upgrade Venezuela P040174 Urban Development SIL 1999 6/30/05 60.70

Total number of Latin America and Caribbean projects 19 Total from 12 countries 1,484.00

Middle East & North Africa Region

Social Fund II Egypt, Arab P043102 Social Protection SIL 1996 6/30/01 120.00

Republic of

NW Mountainous and Tunisia P072317 Rural Sector SIL 2003 12/31/08 34.00

Forest Areas Development

OT - Emergency West Bank P034112 Transport SIL 1994 12/31/98 30.00

Rehabilitation I and Gaza

Palestinian NGO Project II West Bank P071040 Social Protection SIL 2001 8/31/05 8.00

and Gaza

Rural Access Improvement Yemen, P070391 Transport APL 2001 12/31/05 45.00

Program Republic of

Taiz Municipal Development Yemen, P070092 Urban Development SIL 2002 12/31/05 45.20

& Flood Protection Republic of

Total number of Middle East and North Africa projects 6 Total from 4 countries 282.20





7 6



South Asia Region

Social Investment Bangladesh P053578 Rural Sector SIL 2003 6/30/07 18.20

Program Project

Rural Access Roads Bhutan P059481 Transport SIL 2000 4/30/05 11.60

Population Training (VII) India P009940 Health, Nutrition SIL 1990 6/30/98 96.70

and Population

Uttar Pradesh Sodic India P009961 Rural Sector SIL 1993 3/31/01 54.70

Lands Reclamation

Andhra Pradesh Forestry India P010449 Rural Sector SIL 1994 9/30/00 77.40

Blindness Control India P010455 Health, Nutrition SIL 1994 6/30/02 117.80

and Population

Community School Nepal P082646 Education LIL 2003 9/30/06 5.00

Project

Social Action Program Pakistan P010456 Education SIL 1994 12/31/97 200.00

NWFP On-Farm Water Pakistan P071092 Rural Sector SIL 2001 6/30/06 21.40

Management Project

Total number of South Asia projects 9 Total from 5 countries 602.80

Total number of projects in sample 84 Total from 57 countries 4,867.40

7 7

The performance of the completed projects inthe portfolio is analyzed using OED ratings foroutcome, sustainability, and institutional im-pact.1 Of the 3,917 projects approved betweenfiscal 1989 and fiscal 2003, 2,187 were inactive asof end of fiscal year 2003. To make the compar-ison between CBD/CDD and non-CBD/CDD proj-ects robust, a few modifications were made thatresulted in a database of 1,728 inactive projects(table G.1):

• Adjustment lendings were dropped (only 6CBD/CDD adjustment lending projects).

• Projects exiting before 1994 were dropped(the first CBD/CDD project exited in 1994).

The analysis did not attempt an annual com-parison, because only 5 CBD/CDD projects ex-ited in 1994; 9 in 1995, and 4 in 1996. Instead,the entire period from 1994 to 2003 was dividedinto two phases: 1994–98 (phase 1) and1999–2003 (phase 2). Of the 334 CBD/CDD proj-ects, 70 projects exited during phase 1 and 264projects exited during phase 2. Of the non-CBD/CDD projects, 508 exited in phase 1 and 886in phase 2.

Outcome Ratings for CompletedInvestment Projects2

Overall: About 74 percent of the CBD/CDD proj-ects were rated “satisfactory” on outcome forboth phase 1 and phase 2. The correspondingnumbers for the non-CBD/CDD projects are 66percent and 72 percent (figure G.1). Though theCBD/CDD portfolio outperforms the non-CBD/CDD portfolio, the difference between theratings for fiscal years 1999–03 is statistically in-significant and the trend for non-CBD/CDD is a

rising one. Outcome ratings of CBD/CDD projectshave been better when they are disbursement-weighted.

This evaluation also found no evidence tosupport the hypothesis that the CBD/CDD proj-ects in conflict/post-conflict countries outper-form the CBD/CDD projects in non-conflictcountries for the period 1999–2003 by 4 per-centage points (table G.2).3

Regional:4 The Africa Region has the largestCBD/CDD portfolio, but is the lowest-perform-ing Region on outcome. While only 61 percentof the Africa CBD/CDD projects were rated sat-isfactory for the aggregate period 1994–2003,the Region has improved by 4 percentage pointsfrom phase 1 to phase 2 (table G.3). The LatinAmerica and Caribbean Region, with the sec-ond-largest portfolio, is the best-performing Re-gion on outcome. However, the percentsatisfactory rating in Latin American and theCaribbean on outcome has declined by 10 per-centage points for the CBD/CDD projects.5 The

ANNEX G: OED RATINGS OF COMPLETED PROJECTS

CBD/CDD Non-CBD/CDD

Number of projects approved,

1989–2003 847 3,070

Number of closed projects 336 1,851

Number of closed

investment projects 334 1,421

Number of closed investment

projects, 1994–2003 334 1,394Source: World Bank database.

Table G.1: Project Sets Compared

South Asia Region has maintained its perform-ance over time, but that of the East Asia and Pa-cific Region has declined (10 percentage points).

Primary Sector: CBD/CDD projects codedunder the education sector show the best per-formance on outcome, followed by projectsunder the transport, urban development, and so-cial protection sectors. Those under the rural de-velopment sector, with the largest CBD/CDDportfolio, performed below average on outcomein aggregate, as did projects under water supply,health, and environment. The CBD/CDD projectsunder the rural development sector show stableperformance; however, those under the socialprotection sector (second-largest CBD/CDDportfolio) show a marginal decline (table G.4).Projects under the education sector indicate an11 percentage point improvement, and thoseunder the health sector a 7 percentage pointdecline.6

The non-CBD/CDD portfolio, in aggregatefor 1994–2003, performed better than theCBD/CDD portfolio for the environment, health,and transport sectors, and CBD/CDD portfoliofor education, rural development, social pro-tection, and water supply outperformed that ofnon-CBD/CDD portfolio for satisfactory ratingson outcome. However, the differences were notstatistically significant between the two groupsfor any of the sectors.

Sustainability and Institutional Impact

Sustainability. While sustainability ratings haveimproved for both the CBD/CDD and non-CBD/CDD projects, a significantly lower per-centage of CBD/CDD projects were rated “likely”or better on sustainability compared with thenon-CBD/CDD portfolio (figure G.2).7 The Re-gional variation is somewhat similar, as seen forthe outcome ratings—the Africa Region has thesmallest percentage of projects rated “likely” orbetter, and the Middle East and North Africa Re-gion has the highest percentage, followed closelyby Latin America and the Caribbean (table G.3).While both Africa and Latin America and theCaribbean (the two Regions with large CBD/CDDportfolios) indicate improving sustainability forCBD/CDD and non-CBD/CDD projects, theCBD/CDD projects in Africa outperform non-CBD/CDD projects in phase 2 and that of non-CBD/CDD projects outperform CBD/CDDprojects in Latin America and the Caribbean.

Institutional Development Impact.8 The per-centage of projects rated “substantial” or betteron institutional development (ID) impact was


7 8

Figure G.1: Projects—the Gap inSatisfactory Outcome Ratings BetweenCBD/CDD and Non-CBD/CDD ProjectsHas Narrowed

Satis

fact

ory

(%)

1994-98

CBD/CDD

74%66%

72%73%

1999-20030

40

20

60

80

100

Non-CBD/CDD


Note: OED ratings are based on OED reviews of ICRs, 25 percent of which

are subsequently revisited through OED field assessments. The outcome

ratings of the closed investment projects reveal insignificant differences

between CBD/CDD and non-CBD/CDD projects in the two phases. The

differences between CBD/CDD and non-CBD/CDD projects were also

insignificant for each exit year between fiscal 1999 and 2003.

CBD/CDD Non-CBD/CDD(%) (%)

Conflict/post-conflict

countries 76 69

Non-conflict countries 72 73

All countries 73 72Source: World Bank database.

Table G.2: Satisfactory OutcomeRatings Higher for Conflict/Post-Conflict Countries (1999–2003)

7 9

lower for the CBD/CDD portfolio (29 percent)than for the non-CBD/CDD portfolio (36 per-cent) for the exit period 1994–98. However, thedifference between the two groups was negligi-ble for the exit period 1999–2003. While the per-formance has improved for both groups,especially for the CBD/CDD portfolio, the per-centage for both groups is still low (46 percent).

A N N E X G

Figure G.2: Project Sustainability HasBeen Consistently Lower for CBD/CDDProjects But Is Improving

Like

ly o

r bet

ter (

%)

1994-98

Exit fiscal year

CBD/CDD

37%

48%

63%55%

1999-030

40

20

60

80

100

Non-CBD/CDD


Table G.3: Regional Variations (1994–2003)

RegionEast Europe & Latin Middle

Asia & Central America & East & South Africa Pacific Asia Caribbean N. Africa Asia Total

Number of projects 26 9 2 21 3 9 70

Phase 1 Outcome (%) 58 78 50 95 67 78 74

1994–98 Sustainability (%) 31 33 50 43 67 33 37

Institutional development

impact (%) 27 33 50 29 67 11 29


Phase 2 Outcome 62 68 75 85 83 78 73

1999–2003 Sustainability (%) 39 45 55 72 67 66 55


impact (%) 33 29 50 63 61 54 46


Total Outcome (%) 61 70 73 87 81 78 73

1994–2003 Sustainability (%) 37 43 55 65 67 60 51


impact (%) 31 30 50 55 62 46 42Source: World Bank database.


8 0

Table G.4: Education Projects Lead in Percentage of Projects Satisfactory on Outcome (1994–2003)

CBD/CDD Non-CBD/CDD

Non-No. of Phase 1 Phase 2 CBD/CDD No. of Phase 1 Phase 2 CBD/CDD

Sector projects (%) (%) (%) projects (%) (%) (%)

Education 43 80 91 88 139 85 82 83

Environment 21 0 65 62 44 60 68 66

Health 56 64 57 59 83 65 70 69

Rural development 77 69 69 69 249 61 66 63

Social protection 69 82 81 81 34 77 76 76

Transport 11 100 75 82 183 82 85 84

Urban development 27 100 79 81 76 70 61 64

Water supply 19 50 73 68 66 50 67 62

Grand total 323 74 73 73 874 69 74 72


8 1

The sample of 84 projects spans 57 countries.Twenty-eight of these countries had completedPoverty Reduction Strategy Papers (PRSPs) as ofApril 2004. All 28 PRSPs were reviewed to assessthe borrower countries’ focus on CBD/CDD andCDD-related aspects. The borrower countryfocus was compared with the CDD focus of theWorld Bank Country Assistance Strategy for thesame country. Only CASs prepared since fiscal1999 were considered. As a result, 26 CASs werereviewed. The relevant questions and the re-spective responses are presented in the table H.1.Overall, the review concluded:

• There are significant differences between CASsand PRSPs on three issues related to CBD/CDD

approach: donor coordination, enabling envi-ronment, and decentralization.

• There is similarity between CASs and PRSPson many issues, including one basic issue:over four-fifths of each are silent on com-munity management and control of resourcesas a strategy for the CBD/CDD approach (fig-ure H.1).

To assess the evolution of the focus onCBD/CDD and CDD-related aspects, 58 CASswere reviewed, 2 from each of 29 countries (17with a PRSP, and 12 without). The selection wasmade based on the availability of a CAS for acountry from two time periods, one from fiscalyears 1994–98 and one from fiscal years

ANNEX H: FOCUS ON CBD/CDD AND RELATED ASPECTS IN BANK

AND BORROWER STRATEGY

Figure H.1: Fewer Than a Fifth of CASs and PRSPs Identify Community Control over Resources

15%0%

73%

12%

Source: Review of CASs and PRSPs.

8%0%

65%

27%

CASs (n = 26) PRSPs (n = 26)

Identifies community participation in decision making & resource allocation, especially of the poor Identifies partnership (community role in decision making & planning responsibilities) Identifies information & coming closer to communities Top-down approach (no participation in the formal decision making)


8 2

1999–2004. The relevant questions and the re-spective responses are presented in table H.1.Overall, the review concluded:

• The emphasis on the CBD/CDD approach inthe Bank’s country-level strategy has increasedover time (see figure 3.4 in main text).

• Attention has increased over time to issuesrelated to the CBD/CDD approach: capacity en-hancement, dissemination, donor coordina-tion, enabling environment, decentralization,and monitoring and evaluation.

• The emphasis on a multisectoral approach incountry strategies has changed little over time.

Table H.1: Details of the CAS and PRSP Review (percent)

CAS CASDoes the strategy Options CAS PRSP FY99–2004 FY94–98

Identify level of In decision making & resource allocation, especially of the poor 15 8 21 0

community participation In decision making & planning responsibilities 73 65 59 66

as important for poverty In information sharing & coming closer to communities 12 27 7 7

alleviation? No participation in the formal decision making 0 0 14 28

Reflect participatory Reflects with examples of involvement of all stakeholders 12 12 17 7

approach in other Reflects with examples of local government (+) involvement 46 38 41 14

economic, macro, Reflects, but without explicit involvement of local 38 42 41 31

sector work and government and/or grassroots

analysis? Not reflected 4 8 0 48

Emphasize a Yes, with open menu 4 4 3 7

multisectoral approach? Yes, with a positive list on the menu 23 12 21 7

Indicates community choice 31 73 24 21

No mention of multisectoral approach 42 12 52 66

Link decentralization Links to lending and to community participation 15 12 17 3

to participation? Links to community participation 19 54 17 14

Links without explicit linkage to community participation 58 31 52 31

Decentralization not addressed as an issue 8 4 14 52

Focus on improving Yes, with emphasis on communities 15 23 21 3

the dissemination Emphasis either on how or to who, not both 31 42 41 14

of information? Some indication; but not on how and to who 35 31 21 17

No focus on dissemination of information 19 4 17 66

Focus on capacity Emphasis on community & local government CB 12 4 17 0

building (CB)? Emphasis on local government or community CB, not both 46 65 59 45

Refers to capacity building 42 31 24 52

No focus on capacity building 0 0 0 3

Table H.1: Details of the CAS and PRSP Review (percent) (continued)

CAS CASDoes the strategy Options CAS PRSP FY99–2004 FY94–98

Focus on building an Lending programs developed to support enhancing or 8 0 7 0

enabling environment building an enabling environment for CDD projects

within the country for Indication on enhancing or building an enabling 27 69 17 7

supporting projects? environment for CDD projects

Emphasis on enhancing or building an enabling environment 62 27 72 86

for supporting projects

No focus on building an enabling environment within 4 4 3 7

the country for supporting projects

Put emphasis on Links lending to improving M&E and CB to carry out M&E 19 4 14 0

monitoring and Emphasis on improving M&E or M&E capacity, not both 58 69 55 21

evaluation (M&E) Refers to M&E 23 27 28 28

of activities? No emphasis on M&E 0 0 3 52

Address donor Links lending to strategies (or developing strategies) 27 0 28 17

harmonization and dealing with donor coordination issues

coordination issues Developing strategies/multilateral networks to deal 54 27 59 52

to foster cooperation, with donor coordination issues

and less competition Indicates donor coordination; but no explicit 19 58 10 24

(MDG 8)? strategy/multilateral networks

No explicit suggestion to address donor coordination issues 0 15 3 7

Total number 26 26 29 29

Source: Review of CASs and PRSPs.

APPENDIX H : RESULTS FROM THE MAP SELF -ADMINISTERED QUEST IONNAIRE

8 3

8 5

Central GovernmentA structured survey of government officials wasconducted in four case countries—Benin, Brazil,Nepal, and Vietnam—and unstructured inter-views of government officials were conducted inEgypt, to assess, among other things, the extentto which Bank-supported participatory inter-ventions have been relevant to government andcommunity priorities, and to what extent theseinterventions helped improve the institutional ca-pacity of the government at both the centraland local levels. Unstructured interviews of gov-ernment officials were also conducted in Turkeyin conjunction with an OED project assessmentmission.

In Benin, a total of 26 interviews were con-ducted with different ranking officials based inCotonou; in Nepal, 16 central government offi-cials based in Kathmandu were interviewed; andin Brazil, 8 state government officials from Natalwere interviewed (interviews with the Inter-American Institute for Cooperation in Agriculture(IICA) were not considered). In Vietnam, theProject Management Unit (PMU) director (ahigh-ranking official from the ministry) for eachof the participatory projects was interviewed,for a total of 14 interviews. The most pertinentresults for the evaluation are presented below.

The Bank’s comparative advantage andexpertise: Sixty percent of officials agreed thatthe Bank has a comparative advantage in advis-ing government on the basis of analytical andevaluative evidence, rather than in working di-rectly with communities. Thus, predictably, nearly60 percent of the government officials agreedthat the Bank should provide resources to thecentral government to carry out participatoryprojects rather than undertake these interven-

tions on its own. In Benin and Brazil, about 50percent of the officials agreed that the Bank hasthe expertise to build or enhance local govern-ment capacity to support participatory inter-ventions; the percentage was much lower inNepal (only 19 percent).

The Bank’s ability in using participatoryapproaches to address pertinent issues:Fewer than a quarter of the officials surveyed inNepal and Vietnam and fewer than a third inBenin perceive that the Bank can account forsocial and cultural factors influencing out-come, ensure sustainable flow of benefits, orensure downward accountability to the lowestlevel of government using participatory ap-proaches. In Brazil, although 75 percent of theofficials think that the Bank has the ability to en-sure downward accountability to the lowest levelof government using participatory approaches,only one official indicated that the Bank can en-sure a sustainable flow of benefits after projectsfinish.

The change in coordination between gov-ernmental units: A majority of officials in Braziland Vietnam indicated increased frequency ofmeetings within the ministry and among min-istries since the initiation of the Bank’s partici-patory intervention. In Nepal, however, only 44percent reported an increase in the frequency ofmeetings among ministries, compared with 81percent reporting an increase in the frequencyof meetings within ministries.

Responsibility for monitoring and evalua-tion of Bank-funded participatory proj-ects: In Nepal and Vietnam,1 a majority of theofficials indicated that the responsibility of mon-

ANNEX I: CENTRAL GOVERNMENT AND LOCAL GOVERNMENT SURVEYS


8 6

itoring and evaluation of Bank-funded partici-patory projects rests with the central or the re-gional government. In Benin and Brazil, a large

percentage of respondents did not pick any gov-ernment level or communities to be responsible.A majority had picked others, with 11 to 13 per-

Table I.1: Government Officials Survey Results by Country (percent)

Benin Brazil Nepal Vietnam Total

Participation leads to better outcome 65 83 63 86 73

Bank knowledge and expertise. Agree that Bank:

Has comparative advantage in advising government on the basis of analytical

and evaluative evidence rather than work directly with communities 54 63 69 60

Should provide resources to the central government to carry out participatory

projects rather than undertaking them directly 50 50 69 64 58

Has the expertise to build/enhance local government capacity to support

participatory interventions 50 50 19 40

Bank has substantial ability of using participatory approaches

on the following aspects:

Account for social and cultural factors influencing outcome 15 50 6 14 17

Ensure sustainable flow of benefits after projects finish 23 13 13 21 19

Ensure accountability downward to lowest level of government 27 75 13 21 28

Intra-government coordination. Since the initiation of Bank participatory

interventions increase in frequency of meeting:

Within ministry 75 81 64 74

Between ministries 63 44 64 55

Level of empowerment effective and efficient for development approaches:

Community is informed and consulted on government development plan 79 25

Community prepares and/or implements a development plan 0 38

Community prepares a development plan with the help of gov/NGOs 17 19

Community has control over decisions and resources 4 19

More than 75 percent of the communities have the ability to:

Identify needs and prioritize them 17 21

Manage financial resources 0 29

Participatory approaches increase time in involving communities 80 50 81 79 71

M&E responsibility for the Bank-funded participatory projects is with:

Central government 19 13 56 50 34

Regional government 0 13 13 36 13

Local government 4 0 0 21 6

Communities 8 0 0 7 5

NGOs 0 0 6 14 5

Do not know 12 13 6 7 9

Total number of observations 26 8 16 14 64

Source: Government Officials Survey.

cent reporting no knowledge of who was re-sponsible for M&E.

Local Government SurveysSurveys of local government officials were con-ducted in two countries where more intensivefieldwork was carried out. In Brazil, structuredsurveys were conducted with 38 local govern-ment representatives, while in Benin interviewswere conducted with 24 local government rep-resentatives to assess, among other things, theextent to which Bank-supported participatory in-terventions have been relevant to local govern-ment and community priorities. It should benoted that because of Benin’s DecentralizationProgram and recent elections, many of the com-munal representatives were relatively new totheir positions.

Among the interesting findings in these sur-veys: local government representatives in bothcountries appear to be skeptical about the levelof competency of their constituent communitiesto take charge of their own development. InBrazil, for example, only 33 percent of local gov-ernment officials surveyed believed that morethan 75 percent of communities have the ca-pacity to identify and prioritize their needs; inBenin, the figure was 45 percent. It is worth not-ing that in Benin, 32 percent of officials said thatbetween 0 and 50 percent of communities havethis capacity; in Brazil, the figure was 30 percent,with a full 16.7 percent feeling that none of thecommunities had this capacity. (See table I.2.)

Officials in both countries also were askedwhat percentage of communities had the capac-ity to prepare a development plan. In Brazil, 47percent said that fewer than 25 percent of com-munities had this capacity; 24 percent said thatnone of their communities had this capacity. InBenin, the numbers were similar: 27 percent saidthat fewer than 50 percent of communities hadthis capacity and 32 percent believed that fewerthan 25 percent did.

Officials in both countries also were askedwhat percentage of communities had the capac-ity to implement and maintain a subproject. InBrazil, 57 percent of respondents believed thatfewer than half of communities had this capac-ity; in Benin, the figure was closer to 67 percent.

When asked how many communities had thecapacity to manage the financial resources in-volved in a subproject, 52 percent of local officialsin Brazil estimated that fewer than 25 percent ofcommunities were capable; in Benin, 57 percentof local officials gave this same estimate.

Despite these grim assessments of communitycapacity by the local government representa-tives in Benin and Brazil, the survey respon-dents support the participatory process ingeneral. When asked to what extent they feltthe participatory approach should be extended,60 percent of Brazilian government officials re-sponded “all sectors” and 54 percent responded“all communities.” A smaller proportion arguedfor “some” sectors and communities, and onlyone respondent responded “none” to eitherquestion. In Benin, results were similar: 84 per-cent believed that the participatory approachshould be scaled up to more communities, and75 percent believed that participatory approachesshould be scaled up to all sectors.

The reasons for this dichotomy are unclear.While it may be that local governments were of-fering a “politically correct” answer to these ques-tions, it is also possible that local governments arelooking to either increase or maintain their in-volvement in local development activities and maybe threatened by the notion of complete controlof subproject implementation by communities.2

Other issues of interest that were raised dur-ing local government surveys were:

• NGO Capacity. Local officials in Benin weredivided on whether NGOs should be involvedin the implementation of participatory projects.

8 7

A N N E X I

Brazil BeninRange Percentage Range Percentage

Above 75% 33 Above 75% 45

50–75% 37 50–75% 18

25–50% 7 25–50% 18

Below 25% 7 Below 25% 14

None 17 None 0

Total 100 Total 100

Source: Local Government Officials Survey.

Table I.2: Percentage of CommunitiesThat Can Identify and PrioritizeTheir Needs


8 8

A common view expressed was that the qual-ity of NGOs can sometimes be suspect, andthus there seems to be some apprehensionabout NGOs in general

• Central/local government coordination.Local officials in Benin indicated that the levelof coordination has increased substantiallysince the onset of participatory projects intheir area. However, in Brazil, only 43 percentof local officials agreed that this was the case,and 24 percent believe that coordination hasactually decreased.

• Increased time needed to involve com-munities. The vast majority of local officials

in Benin and Brazil agreed that involving com-munities in participatory development ap-proaches requires an increase in time, whichthereby indicates increased costs associatedwith the approach. In the case of Brazil, it isinteresting to note that the response rates tothis question at the central level (50 percent)are significantly lower than the response ratesat the local level (80 percent). In Benin, re-sponse rates to this question between thecentral and local level were similar, with the 75percent at the local level and 80 percent at thecentral agreeing that participatory approachesrequire more time.

8 9

This annex explores the costs of adopting aCBD/CDD approach to multiple actors and thebenefits for poverty impact in an attempt to bet-ter understand efficiency in CBD/CDD projects.The data are limited, but CBD/CDD projectsseem to cost more to design and implement forall the players, but may provide offsetting savingsin infrastructure costs. Whether a sufficient en-hanced poverty impact occurs to justify the extracosts incurred is not evident in the cases stud-ied, but poverty impact is not well evaluated.

A typical project has multiple layers of oper-ational costs. These are incurred by the Bank aslender, by the borrower (perhaps at several lev-els) as implementer, possibly by a contractor, andfinally by the households of the community ofbeneficiaries. An efficient system would be onethat, for a given resource transfer and project out-come, would be “least cost,” with due socialweighting of costs and benefits in favor of anypoverty objective. Presumably the system shouldmaximize the incentives down through the chainof actors.

There are four main categories of cost1 thatcan be compared between CBD/CDD and non-CBD/CDD interventions:

• Operational costs to the Bank for appraisaland supervision

• Operational costs to the borrower for appraisaland supervision

• Unit costs of project investments, such as costsof contracted construction per kilometer ofroad

• Opportunity costs to beneficiaries of partici-pation.

Benefits can be divided into primary benefitsfrom investments, such as productivity or wel-

fare gains, socially weighted as appropriate forpoverty objectives, and secondary benefits thatmight arise at a later date from improved ca-pacity. Each of the above will be reviewed in thisannex.

Operational Costs to the BankThe Bank’s operational costs have been assessedby three means: (i) actual Bank operational costdata against project commitment size and bytype of project—CBD/CDD or non-CBD/CDD;(ii) a staff survey to assess staff perceptions aboutBank costs; and (iii) an earlier study that alsoused staff interviews and actual cost data.

Bank Costs Based on DataBank costs for projects with a CBD/CDD ap-proach are higher than for non-CBD/CDD ap-proaches. The Bank cost graphs in the mainreport (Chapter 3, figure 3.8), read in conjunc-tion with table J.1, show, for the project uni-verse, the lending costs up to Board approval,and the supervision costs thereafter. For the su-pervision costs, only the completed projectswere taken, leaving 1,493 non-CBD/CDD and374 CBD/CDD from the total of 2,361 and 839,respectively. They also show that CBD/CDD proj-

ANNEX J: EFFICIENCY

Non-CBD/CDD CBD/CDD

Average costs to approval 395 355

Average supervision costs 430 356

Total Bank operational costs 825 711

Source: World Bank database and calculations.

Table J.1: Mean Bank OperationalCosts by Type of Lending (US$’000)for the Mean $50 to $60 MillionCommitment Size

ects cost the Bank more to prepare, appraise, andsupervise across the whole size range. For Bankcosts up to Board approval, at the CBD/CDDproject mean commitment size of $50 to $60million (the average CBD/CDD project is a $57million commitment), the cost of CBD/CDD isabout 11 percent higher than non-CBD/CDD($430,000 compared with $356,000, a differenceof $74,000). For supervision, in the same sizebracket, CBD/CDD costs the Bank about 21 per-cent more ($430,000 compared with $356,000 fornon-CBD/CDD). The aggregate difference of op-erational costs, including costs before and afterapproval, is 16 percent for the relevant com-mitment size. These costs include trust funds. Tolook at it another way, the average non-CBD/CDDproject of $100 million commitment could beprepared for about the same cost as a $65 mil-lion CBD/CDD project. The cost gap is largelysustained across project commitment sizes. Butthe gap in supervision costs is narrower for thesmaller projects and widens with size, perhapsindicating some added challenge with scalingup of CBD/CDD.

Does the cost difference matter? An averageoperational cost increase of about 16 percentacross the Bank as a whole would certainly besignificant.

Staff Perceptions Drawn from SurveysStaff perceive the costs of CBD/CDD to be higher.The staff survey asked questions about staff per-ceptions of such relative costs. In response to thestatement (Survey Question 6) that implemen-tation costs per dollar lent for CBD/CDD proj-ects are higher than other more traditional typesof projects, 41 percent of staff either agreed orstrongly agreed, and 27 percent disagreed orstrongly disagreed, with 31 percent either neu-tral or saying they did not know. This suggeststhat a majority of those taking a position per-ceived what the data show—that Bank costs arehigher for CBD/CDD. In response to the relatedbut more specific statement (Question 7) thatCBD/CDD approaches across the whole proj-ect cycle, from identification to completion, takemore Bank staff resources per dollar of lendingthan other types of investment projects, 49 per-cent of staff agreed or strongly agreed, with 23

percent disagreeing or strongly disagreeing.This answer is consistent with the previous an-swer.

The 1994 Hentschel PaperHentschel (1994) found higher costs for partic-ipatory projects based both on interviews withstaff associated with 21 participatory operationsand on data drawn from the Bank managementinformation system. Hentschel compared a sam-ple of 42 participatory projects between 1987and 1994 with a Bankwide control group. Butcosts were compared on a per project basis, withno attempt to analyze cost per dollar lent or perdollar of total project cost. Interestingly, thepaper stopped short of aggregating the two setsof budget-origin data from Bank and non-Bank,mostly trust fund, sources. This OED study hassomewhat extended the analysis. Taking the totalresources given in the Hentschel study from allbudget sources and for all stages of the projectcycle, and assuming that both the participatoryand the Bankwide control group projects wouldbe five-year projects, suggests a total of staffweeks for the full cycle of 313 for the participa-tory sample and 223 for the Bankwide control.Under that assumption, the costs would be about40 percent higher for CBD/CDD on a per-projectbasis. The mean project sizes in the Hentschelsample are not given, so it is not possible to nor-malize for the costs per dollar lent/project sizerelationship.

Operational Costs to the BorrowerThe evidence suggests that costs to the bor-rower for CBD/CDD operations are higher thanfor non-CBD/CDD.2 However, the evidence isscattered and limited. It is drawn from twosources: first, surveys of borrower perceptionsin four case study countries and, second, somedata from Indonesia and Egypt.

About 80 percent of borrower officials whowere asked in case study country surveys whetherCBD/CDD projects took more staff time re-sponded “yes” (Benin, 80 percent; Vietnam, 73percent; Brazil [state], 50 percent; Brazil [mu-nicipal], 79 percent; and Nepal, 81 percent).The sample size by country was in the range of7 to 15. So the perception seems to be quite


9 0

strong that CBD/CDD costs more in borrowerstaff time.

Data from the Kecamatan Development Pro-ject (KDP) in Indonesia, being assessed by OED,suggest the following:

• At the subdistrict level, the operational costsof the Financial Management Units, which wasdeducted from the grants, was 5 percent ofgrants/loans. This proved just enough, butbarely, to keep the units funded.

• In addition, $61.9 million was provided for fa-cilitators, implementation technical assistance,and government administrative costs for agrant component of $189 million (about 33percent). However, a modest portion of thetechnical assistance costs could be consideredoutside of the normal operational costs. Nev-ertheless, including the costs of Financial Man-agement Units, the total operational costappears to have been not less than 30 per-cent. This is somewhat higher than typicalbreak-even costs of operating microfinance,which has been found to be around 25 percentglobally, including cost of funds at around 7 per-cent, but which has often ridden on the localinstitutional support of other community de-velopment project expenditures.

The Indonesia KDP cost can be comparedwith the non-CBD/CDD Indonesia Sulawesi Agri-cultural Area Development Project, a more tra-ditional project that did not perform well(although it had some elements of consultationin one component). The actual operating costsin that project added to half the consultant costs(since some were technical agriculture support)comes to about 25 percent of the total projectcosts, notwithstanding its much smaller size.So here there appears to be a difference of atleast 5 percent, perhaps more if normalized forsize.

In Egypt, drawing from the OED case studyanalysis, data were limited. However, operatingcosts as a percentage of the total project costsacross 8 CBD/CDD projects lay in the range of0.9 percent to 8.3 percent, with the modal fig-ure around 6 percent, while for 3 non-CBD/CDDcomparators, the operating costs were between

1.0 percent and 3.0 percent, with a modal figureof 2.6 percent. Although a very small sample, thissuggests a difference of about 3 percent, withCBD/CDD being the more costly. However, inEgypt it is probable that a number of costs werecarried by government outside the defined proj-ect funding, making a comparison with Indonesiadifficult. Also in Egypt, within the Public WorksProgram of the Social Funds III Project, the moreCBD/CDD-oriented Community DevelopmentProgram component had administrative coststhat, at 8 to 10 percent, were about 6 percenthigher than the parallel non-CBD/CDD PublicWorks Program, at 2 to 4 percent.

Unit Costs of Project InvestmentThe evidence on the costs of construction pro-vides a mixed picture. In four of the cases re-viewed, unit costs of investment, such as villageroad construction costs, have fallen with partic-ipatory approaches. In no study cases have costsrisen, although questions have been raised aboutconstruction quality, and therefore whether it isa fair comparison. In Indonesia, in both the Vil-lage Infrastructure 2 Project and the KecamatanDevelopment Project, the evidence suggests thatcosts are about 20 to 30 percent lower in com-munity-managed infrastructure than in the sameinfrastructure built by previous top-downprocesses, often using public agency force ac-count or poorly supervised or corrupt and col-luding contractors. In Brazil, a comparison byOED of the estimated cost per beneficiary ofMossoro Municipality Pipeline with the NortheastRural Poverty Alleviation Program showed thatthe investment cost of the latter was about 40percent of the former.

In Nepal, evidence shows lower subprojectunit costs from CBD/CDD projects comparedwith more traditional government agency proj-ects. For example, quoted rates in person daysper cubic meter of earth moved for roads andbridges in ordinary soil was 0.47 for the RuralCommunity Infrastructure Project and 0.70 forgovernment projects, indicating costs that wereabout 50 percent higher for the conventionalgovernment project. Also, in Nepal,3 unit costsof service delivery under community programswere found to be significantly lower than under

9 1

A N N E X J


9 2

agency programs, exhibiting in many cases over100 percent differences. However, it is unclearhow comparable the different programs were intechnical difficulty.

However, OED’s 2002 Social Fund Evalua-tion did not find any clear advantage in cost ef-fectiveness between social funds, localgovernment, other central agencies, and NGOsacross the 27 countries studied. It found the datato be highly variable, as might be suggested bythe differences between the Indonesia and Egyptdata quoted above. There were problems innormalizing for quality. That study warrantsbeing given more weight than the other casesquoted because it represents a larger samplewith a comparable methodology across countrycases. The study found some indication thatunit costs tend to be somewhat lower wherecommunity contributions were high and/orwhere there was community management andcontracting. Overhead expenses were found tobe in the range of 7 to 14 percent of total pro-gram costs.

Opportunity Costs of BeneficiaryParticipationCosts of participation are higher by definition inparticipatory projects; the question is at whatlevel are the costs of participation in relation tothe benefits and, at household level, the proba-bility of benefits. No cases were found where thecost of participation had been analyzed either exante or ex post. Indeed, in the OED Egypt casestudy, it was noted that, with the many differentparticipatory approaches being followed, an op-portunity had been squandered to compare pro-gram efficiency. Given the lack of data, we drawfrom only one project case.

The OED PPAR for the Indonesia KecamatanDevelopment Project offers an example of thecosts to a representative household of the meet-ings needed to actively participate in the eco-nomic loans component and compared it to thebenefits of the group credit provided. (See boxJ.1.) The costs were substantial. If all meetingsin the KDP project were attended, it would bepossible to go to about 16.

In a typical village in North Lampung, Sumatra, for an active par-ticipant who took an economic loan, there were 5 decision meet-ings and a Verification Team meeting. Two of the decisionmeetings were 2 hours long and involved 1 hour of travel. Threeof the meetings were 4 hours long and were farther away, tak-ing 2 hours of travel. The Verification Team meeting was a wholeday long plus 1 hour of travel (9 hours total). Thus, the total timewas 33 hours, or about 4 working days. At a minimum wage inplantations in this area of Rp21,000 per day, the opportunity costwas Rp84,000. The total time from initiation of the discussions toreceipt of grants/loans was 1 year and 4 months. Thus, for an av-erage economic loan size of Rp350,000, this person was spend-ing about 25 percent of the value of the economic loan inmeetings, with that investment not paying off in terms of re-ceipt of the money for over a year. However, there is also aprobability factor. Since KDP funding was competitive, therewas a significant chance of not receiving benefits at all. In thiskecamatan, 18 of 42 KDP proposals were accepted in the year

in question. Thus, the probability of not getting any reward in thiscase was about 0.4. Applied to the economic loan size (Rp350,000* 0.4 = 140,000), this suggests an opportunity cost in terms of timeof about 60 percent of the loan size (Rp84,000/Rp140,000). How-ever, there would be other gains on the positive side. Some ofthe time given would have gained respect and position in the com-munity. Some may have contributed to other infrastructure ben-efits relevant to the participant’s hamlet. Also, there was aprobability of not having to repay the loan at all. (In this partic-ular village loan repayments were mostly between 80 and 100percent, well above the project average.) However, it is con-cluded that, overall, the costs of full participation were sub-stantial. This probably worked against the full participation of thepoorer households who could least afford to give time at the riskof no benefits. While the case given here is a composite individualcase, a village-level calculation, assuming the levels of atten-dance at meetings reported and the types of meetings, gener-ally supports the estimate presented.

Box J.1: Costs of Household Time in a KDP Village

BenefitsThe primary benefit expected from a CBD/CDDintervention would be its impact on poverty inthe broadest sense, which would call for esti-mating the benefits reaching the lower quintilesand might also place some social weighting onthose benefits. Here we explore two types ofevidence, the evidence on poverty impact and,more broadly, the project outcome data relativeto costs to assess the development efficacy ofCBD/CDD projects relative to non-CBD/CDDwhich, among other things, accommodates thedifferent project objectives.

With respect to poverty, in the four studycases where household surveys were done4

(Benin, Brazil, and Madhya Pradesh and UttarPradesh in India), the impact on the poorestCBD/CDD quintile over the non-CBD/CDD quin-tile on consumption and expenditure was small.It was statistically insignificant in all cases, exceptfor consumption (but not expenditure) in theMadhya Pradesh case.

With respect to Bank project outcome per-formance, CBD/CDD has a slight edge over non-CBD/CDD, but only 74 percent satisfactory orbetter compared with 72 percent (from 1989 to2003)—not a large difference. Moreover, in re-cent years CBD/CDD has not improved per-formance as much as non-CBD/CDD, which hasbeen closing rapidly. This perhaps suggests thatthe Bank has learned more about how best to de-sign and implement non-CBD/CDD thanCBD/CDD. However, it is probably also a func-tion of the fact that CBD/CDD performance wascloser to a reasonable ceiling of expectation.

The Net Effect of All Cost Differencesbetween CBD/CDD and Non-CBD/CDD

CostsThe above data are indicative of costs at differ-ent levels in the system. What the data appear toshow is the following:

• Costs to the Bank are about 16 percent higherfor CBD/CDD. Given the large sample, this isa fairly robust figure. Bank operational coststhemselves are small compared with those ofgovernment or communities. Nevertheless,

the extra cost to the Bank is significant: anoverall 16 percent increase in Bank costs for thesame output across the whole Bank programwould be substantial.

• Costs to the borrower at the government levelare perceived to be higher by most officials, butit has not been possible to find comparable ac-tual cost data. Approximate costs are known insome individual cases, although there are ques-tions about cost categories. They seem to sup-port the perception of higher CBD/CDDoperational costs to government, but the sam-ple is very small.

• Costs of construction of subprojects appear tobe lower, perhaps typically around 20 percentlower, although there are cases (Nepal) wherecost savings are claimed to be much greaterthan 20 percent, and recent data from In-donesia are showing savings of over 50 percentin some cases compared with government-managed contracts. There is some questionabout the robustness of the data in some coun-tries.5

• Opportunity costs to the beneficiary of timegiven both for consultation and implementa-tion appear to be significantly higher, in somecases as high as 10 to 20 percent of the in-vestment resources provided to the house-hold, but again the data are very limited.6

Based on the above, the reduced unit costs ofinvestment would need to be substantial, perhapsas much as 30 percent, to cover the extra con-sultation and management costs, or, alterna-tively, benefits in terms of poverty impact wouldneed to be high. But this conclusion would bedifferent, particularly if operational costs to gov-ernment are not actually as high as surveyed of-ficials seem to suggest.

BenefitsCBD/CDD projects do not significantly outshinenon-CBD/CDD in outcome performance and,so far, there is limited evidence of significantpoverty reduction differences. With respect to theoutcome rating, CBD/CDD projects were ratedsatisfactory in 74 percent of cases over the pe-riod 1989–2003, compared with 72 percent fornon-CBD/CDD. Thus, CBD/CDD has performed

A N N E X J

9 3


9 4

marginally better. But this does not offer a de-cisive outcome performance edge. Moreover,as noted, the performance trend for non-CBD/CDD has gained steadily, while CBD/CDDhas remained almost static.

With respect to poverty impact, the case stud-ies and surveys found little evidence thatCBD/CDD projects have realized significantpoverty impact gains, despite their poverty ob-jectives. As found in the OED Egypt Matrouh Re-source Management Project, this is partlybecause many investments are land-related, sothat benefits are almost bound to reflect theexisting inequity of land ownership. Although inthe Matrouh case OED believed that, becausewater cisterns were targeted to the poor, therehad been some reduction in the level of re-gressiveness due to the project. In the TurkeyEastern Anatolia Watersheds Management Pro-ject, some of the poorer herders, who were notclosely linked to the decision communities, ap-peared to have been losers rather than gainersdue to grazing land closure. On the related issueof gender, while there is evidence of someprogress, there is still far to go, especially invery conservative situations such as the EgyptMatrouh Project.

CBD/CDD projects only perform 2 percentagepoints better than non-CBD/CDD on the insti-tutional development (ID) rating (44 percentversus 42 percent). This is a very small difference,

and the rating is still low in absolute terms. Thisis important in assessing CBD/CDD participa-tory performance, since the performance ofcommunity processes is a significant elementof the overall ID performance rating. In otherwords, substantial gains in participatoryprocesses at the community level should partlyshow up in gains in ID rating.

As noted earlier, secondary benefits toCBD/CDD may be relevant here as well as sec-ondary costs. On the benefits side, there may beimproved efficiency in consultative processeswith payoffs outside the project. On the costsside, there may be costs such as the cost of di-verting an NGO from a more important task toone that is less important, but more immediatelyrewarding. However, non-CBD/CDD projectsthat mostly seem to be focused more on growththan equity may have substantial poverty im-pacts through growth.

Lack of DataThe many data and analysis gaps in the efficiencystory need to be filled, particularly regarding bor-rower costs. Indeed, in the Egypt case study, asnoted above, OED pointed out a missed oppor-tunity to compare poverty impact efficiency acrossa range of Bank-funded project approaches, fromthe very intensive CBD/CDD approaches to theless-intensive social fund approaches.

The Formal SystemIn Nkayi District, western Zimbabwe, water issupposedly managed at the community levelthrough formal waterpoint committees, usuallymade up of three women (representing users)and one man (representing authority). The com-mittee is technically a subcommittee of the vil-lage development committee, and is part of atiered maintenance system involving structuresat the ward and district levels. The system isbased on the concept of establishing one com-mittee for each waterpoint, representing theusers of that point, and great emphasis is placedin training on encouraging a sense of “owner-ship” for the waterpoint. Training also empha-sizes the requirement that committee membersare elected, that meetings are held regularly,and that proper minutes are taken. The com-mittee is expected to undertake routine pre-ventive and minor corrective maintenance, andto guide the community in agreeing rules orbylaws relating to the waterpoint. Models ofsuch bylaws are given at training sessions. Thisformal management system is based on the as-sumption that people will use and manage onewater source only, and that there is a need to re-strict irresponsible use of the water source. Butresearch in Nkayi District has uncovered localpractices of water use and decision making thatare contrary to the formal system as manifestedthrough committee structures.

Local PrinciplesMany local principles of water use and manage-ment are not explicit rules or regulations, butrather customs and conventions, or what peoplesuggest is the “right way of doing things.” Theseoften predate (by many decades) the establish-ment of waterpoint committees.

Open access and the use of multiplewater sources: People prefer to maintain ac-cess to a number of different water sources overa wide area, not just to the local one that they“own.” This is partly because certain sourcesare preferred for particular purposes. It is alsofor “insurance” reasons, because if one sourcedries up, breaks down, or access to it is re-stricted, the users want to be sure of being ableto draw water elsewhere. The Nkayi people be-lieve strongly that everyone should have accessto water sources to secure at least the mini-mum necessary for survival. But such universalaccess becomes increasingly disputed duringdry months. As water sources diminish, someusers (often committee members) try to con-serve the remaining supplies by restricting ac-cess to community members in the immediatevicinity and to those who have participated inimplementation. Such action is reinforced inmany cases by “ownership” messages intro-duced by project mobilization and implemen-tation activities.

Scarcity, conventions: People in Nkayi usevery small quantities of water for domestic pur-poses (estimated at a maximum of 8 to 10 litersper person per day in the dry season—a desir-able amount would be 15 to 20 1/p/d). Evenwhen water is relatively plentiful (for example,at a fast-flowing borehole), people do not in-crease the amount they use substantially. Thereare two likely reasons for this: first, the deeplyrooted fear of drought and the perception ofwater as a scarce resource means that people ha-bitually employ practices that are water con-serving. Secondly, water use is partly determinedby who and how many in the family can collectwater—those households with lots of small chil-

ANNEX K: NKAYI DISTRICT FORMAL AND INFORMAL SYSTEMS

9 5


9 6

dren and only one adult to carry water use rel-atively small quantities.

Water-use preferences: The Nkayi men andwomen have markedly different priorities wherewater use is concerned—men want to ensurethat they can water their cattle, while women aremore preoccupied with having enough waterfor drinking, washing, and cleaning.

Ownership equals access? As people usemultiple water sources over a wide area, the ad-ministrative boundaries through which water ismanaged are not necessarily appropriate. The wa-terpoint committees are largely established onthe basis of village boundaries and are ineffec-tive in area wide resource management, as theyhave no remit outside their own restricted area.This is the case even if people of that commu-nity depend on “external” sources of water (a dis-tant borehole or a dam) for their livelihoods.Attempts to introduce greater “ownership” ofnew water supplies may result in restricting ac-cess. It is generally the poorer households andfamilies living on the outskirts that suffer fromsuch restricted access. So, under such ownershippolicies, improved management of the water-point can on occasion be achieved at the ex-pense of equity. People’s preferences regardingdifferent sources of water are complex and theirchoice of waterpoint not attributable to single fac-tors such as cleanliness or time. Additionally,

people do not generally use water sources irre-sponsibly as their proper use is defined by cus-tom and practice. Project mobilization needs totake account of such complexities and to rec-ognize local cultures of water use.

Decision making: Committees are not nec-essarily the Nkayi villagers’ preferred way of con-ducting local business. In fact, most decisions ofimportance (such as restricting access to thewater source, or rationing the amount of wateravailable, or deciding to make cash contribu-tions for maintenance), are made at “meetingsof the people” nominally held under the auspicesof the village development committee. A num-ber of decision-making principles are apparentat community level. The villagers believe thateveryone potentially affected by a decisionshould be present when it is made; thereforemeetings of all available adults in the communityare held to discuss issues of water-resource man-agement (and other related issues, such as graz-ing). Wherever possible the use and regulationof local resources is conducted both throughinformal decision-making and through adher-ing to custom and practice. Meetings are onlyheld when a problem arises and action takenonly when absolutely necessary. Many of the re-source use management and decision makingarrangements are strongly influenced by the de-sire to avoid conflict between neighbors.

Source: From Cleaver 1998 (also cited in Kumar 2003).

9 7

A survey was conducted to seek the perceptionsof selected Bank staff and managers on such is-sues as CBD/CDD project performance, incen-tives, process, and resources. A total of 400surveys were electronically mailed to a selectbut varied set of staff. The response rate was 38percent. The following analysis is based on the152 completed surveys received by OED by thespecified date (tables L.1 to L.4). Both responserate and multivariate analysis were conducted.The most pertinent results for the CBD/CDDevaluation are presented below.

Targeting: Fifty-two percent of the respondentsindicated that they agree with the statement

that the Bank-supported CBD/CDD projects aregenerally sufficiently targeted at the poorest, 16percent disagreed with the statement, and 20 per-cent neither agreed nor disagreed.

Bank strategy, processes, and products:The Bank is increasingly decentralizing its op-erations to field level. Thirty-nine percent of therespondents were satisfied with the impact ofBank decentralization on the efficacy of Bank sup-port for CBD/CDD projects. However, only 27percent of the respondents were satisfied withguidance and support from either the manage-ment in the region over the project cycle, orthe relevant thematic groups, or the Social An-

ANNEX L: RESULTS OF BANK STAFF SURVEY

Figure L.1: Fewer Than a Third of the Respondents Are Satisfiedwith the Bank’s Support and Guidance

Bank’s managementover project cycle

Thematic group s

CDD Social Anchorteam in ESSD

Bank matrix managementorganizational structure

0 4020 60 80 100

Satisfied Somewhatsatisfied

Not satisfied No response

Guidance / support for CBD/CDD projects from

Percent

Source: Bank staff survey.


9 8

chor team in ESSD, or the matrix-managementorganizational structure for the needs ofCBD/CDD projects (figure L.1). Only 9 percentwere satisfied with coordination within the Bankacross sectors.1 There were no significant varia-tions in the response to the above queries by re-spondent profession/specialty or associationwith CBD/CDD projects (table L.5).

Cost of doing business: Thirty-nine percent ofthe respondents agreed that implementationcosts per dollar lent for CBD/CDD projects arehigher than for more traditional projects (26 per-cent of the respondents disagreed, and 13 per-cent neither agreed nor disagreed).2 Also, onlyabout a quarter agreed that sufficient resourceswere made available to effectively appraise andimplement CBD/CDD projects or address safe-guard issues.3 Despite the lack of resources to ef-fectively implement participatory projects, 35percent indicated that the Bank has scaled up theproject in over 60 percent of the projects.

Sustainability: Only 18 percent of the re-spondents indicated that they agree with thestatement that community maintenance contri-butions are sufficient to sustain infrastructure in-vestment for Bank-funded CBD/CDD projects.Further, only about a quarter agreed with thestatement that the Bank generally contributesfunding for CBD/CDD projects long enough toreach a satisfactory level of sustainability of com-munity processes.

Knowledge and skills: Nearly a quarter of therespondents disagreed with the statement that

task managers of CBD/CDD projects could mon-itor fiscal accountability as satisfactorily as inmore traditional projects; 36 percent agreed,and 28 percent were neutral. Less than a fifth ofthe respondents were satisfied with the quantityand quality of training on CBD/CDD over the past2 years.

Coordination with other players: About afourth of the respondents indicated that an-other agency collaboration (NGO or bilateraldonor) significantly enhances the quality of theCBD/CDD project in over 60 percent of the proj-ects. Thirty-nine percent of the respondents alsoagreed with the statement that inadequate donorcoordination in a cofinanced project is likely tohave a greater negative impact on outcomes ina CBD/CDD project than in a more traditionalproject (18 percent disagreed and 22 percent nei-ther agreed nor disagreed).4

Comparative advantage: Thirty-eight per-cent of the respondents disagreed that bilateraldonors’ interventions generally achieve a betterCBD/CDD outcome than Bank interventions.Thirty-one percent also disagreed that NGO-supported interventions generally achieve a bet-ter CBD/CDD outcome than Bank interventions.5

Forty-four percent of the respondents indicatedthat they agree with the statement that the Bankhas a comparative advantage over bilaterals toachieve development impact in CBD/CDD in-terventions; 13 percent disagreed with the state-ment, and 27 percent neither agreed nordisagreed.

9 9

A N N E X L

Table L.1: Response Rates for Bank Staff Survey (percent)

Agree or Neither Disagree strongly agree nor or strongly

agree disagree disagree

The Bank sufficiently addresses the policy issues needed to support successful CBD/CDD interventions. 38.82 23.03 26.97

Bank CBD/CDD projects generally have been sufficiently targeted at the poorest. 51.97 19.74 16.45

Bank CBD/CDD projects have addressed sufficiently, and had been consistent with, broader

institutional and fiscal decentralization. 32.89 22.37 30.26

Sufficient resources (relative to non-CBD/CDD projects) are made available by country directors to

effectively appraise and implement CBD/CDD projects. 25.00 21.71 36.18

Sufficient resources are made available by country directors to effectively address safeguard

issues related to CBD/CDD projects. 21.71 26.97 30.92

Implementation costs per dollar lent for CBD/CDD projects are higher than other traditional

types of projects. 38.82 12.50 25.66

CBD/CDD approaches across the whole project cycle, from identification to completion, take more

Bank staff resources per dollar of lending than other traditional types of investment projects. 46.71 12.50 21.71

CBD/CDD projects are more risky than traditional non-CBD/CDD projects. 24.34 21.71 44.74

Task managers of CBD/CDD projects can monitor fiscal accountability as satisfactorily as more

traditional non-CBD/CDD projects. 35.53 27.63 23.03

In Bank-funded CBD/CDD projects, community maintenance contributions generally are sufficient

to sustain infrastructure investments. 17.76 27.63 36.84

The Bank generally continues funding CBD/CDD projects for long enough (e.g., if necessary, into a

second or third phase) to reach a satisfactory level of sustainability of community processes. 24.34 23.68 26.97

Sectoral technical standards (e.g., irrigation or curriculum design standards) in CBD/CDD operations

have not been excessively compromised by CBD/CDD approach compared with traditional operations. 43.42 23.03 11.84

Bilateral donor interventions generally achieve a better CBD/CDD outcome than the Bank

interventions. 10.53 25.66 38.16

NGO-supported interventions generally achieve a better CBD/CDD outcome than the Bank

interventions. 26.32 24.34 30.92

Inadequate donor coordination in a co-financed project is likely to have a greater negative impact

on outcomes in a CBD/CDD project than in a more traditional non-CBD/CDD project. 38.82 21.71 17.76

The Bank has a comparative advantage over bilaterals to achieve development impact in

CBD/CDD interventions. 44.08 26.97 12.50

Your Region has sufficient CBD/CDD-related skills to achieve satisfactory CBD/CDD

project performance. 48.68 15.79 18.42

CBD/CDD projects that use program specific committees to make investment allocation decisions,

while perhaps facilitating project implementation in the short term, may often fail in the long

term to significantly strengthen decentralized local institutions because they operate—parallel

to, rather than integrated with, local government. 44.74 20.39 21.05Note: Based on 152 responses.


1 0 0


Satisfied Somewhat Notor better satisfied satisfied

Understanding by the management in your region of the objectives and design of CBD/CDD projects. 40.13 28.95 15.13

Guidance by the management in your Region over the project cycle on CBD/CDD projects. 26.97 35.53 16.45

Guidance for CBD/CDD work provided by the CDD Social Anchor team in ESSD. 21.05 29.61 19.74

Support for CBD/CDD work from the relevant Thematic Groups. 26.97 29.61 17.11

Support by the current Bank matrix-management organizational structure for the needs of

CBD/CDD projects. 13.82 29.61 27.63

Coordination within the Bank across sectors in CBD/CDD interventions. 9.21 34.21 36.18

Coordination within government in borrowing countries for CBD/CDD interventions. 19.08 30.26 30.92

Amount of training on CBD/CDD over the last two years (including clinics, brown-bag lunches,

etc. as well as longer training). 19.08 26.97 22.37

Quality of training on CBD/CDD over the last two years (including clinics, brown-bag lunches,

etc., as well as longer training). 17.11 23.68 9.87

The impact of Bank decentralization to field offices on the efficacy of Bank support for

CBD/CDD projects. 39.47 21.71 12.50

Emphasis placed by the Bank on donor coordination in CBD/CDD projects. 28.29 36.84 14.47

Relevance of current Bank safeguards for CBD/CDD projects. 16.45 35.53 25.66Note: Based on 152 responses.


More Less than 80% 60–80% 40–60% 20–40% than 20%

In what percentage has the Bank committed to continuing support to

community groups to the point of satisfactory sustainability of

those group processes? 9.21 18.42 19.74 17.11 8.55

In what percentage have community groups and associated

community processes reached to a level you would rate

as “likely” for sustainability? 2.63 25.00 27.63 16.45 9.21

In what percentage has another collaborating development

agency (e.g., other donor, NGO, etc.) significantly enhanced

the quality of the CBD/CDD elements of design in the project? 6.58 18.42 21.05 11.84 10.53

In what percentage has the Bank scaled up the project? 10.53 24.34 19.74 5.92 9.87Note: Based on 152 responses.


1-5 6-10 11-15 16-20 20+

In your experience, what would be the average number of years needed

for project support of community groups initially formed under the

project to reach a level of sustainability of community processes

requiring very limited outside support (such as simply a

supporting/maintenance visit once a year). 23.68 51.97 11.84 0.00 0.66Note: Based on 152 responses.


1 0 1


1 0 2

Tab

le L

.5: M

ult

ivar

iate

An

alys

is f

or

Ban

k S

taff

Su

rvey

Sufficient resources are made available bycountry directors to effectively address safe-guard issues related to CBD/CDD projects

Implementation costs per dollar lent forCBD/CDD projects are higher than othertraditional types of projects

NGO-supported interventions generally achievea better CBD/CDD outcome than the Bankinterventions

Inadequate donor coordination in a co-financedproject is likely to have a greater impact on out-comes in a CBD/CDD than non-CBD/CDD project

Support by the current Bank matrix-manage-ment organizational structure for the needs ofCBD/CDD projects

Coordination within the Bank across sectors inCBD/CDD interventions

Coordination within government in borrowingcountries for CBD/CDD interventions

The impact of Bank decentralization to fieldoffices on the efficacy of Bank support forCBD/CDD projects

Coef

.Co

ef.

Coef

.Co

ef.

Coef

.Co

ef.

Coef

.Co

ef.

Dum

my

for A

frica

Reg

ion

–0.1

3–0

.48

–0.7

4 *

0.07

–0.0

5–0

.54

0.33

–0.0

5

Dum

my

for E

ast A

sia

& P

acifi

c Re

gion

–0.2

0–0

.63

0.00

0.12

–0.2

7–0

.86

*0.

04–0

.12

Dum

my

for E

urop

e &

Cen

tral A

sia

Regi

on0.

98 *

–0.9

6 *

*–0

.64

0.39

–0.1

8–0

.66

0.58

–0.7

1

Dum

my

for L

atin

Am

eric

a &

Car

ibbe

an R

egio

n0.

21–0

.50

–0.8

0 *

0.51

0.18

–1.0

4 *

*0.

540.

09

Dum

my

for S

outh

Asi

a Re

gion

0.26

0.22

0.09

0.62

–0.1

5–0

.39

0.54

0.19

Dum

my

for C

DD s

peci

alis

t0.

31–0

.20

–0.4

80.

83–0

.02

–0.6

70.

170.

34

Dum

my

for e

cono

mis

t–0

.60

*–0

.13

0.27

–0.4

30.

420.

060.

60 *

0.06

Dum

my

for s

ecto

ral t

echn

ical

or t

hem

atic

spe

cial

ist

–0.2

7–0

.15

–0.1

6–0

.03

0.33

0.05

0.35

0.23

Dum

my

for s

ocio

logi

st/s

ocia

l ant

hrop

olog

ist

0.03

–1.0

4 *

*–0

.20

–0.2

30.

130.

450.

580.

16

Dum

my

for m

anag

er (s

ecto

r lev

el o

r abo

ve)

–0.3

91.

14

***

–0.4

00.

410.

760.

30–0

.27

0.61

Num

ber o

f CDD

pro

ject

s w

ith >

2 w

eeks

–0

.02

–0.0

8 *

*–0

.04

–0.1

1 **

*–0

.02

0.04

0.00

–0.0

2

of p

artic

ipat

ion

Ratio

of n

o. o

f CDD

pro

ject

s ta

sk m

anag

ed a

nd n

o. o

f 0.

110.

26–0

.29

–0.2

1–0

.09

0.44

0.39

0.12

non-

CDD

proj

ects

task

man

aged

in p

ast 5

yea

rs

Days

of t

rain

ing

in C

DD-re

late

d to

pics

in la

st 2

yea

rs0.

00–0

.01

–0.0

4 *

*0.

010.

05 *

*0.

000.

020.

01

Obse

rvat

ions

121.

0011

7.00

124.

0011

9.00

108.

0012

1.00

122.

0011

2.00

Pseu

do R

-squ

ared

0.05

0.07

0.09

0.05

0.05

0.04

0.04

0.03

Chi2

16.4

226

.17

31.8

017

.69

12.2

410

.67

12.5

08.

52

Not

e: *

Sig

nific

ant a

t 10%

; **

sign

ifica

nt a

t 5%

; ***

sign

ifica

nt a

t 1%

.

1 0 3

Different stakeholders can have different (evenopposing) perspectives on the various aspectsand outcomes of a project. Hence, it is impor-tant to collect information from different stake-holders to get a complete picture of alternativeperspectives. Both qualitative and quantitativetools were used to collect data for this evaluation.In each of the four project areas, a total of 30communities were selected, and roughly 40households were interviewed per community.Two focus groups and two key informant inter-views were also conducted in nearly all com-munities. This annex is organized in three mainsections. The first one presents the methodol-ogy used for the selection of communities andhouseholds within communities for each of thefour project areas. The second presents detailsof the survey instruments and timing of the field-work. The final section describes the method-ology adopted for household data analysis.

Four CBD/CDD projects (three of which wereCDD) were selected for extensive fieldwork at thecommunity level: the Rural Poverty AlleviationProject (RPAP) in Rio Grande do Norte, Brazil(henceforth Brazil); the Borgou Pilot Project inBenin (henceforth Benin); the Uttar PradeshSodic Land Reclamation Project in India (hence-forth Uttar Pradesh), and the Madhya PradeshForestry Project in India (henceforth MadhyaPradesh). The two projects in India were single-sector interventions, while the projects in Beninand Brazil were multisectoral. Fieldwork in Brazilcovered largely water supply subprojects; whilein Benin it covered largely primary schools,health facilities, and storage houses.

The community-level data collection was un-dertaken primarily to assess two issues: first,the association between Bank-supportedCBD/CDD interventions and social capital en-

hancement and empowerment of communities(Annex N); second, the sustainability of projectinvestments (Annex P). The household data wasalso used to assess the extent to which CBD/CDDproject investments met the priority needs ofbeneficiary communities (figure 3.5, Chapter 3).

Sample SelectionOED’s fieldwork adopted a non-experimentalevaluation design that compared randomly se-lected CBD/CDD communities with compara-tor communities in the four projects using acomparison group methodology. In all four casesthe comparator group exhibited similar problemsor issues as the project group and had similar so-cioeconomic and cultural characteristics. Theselected comparators varied according to proj-ect and country context. In two project areas(Benin and Brazil), the comparison group com-munities had benefited from similar subprojectsas the CBD/CDD communities but through anon-participatory approach adopted either by thegovernment or a religious organization. Thesetwo cases allowed the evaluation to assesswhether a program that involves communities ismore effective than one that does not. In anotherproject area (Madhya Pradesh) comparator com-munities benefited from a similar activity(forestry) carried out through a participatoryapproach, but supported by the government inIndia. Here the evaluation assessed whetherthere is any difference in outcomes because ofthe participatory approach pursued by the Bankas compared with the participatory approachpursued by the government. Finally, in the fourthproject area (Uttar Pradesh), comparison com-munities did not benefit from a similar activityas project communities. Here the evaluation as-sessed the overall outcomes of the Bank

ANNEX M: METHODOLOGY FOR COMMUNITY-LEVEL DATA COLLECTION

AND ANALYSIS


1 0 4

CBD/CDD project, and not only that of its par-ticipatory approach.

Community SelectionThe process of community selection varied ac-cording to the project context and is detailedbelow. In all four project areas, a larger samplethan actually required was selected to allow forreplacement of any community that was foundnot to satisfy the selection criteria after fieldverification.

BeninIn the Borgou region of Benin, the Bank fi-nanced three CBD/CDD projects in the pastdecade: the Borgou Regional Pilot Project(PAMR); the Social Fund Project (AgeFIB); andthe Food Security Project (PILSA). While thefocus of the evaluation was on PAMR (hencethe focus on the Borgou region), as project doc-uments revealed it to be a CDD project, com-munities that benefited from AgeFIB and PILSAwere also surveyed to allow for comparisonamong the three projects.1 The selection of thePAMR, AgeFIB, and PILSA communities requiredfirst of all identifying, based on the subprojectrecords of the three projects, communitieswithin the Borgou region that had benefitedfrom only one of these three Bank CBD/CDD in-terventions.2 The selection was further restrictedto communities that (a) were located in ruralareas, (b) benefited from no more than threesubprojects, (c) benefited from specific type ofinvestments. For PAMR communities, the se-lection was restricted to those that benefited

from any of the following types of subprojects:construction of a school or other infrastructure(such as well, storage house, hangar, and thelike), functional literacy, training in beekeep-ing, hygiene and nutrition, and provision of es-sential drugs. The selection of the fivecomparator communities was undertaken bythe local expert contracted for the community-level fieldwork, based on the following threecriteria: communities that: (a) did not benefitfrom any of the Bank CBD/CDD projects; (b)benefited from similar subprojects as the proj-ect communities; and (c) were located in ruralareas. Comparator communities benefited fromnon-participatory projects funded by the Stateand various religious organizations. These com-munities received investments for the con-struction of schools (two communities), wells(two communities), and a health center (onecommunity).

Inaccuracies in subproject records requiredchanges to the original sample of communitiesas well as dropping some communities fromthe analysis. After verification in the field of theproject status of selected communities, a fewcommunities were replaced because they had ei-ther benefited from more than one BankCBD/CDD project, and/or had benefited frommore than three subprojects. These communi-ties were replaced with randomly selected com-munities of the same project status. A fewcomparator communities were also replaced asthey were found to have benefited from one ofthe three Bank CBD/CDD projects. Further, notall subprojects could be covered in the desired

Table M.1: Coverage of Fieldwork in Benin

Participatory approaches Top-down approaches

PAMR AgeFIB PILSA Other donor State-funded Religious org.

Fieldwork coverage

Communities 17 7 2 1 3 2

Households 736 304 85 45 118 88Analysis coverage

Communities 13 4 2 0 3 2

Households 566 177 85 0 118 88

proportion since it was not uncommon to arrivein a community that had, according to records,received funding for, say, a school, only to findthat what was actually funded was a storehousethat had not been accurately registered in therecords. The inaccuracies in subproject-levelrecords were detected by visiting sampled com-munities and asking information of local leaders.However, in some cases, community leaders’recollections also proved inaccurate, as focusgroup sessions later revealed the presence ofother Bank projects that were unknown to localleaders. Such communities, though surveyed,were dropped from the analysis. Further, onecomparator community was also dropped fromthe analysis because it was the only one to haveparticipated in a non-Bank CBD/CDD project.Table M.1 presents the final number of house-hold surveys conducted and the number actu-ally used for the analysis. Two key informantinterviews and one mixed (male and female)focus group were conducted in each of the com-munities.

BrazilThe RPAP (and the follow-on RPRP) adoptedthree distinct community-driven implementa-tion modalities; in increasing order of decen-tralization, these were:

• PAC: The Community Association (CA) submitsa subproject proposal to the State TechnicalUnit. Using a statewide vetting process, theState Technical Unit chooses the soundest pro-posals, with some reference to the evenness ofdistribution among the various municipalities.Once approved, project funds flow directly toa bank account set up locally by the Commu-nity Association.

• FUMAC: A municipal council (called theFUMAC Council), with representatives of civilsociety and the government, is set up by theproject at the municipal level. The proposalsprepared by the CAs are first reviewed andranked by the FUMAC Council, and only thensubmitted to the State Technical Unit. Thecouncil chooses among subproject proposalswith reference to an indicative budget com-municated by the State Technical Unit. Vetting

by the State Technical Unit is more of a formalitycompared to PAC; providing the subprojectsmeet the required technical specifications, theState Technical Unit signs off on the proposalmade by the FUMAC Council.

• FUMAC-P: The procedures are the same as forFUMAC, except that the FUMAC-P Council isgiven an annual budget, which it administersitself. The council signs agreements with theCAs, transfers project funds to them, keepstrack of receipts, and monitors physicalprogress. It is accountable to state govern-ment auditing procedures. If one CA fails toprovide the necessary receipts, disbursementsto all other CAs in that municipality may befrozen, paralyzing the project process.

The selection of project communities wasbased on the project’s monitoring and infor-mation system (MIS). To keep logistic and trans-port costs within the budget, fieldwork wasrestricted to the two regions (out of four) thathad the highest number of communities whereonly one subproject had been financed by theRPAP—Agreste and Oeste Potiguar. The crite-rion of one subproject per community was cho-sen for two main reasons. First, 79 percent of thecommunities that benefited from the RPAP in RioGrande do Norte received only one subproject.Second, we wanted to avoid comparing com-munities that had received only one subprojectwith those that had benefited from more. The se-lection of project communities was further re-stricted to those that: (a) were located in the ruralareas, and (b) had benefited from one the fol-lowing investments: water supply, electricity, ir-rigation, or small bridges.3 The selection ofproject municipalities was limited to those thathad at least two communities that met the abovecriteria. A stratified random sample of 11 proj-ect municipalities was selected, with each of thethree implementation modalities being repre-sented in proportion to the number of munici-palities under each modality. A total of 24communities were selected within these mu-nicipalities using a table of random numbers.

The selection of comparator communities re-quired, first of all, identifying municipalities thatwere targeted by the RPAP and the ongoing RPRP,

1 0 5

A N N E X M


1 0 6

but that had not yet benefited from either proj-ect.4 Drawing on the MIS data for the RPAP andongoing RPRP (updated to October 15, 2003), sixcomparator municipalities were identified—three in each of the two regions. The selectionof suitable comparator communities was un-dertaken by the local expert contracted for com-munity-level fieldwork. Comparator communitieshad to satisfy four main criteria: (a) they had tohave benefited from a similar service as projectcommunities around the same time as thesedid; (b) they could not have benefited from aBank-financed CBD/CDD intervention; (c) theyhad to be located in rural areas; and (d) they hadto have more than 40 households at the time ofthe survey.

Inaccuracies in the project’s MIS requiredchanges to the original sample of communities,as well as dropping some communities from theanalysis. Two project communities were dropped;qualitative data revealed that one had recentlyapplied for funds under the ongoing RPRP, whilethe other was the only one to have benefitedfrom a rural electrification investment. The ma-jority of project communities used for the analy-sis benefited from water supply investments,while three benefited from irrigation invest-ments and two from small bridges. Three of thesix comparator communities surveyed were alsodropped. The qualitative data revealed that twoof them had recently applied for funds under theongoing RPRP, while one of them was the onlyone in the sample to have benefited from agovernment water supply program that had a par-

ticipatory component. All comparator commu-nities included in the analysis benefited from agovernment-funded water pipeline constructedat the time when the RPAP was being imple-mented. None of them benefited from eitherthe RPAP or the ongoing RPRP, while one bene-fited in 1994 from another Bank CBD/CDDproject—the reformulated Northeast Rural De-velopment Program (NRDP; 1993–96). This com-munity was, however, retained for the analysis,as it did not differ from the other two compara-tor communities, and would therefore not biasthe results.5

Some of the selected communities were foundto have fewer than 40 households. In these cases,a census was taken and, where possible, adjacentCBD/CDD communities that received only onesimilar subproject were selected to make up forthe missing number of respondents. Some mu-nicipalities that figured as PAC in the MIS hadbeen “upgraded” to FUMAC under the on-goingRPRP. These municipalities continued to be con-sidered as PAC for the purpose of this evaluationonly if no subproject had yet been financedthrough the FUMAC implementation modality.Table M.2 presents the number of householdsurveys conducted and the number used in theanalysis.

Two key informant interviews were conductedin all except five communities, where only thecommunity leader was interviewed. With a fewexceptions, two focus group interviews wereconducted in each community. In six commu-nities only one focus group interview was carriedout, and in two communities no focus groupsession was held.

Madhya PradeshTwenty project villages were randomly selectedfrom two distinct types of forest zones under theJoint Forest Management (JFM) project in Mad-hya Pradesh—10 from each zone. One of thezones is characterized by dense forests (ANR), theother by degraded forests (VRDP). In order tokeep logistic and transport costs within thebudget, fieldwork was restricted to two districts:Betul and Bilaspur, one from the western part ofthe state and the other from the eastern part(now in the state of Chattisgarh). Bilaspur was

Table M.2: Coverage of Fieldwork in Brazil

Modality FUMAC FUMAC-P PAC Non-Bank

Fieldwork coverage

Municipalities 5 2 3 3

Communities 15 8 4 6

Households 514 240 118 225

Analysis coverage

Municipalities 5 2 3 2


Households 485 211 118 117

1 0 7

chosen because it had been surveyed in March2000 by the World Bank’s Environmentally andSocially Sustainable Development (ESSD) Net-work. Betul was selected from a list of four dis-tricts prepared by the implementing agencybased on security concerns because it had agood number of both ANR and VRDP villages.Random selection of villages within each forestzone and district was done using a table of ran-dom numbers.

The selection of comparator villages required,first, identifying areas in the districts of Betul andBilaspur under the government-supported JFM.This strategy was very similar to that supportedby the Bank. Under JFM, villagers cooperate toprotect forests in exchange for a share in theusufruct and the final harvest. The selection ofcomparator villages from the government JFMarea was undertaken by the project’s imple-menting agency based on the following five cri-teria: villages that (a) were located in rural areas;(b) did not benefit from any Bank CBD/CDD in-tervention; (c) were within 5 kilometers of thesame forest block as project villages; (d) hadsimilar poverty levels as project villages; and (e)had between 40 and 80 households.

Inaccuracies in subproject records requiredchanges to the original sample of communities,as well as dropping some communities from theanalysis. One project village selected in the dis-trict of Betul was not found on the map of Betulby the Forest Department. This village had to bereplaced by one in the same area with similar for-est cover. Another village in the same district,which the records showed as government-JFM,was actually a Bank-JFM village, bringing thetotal number of project villages to 21. Further,qualitative data revealed that two of surveyedcomparator villages did not have government-funded JFM, and were hence dropped from thefinal analysis. The survey data revealed that vil-lagers did not perceive any difference betweena Village Protection Committee, which is set upin degraded forest zone, and a Forest ProtectionCommittee, which is set up in dense forest zone.Hence, data analysis was not differentiated bytype of forest zone. Table M.3 presents the num-ber of household surveys conducted and thenumber used in the analysis.

Two focus group interviews and two key in-formant interviews were conducted in each vil-lage, with the exception of one where no focusgroup interview was conducted.

Uttar PradeshIn March-June 2000, the World Bank’s ESSD Net-work surveyed 19 villages in the district ofRaibareli, which had benefited from the Bank’sSodic Land Reclamation Project. In order toallow for comparison over time, fieldwork for theOED study was conducted in the same area. TheESSD survey covered villages that were treatedduring four of the five annual project phases(1993–98); none of the 19 villages was treatedduring phase II (1994–95). In order to cover allproject phases, 4 of the 19 villages were replacedwith randomly selected villages within Raibarelidistrict that benefited from the project duringphase II. The random selection of these fourvillages was based on the project’s database.

The selection of comparator villages requiredidentifying areas in Raibareli district that facedproblems of sodicity of land similar to thosefaced by project villages prior to the Bank’s in-tervention, but which never benefited from anysodic land reclamation activity. The 11 com-parator villages were selected by the imple-menting agency of the Bank’s Sodic LandReclamation Project (UPBSN) from those thatwere in the pipeline to be treated if additionalfunds to address sodicity became available.UPBSN selected comparator villages based on thefollowing three criteria: villages that (a) werelocated in rural areas, (b) were located in the

A N N E X M

Table M.3: Coverage of Fieldworkin Madhya Pradesh

Type of GoMP Non-forest ANR VRDP JFM JFM

Fieldwork coverage


Households 414 393 261 79

Analysis coverage

Communities 11 10 7 —

Households 414 393 261 —

same three blocks as project villages; and (c)had similar population size as the project vil-lages. One comparator village was dropped fromthe analysis, as the comparison of baseline in-formation collected for this study revealed sig-nificant difference between this and the othervillages. Table M.4 presents the number of house-hold surveys conducted and the number used inthe analysis..

Two focus group interviews and two key in-formant interviews were conducted in each vil-lage, with the exception of one where only oneof the two key informant interviews was con-ducted.

Household SelectionWherever possible, 40 households were selectedfrom each community. Two slightly different ap-proaches were adopted for household selec-tion. In communities where the team had theinformation on the total number of households,these were divided by the number of interviewsto be conducted (40) to get an interval of R.The households were then arranged in a con-centric manner on the drawing board and a ran-dom starting household was selected. Every Rthhousehold was selected until the required num-ber of interviews was complete. In rural dis-persed communities where there was low initialknowledge of the number of households, thecommunity was divided in 4 zones, and 10 house-holds were covered in each zone. A similar strat-egy as above was adopted for each zone, butwith a rough estimate from the local leader onnumber of households in each zone. In com-munities with 40 or fewer household, all house-holds were surveyed.

Survey Instruments and Timingof FieldworkInformation was collected at the communitylevel using three instruments. These were allpilot tested in the field in each of the four proj-ect areas before being launched.

• First, a pre-coded household questionnaire,which was applied to one adult (25 years orolder) from each randomly selected house-hold who had resided in the community for the

past eight years. The household survey en-quired about respondents’: (a) demographiccharacteristics—age, education, gender, oc-cupation, marital status, etc; (b) householdcharacteristics, including variables capturingeconomic status at the time of the survey andbefore subproject implementation; (c) aware-ness of community problems and participa-tion in community-level project organizations;(d) perception of sustainability of project in-vestments; and (e) perceptions of the levels ofand the changes in social capital and empow-erment.6

• Second, semi-structured focus group inter-views held with two groups in each community(one all-female and one all-male) of 10–15 self-selected participants. Focus group sessionsattempted to explore, among other things,the following issues: (a) the process of sub-project selection, implementation, and oper-ation; (b) communities’ access to information;(c) the leadership structure within communi-ties; (d) the levels and changes in empower-ment; (e) the priority needs of the communityat the time of the survey and before subpro-ject implementation.

• Third, structured key informant interviewsheld with a community leader and a memberof the community organization set up by theproject. Key informant interviews used a struc-tured, open-ended questionnaire. The com-munity leader interview consisted of questionsabout community facilities, ethnic make-up,and the like. The interview with a member ofthe community organization set up by the proj-ect addressed issues of community trust, co-hesion, and solidarity, as well as providing


1 0 8

Project Comparator

Fieldwork coverage

Communities 19 11

Households 757 440

Analysis coverage

Communities 19 10

Households 757 400

Table M.4: Coverage of Fieldworkin Uttar Pradesh

information on the functioning of the com-munity organization set up by project.

In Benin, fieldwork was conducted in Octo-ber and November 2003 by a team headed byRoch Mongbo from the University of Abomey-Calavi. The fieldwork in Brazil was conducted be-tween November 2003 and January 2004 by ateam headed by Alberto Costa from the Univer-sity for the Development of the Itajaí River Val-ley. In India, fieldwork was conducted inDecember 2003 and January 2004 by the Centerfor Development Economics, Delhi School ofEconomics. In all four project areas, an OEDteam member supervised fieldwork activity to en-sure quality.

Methodology for Household Data Analysis

Comparison of ex-ante characteristics ofproject and comparator communities. Re-spondents’ demographic and socioeconomicinformation before the Bank intervention wereaggregated at the community level to provide ageneral profile of the communities covered byfieldwork. A Student t-test was performed onthese aggregated variables to check whetherthe project and the comparator groups had thesame mean. While the difficulty of getting per-fect matches between the project and the com-parator group must be acknowledged, anattempt was made to get as close a match as pos-sible. As tables M.5–M.8 show, only minimal dif-ferences were found between the project andcomparator groups in the four project areas. InBenin the project group reported a significantlygreater number of children below the age offour than the comparator group. In Brazil, agreater number of women were interviewed inthe comparator group than in the project group.In Madhya Pradesh, slightly more respondentsin the comparator group reported knowingthe chairman of the Farmer’s Club, while theproject group reported greater ability to or-ganize self-help groups and raise resources fromwithin the community. In Uttar Pradesh, re-spondents in the project group were slightlyolder than those in the comparator group, whilemore respondents in the latter than the former

reported knowing local leaders and local electedofficials.

Bivariate analysis was used to compare therespondents’ perceptions of levels of andchanges in social capital and empowerment be-tween the project and the comparator groups.A test of proportion was performed for binaryvariables and the Kruskall-Wallis nonparametrictest of differences for categorical variables (tablesN.3–N.6).

Multivariate analysis. In order to control fordifferences in geographic, demographic, and so-cioeconomic factors between the project andthe comparator groups, multivariate analysis wasperformed on the variables that capture re-spondents’ perceptions of the changes in socialcapital and empowerment (see tables M.9–M.12for the list of dependent and independent vari-ables used in each country). An Ordered Probitmodel was chosen because the dependent vari-ables are ordinal ranging from least to most,with most capturing greater outcome. The esti-mation was performed using population weightsand adjusting for cluster effects.7

Two specifications of the same model wereused; with and without interactive terms. The dis-cussion of the overall association between theproject and the dependent variables is based onthe specification without interactions. The spec-ification with interactions was used in order toexplore the association between the dependentvariables and the project for women, the poor,and members of project organizations.8 The re-sults of the specification with interactive variablesare presented in full (tables N.9–N.16), while asummary of the results of the project dummiesfor the specification without interactions is pre-sented in tables N.7–N.8.

As already mentioned, all dependent vari-ables represent changes over time. It is how-ever important to note that there are two typesof change variables: (a) changes as perceivedand directly reported by respondents, and (b)changes derived from respondents’ assessmentof the situation in two points in time—before andafter subproject implementation. All dependentvariables that capture changes in social capital

1 0 9

A N N E X M


1 1 0

and empowerment are of the first type, with thesole exception of the variable that capture re-spondents’ mobilization skills in the Brazil proj-ect, which is of the second type.

The independent variables include commu-nity characteristics (such as dummy for regions,population of the community, and the like),household characteristics (such as householdsize and the index of economic status), and re-spondent characteristics (such as age, level ofeducation, and so forth). The model used forBenin and Brazil also controls for the type of sub-project financed. Household and respondentcharacteristics were created drawing on demo-graphic and socioeconomic information beforethe Bank intervention as reported by respon-dents. The model includes two variables repre-senting the respondent’s economic status: (a)the index of economic status, and (b) a dummyvariable for poor. The reason for including both(a) and (b) is that these are defined differently.While the index for economic status is an ab-solute figure calculated across all respondents,

the dummy variable for poor captures the bot-tom quartile of economic status within eachcommunity. Therefore, while (a) is a measure ofeconomic status across the entire sample, (b)represents the relatively poor households withineach community.

As already mentioned above (pp. 104, 105),the studies conducted in Benin and Brazil in-cluded three types of communities that bene-fited from Bank-supported CBD/CDD projects.In Benin, only PAMR communities were con-sidered as project communities; AgeFIB andPILSA communities were included to elicit dif-ferences in performance between the threeCBD/CDD projects. In Brazil, the RPAP includedthree implementation modalities—PAC, FUMAC,and FUMAC-P. These could not be combined inone project group, as they were differently as-sociated with some of the dependent variables.9

For clarity of exposition, the bivariate analysis re-ports only the response rate of FUMAC com-munities, which account for 60 percent ofrespondents in project communities.

Table M.5: Benin: Comparison of Ex-Ante Characteristics of Project and Comparator Communities

CBD/CDD Comparator

Household size (member above the age of 4) 8.75 7.39

Number of children above the age of 4 3.00 2.50

Number of children below the age of 4 1.93 1.07 ***

Schooling of the respondent 0.26 0.33

Age of the respondent 42.71 45.68

Dummy for female 0.31 0.31

Knew community leaders 0.91 0.94

Knew religious leaders 0.86 0.89

Knew the mayor 0.47 0.52

Participation in traditional events 2.86 2.73

Participation in political events 2.33 2.40

Ability to raise resources from within the community 0.58 0.67

Ability to raise funds outside the community 0.38 0.33

Ability to speak freely with community leaders 0.64 0.59

Ability to express community needs to local government officials 0.50 0.55

Blue collar skills 0.46 0.51Note: Test of significance based on Student t-test. * Significant at 10%; ** significant at 5%; ***significant at 1%.

1 1 1

A N N E X M

Table M.6: Brazil: Comparison of Ex-Ante Characteristics of Project and Comparator Communities

CBD/CDD Comparator

Municipal Human Development Index 0.62 0.62

Score for community 0.23 0.27

Population of community 59.56 100.00

Household size 4.42 4.83

Number of children 1.56 1.78

Medium consumer durables 1.74 1.80

Large consumer durables 0.23 0.32

Large animals 4.27 1.95

Small animals 7.91 4.14


Dummy for female 0.43 0.71 ***


Dummy for agricultural laborer 0.56 0.38

Participation in political events 1.91 2.00


Ability to raise resources from within the community 0.39 0.36


Ability to speak freely with community leaders 0.50 0.35

Ability to express community needs to local government officials 0.58 0.60Note: Test of significance based on Student t-test. * Significant at 10%; ** significant at 5%; ***significant at 1%.


1 1 2

Table M.7: Madhya Pradesh: Comparison of Ex-Ante Characteristics of Project and ComparatorCommunities

CBD/CDD Comparator





Land owned 1.58 1.36

Small consumer durables 0.09 0.07








Knew village leader 0.59 0.47

Knew Sarpanch 0.80 0.80

Knew Farmers’ Club chairperson 0.02 0.06 ***

Knew local elected officials 0.15 0.16

Knew forest staff 0.63 0.67


Participation in non-traditional events 2.27 2.28

Ability to organize self-help groups/raise resources from within the community 0.21 0.12 ***


Ability to express community needs to local government officials 0.37 0.39Note: Test of significance based on Student t-test. * Significant at 10%; ** significant at 5%; ***significant at 1%.

1 1 3

A N N E X M

Table M.8: Uttar Pradesh: Comparison of Ex-Ante Characteristics of Project and ComparatorCommunities

CBD/CDD Comparator





Amount of land owned 81.72 68.72

Small consumer durables 0.26 0.43





Schooling of respondent 0.46 0.45


Age of the respondent 44.61 42.42 **

Knew local leaders 0.79 0.90 **

Knew Farmers’ Club chairperson 0.01 0.01

Knew local elected officials 0.41 0.60 ***


Participation in non-traditional events 2.24 2.10


Ability to express community needs to local government officials 0.45 0.37

Blue collar skills 0.36 0.34Note: Test of significance based on Student t-test. * Significant at 10%; ** significant at 5%; ***significant at 1%.


1 1 4

Table M.9: Definition of Variables: Benin

Dependent variable Definition

Change in access to information Changes in access to information regarding development activities in the community (more=3, same=2,

less=1)

Change in mobilization skills Composite variable equal to the sum of four dummy variables (1=more, 0=everything else) that capture

change in the respondent’s ability to (a) mobilize community efforts and resources; (b) raise funds out-

side the community; (c) speak freely with community leaders; (d) express the needs of the community

to local government officials.

Change in ability to reach agreement Change in the community’s ability to reach an agreement (more=3, same=2, less=1)

Change in community leaders’ Change in the extent to which community leaders and local government officials listen and respond to

responsiveness to community needs community needs (Listen and respond more=4, listen more=3, same=2, listen less=1).

Change in trust Composite variable equal to the sum of four dummy variables (1=more, 0=everything else) capturing change

in trust in: (a) community members, (b) community organizations, (c) local leaders, (d) government

officials.

Change in associational life Composite variable equal to the sum of two dummy variables (1=more, 0=everything else) capturing change

in: (a) villagers’ participation in groups and associations, (b) cooperation between groups and

individuals.

Change in participation in Change in the respondent’s participation in community’s traditional events (more=3, same=2, less=1)

traditional events

Change in participation in Change in the respondent’s participation in community’s political events (more=3, same=2, less=1)

political events

Change in circle of friends Change in the respondent’s circle of friends (improved=3, same=2, deteriorated=1)

Independent variable

PAMR Equals one if PAMR community, zero otherwise

AgeFIB Equals one if AgeFIB community, zero otherwise

PILSA Equals one if PILSA community, zero otherwise

Female in PAMR Equals one if respondent is a female from PAMR community, zero otherwise

School construction subproject Equals 1 if school construction subproject, zero otherwise

Training subproject Equals 1 if training subproject, zero otherwise

Difficult access to community Equals one if community is geographically difficult to access, zero otherwise

Household size Number of people living under the same roof

Number of children Number of children below the age of 16

Dummy for female Equals 1 if the respondent is a female and zero if male

Schooling Equals one if the respondent has attended school, zero otherwise

Age Age of the respondent

Age squared Age squared

Number of leaders known Number of leaders the respondent knew prior to the Bank intervention

Participation in traditional events Frequency of participation in traditional events prior to Bank intervention

Participation in political events Frequency of participation in political events prior to Bank intervention

Blue collar skills of the respondent Equals 1 if the respondent was able to do blue collar activities (such as masonry, carpentry, stitching,

etc.) prior to Bank intervention

Mobilization skills of the respondent Number of skills the respondent reported to have prior to Bank intervention. Composite variable equal

to the sum of four dummy variables (1=able, 0=everything else) that captures respondent’s ability to (a)

mobilize community efforts and resources; (b) raise funds outside the community; (c) speak freely with

community leaders; (d) express the needs of the community to local government officials.

1 1 5

A N N E X M

Table M.10: Definition of Variables: Brazil

Dependent variables Definition

Change in access to information Changes in access to information regarding development activities in issues of interest to the commu-

nity (more=3, same=2, less=1)

Change in mobilization skills Composite variable equal to the sum of four dummy variables (1=more, 0=everything else) that capture

changes in the respondent’s ability to (a) raise resources from within the community; (b) raise funds out-

side the community; (c) speak freely with community leaders; (d) express the needs of the community

to local government officials. The dummies for change were derived from respondent’s assessment of

their skills in two points in time—before and after subproject implementation.


Change in leaders’ responsiveness Change in community leaders’ responsiveness to communities demands (more=3, same=2, less=1)

Change in trust Composite variable equal to the sum of four dummy variables (1=more, 0=everything else) capturing change

in trust in: (a) community members, (b) community associations, (c) municipal government officials, and

(d) state government officials.


in: (a) people’s participation in groups, (b) cooperation between groups and individuals.


traditional events

Change in participation in Change in the respondent’s participation in community’s political events (more=3, same=2, less=1)

political events


Independent variables Definition

PAC Equals 1 if RPAP implemented through PAC modality, zero otherwise

FUMAC Equals 1 if RPAP implemented through FUMAC modality, zero otherwise

FUMACP Equals 1 if RPAP implemented through FUMAC-P modality, zero otherwise

Poor in PAC Equals 1 if respondent is poor and in a PAC community, zero otherwise.

Poor in FUMAC Equals 1 if respondent is poor and in a FUMAC community, zero otherwise.

Poor in FUMAC-P Equals 1 if respondent is poor and in a FUMAC-P community, zero otherwise.

Irrigation subproject Equals 1 if irrigation subproject, zero otherwise

Small bridge subproject Equals 1 if small bridge subproject, zero otherwise

Agreste region Equals 1 if Agreste region, and zero if Oeste region

Municipal Human Development Index Municipal Human Development Index 2000

Score for community Level of basic infrastructure in a community (such as primary school, basic health post, water supply

system, electrification, telephone boots, etc.) prior to Bank intervention (based on village leader inter-

view).

Economic status index Composite variable equal to the sum of two rebased variables that capture household’s ownership of

the following items prior to subproject implementation: (a) large animals (horse, cow, and ox), and (b)

consumer durables (car, motorcycle, bicycle, freezer, television, satellite dish)

Dummy for poor Equals 1 if respondent is from the bottom-quartile of the distribution along the Economic Status Index

in his/her community, zero otherwise



Member of CA Equals 1 if member of Community Association set up by the Bank CBD/CDD project, zero otherwise

Dummy for female Equals 1 if respondent is a female, zero otherwise



1 1 6

Table M.10: Definition of Variables: Brazil (continued)

Independent variable Definition

Schooling Level of education attained by the respondent (5=some secondary and above, 4=completed primary, 3=some

primary, 2=literate, 1=illiterate).

Agricultural laborer Equals 1 if the respondent is an agricultural laborer, zero otherwise



Participation in political events Frequency of participation in political events prior to Bank intervention

Participation in traditional events Frequency of participation in traditional events prior to Bank intervention

Mobilization skills Number of skills the respondent reported to have prior to Bank intervention. Composite variable equal

to the sum of four dummy variables (1=able, 0=everything else) that captures respondent’s ability to (a)

raise resources from within the community; (b) raise funds outside the community; (c) speak freely with

community leaders; (d) express the needs of the community to local government officials.

Table M.11: Definition of Variables: Madhya Pradesh


Change in access to information Changes in access to information regarding issues of interest to the community (more=3, same=2,

less=1)

Change in mobilization skills Composite variable equal to the sum of three dummy variables (1=more, 0=everything else) that cap-

ture change in the respondent’s ability to (a) organize self-help groups and raise resources from within

the village; (b) raise resources outside the village; (c) express the needs of the village to local govern-

ment officials.


Change in community leaders’ Change in the extent to which community leaders listen and respond to community needs (listen and

responsiveness respond more=4, listen more=3, same=2, listen less=1).

Change in trust Composite variable equal to the sum of five dummy variables (1=more, 0=everything else) that capture

change in respondent’s trust in: (a) village members, (b) village organizations, (c) village leaders, (d) local

government officials, (e) staff of the forest department (implementing agency).

Change in associational life Composite variable equal to the sum of two dummy variables (1=more, 0=everything else) that capture

change in: (a) people’s participation in groups, (b) cooperation between groups and individuals.

Change in participation in Change in the respondent’s participation in community’s traditional events. (more=3, same=2, less=1).

traditional events

Change in participation in Change in the respondent’s participation in community’s non-traditional events. (more=3, same=2,

political events less=1).



Project village Equals 1 if project village, and zero if comparator

Poor in project village Equals 1 if respondent is poor and in a project village, zero otherwise

Betul district Equals 1 if Betul district, and zero if Bilaspur district

Score for community Level of basic infrastructure in a community (such as primary school, basic health post, wells, hand-pumps)

and distance to the closest market prior to Bank intervention (based on village leader interview)

Table M.11: Definition of Variables: Madhya Pradesh (continued)

Independent variable Definition

Economic status index Composite variable equal to the sum of three rescaled variables that capture household’s ownership of

the following items prior to subproject implementation: (a) land (linear log), (b) large animals (horse, cow,

and ox), (c) consumer durables (car, bicycle, fan, radio)





Member of forest committee Equals 1 if member of forest committee set up by Bank CBD/CDD project, zero otherwise

Dummy for female Equals 1 if respondent is female, zero otherwise



Schooling Equals 1 if the respondent has attended school, zero otherwise

Number of leaders known Number of leaders a respondent knew prior to the Bank intervention

Participation in traditional events Frequency of participation in the traditional events prior to Bank interventions

Participation in non-traditional events Frequency of participation in the non-traditional/political events prior to Bank interventions

Mobilization skills Number of skills the respondent reported to have prior to Bank intervention. Composite variable equal

to the sum of three dummy variables (1=able, 0=everything else) that captures respondent’s ability to

(a) organize self-help groups and raise resources from within the village; (b) raise resources outside the

village; (c) express the needs of the village to local government officials

Table M.12: Definition of Variables: Uttar Pradesh


Change in access to information Changes in access to information regarding issues of interest to the community (more=3, same=2,

less=1)

Change in mobilization skills Composite variable equal to the sum of three dummy variables (1=more, 0=everything else) that cap-

ture change in the respondent’s ability to (a) organize self-help groups and raise resources from within

the village; (b) raise resources outside the village; (c) express the needs of the village to local govern-

ment officials.


Change in community leaders’ Change in the extent to which community leaders listen and respond to community needs (listen and

responsiveness respond more=4, listen more=3, same=2, listen less=1)

Change in trust Composite variable equal to the sum of five dummy variables (1=more, 0=everything else) capturing change

in trust in: (a) village members, (b) village organizations, (c) village leaders, (d) local elected officials,

(e) implementing agency


in: (a) people’s participation in groups, (b) cooperation between groups and individuals


traditional events

Change in participation in Change in the respondent’s participation in community’s non-traditional events. (more=3, same=2,

non-traditional events less=1).


1 1 7



1 1 8

Table M.12: Definition of Variables: Uttar Pradesh (continued)

Dependent variables Definition



Project village Equals 1 if project village, and zero otherwise (if comparator)

Poor in project village Equals 1 if respondent is poor and in project village, zero otherwise

Female in project village Equals one if the respondent is a female in a project village, zero otherwise

Amawa block Equals 1 if Amawa block, zero otherwise

Maharajganj block Equals 1 if Maharajganj block, zero otherwise

Rural community Equals 1 if it is a rural community, zero if rural dispersed

Score for community Level of basic infrastructure in a community (such as primary school, basic health post, wells, hand-pumps)

and distance to the closest market prior to Bank intervention (based on village leader interview)

Economic status index Composite variable equal to the sum of three rescaled variables that capture household’s ownership of

the following items prior to subproject implementation: (a) land (linear log), (b) large animals (horse, cow,

and ox), (c) consumer durables (car, bicycle, fan, radio)





Member of SIC Equals 1 if member of village organization set up by the Bank CBD/CDD project, zero otherwise

Dummy for female Equals 1 if respondent is a female, zero otherwise



Schooling Equals 1 if the respondent has attended school, zero otherwise

Number of leaders Number of leaders the respondent knew prior to the Bank intervention

Participation in traditional events Frequency of participation in the traditional events prior to Bank interventions

Participation in non-traditional events Frequency of participation in the non-traditional/political events prior to Bank interventions

Blue collar skills of the respondent Equals 1 if respondent was able to do boring, construct field drains, link drains and/or construct irriga-

tion channels prior to Bank intervention

Mobilization skills of the respondent Number of skills the respondent reported to have prior to Bank intervention. Composite variable equal

to the sum of three dummy variables (1=able, 0=everything else) that captures respondent’s ability to

(a) organize self-help groups and raise resources from within the village; (b) raise resources outside the

village; (c) express the needs of the village to local government officials

1 1 9

One of the premises of the CDD approach is thatit fosters the formation of social capital at thecommunity level and empowers communities totake charge of their own development.1 The as-sessment of the extent to which the CDD proj-ects supported by the World Bank haveimproved communities’ capacity focuses onthese two processes. This annex presents thefindings of community-level fieldwork under-taken for this evaluation in the Borgou regionof Benin (henceforth Benin); the state of RioGrande do Norte in Brazil (henceforth Brazil);the Betul and Bilaspur districts in MadhyaPradesh, India (henceforth Madhya Pradesh);and Raibarelli district in Uttar Pradesh, India(henceforth Uttar Pradesh). Three of these proj-ects were CDD and one was CBD. Henceforth,when reference is made to the four projects, thebroader term CBD/CDD is used. In addition,where relevant, this annex draws on a review ofappraisal documents of 84 sampled CBD/CDDprojects and on the literature on participatorydevelopment.

The field research at the community level in-cluded household surveys, focus group inter-views, and key informant interviews (see AnnexM for details). The methodology adopted forthe analysis of the household data as well as de-tails of the model used for multivariate analysisare also discussed in Annex M . Results of the bi-variate analysis are presented in tables N.3–N.6,while those of the multivariate analysis are pre-sented in tables N.7–N.16.2 It should be notedthat in the first two sections, discussion of theresults of the multivariate analysis refers to themodel without interactive terms, while the lasttwo sections draw on the model with interactiveterms (Annex M). As already mentioned in AnnexM, the results of the specification with interac-

tions are presented in full (tables N.9–N.16),while a summary of the results of the projectdummies for the specification without interac-tions is presented in tables N.7 and N.8. Unlessotherwise specified, the discussion of the sta-tistical significance is always based on the prob-ability value of the regression coefficients of therelevant specification.

EmpowermentThe World Bank’s (2002d) sourcebook on em-powerment and poverty reduction identifiesfour key elements for a successful empower-ment strategy: (a) inclusion and participation, (b)access to information, (c) accountability, (d) andlocal organizational capacity. This understandingof empowerment has informed data collectionfor this study, which explores both the levels ofempowerment at the time of fieldwork andrespondents’ perceptions of changes in em-powerment before and after subproject imple-mentation. While a comprehensive assessmentof the levels of empowerment and the empow-ering effects of the Bank’s CBD/CDD initiativesat the community-level was beyond the scope ofthis study, our analysis focuses on some aspectsof the four above-mentioned elements. It shouldalso be noted that empowerment is multidi-mensional and it is hence possible for a personto experience empowerment in one dimensionand disempowerment in another.

Inclusion and ParticipationCBD/CDD projects are operationalized at thecommunity level through community organiza-tions. While at times these predate CBD/CDD in-terventions, new ad hoc organizations are oftencreated by CBD/CDD projects, as was the casein Benin, Brazil, Madhya Pradesh, and Uttar

ANNEX N: ENHANCING COMMUNITY CAPACITIES


1 2 0

Pradesh.3 Because these organizations consti-tute the locus of decision making at the locallevel, becoming a member (or attending meet-ings) is extremely important in order to attain in-clusion in decision making. This, however, isnot in itself sufficient. Drawing on the literatureon participatory development, we make a dis-tinction between formal inclusion, which con-cerns the extent to which community membersare able to enter decision-making arenas, andsubstantive inclusion, which captures the ex-tent to which different participants are able toexert influence over decisions.4 While assessingformal inclusion is a fairly simple exercise, as-sessing substantive inclusion is far more complex,as it requires a detailed analysis of the veryprocess through which decisions are made. Thistype of investigation was beyond the reach of ourfield research. However, based on our data, wecan assess the extent to which villagers werelikely to attain substantive inclusion in decisionmaking.

Our household data reveal that in Benin, alarge share of respondents attained formal in-clusion in subproject decision making, while theopposite holds for Brazil and Uttar Pradesh. InMadhya Pradesh the picture is somewhat mixed.As figure N.1 shows, in Benin, 72 percent of re-spondents attended the meetings for subprojectselection, while in Brazil, only 37 percent of re-spondents in FUMAC communities attendedthese meetings, and in Uttar Pradesh only 16percent of respondents were members of proj-ect organizations and only 13 percent attendedmeetings regularly.5 In Madhya Pradesh, overhalf of the respondents were members of proj-ect organizations, but only a third of respon-dents attended meetings regularly. In Brazil, themajority of respondents were also likely to haveexerted only minimal influence over subprojectdecision making. Only 22 percent of the re-spondents in FUMAC communities spoke duringthe meetings (figure N.1), and a large share ofrespondents would refrain from expressing griev-ances with the subproject being implemented ifthis risked losing projects funds or compromis-ing relations with other villagers.6 The OED So-cial Fund Evaluation (OED 2002b) reports similarresults; only around 15 percent of the benefici-

aries reported speaking at the meetings in thefour countries surveyed. Focus group sessionsheld with villagers across the four project areasindicate that decision-making processes relativeto the Bank-funded subproject lacked broadcommunity participation.

It can be argued that a low level of inclusionin community organizations responsible for sub-project selection and management is not in itselfproblematic. It is unrealistic and perhaps ineffi-cient to expect communities to collectively un-dertake such activities, and a group of villagerscould instead be chosen to do so on behalf of thecommunity. Though valid, this argument raisesconcerns regarding the ways in which commu-nity representatives are selected and the inclu-sion of weaker social groups. In three of thefour project areas for which information isavailable, respondents who were members ofcommunity organizations set up by the Bank in-tervention had a higher socioeconomic profile,including greater mobilization skills and a moreextensive social network than non-membersprior to subproject implementation.7 Similarly,a large share of focus group interviews in Indiaand Benin pointed out that decision making re-garding the subproject was largely controlledby local leaders (figure 3.6 in Chapter 3). Variousstudies in the literature on participatory devel-opment also point out that the better-educatedmembers of the community and the relativelybetter-off are often the ones who represented thecommunity in participatory intervention (Desai1996; Gibson and Marks 1995; van der Linden1997; Ribot 1998).8

Access to InformationAs the World Bank’s (2002d) sourcebook on em-powerment puts it, “information is power.…Without information that is relevant, timely, andpresented in forms that can be understood, it isimpossible for poor people to take effective ac-tion.” Ensuring people’s access to information isparticularly important in CDD projects, in whichcommunities are expected to take a proactiverole in initiating the subproject cycle. A reviewof the appraisal documents for our sample of 84interventions reveals that fewer than half of themincluded an extensive campaign to disseminate

1 2 1

project information.9 When information is notdisseminated widely, communities are likely tobe dependent on a few informed individuals foraccessing development opportunities, and asthe focus group interviews in Benin and India re-veal, these tend to be the local leaders. Con-trolling information reinforces the position ofpower of these leaders, and creates opportuni-ties for strengthening their clientelistic network(Kumar and Corbridge 2002; Desai 1996; DasGupta and others 2000).

Household data also reveal communities’ lackof information regarding the subproject imple-mented. In Benin and Brazil, the vast majority ofrespondents in project communities had no in-formation on the cost of the subproject—86 and82 percent, respectively.10 Evidence from the lit-erature on northeast Brazil supports these find-ings (Tendler 2000).11 Communities’ lack ofinformation on the subproject affects the abilityof the community to hold to account the peo-ple who managed the subproject investments on

A N N E X N

Figure N.1: Beneficiaries’ Inclusion and Participation in Subproject Decision Making

37%

72%

22%

62%

96%

34%

Perc

enta

ge o

f res

pond

ents

Benin [N = 566] Brazil [N = 485]0

40

20

60

80

100

Aware of meetings Attended meetings Spoke at meetings

16%

52%

13%

50%

92%

33%

Perc

enta

ge o

f res

pond

ents

Madhya Pradesh [N=807] Uttar Pradesh [N=757]0

40

20

60

80

100

Aware of project organizations

Member of project organizations

Attended meetings regularly


1 2 2

its behalf, and provides further evidence of thegeneral lack of broad-based community partici-pation in Bank-funded initiatives.

In addition to exploring the level of informa-tion at the time of our fieldwork, the householdsurveys also captured respondents’ perceptionsof the change in access to information on is-sues of interest to the communities before andafter subproject implementation. Our findings re-veal a mixed picture (figure N.2). A significantpositive association was found between theBank’s CBD/CDD projects in Madhya Pradeshand respondents’ access to information on issuesof relevance to the community, while no signif-icant association was found in Benin and UttarPradesh. In Brazil, respondents in comparatorcommunities reported a significantly higher in-crease in access to information than did re-spondents in FUMAC communities.12

AccountabilityThe notion of accountability has a range of con-notations. It is used here to refer to citizens’ abil-ity to hold local leaders and public officialsaccountable. The availability and accessibility of

information are critical for accountability, and thefindings discussed above on access to informa-tion already indicate weak accountability to com-munities. Although necessary, access toinformation is not in itself sufficient, as it does notautomatically result in accountability; citizensmust act upon the information they acquire (Jenk-ins and Goetz 1999). While it was beyond thescope of our field research to explore accounta-bility issues extensively, a few aspects of these is-sues were captured by our questionnaires.

A large share of respondents in project com-munities in Benin, Uttar Pradesh, and MadhyaPradesh and over half of those in FUMAC com-munities in Brazil agreed that if dissatisfied withthe performance of community leaders, villagerswould call a meeting to discuss it. More than halfof the respondents in project communities inBenin and Madhya Pradesh also agreed that if dis-satisfied with community leaders, the communitywould replace them. Fewer respondents agreedwith this statement in project communities inUttar Pradesh and FUMAC communities inBrazil—respectively 42 and 24 percent. Ourhousehold data also capture respondents’ per-

Figure N.2: The Bank’s CBD/CDD Projects and Access to Information

Resp

onde

nts

repo

rtin

g an

incr

ease

(%)

Project Comparator

0

40

20

60

80

100

Benin [N = 772] Brazil

[N = 602]Madhya P. [N = 871]

Uttar P. [N = 992]

Difference is statistically significant★

★

★

Note: Statistical significance based on the model without interactions.

1 2 3

ceptions of the changes in community leaders’responsiveness to community needs before andafter subproject implementation. Multivariateanalysis indicated no statistically significant as-sociation between the Bank’s projects in Beninand Madhya Pradesh and community leaders’responsiveness to community needs. In Brazil,respondents in all three types of project com-munities reported a significantly smaller increasein community leaders’ responsiveness than didrespondents in comparator communities.

Local Organizational CapacityAccording to the Bank’s empowerment source-book, local organizing capacity “refers to theability of people to work together, organizethemselves, and mobilize resources to solveproblems of common interest” (World Bank2002b). In order to assess the extent to which theBank’s interventions succeeded in fostering com-munities’ organizational capacity, our surveyscaptured respondents’ perceptions of thechanges in their mobilization skills, and in theability of the community to reach an agreementbefore and after subproject implementation.13 A

third aspect relevant to local organizational ca-pacity (which is also a dimension of social cap-ital) is the change in associational life, whichcaptures changes in respondents’ participationin community groups and changes in coopera-tion between community groups.

Multivariate analysis indicated no statisticallysignificant association between the Bank’s proj-ects and the changes in respondents’ mobiliza-tion skills, with the exception of Uttar Pradesh,where respondents in project communities re-ported a significantly greater increase in mobi-lization skills than did respondents in comparatorcommunities. The relation between Bank’sCBD/CDD projects and communities’ ability toreach an agreement was mixed (figure N.3). Theprojects in Benin and Uttar Pradesh were posi-tively associated with communities’ ability toreach an agreement, while no statistical associ-ation was found in Madhya Pradesh. In Brazil, re-spondents in all three types of projectcommunities reported a significantly smaller in-crease in their ability to reach an agreementthan did respondents in comparator communi-ties. A more positive picture emerges from the

A N N E X N

Figure N.3: The Bank’s CBD/CDD Project Communities’ Ability to Reach an Agreement

Resp

onde

nts

repo

rtin

g an

incr

ease

(%)

Project Comparator

0

40

20

60

80

100

Benin [N = 772] Brazil

[N = 602]Madhya P. [N = 1,068]

Uttar P. [N = 1,157]

Difference is statistically significant

★

★

★

★



1 2 4

analysis of the changes in associational life (fig-ure N.4). All four Bank’s projects are positively

associated with respondents’ perceptions of thechanges in associational life.14

Figure N.4: The Bank’s CBD/CDD Projects and Social Capital

★

★

★

★★

★

★

★★

★

★

★

★

★

★★

Resp

onde

nts

repo

rtin

g an

incr

ease

(%)

PAMR Comparator

0

40

20

60

80

100

Trus

t

Asso

cia-

tiona

l life

Trad

ition

al ev

ents

Polit

ical

even

ts

Frien

ds

Difference is statistically significant

Benin [N = 772]

Resp

onde

nts

repo

rtin

g an

incr

ease

(%)

Project Comparator

0

40

20

60

80

100Madhya Pradesh [N = 1,068]

Resp

onde

nts

repo

rtin

g an

incr

ease

(%)

FUMAC Comparator

0

40

20

60

80

100Brazil [N = 602]

Resp

onde

nts

repo

rtin

g an

incr

ease

(%)

Project Comparator

0

40

20

60

80

100Uttar Pradesh [N = 1,157]

Trus

t

Asso

cia-

tiona

l life

Trad

ition

al ev

ents

Polit

ical

even

ts

Frien

ds

Trus

t

Asso

cia-

tiona

l life

Trad

ition

al ev

ents

Polit

ical

even

ts

Frien

ds

Trus

t

Asso

cia-

tiona

l life

Trad

ition

al ev

ents

Polit

ical

even

ts

Frien

ds


1 2 5

Social CapitalSocial capital refers to the norms and networksthat enable collective action.15 By drawing peo-ple together to collectively decide and manageproject activities and outputs, these projects areexpected to expand the depth and range of com-munities’ social networks. In order to assess theextent to which Bank-funded interventions havesucceeded in enhancing social capital at thecommunity level, we draw on five variables,which capture respondents’ perceptions of thechanges in (a) trust, (b) associational life, (c) par-ticipation in traditional events, (d) participationin non-traditional/political events, and (e) circleof friends before and after subproject imple-mentation.16 It is important to bear in mind thatthese variables capture only some of the multi-ple dimensions of social capital and that ouranalysis of the association between Bank-sup-ported projects and social capital was limitedto the changes observed in these five dimensions.

Multivariate analysis indicated a statisticallysignificant and positive association between theprojects in Uttar Pradesh, Madhya Pradesh, andBenin and respondents’ perceptions of thechanges in social capital.17 Results for the Brazilprojects are mixed. While respondents in projectcommunities reported a significantly greater in-crease in associational life than did respondentsin comparator communities, the opposite holdsfor respondent’s perceptions of changes in theircircle of friends (figure N.4).18 There are three rea-sons for the different levels of change in socialcapital in the four project areas, which mightalso explain the differences in the changes inempowerment. First, communities in the fourproject areas are likely to have different capacitylevels; therefore, the change that can be expectedas a result of exposure to a Bank intervention isalso likely to be different. Second, the socio-po-litical setting in which these initiatives are im-plemented affects their impact on social capitalenhancement. The literature notes that the per-vasive clientelism in the northeast of Brazil cre-ates a social system in which vertical ties of mutualdependence prevail, and hinder the develop-ment of strong horizontal links of solidarity be-tween communities (Costa and others 1997;Tendler 2000). This probably explains the lack of

influence on social capital of the Bank’s initiativein Rio Grande do Norte. Third, a lengthy en-gagement with a consistent capacity-buildingstrategy with the same communities is likely toyield better results than a brief one. This mightexplain why the Borgou Pilot Project, which wasintroduced in communities where the Bank’sVillage Level Participatory Approach had beenimplemented in the 1990s and shared its ap-proach, outperformed the Social Fund (AgeFIB)and the Food Security Project (PILSA).19

Bank’s Projects and Membersof Community Organizations As already mentioned above, new ad hoc com-munity organizations were created by the Bank’sprojects for their operationalization at the com-munity level (see endnote 3). This section ex-plores the institutional development impact ofthe CBD/CDD projects in Brazil, Madhya Pradesh,and Uttar Pradesh on the members of these or-ganizations. In order to do so, a variable inter-acting membership in project organizations andthe project dummy was included in the model.This variable estimates the association betweenthe dependent variable and membership in proj-ect organizations relative to the project dummy.

The multivariate analysis indicated a statisti-cally significant and positive association betweenthe projects in Brazil, Madhya Pradesh, and UttarPradesh, and the changes in empowerment re-ported by members of community organizationsset up by these projects (table N.1). In Brazil,members reported a greater increase in access toinformation and community leaders’ responsive-ness to community needs than non-members. InMadhya Pradesh, members reported a greater in-crease in access to information and mobilizationskills than non-members, while in Uttar Pradesh,they reported a greater increase in mobilizationsskills and communities’ ability to reach an agree-ment. A statistically significant and positive asso-ciation is also found between the projects inMadhya Pradesh and Uttar Pradesh, and respon-dent’s perceptions of the changes in social capi-tal. These two Bank projects are positivelyassociated with four of the five dimensions of so-cial capital considered in this study (table N.2). Aweaker association is found between the Brazil

ANNEX N


1 2 6

project and changes in social capital, with mem-bers of Community Associations reporting a greaterincrease in only one of the five dimensions of so-cial capital considered—associational life.

While encouraging, these findings raise im-portant concerns, because, as pointed out ear-lier (see endnote 7), respondents who weremembers of project-induced community or-ganizations had a higher socioeconomic profile,including greater mobilization skills and a moreextensive social network than non-membersprior to subproject implementation.

Bank’s Projects and Vulnerable GroupsBank’s CBD/CDD projects are aimed at em-powering and enhancing the social capital ofvulnerable groups, including women and thepoor (endnote 1). Two interactive variables wereincluded in the regression model in order toexplore the associations between the changes insocial capital and empowerment and the poor inproject communities, on the one hand, andwomen in project communities, on the other.The first variable interacts belonging to the bot-tom quartile of the index of economic statuswith the project dummy, while the second in-teracts being a women with the project dummy.These variables estimate the association betweenthe dependent variable and women or the poor

in project communities relative to the projectdummy.

The multivariate analysis indicated that thepoor in project communities in Madhya Pradeshreported a significantly greater increase in twodimensions of social capital—trust and associa-tional life—as well as a greater increase in theirmobilization skills than did the relatively better-off in project areas. In Uttar Pradesh, the asso-ciation between changes in empowerment andthe poor in project communities is mixed. Whilethe poor in project areas reported a greater in-crease in the community’s ability to reach anagreement than did the relatively better-off inCBD/CDD communities, they also reported asignificantly smaller increase in their access to in-formation. In Brazil, no significant association isfound between changes in social capital and em-powerment and the poor in project communi-ties.

The projects in Benin and Uttar Pradesh,which explicitly targeted women, do not appearto have enhanced women’s capacities over andabove other respondents in project communities.The only exceptions are women in project areasin Uttar Pradesh who reported a significantlygreater increase in their mobilization skills andassociational life than did men in project com-munities.

Table N.1: Empowering Members of Community Organizations

Change in… Brazil Madhya Pradesh Uttar Pradesh

Access to information positive positive

Mobilization skills positive positive

Ability to reach agreement positive

Community leaders’ responsiveness to community needs positive

Table N.2: Enhancing the Social Capital of Members of Community Organizations

Change in… Brazil Madhya Pradesh Uttar Pradesh

Trust in individuals/organizations positive positive

Associational life positive positive positive

Participation in traditional events

Participation in non-traditional/political events positive positive

Circle of friends positive positiveNote: Results for the dummy for members of project organizations in CBD/CDD communities estimated in the model with interactions (tables N.10–N.16).

1 2 7

ANNEX N

Table N.3: Benin: Bivariate Analysis of Variables Relevant to the Discussion on Empowermentand Social Capital (percent)

PAMR, 566 Comparator, 206observations observations

Change in access to information Worse 1 2 **Same 12 17 Better 86 80

Change in mobilization skills Same 10 16 **Increase in 1 of 4 10 8 Increase in 2 of 4 16 14 Increase in 3 of 4 20 29 Increase in all 4 45 34

Change in ability to reach an agreement Worse 1 9 ***Same 14 22 Better 85 68

Change in community leaders’ responsiveness to community needs Less 3 1 **Same 25 35 Listen more 44 40 Listen and respond more 28 24

Change in trust in individuals and organizations Same 32 43 ***Increase in 1 of 4 11 12 Increase in 2 of 4 17 24 Increase in 3 of 4 26 12 Increase in all 4 14 8

Change in associational life Same 18 25 ***Increase in 1 of 2 43 50 Increase in 2 of 2 39 25

Change in participation in traditional events Less 9 8 Same 47 55 More 44 37

Change in participation in political eventsa Less 12 11 Same 54 50 More 35 38

Change in circle of friends Less 1 2 **Same 25 32 More 74 66

Express grievances if this risks losing project funds or No 50 58 compromising relations with other villagers Yes 49 42

If unhappy with community leaders, villagers call a Disagree 6 9 ***meeting to discuss itb Somewhat disagree 5 6

Somewhat agree 15 23 Agree 74 62

If unhappy with community leaders, Disagree 19 27 ***villagers replace them c Somewhat disagree 11 13

Somewhat agree 13 13 Agree 57 47



Table N.3: Benin: Bivariate Analysis of Variables Relevant to the Discussion on Empowermentand Social Capital (percent) (continued)

PAMR, 566 Comparator, 206observations observations

Participation at community meetings for subproject selection Unaware of meetings 4

Aware of meetings 96

Attended meetings 72

Attended and spoke at meetings 34 Note: Significance based on a test of proportion for binary variables and the Kruskall-Wallis test for categorical variables. * Significant at 10%; ** significant at 5%; ***significant at 1%.

a. No. observations: project = 562

b. No. observations: project = 561

c. No. observations: project = 555; comparator = 202

Table N.4: Brazil: Bivariate Analysis of Variables Relevant to the Discussion on Empowermentand Social Capital (percent)

FUMAC, 485 Comparator, 117observations observations

Change in access to information a Worse 16 3 ***

Same 38 41

Better 46 56

Change in mobilization skills Same 60 53

Improve in 1 of 4 20 33

Improve in 2 or more of 4 20 14

Change in ability to reach an agreement Worse 13 2 ***

Same 32 27

Better 54 69

Change in community leaders’ responsiveness to community needsb Less 19 9 *

Same 53 58

Listen more 28 32

Change in trust in individuals and organizations Same 66 59

Increase in 1 of 4 22 34

Increase in 2 and above of 4 12 7

Change in associational life Same 49 58 ***



Change in participation in traditional events c Less 18 17

Same 73 76

More 9 7

Change in participation in political events d Less 13 2

Same 80 97

More 7 1

Change in circle of friends Less 4 3 *

Same 51 44

More 44 54

1 2 8


1 2 9

Table N.4: Brazil: Bivariate Analysis of Variables Relevant To the Discussion onEmpowerment and Social Capital (percent) (continued)

FUMAC, 485 Comparator, 117observations observations

Express grievances if this risks losing project funds No 66 77 **

Yes 34 22

Express grievances if this risks compromising relations No 61 66

with other villagers Yes 39 33

If unhappy with community leaders, villagers call a meeting No 38 44 ***

to discuss ite Yes 56 14

If unhappy with community leaders, villagers replace theme No 64 49 ***

Yes 24 4

Participation at community meetings for subproject selection Unaware of meetings 38

Aware of meetings 62

Attended meetings 37

Attended and spoke at meetings 22 Note: Significance based on a test of proportion for binary variables and the Kruskall-Wallis test for categorical variables. * Significant at 10%; ** significant at 5%; ***significant at 1%.

a. No. observations: project = 482; comparator = 115

b. No. observations: project = 452; comparator = 96

c. No. observations: project = 481; comparator = 117.

d. No. observations: project = 474; comparator = 117

e. No. observations: project = 484; comparator = 111.

Table N.5: Madhya Pradesh: Bivariate Analysis of Variables Relevant to the Discussion onEmpowerment and Social Capital (percent)

Project, 807 Comparator, 261observations observations

Change in access to information a Worse 16 21 ***

Same 35 41

Better 49 38

Change in mobilization skills Same 35 40 *



Improve in all 3 15 12

Change in ability to reach an agreement Worse 23 25

Same 29 31

Better 19 15

Change in community leaders’ responsiveness to community needs b Less 17 16

Same 37 38

Listen more 20 27

Listen and respond more 19 15

Change in trust in individuals and organizations Same 62 67 *




1 3 0

Table N.5: Madhya Pradesh: Bivariate Analysis of Variables Relevant To the Discussion onEmpowerment and Social Capital (percent) (continued)





Increase in all 5 6 4

Change in associational life Same 60 67 **



Change in participation in traditional eventsc Less 9 13

Same 65 63

More 26 23

Change in participation in non-traditional eventsd Less 11 8

Same 55 63

More 34 29

Change in circle of friendse Less 10 12 ***

Same 47 54

More 43 34

Express grievances if this risks losing project funds No 48 46

Yes 43 47

Express grievances if this risks compromising relations No 40 34 *


If unhappy with community leaders, villagers call Disagree 14 17

a meeting to discuss itf Somewhat disagree 6 3

Somewhat agree 19 20

Agree 61 59

If unhappy with community leaders, villagers replace themg Disagree 23 26

Somewhat disagree 8 7


Agree 51 47

Participation in community organizations set up by the bank project Unaware of project organizations 8

Aware of project organizations 92

Member of project organizations 52

Member and attended 33

meetings regularly Note: Significance based on a test of proportion for binary variables and the Kruskall-Wallis test for categorical variables. * Significant at 10%; ** significant at 5%; ***significant at 1%.

a. No. observations: project = 653; comparator = 218.

b. No. observations: project = 581; comparator = 191.


d. No. observations: project = 787; comparator = 258.

e. No. observations: project = 716; comparator = 249

f. No. observations: project = 688; comparator = 229.

g. No. observations: project = 667; comparator = 227.

1 3 1

ANNEX N

Table N.6: Uttar Pradesh: Bivariate Analysis of Variables Relevant to the Discussion onEmpowerment and Social Capital (percent)


Change in access to informationa Worse 8 8 ***

Same 29 42

Better 63 50

Change in mobilization skills Same 22 33 ***



Improve in all 3 30 16

Change in ability to reach an agreement Worse 24 46 ***

Same 15 39

Better 56 8

Change in community leaders’ responsiveness to community needs Less 26

Same 27

Listen more 23

Listen and respond more 22

Change in trust in individuals and organizations Same 33 51 ***





Increase in all 5 4 0

Change in associational life Same 53 68 ***



Change in participation in traditional events Less 12 20 ***

Same 36 42

More 52 38

Change in participation in non-traditional Less 12 15 ***

Same 40 52

More 47 34

Change in circle of friendsb Less 5 4 ***

Same 32 42

More 63 54

Express grievances if this risks losing project funds No 45 36 ***

Yes 54 64

Express grievances if this risks compromising relations No 33 30


If unhappy with community leaders, villagers call a Disagree 11 11 *

meeting to discuss itc Somewhat disagree 5 4


Agree 63 55



1 3 2

Table N.6: Uttar Pradesh: Bivariate Analysis of Variables Relevant to the Discussion onEmpowerment and Social Capital (percent) (continued)


If unhappy with community leaders, villagers replace themd Disagree 25 25

Somewhat disagree 7 7


Agree 42 37

Participation in community organizations set up by the bank project Unaware of project organizations 50

Aware of project organizations 50

Member of project organizations 16

Member and attended

meetings regularly 13 Note: Significance based on a test of proportion for binary variables and the Kruskall-Wallis test for categorical variables. * Significant at 10%; ** significant at 5%; ***significant at 1%.

a. No. observations: project = 686; comparator = 306.

b. No. observations: project = 705; comparator = 386.


d. No. observations: project = 722; comparator = 393.

Table N.7: Coefficients and Significance of Project Dummies in the Model withoutInteractive Terms: Change in Empowerment (Ordered probit)

Change inChange in Change in Change in community leaders’access to mobilization ability to reach responsiveness to

information skills agreement community needs(Coef.) (Coef.) (Coef.) (Coef.)

Benin (PAMR) 0.19 0.25 0.73 *** 0.13

Brazil (PAC) –0.04 –0.44 –0.53 *** –0.42 *

Brazil (FUMAC) –0.40 *** –0.11 –0.50 *** –0.38 *

Brazil (FUMAC-P) –0.37 ** –0.42 *** –0.83 *** –0.65 **

Madhya Pradesh 0.27 ** 0.13 0.19 0.07

Uttar Pradesh 0.08 0.61 *** 0.93 *** —Note: Weighted estimation (except for Benin) adjusted for cluster effects. * Significant at 10%; ** significant at 5%; ***significant at 1%.

1 3 3

ANNEX N

Table N. 8: Coefficients and Significance of Project Dummies in the Model withoutInteractive Terms: Change in Social Capital (Ordered probit)

Change in Change in Change intrust in Change in participation participation in Change in

individuals & associational in traditional non-traditional/ circle of organizations life events political events friends

(Coef.) (Coef.) (Coef.) (Coef.) (Coef.)

Benin (PAMR) 0.35 *** 0.27 ** 0.17 0.08 0.35 ***

Brazil (PAC) –0.27 –0.26 0.11 –0.08 0.07

Brazil (FUMAC) –0.09 0.27 * 0.13 –0.10 –0.27 ***

Brazil (FUMAC-P) –0.34 ** –0.57 ** 0.01 –0.32 ** –0.45 ***

Madhya Pradesh 0.36 *** 0.36 ** 0.34 ** 0.20 ** 0.26 ***

Uttar Pradesh 0.55 *** 0.48 * 0.50 *** 0.25 ** 0.34 **

Note: Weighted estimation (except for Benin) adjusted for cluster effects. * Significant at 10%; ** significant at 5%; ***significant at 1%.

Table N.9: Benin: Change in Empowerment (Ordered probit)



Dummy for PAMR 0.19 0.16 0.73 *** 0.08

Dummy for AgeFIB 0.01 –0.11 0.54 0.05

Dummy for PILSA –0.46 * 0.37 –0.19 –0.11

Dummy for female in PAMR 0.01 0.26 –0.01 0.16

Dummy for school construction subproject 0.32 *** 0.05 –0.07 0.09

Dummy for training subproject 1.23 *** 0.07 1.43 *** 0.51 ***

Dummy for difficult access to community –0.05 –0.05 –0.16 –0.07

Household size 0.02 0.01 * 0.00 0.02 ***

Number of children –0.03 –0.01 0.01 –0.01

Dummy for female –0.22 –0.36 ** –0.15 –0.14 *

Schooling of the respondent –0.07 0.09 –0.11 0.01

Age 0.00 –0.04 *** –0.03 –0.01

Age square 0.01 0.18 0.27 0.10

Number of leaders known 0.17 *** 0.11 ** 0.15 ** –0.03

Participation in traditional events 0.21 *** –0.03 0.13 * –0.02

Participation in political events 0.00 0.05 –0.02 0.12 ***

Blue collar skills of respondent 0.13 –0.05 –0.10 0.10

Mobilization skills of the respondent –0.11 *** 0.38 *** 0.01 –0.02

Observations 1,028 1,028 1,028 1,028

Pseudo R-squared 0.08 0.12 0.09 0.02

Chi2 472.67 489.58 276.21 361.93

Note: Estimation adjusted for cluster effects. * Significant at 10%; ** significant at 5%; ***significant at 1%.


1 3 4

Table N.10: Brazil: Change in Empowerment (Ordered probit)



Dummy for PAC –0.25 –0.44 –0.62 *** –0.44 *

Dummy for FUMAC –0.62 ** –0.23 –0.55 *** –0.51 *

Dummy for FUMACP –0.62 ** –0.51 ** –0.95 *** –0.69 **

Dummy for poor in PAC 0.43 –0.09 0.17 –0.10

Dummy for poor in FUMAC 0.11 0.25 –0.04 0.11

Dummy for poor in FUMACP 0.31 0.20 0.21 –0.27

Dummy for irrigation subproject –0.05 0.04 –0.05 –0.39

Dummy for small bridge subproject 0.06 0.53 ** –1.04 *** –0.67 ***

Dummy for Agreste region 0.08 –0.02 –0.34 –0.43 ***

Municipal Human Develop Index 1.35 –2.44 –6.33 ** 0.85

Score for community –0.80 –0.91 * –0.65 –0.31

Economic status 0.44 0.66 *** 0.37 0.07

Dummy for poor –0.25 –0.09 –0.07 0.01

Household size 0.01 0.04 0.05 0.00

Number of children 0.01 –0.04 –0.04 0.01

Dummy for member of the CA 0.49 *** 0.13 0.19 0.26 **

Dummy for female 0.04 –0.10 0.11 –0.17

Schooling of the respondent 0.04 –0.04 0.03 0.08

Dummy for agricultural laborer –0.26 ** –0.24 0.01 –0.13

Age –0.01 –0.01 0.02 –0.01

Age square 0.03 0.02 –0.23 0.12

Participation in political events 0.10 –0.22 0.11 –0.02

Participation in traditional events 0.14 0.11 0.08 0.26 ***

Mobilization skills of the respondent 0.01 –0.09 ** 0.03 0.06 *

Observations 916 925 915 771

Pseudo R-squared 0.05 0.03 0.06 0.07

Chi2 1,636.00 418.59 2,147.15 3,005.88

Note: Weighted estimation adjusted for cluster effects. * Significant at 10%; ** significant at 5%; ***significant at 1%.

1 3 5

ANNEX N

Table N.11: Madhya Pradesh: Change in Empowerment (Ordered probit)



Dummy for project village 0.22 0.01 0.21 0.05

Dummy for poor in project village 0.11 0.30 ** –0.13 0.11

Dummy for Betul district 0.27 * 0.12 0.55 *** 0.38 **

Score for community –0.25 –0.48 –1.42 *** –1.20 ***

Economic status 0.21 0.05 0.17 0.04

Dummy for poor –0.08 –0.22 * 0.00 –0.09

Household size -0.03 0.02 0.01 –0.02

Number of children 0.08 ** 0.00 –0.01 0.06

Dummy for member of forest committee 0.28 *** 0.35 *** 0.06 –0.02

Dummy for female –0.14 –0.11 * –0.03 –0.07

Age of respondents 0.01 0.00 0.04 * 0.00

Age square –0.05 –0.11 –0.51 ** 0.07

Schooling of the respondent 0.34 *** 0.34 *** 0.18 0.10

Number of leaders known 0.00 0.04 0.00 0.03

Participation in traditional events 0.05 0.17 *** 0.31 *** 0.19 ***

Participation in non-traditional events –0.04 0.05 0.03 –0.07

Mobilization skills of the respondent –0.06 0.18 *** 0.02 –0.08

Observations 859 1,046 756 712

Pseudo R-squared 0.05 0.07 0.10 0.05

Chi2 155.53 490.72 216.07 289.32



1 3 6

Table N.12: Uttar Pradesh: Change in Empowerment (Ordered probit)


information skills agreement community needsa

(Coef.) (Coef.) (Coef.) (Coef.)

Dummy for project village –0.09 0.34 ** 0.80 *** —

Dummy for poor in project village –0.28 * 0.23 0.42 *** –0.17

Dummy for female in project village 0.37 0.34 *** –0.05 –0.04

Dummy for Amawa district 0.25 0.08 0.14 0.49 ***

Dummy for Maharajganj district –0.17 –0.29 –0.25 0.22 ***

Dummy for rural –0.52 ** –0.16 –0.25 –0.34 ***

Score for community –0.25 0.06 0.74 * –0.17

Economic status 0.17 –0.25 –0.75 * 0.19

Dummy for poor –0.11 –0.48 ** –0.66 ** —

Household size 0.02 0.01 0.00 0.01

Number of children –0.08 –0.05 ** –0.02 –0.04

Dummy for member of SIC 0.25 0.25 ** 0.32 ** 0.30

Dummy for female –0.20 –0.41 *** –0.24 ** —

Age of the respondent –0.03 –0.01 0.02 –0.07 ***

Age square 0.20 0.02 –0.29 0.71 ***

Schooling of respondent 0.29 ** 0.37 *** 0.00 –0.01

Number of leaders known 0.19 *** 0.35 *** 0.07 0.19 ***

Participation in traditional events 0.07 –0.05 0.03 —

Participation in non-traditional events 0.17 ** 0.24 *** –0.03 —

Blue collar skills of respondent –0.02 0.18 0.40 *** —

Mobilization skills of the respondent 0.14 0.18 *** 0.12 –0.04

Observations 986 1,148 1,082 747

Pseudo R-squared 0.09 0.10 0.13 0.03

Chi2 757.04 2,022.38 2,407.09 408.07


a. This question was only asked in project villages. Three variables were dropped from this regression because in order to be estimated adjusting for cluster effects, the model requires

the number of clusters to be greater than the number of constraints. The variables dropped were found not significant in the model estimated without adjusting for cluster effects.

1 3 7

ANNEX N

Table N.13: Benin: Change in Social Capital (Ordered probit)

Change in Change in trust in Change in participation Change in Change in

individuals & associational in traditional participation in circle of organizations life events political events friends


Dummy for PAMR 0.33 *** 0.29 ** 0.12 0.01 0.34 ***

Dummy for AgeFIB 0.28 ** 0.05 0.07 –-0.04 –0.14

Dummy for PILSA 0.30 *** 0.32 * –0.08 0.03 –0.06

Dummy for female in PAMR 0.07 –0.06 0.13 0.23 0.03

Dummy for school construction subproject 0.43 *** 1.32 *** –0.12 –0.01 0.19

Dummy for training subproject 0.16 0.03 –0.01 –0.36 *** 0.45

Dummy for difficult access to community –0.10 –0.25 –0.09 –0.18 * –0.34 **

Household size 0.02 *** 0.00 –0.02 ** –0.01 0.01

Number of children –0.02 –0.01 0.04 *** 0.02 0.00

Dummy for female –0.16 –0.17 –0.01 –0.11 –0.13

Schooling of the respondent –0.08 0.03 –0.01 0.11 0.08

Age –0.04 ** –0.01 0.00 0.01 –0.01

Age square 0.34 ** 0.10 0.00 –0.18 0.11

Number of leaders known 0.13 * 0.20 *** 0.04 0.08 0.14 **

Participation in traditional events –0.05 0.11 ** 0.06 –0.03 0.14 ***

Participation in political events 0.11 ** –0.01 0.10 * 0.39 *** –0.06

Blue collar skills of respondent 0.06 0.15 –0.25 ** –0.35 *** 0.02

Mobilization skills of the respondent 0.05 0.02 –0.05 –0.03 –0.01

Observations 1,028 1,028 1,028 1,026 1,028

Pseudo R-squared 0.03 0.15 0.02 0.08 0.04

Chi2 151.03 629.10 491.85 303.81 768.10

Note: Estimation adjusted for cluster effects. * Significant at 10%; ** significant at 5%; ***significant at 1%.


1 3 8

Table N.14: Brazil: Change in Social Capital (Ordered probit)

Change in Change in trust in Change in participation Change in Change in

individuals & associational in traditional participation in circle of organizations life events political events friends


Dummy for PAC –0.21 –0.23 0.07 –0.17 0.20

Dummy for FUMAC –0.10 0.24 0.15 –0.14 –0.37 ***

Dummy for FUMACP –0.28 –0.70 ** –0.05 –0.39 ** –0.61 ***

Dummy for poor in PAC –0.23 –0.35 0.18 0.28 –0.49 **

Dummy for poor in FUMAC 0.03 –0.37 0.02 0.11 0.14

Dummy for poor in FUMACP –0.22 0.01 0.28 0.23 0.38 *

Dummy for irrigation subproject 0.01 0.08 –0.25 –0.36 –0.25

Dummy for small bridge subproject 0.35 –0.69 *** –1.19 *** –0.54 –0.81 ***

Dummy for Agreste region 0.00 –0.55 ** –0.21 –0.04 –0.23

Municipal Human Develop Index –2.37 –8.18 ** –4.90 ** –1.73 –4.65

Score for community –0.68 0.17 0.22 0.11 –0.90 *

Economic status 0.33 0.55 ** –0.47 ** 0.05 0.42

Dummy for poor 0.11 0.34 –0.31 –0.17 –0.10

Household size 0.05 * 0.05 0.05 * 0.02 0.02

Number of children –0.04 –0.07 –0.09 ** –0.02 –0.03

Dummy for member of the CA 0.00 0.38 ** –0.06 0.02 0.17

Dummy for female –0.13 0.02 –0.11 –0.16 0.08

Schooling of the respondent –0.05 * 0.01 0.08 * 0.05 –0.07 **

Dummy for agricultural laborer –0.43 ** 0.16 0.00 –0.10 –0.02

Age –0.01 –0.01 –0.01 –0.02 0.01

Age square –0.04 0.00 0.02 0.20 –0.10

Participation in political events –0.22 0.04 0.00 0.00 0.12

Participation in traditional events 0.12 0.17 0.20 *** 0.17 *** 0.20 ***

Mobilization skills of the respondent 0.03 0.00 0.11 *** 0.06 ** 0.06 **

Observations 925 925 919 909 917

Pseudo R–squared 0.03 0.07 0.05 0.03 0.05

Chi2 1,043.50 986.26 1,945.22 283.91 874.42


1 3 9

ANNEX N

Table N.15: Madhya Pradesh: Change in Social Capital (Ordered probit)


individuals & associational in traditional non-traditional circle of organizations life events events friends


Dummy for project village 0.25 * 0.20 0.41 *** 0.11 0.20 *

Dummy for poor in project village 0.33 * 0.54 ** –0.25 0.22 0.11

Dummy for Betul district 0.13 0.18 –0.01 0.26 ** –0.09

Score for community –0.83 ** –0.50 * 0.33 0.05 0.21

Economic status 0.17 –0.03 0.37 * –0.13 –0.27 *

Dummy for poor –0.38 * –0.50 0.31 –0.22 –0.37 **

Household size –0.02 –0.01 0.00 –0.02 0.01

Number of children 0.03 0.01 0.05 ** 0.04 * 0.03

Dummy for member of forest committee 0.33 *** 0.29 *** 0.13 0.30 *** 0.22 **

Dummy for female –0.18 ** –0.21 * –0.15 * –0.20 ** –0.11

Age of respondents 0.00 –0.01 0.07 ** 0.04 –0.02

Age square –0.06 0.12 –0.76 *** –0.43 * 0.18

Schooling of the respondent 0.45 *** 0.26 0.00 0.19 0.17

Number of leaders known 0.00 –0.02 –0.03 –0.03 0.03

Participation in traditional events 0.00 0.11 * –0.04 0.00 0.03

Participation in non-traditional events 0.02 0.03 –0.02 0.00 0.05

Mobilization skills of the respondent 0.05 0.12 *** 0.00 –0.14 ** –0.11 **

Observations 1,046 1,046 1,045 1,042 956

Pseudo R-squared 0.06 0.06 0.03 0.04 0.04

Chi2 721.66 113.39 80.62 100.59 327.13



1 4 0

Table N.16: Uttar Pradesh: Change in Social Capital (Ordered probit)


individuals & associational in traditional non-traditional circle of organizations life events events friends


Dummy for project village 0.44 *** 0.18 0.33 ** 0.09 0.15

Dummy for poor in project village 0.19 0.26 –0.10 0.05 0.22

Dummy for female in project village –0.01 0.31 * 0.31 0.17 0.10

Dummy for Amawa district 0.24 ** 0.15 0.09 –0.12 –0.09

Dummy for Maharajganj district –0.13 –0.31 * 0.04 –0.24 –0.09

Dummy for rural –0.09 –0.08 0.14 –0.09 0.15

Score for community –0.29 0.65 1.09 *** 0.10 0.31

Economic status –0.11 –0.10 –0.13 –0.25 0.17

Dummy for poor –0.24 –0.31 ** –0.06 –0.30 * –0.28 ***

Household size 0.01 0.00 –0.01 0.01 0.02

Number of children –0.02 –0.05 * –0.01 –0.04 –0.04

Dummy for member of SIC 0.37 ** 0.45 *** 0.25 0.35 *** 0.61 ***

Dummy for female –0.05 –0.31 * 0.08 0.10 0.08

Age of the respondent –0.02 –0.02 –0.03 –0.01 –0.02

Age square 0.26 0.18 0.27 0.11 0.13

Schooling of respondent 0.44 *** 0.30 *** 0.04 0.03 0.03

Number of leaders known 0.18 *** 0.14 0.14 0.11 0.21 ***

Participation in traditional events 0.04 –0.13 ** 0.04 0.02 –0.03

Participation in non–traditional events 0.01 –0.08 0.10 0.16 0.14 ***

Blue collar skills of respondent 0.13 ** –0.09 0.01 0.07 0.28 ***

Mobilization skills of the respondent 0.07 0.12 –0.03 –0.03 0.08

Observations 1,148 1,148 1,148 1,144 1,083

Pseudo R-squared 0.06 0.07 0.03 0.03 0.06

Chi2 1,120.56 218.88 194.52 511.74 2,091.91


1 4 1

Poverty Targeting MechanismsBank projects use a variety of methods to targetproject beneficiaries. The most popular methodis geographic targeting. A project can focus ona poor region or province, usually using gov-ernment figures/criteria, or at a more local levelsuch as a municipality or community, often usingthe results of participatory planning processes.Another mechanism that has been increasinglyused, especially in social funds, is self-targeting.In this method the project supports basic infra-structure and services that are likely to be amongthe priority needs of poor communities or house-holds, such as basic schools, health posts, waterpumps, and similar infrastructure. A third mech-anism becoming popular with CDD projects issocial targeting, whereby the project targets par-ticular social groups (women, handicapped, dis-advantaged, small and marginal farmers, herders,and others).

The approach to poverty targeting for anyproject is largely based on the project objec-tives, availability of data, and institutional con-siderations. For example, the developmentobjective of the Natural Resources Managementand Poverty Reduction Project for Armenia (ap-proved in fiscal 2002) was to adopt sustainablenatural resource management practices and toalleviate rural poverty in mountainous areaswhere degradation is now reaching a criticalpoint. Provincial-level geographic targeting alonewas sufficient given the nature of the project: theproject selected two marzes in the poorest moun-tainous regions of Armenia. In contrast, the de-velopment objective of the Rural Developmentin Marginal Areas Project in Mexico (fiscal 1998)was to improve the well-being and the incomeof smallholders in about 24 targeted marginalareas—among the poorest of the country—

through sustainable increases in productivityand better food security. Clearly, project objec-tives required geographic targeting to identifymarginal areas, social targeting to focus on small-holders, and self-targeting to focus on subpro-jects that define basic needs of the poor (foodsecurity). The project used a combination of allthree poverty-targeting mechanisms. Over time,Bank-supported CBD/CDD projects are em-ploying increasingly sophisticated mechanismsto target beneficiaries.1

Poverty Targeting for CBD/CDD ProjectsCovered by the FieldworkThe evaluation studied in depth four targetedCBD/CDD projects, one each in Benin and Brazil,and two in India. One or more targeting mech-anisms were adopted in all of them.

• The Borgou Pilot Project in Benin applied ge-ographic targeting in 250 villages using a par-ticipatory diagnosis. The eligibility criteria forbeneficiaries included the definition of clearpriorities, as established by the Comité Villa-geois de Concertation, and the capacity of thecommunity to contribute financially to cer-tain types of investments. Specific emphasiswas to be placed on women’s groups and Fu-lani herders.2

• The RPAP project in Brazil’s Rio Grande doNorte was to apply three targeting methods: (a)geographic by poverty level and other charac-teristics of the municipality; (b) geographicwithin municipalities to target rural settle-ments and communities; and (c) community-based selection of poor beneficiaries andparticularly vulnerable groups by the projectMunicipal Councils. The communities them-selves, through their majority participation in

ANNEX O: POVERTY TARGETING


1 4 2

the Councils, were to determine where proj-ect resources would best be applied.

• In the Forestry Project in Madhya Pradesh,India, the target group consisted of tribal peo-ples and forest fringe villagers. The group wasto be to be identified by the government for-est department. The project incorporated spe-cific measures to safeguard the interests of thelandless and women.

• The Sodic Land Reclamation project in UttarPradesh, India, was to target small and marginalfarmers or previously landless allottees in thesodic land area identified by the governmentagency. Women were to be targeted and sup-ported as a special group.

Disaggregated data were not available formost projects (except for Brazil) to allow com-ment on the success with which the project tar-

geted the poorest and most disadvantaged. InBrazil, although the project was a targeted in-tervention, 136 of the 166 municipalities in thestate were covered. The justification for the vastcoverage was that all of the rural areas of the statewere deemed sufficiently poor to warrant in-clusion in the project. Only the state capital andits surrounding area were considered ineligible.Using a Municipal Human Development Index(MHDI), the evaluation attempted to assesswhether municipalities with lower MHDI re-ceived more Bank funds compared to munici-palities with higher MHDI. There was no relationbetween the level of MHDI and per capita in-vestment in the municipality as a whole. In otherwords, there appears to be no concerted effortto target greater resources (subprojects) topoorer municipalities.

1 4 3

This annex draws on the household surveys andthe qualitative data gathered in four project areasin the context of this evaluation to explore issuesrelevant to sustainability of subprojects fundedby Bank interventions.

BeninBoth the PAMR and AgeFIB projects providedsupport for construction of small infrastructure,a very large percentage of which was primaryschools, in hundreds of communities. Typically,the government pays the salaries of certifiedschoolteachers for village primary schools. How-ever, the fieldwork in the Borgou region revealsthat the government has not been in a positionto provide for paid certified teachers in the nu-merous schools that have been constructedunder both the projects. Interviews with villageleaders revealed that over 50 percent of PAMRschools and 80 percent of AgeFIB schoolteach-ers were community teachers, in comparisonwith comparator villages, where only a third ofthe teachers were community teachers. Becauseof a shortage of government teachers, the com-munities have been forced to hire teachers andpay their salaries from their own resources. Thefieldwork was undertaken shortly after the proj-ects closed, and new infrastructure generallydoes not entail significant maintenance costs. Itis not clear whether poor communities will beable to bear the cost burden of maintenanceand teachers’ salaries from their own resourcesover the long run. The expectation among thecommunities is that the government will be re-sponsible for the salary portion of the provisionof education services. The majority of the house-hold survey respondents considered repairs tobe a responsibility of parent-teacher associa-tions, but considered staff salaries a central gov-

ernment responsibility. These interventions haveput a resource burden on the communities,which they may not have anticipated and mayfind difficult to sustain.

There is also the issue of the quality of educa-tion services imparted, which requires coordina-tion with the education department in terms ofadherence to a centrally planned curriculum,among other things. In the absence of certifiedschoolteachers, it is not clear how education inthese schools will conform to a national standard.Poor communities may not be able to pay adequatesalaries and benefits to attract qualified teachers.

BrazilIn Brazil, project communities benefited fromthree types of investments: water supply (20communities), irrigation (three communities),and small bridges (two communities); while thethree comparator communities benefited froma government water pipeline. The comparisonbetween project and comparator communitiesis restricted for the analysis on sustainability tothose project communities that received watersupply investments.

Water supply investments: The RPAP financeda variety of water supply systems, including wells,cisterns, and small dams. While the majority ofthese systems are community-based, and hencerequire the community to collectively organizefor its O&M, three communities benefited fromhousehold-based water supply systems, such ashousehold water tanks and boxes, whose O&Mfalls solely on the individual household. Thesethree project communities were dropped fromthe comparative analysis between the projectand the comparator group. As figure P.1 shows,a larger share of respondents in comparator

ANNEX P: EVIDENCE FROM FIELDWORK ON SUSTAINABILITY


1 4 4

communities than in project communities paiduser fees, and were satisfied with the services pro-vided and with O&M of their water system. How-ever, these aggregated figures hide pronounceddifferences among project communities. As thequalitative data reveal, while almost half of themcollect monthly fees to cover the cost of theelectricity, and in some cases maintenance ofthe equipment, a few have yet to set up adequateO&M systems, and others have transferred thesystem to the state water company, which is nowresponsible for O&M. The quantitative data re-flect this great variation among project com-munities. The percentage of respondents that payuser fees varied widely, from 100 percent in fourcommunities to zero percent in six communities.Similarly, satisfaction with the services providedand with O&M varied, respectively, between 8and 94 percent and 5 and 100 percent. Con-versely, variation between comparator commu-nities is very limited, and all display similarlyhigh level user-fee payment and satisfaction withthe services provided and with O&M.

Irrigation investments: Fewer than half ofthe respondents who are members of the threecommunity associations (CA) that benefited fromirrigation investments pay user fees.1 Only 37 per-cent of them rated the services provided asgood, while they hold divergent opinions onO&M, with 42 percent rating it as poor and an-other 42 percent rating it as good. These ag-gregate figures, however, hide pronounceddifferences among the three CAs. In one of them,the subproject is paralyzed by the high cost ofelectricity to operate the pump, while anotheris temporarily suspended by a shortage of water.A large share of the CA members interviewedwhere the irrigation system is functional andwhere it has been temporarily suspended ratedthe service provided and O&M as good. While themajority in the former pay user fees, half in thelatter do so. Small bridges: Fifty-seven percent of the re-spondents in the two communities that benefitedfrom the construction of a small bridge are sat-isfied with the service provided by the infra-

Figure P.1: User-Fee Payments and Satisfaction Rates with Water Services in Projectand Comparator Communities in Rio Grande do Norte

Resp

onde

nts

(%)

Pay user fees Satisfied w/services Satisfied w/O&M

57%

***

*** ***95%

89%

59%

89%

59%

0

40

20

60

80

100

Project [N = 560] Comparator [N = 117]

Note: Significance level based on test of proportion. * Significant at 10%; ** significant at 5%; ***significant at 1%.

1 4 5

structure, while 52 percent rated O&M as poor.None of the respondents pays any kind of feesfor the upkeep of the small bridges. As the qual-itative data reveal, the choice of these invest-ments, which did not result from a process ofbroad community participation, is not perceivedas a solution to one of their main priorities.

Madhya PradeshThe World Bank project aimed to assist imple-mentation of the government strategy for de-velopment of the forest sector using joint forestmanagement (JFM). The Bank project also pro-vided improved information, extension services,and complementary investments in communalinfrastructure to give villagers an incentive to co-operate. The appraisal document identified theneed for the Bank to support the forest sectorin the state for about 10 years with investmentsthat could total more than US$200 million. Ac-tual support provided was about a quarter of thisamount over a period of four years. The Bank didnot follow through with a second interventionin the state.

It appears that this is the main reason why ithas not been possible to build the elements ofsustainability in this short time. While house-

hold data indicate that Bank JFM villages seemto have experienced less of a decline in forestcover than comparator communities (table P.1),this gain has not been sufficient to provide ad-equate returns to the communities. At the timeof the survey only a small percentage of the ben-eficiaries from Bank JFM communities reportedcollection of forest products, a large percent-age reported having less access to forest prod-ucts than previously, and a very small percentagereported collection of the forest products for in-come (table P.2). Focus group sessions revealincreasing hardship and lack of income-gener-ating activities in the Bank JFM villages. Further,communities that have received support from theBank appear to be receiving less support fromthe government for other development activities.Nearly two-thirds of the Bank beneficiaries reportno micro-project in their village, compared witha third of government beneficiaries.

Uttar PradeshThe World Bank project aimed to remove sodic-ity of land. The project also provided gypsum andother inputs, helped construct boring for irriga-tion, set up formal extension services, and intro-duced loans for cropping. Maintenance of drains,especially the main drains, is critical to the sus-tainability of reclaimed lands. Most respondentsfrom the Bank project communities character-ized maintenance of the drains as “bad,” and saidthat O&M of the drains had deteriorated overtime (figures P.2 and P.3). Focus group sessions re-iterated these findings, emphasizing dysfunctionalmain drains. An interesting fact revealed by theopen-ended discussion was that most farmers no

ANNEX P

Table P.1: Declining Forest Cover(percent)

Forest Less than Same as More thancover before before before

Bank-JFM 35 10 22

Govt.-JFM 62 14 16

Non-JFM 90 3 1

Table P.2: Forest Product Use

Percent reporting collection of forest products Percent reporting decline in collectionForest product Bank Government Bank Government

Fuelwood 65 94 47 52

Fodder 16 22 15 13

Tendu 47 54 39 34

Mahua 36 43 29 26

Amla 8 15 13 16

Grass 6 11 10 15


1 4 6

longer have field drains. Farmers have used thatpart of the land for cropping and are using irri-gation channels for drainage purposes.

The Irrigation Department is responsible formaintaining the main drains, the village organi-zation is responsible for link drains using inter-nal funds, and farmers are responsible for fielddrains. The OED assessment notes that the im-plementing agency believed that political pres-sure from farmers would ensure that thegovernment provides sufficient resources to theIrrigation Department for this activity. However,most beneficiaries are not even aware of thecritical importance of drainage for containingsodicity, or that it is the responsibility of the Ir-rigation Department to maintain the drains. Over80 percent of the respondents thought that con-tinuous application of gypsum and water supplyfor irrigation will prevent the land from becom-ing sodic. Further, fewer than 4 percent of therespondents were aware that the responsibility

of O&M of main drains lies with the Irrigation De-partment, and that responsibility of O&M of linkdrains lies with the village organization (table P.3).

The project constructed many borings, andmost respondents agree that O&M of that bor-ing is the responsibility of the pump owner. How-ever, most water pump owners indicated thatthe money received from farmers for water rightswas insufficient to maintain the pump. At thesame time, a majority of the respondents who paythe pump owner for water for their fields be-lieve they pay enough to cover the O&M of thepump. Eight of the 18 village leader interviews in-dicated that boreholes have not helped improveirrigation; and the other 10 raised concerns aboutthe continued flow of service from the boringsbecause of erratic water supply, prohibitive costsof diesel, temperament of the pump owners,and deteriorating conditions of some boreholes.

The project committee (site implementationcommittee) setup is functional in only one of theproject villages, according to interviews with proj-ect committee members in each of the project vil-lages. Focus group sessions revealed that mostrespondents were also unaware of the existenceof any MK/MMK.

Field

Link

Main

0 5025

Percentage of respondents [N = 757]

75 100

Good Fair Bad Do not know

Type

of d

rain

s

Figure P.2: O&M for Drains Is Bad Figure P.3: O&M for Drains Deteriorating

Field

Link

Main

0 5025

Percentage of respondents [N = 757]

75 100

Improved Same Deteriorated Do not know

Type

of d

rain

s

Table P.3: Many Unaware of Who Is Responsiblefor Drains (percent)

UPBSN/ Bene-Irrigation Village ficiaries/ Do not

Department leaders villagers know

Main drains 15 20 14 36

Link drains 9 17 28 32

Field drains 5 9 51 24

1 4 7

The Safeguards Policy Review is one of twothematic studies conducted for the OED evalu-ation of the World Bank’s support for commu-nity-based and community-driven development.The study reviewed project appraisal, supervi-sion, and completion documents for a sample of84 projects to assess their compliance with theBank’s safeguard policies.1 The desk reviewswere supplemented by interviews with task teamleaders, the Quality Assurance and ComplianceUnit team, and Regional safeguards coordinatorsin selected cases. Selected items of direct rele-vance from the literature on safeguard policiesand CBD/CDD projects were also reviewed.

Detailed findings on each project were con-densed into a set of ratings on quality of com-pliance and analyzed with respect to:environmental assessment (EA) category, sec-tor, Region, project type, and age. Findings andrecommendations were developed from thisanalysis. Examples of best practice and missedopportunities were also identified. A special re-view was made of the 10 CBD/CDD projects inBenin as part of a country study. Finally, 473headquarters and field staff were sent a ques-tionnaire that included questions on safeguardissues to assess Bank performance in the area ofCBD/CDD projects and how Bank capacity to un-dertake CBD/CDD interventions has evolved.2

Because of the broad definition of CBD/CDDused, which includes some projects with onlyminor CBD/CDD aspects, the sample projects areheterogeneous. Therefore, the projects were di-vided into two broad groups:

• CBD/CDD with subprojects (CBD/CDD-S)(76 percent of the sample): CBD/CDD proj-ects for which the majority of investment fund-ing is for a large number of small and scattered

subprojects. Such subprojects may be multi-sectoral or may be limited to a single sector,such as health or education.

• Other Projects (CBD/CDD-NS): Projectsthat have CBD/CDD aspects or components butdo not fit the definition above.

Quality at EntryThe study found that the EA category was cor-rectly assigned for 80 percent of the sample proj-ects but, given the nature and extent of potentialimpacts, it was judged that 9 percent of CategoryBs should have been As and 38 percent of Csshould have been Bs (see box 5.1 in Chapter 5for definitions of these categories). The qualityof appraisal was rated moderately satisfactoryand above for 70 percent of the total sample, withnewer projects scoring higher. The quality ofEA documents was mixed: only two of the fiveA projects and 74 percent of B projects wererated moderately satisfactory or above. The spe-cial requirements for IDA B projects with a sep-arate EA report were generally observed. Thequality of Resettlement Action Plans and In-digenous Peoples Development Plans was gen-erally high. The number of cases where thepotential applicability of one or more of the safe-guards policies should have been discussed butwas not was high—about equal to the caseswhere policies were triggered. Compliance withthe public disclosure and consultation require-ments of the safeguard policies was good forresettlement and indigenous peoples issues, butless so for EAs. In contrast, provisions for capacitybuilding were well developed, with monitoringsomewhat less so.

Overall, quality at entry was rated moderatelysatisfactory and above for 70 percent of the sam-ple. The small group of FI (Financial Intermedi-

ANNEX Q: SAFEGUARD THEMATIC STUDY: A SUMMARY

ary) projects were rated much better than aver-age, while As were distinctly worse than average.Newer projects are markedly better than the oldergroup—81 percent versus 54 percent moderatelysatisfactory and above. CBD/CDD-S projects alsoscore better than CBD/CDD-NS—77 percent ver-sus 50 percent moderately satisfactory and above.Adjustable Program Loans (APLs) scored some-what better than conventional projects. In termsof Regions, Africa, Europe and Central Asia, andthe Middle East and North Africa have the high-est percentages of moderately satisfactory andabove, while Latin America and the Caribbeanand South Asia have the lowest. Among sectors,transport, social, and environment had the bestresults, while the ratings for the rural and urbansectors were well below average.

The study found that internal guidance on theuse of the FI category and, consequently, thepractices of the Regions have not been entirelyconsistent since that category was introduced inJanuary 1999. Discussion about the assignmentof the FI category for most CBD/CDD projectsis ongoing, but full guidance has yet to be issued.

Quality during ImplementationDespite format changes in the Project Status Re-port (PSR) that encourage detailed reportingon the implementation of safeguard measures,such reporting remains sparse and inadequate.This is true especially for Category A projects,which should receive particular scrutiny duringimplementation, and FIs, where the real work ofscreening subprojects and designing mitigationmeasures falls into the project implementationphase. There was no specialist follow up forcases where the dam safety and pest manage-ment policies were triggered. There was almostno reporting on capacity building or monitoringsystems. Most of the Implementation Comple-tion Reports (ICRs) were also less than satisfac-tory on reporting safeguard compliance, withthe majority containing no discussion at all. Ofthe four Project Performance Assessment Re-ports (PPARs) available for this sample, two pro-vided good analysis of safeguard issues, while theother two said nothing.3

The overall quality of implementation wasrated moderately satisfactory and above for only

35 percent of cases, with A projects at 40 percent.While newer projects score much better thanolder ones, at 44 percent moderately satisfactoryand above, they are still far from meeting Bankstandards. As at appraisal, CBD/CDD-S projectsare distinctly better than CBD/CDD-NS (38 per-cent versus 25 percent). In contrast to the qual-ity at entry ratings, East Asia and the Pacific andSouth Asia score highest for quality of supervi-sion of safeguard issues, with Latin America andthe Caribbean and the Middle East and NorthAfrica scoring lowest. Sectorally, water supplyand sanitation and transport had the best record,with energy, mining & private sector, social, andeducation scoring lowest.

Overall Project QualityWhen the ratings for entry and implementationare combined, the overall proportion of proj-ects moderately satisfactory and above is 70 per-cent, the same outcome as for quality at entry.However, this disguises the fact that the proj-ects rated (fully) satisfactory and above slip from52 percent at entry to 17 percent when imple-mentation is considered, and those rated unsat-isfactory and highly unsatisfactory climb from 20percent to 33 percent. Regardless of statisticalquirks, the result of 70 percent falls well belowBank expectations. Between EA categories, theresult for the small FI sample is 100 percent mod-erately satisfactory and above, while Bs and Cs areclose to the average. The main concern is the verylow percentage of A projects (40 percent) that isbeing handled well. Nevertheless, newer proj-ects are closer to compliance than the oldergroup—87 percent to 45 percent. While 87 per-cent is an encouraging result, it includes 65 per-cent in the moderately satisfactory category,indicating considerable room for improvement.The Europe and Central Asia and Middle Eastand North Africa Regions achieved the best resultsoverall, with the other Regions close to eachother at a lower level. Among the sectors, trans-port, social, and environment scored highest,and urban lowest.

The review also found that difficulties exist inapplying the safeguard policies to multicompo-nent projects and that the potential for cumu-lative impacts from large numbers of small


1 4 8

subprojects is sometimes overlooked. The im-portance of adequate collection and disposal ofmedical waste was not recognized in some ear-lier health projects, but recent practice has im-proved.

The review found that 6 of the 11 projectsrated unsatisfactory on overall quality were in theBank’s largest borrower countries.

The survey of Bank staff showed that only aquarter of respondents agreed that resourcesfor addressing safeguard issues were sufficient,though about half felt that current policies wererelevant for CBD/CDD projects.

ConclusionsFour broad themes emerge from the analysis:

• Although there has been clear improvement,safeguard compliance in CBD/CDD projectsdoes not yet fully meet Bank standards.

• While quality at entry needs improvement,safeguards compliance during implementa-tion warrants much greater attention by theBank and borrowers, and may indicate theneed for greater allocation of supervision re-sources.

• Gaps in the compliance system may be lead-ing to significant environmental and social im-pacts, which may not be caught by themonitoring and reporting systems typicallyused.

• The Bank appears to have particular difficultyin ensuring safeguard compliance in its largestborrowers.

Based on its findings and conclusions, theReview makes the following recommendations.

At the level of policy development, Regionalcoordination, staff guidance, and training:

• Guidance is urgently needed on the appro-priate EA categorization of CBD/CDD projects,especially on the use of the FI category and onthe special requirements for IDA B projectswith a separate EA report.

• Training of task teams in the application of thesafeguard policies to CBD/CDD projects shouldbe intensified and should rely heavily on “best

practice” examples, of which this Review hasidentified some.

• A thematic study of the environmental and so-cial implications of changes in land use may bewarranted.

• In any planned revision of Operational Policy4.01 (and/or the other safeguard policies), spe-cial attention should be given inter alia to:streamlining the IDA B with separate EA re-port procedures; defining “financial interme-diary”; dealing with multicomponent A projects;defining standards for supervision and com-pletion reporting on safeguards compliance;and a possible mandated role for the Regionalenvironmental and social units in the supervi-sion of A projects.

• The experience of the Poland: Rural Develop-ment Project should be thoroughly reviewedfor examples of the issues that may arise fromthe use of country systems for safeguards com-pliance.

At the level of Regional safeguards compli-ance assurance:

• Evaluating the recent transfer of sign-offauthority for Category B and FI projects inlight of the above findings and those of otherOED studies, including staffing and budgetingissues.

• Ensuring full compliance at entry with safe-guard policies, especially in the Bank’s largestborrower countries.

• Ensuring that policies other than OperationalPolicy 4.01 are triggered in appropriate casesand necessary follow up actions taken.

• Obtaining resources for and carrying out spe-cial reviews of safeguard compliance forCBD/CDD projects under supervision, withspecial attention to the adequacy of agreedprovisions, the effectiveness of their imple-mentation, and the success of capacity build-ing and monitoring activities.

• Developing standard document packages(cf. procurement documents) for safeguardinstruments such as EAs, EnvironmentalManagement Plans, Resettlement ActionPlans, and Indigenous Peoples DevelopmentPlans.

1 4 9

ANNEX Q


1 5 0

• Reviewing the potential for delegation of safe-guard management authority to national agen-cies.

At the level of project development,approval and supervision:

• Identification of potential safeguard issues, forexample, by use of Strategic EnvironmentalAssessment.

• Mainstreaming environmental and social safe-guards into the preparation process forCBD/CDD projects, for example, in developing,planning, programming, and monitoring pro-grams, as well as staff training.

• Collaborating closely with the Regional envi-ronmental and social units in assigningEA categories appropriately and in usingthe Integrated Safeguards Data Sheet as a“contract” for actions needed between theproject concept development and appraisalstages.

• Being sensitive to the special disclosure and

consultation requirements of Categories A andB (and agreeing with management on com-monsense waivers where process requirementsmay impede project quality or timeliness).

• Obtaining sufficient financial and staff re-sources to allow adequate supervision of theimplementation of agreed safeguard meas-ures, especially for As and Bs with an EMP,including periodic review of a sample of sub-projects.

• Using the comment boxes in the PSR form toexplain the reasoning behind the ratings given,the progress of capacity building or monitor-ing programs, and any unforeseen problemsencountered, with special attention to CategoryA projects.

• Using the Midterm Review to look in greaterdepth at safeguard compliance, with the as-sistance of environmental and/or social spe-cialists.

• Following the guidelines for the ICR in re-porting on safeguard compliance at projectcompletion.

1 5 1

Robert ChambersThe OED team deserves congratulations on theeffort put into the monumental task of this eval-uation. The subject matter is vast, scattered anddifficult to assess. The documents made availableto the Advisory Committee and our discussionsin December 2004 made it clear that the re-search was carried out with conventional rigourand care, and that the conclusions are credible,based on and emerging from careful and bal-anced analysis of the evidence. The extensive andvaluable literature review also drew on and col-lated much other relevant experience. Other aidagencies would do well to conduct evaluationssimilar in their independence, breadth and depthas those of the OED. The conclusions, as far asthey go, resonate with and are confirmed by myown experience. The recommendations, how-ever, fall short of what the evidence implies.

I note that CBD/CDD approaches withHIV/AIDS are not included and are the subjectof a separate evaluation. Given the delicacy andcomplexity of AIDS-related issues, and my par-ticipation in a Bank-led workshop on CDD andHIV/AIDS in Africa, I expect its findings to be evenmore negative than those presented in this pres-ent evaluation.

There is much in the report which merits en-dorsement, presenting aspects which are bothpositive and negative. Rather than list suchpoints, let me highlight four issues which qual-ify the conclusions of the evaluation. I do notmake these comments with any pleasure, orlightly, but given what I have experienced, andgiven the Bank’s commitment to professionalismin the service of poor people, I have to makethem. All four suggest that this evaluation isover-favourable:

1. Picking winners. This is not a criticism of theselection of the large sample of projects whereI agree with the points about this made by Nor-man Uphoff. There are two other points.

First, success with “indigenously matured or-ganisations” is the result of no doubt rationalcherry-picking by the Bank. These were out-standing organisations with exceptionally highcalibre, continuity and commitment of man-agement with charismatic and inspiring leadersand which had existed for a decade or more andalready successfully gone to very large scale: forexample, AKRSP Pakistan, the NDDB in India, andSEWA in Gujarat. They were highly successful be-fore the Bank became involved. They are cor-rectly distinguished as a separate category fromother CBD/CDD. Their performance is irrele-vant to the evaluation of other CBD/CDD proj-ects, which are by far the majority. In earlierstages of their development they did not need,and might have been hampered by, support fromthe World Bank.

Second, the Matrouh Project in Egypt is world-renowned as perhaps the most famous Bankflagship participatory project. It is several timesmentioned. We know that it benefited from ex-ceptional continuity of exceptional staff andhigh-level support from James Wolfensohn. Buteven it, one of the most favoured and best ex-amples that could be found, is noted in this eval-uation for its serious downsides, including theeffects of creating a parallel organisation andoverlooking changes in land tenure whichharmed the Bedouin. If one of the very bestcherries has such flaws, one may wonder aboutthe rest.

ANNEX R: ADVISORY COMMITTEE COMMENTS

2. Positive bias. Any evaluation of Bank projectsinvolving interviews with Government staff andNGO beneficiaries of Bank funding is vulnerableto positive biases. The power and prestige ofthe Bank, the careful respect with which it istreated, and the tendency to try to please withfavourable feedback, present systemic difficultiesin knowing what is really happening. All powerdeceives (see Chapter 5 of my book Whose Re-ality Counts?). However careful the research,there will always be questions about prudent, def-erential and self-serving responses. It is as muchas some officials’ jobs are worth to say anythingnegative about the Bank.

3. Hidden negative externalities. The fol-lowing negative effects are either not mentionedor understated. The fact that they are half-hid-den to conventional research does not makethem any less plausible or less real. Some ofthem are part of another and more inclusive re-search agenda.

• Diversion of progressive NGOs (both INGOsand NNGOs) from rights-based and empow-ering activities which would do more for poorpeople than the provision of infrastructurewhich does less, and/or may even be negative,and/or may drag the NGOs back into activitiesthey were attempting to move on from. (36 percent of projects had some form of NGO in-volvement). In pro-poor terms, this is likely toreduce NGO additionality, leaving poor peoplenet losers.

• Undermining other more participatory, lesstarget- and disbursement-driven, less infra-structure-focused, and more sustainable pro-grammes supported by other organisations inneighbouring areas. (“Why should we do itourselves when our neighbours are getting somuch done for them or for free?”)

• Diversion of government recurrent funds,staff and materials from other places and ser-vices (schools, clinics etc) to the new infra-structure, with hidden costs to services inthose other places. With schools and clinics,for example, resources are most likely to bediverted to communities which are accessibleto government and Bank staff inspection, to

show success. Where government staff and re-current funds are, as so often, limiting, this willdeprive poor communities that are less ac-cessible.

• Risks and costs associated with top-down time-bound disbursement-driven capital projects.These include scope for petty and not so pettycorruption, and the proliferation of oppor-tunistic NGOs as noted on page 35.

• The long-term disempowering effects of de-pendence and disillusion created at the com-munity level (see, e.g., box 4.2). Communitiesbecome, as so many have, less self-reliant andmore inclined to lobby, beg, and wait

These are general tendencies. There will beexceptions. But together on balance they meanthat the findings of the evaluation should bemore negative. Given the goodwill, energy andcommitment of many Bank staff, I regret havingto say this. But it is quite possible that, overall,the Bank’s CBD/CDD initiatives do more harmthan good to poor communities and people. Inaddition there are the opportunity costs of al-ternatives foregone.

4. The comparative disadvantages of theBank. While the conclusions of the report fol-low from the evidence and analysis, they do notadequately confront the comparative disadvan-tage of the Bank with CBD/CDD, nor the fullrange of what would be required if performancewere to justify continuing to try to support it. Thisis alarming, especially when CBD/CDD-relatedlending, far from prudently diminishing, is in-creasing beyond its already remarkably highlevel.

The comparative disadvantages are institu-tional and paradigmatic and related to:

• Reliance on loans, the future repayment ofwhich may impact adversely on governmentservices and so on poor people. If loan-basedrather than grant-based projects are to be jus-tified, repayments will be at the cost of othergovernment expenditure. The bar, therefore,has to be higher.

• Disbursement pressures and the typical one-year sub-project cycle (pp. 20–21) with top-


1 5 2

down one-shot interventions, an approach an-tithetical to participation and to assuring ben-efits to those who are poorer. We know and donot need to learn again how badly this works.

• The management and staff-intensity of em-powering and participatory development. Theextra cost of preparation of CBD/CDD projectsis only 10 percent higher than non-CBD/CDD.For effective pro-poor participation, it wouldneed to be far higher than this.

• The staff incentive system of the Bank whichrewards high and fast disbursements. Thiswas a major factor which emerged from a par-ticipatory workshop for task managers whichI facilitated a few years ago. Nothing I haveheard suggests that this has changed signifi-cantly.

• Inability to learn and change. That the ratingsof CBD/CDD projects are stagnating suggeststhat institutional learning and change are nottaking place. A likely reason is that the Bank isnot looking hard enough at itself or is simplyunable to perceive, learn, and change.

These factors combine to disable the Bank,making it inherently difficult for it to do wellwith CBD/CDD. The question then is whetherthey can be changed.

The report correctly points to the need forradical institutional change in government bu-reaucracies: “The literature shows that the in-stitutionalization of a CBD/CDD approachrequires a radical reorientation in the way gov-ernments and bureaucracies operate.” This ap-plies if anything more to the Bank itself, as thedominant partner, than to governments andtheir bureaucracies. This is not rocket science.It is common sense and common experience. Inpractice, the disabling culture, incentives, pro-cedures and imperatives of the Bank are passedon “downwards” to governments and NGOs. Itis no good saying “Do as I say but not as I do.”The Bank may not be able to become more par-ticipatory. But unless it does, it cannot expect theCBD/CDD it funds to be cost-effective in em-powering and benefiting poor people. Thismeans that the Bank itself must walk the talk, andtake on board “physician heal thyself ” and “dono harm.”

The Recommendations The recommendations in this final version ofthe evaluation fall far short of what is demandedby the evidence. If the Bank is serious aboutpoverty and empowerment, more radical actionis required. In effect, the recommendations asthey stand leave the door open to going on withmore of much the same. The evidence of thisOED evaluation, combined with other studiesand insights, shows the CBD/CDD initiatives ofthe Bank to be of such questionable value thatthe approach now should be damage limitation,intensive learning and finding out whetherchange is possible in the Bank. I hesitate to saywhat I believe the Bank should do, but the stakesfor poor people and communities are so high andon such a scale that it would be wrong for me notto do so. I have agonised over this. And I recog-nise that there is no way I can assess fully the im-plications or modalities for what follows. But onthe basis of the evidence of this evaluation andof other experience, my own best judgement isthat it the Bank should now, and decisively:

• Rein back on and/or slow down existingCBD/CDD projects, where this is legally andethically feasible.

• Impose a moratorium on new ones.• Learn more about what happens and what

might be made to happen by selecting on-going projects for intensive learning throughaction research, including investigating hid-den externalities.

• As part of this examine the Bank itself—its cul-ture, procedures, norms, incentives and be-haviours—and its impact on governments,NGOs and communities, and analyse the con-tradictions between these and empowering,pro-poor community development and howthese play out.

• And then review how the Bank must changeif the short and long-term effects for poor peo-ple, communities and countries of Bank-drivenCBD/CDD are to be positive and to justify thecosts, and whether and how such change couldbe achieved.

Robert Chambers 5 September 2005

ANNEX R

1 5 3

Norman Uphoff This evaluation is originally planned to evaluateWorld Bank support for community-drivendevelopment (CDD) projects. However, whenreviewing earlier drafts of this report, this and theother reviewers saw some problem with char-acterizing these projects “community-driven.”This terminology had been was introducedwithin the Bank to distinguish newer, more par-ticipatory project initiatives from what werebeing called “community-based” developmentprojects. The term “CDD” seemed rathergrandiose to the reviewers for what was actuallyhappening in these projects. They were hardly“community-driven” when the project design,what kinds of things could be done, within whattime frame, on what financial terms, were all de-cided unilaterally by Bank staff. The projectsthemselves were not open to local participatoryinputs, only subprojects. The main decisionsleft to communities were whether or not theywould make a proposal to gain access to Bankproject funds for something they wanted to do(within the non-negotiable framework set byBank or government personnel) and how theywould carry out the work once it was approved.This issue of how these projects should be de-scribed does not make the evaluation that hasbeen done less relevant or meritorious, since thepurpose of the assessment was to learn moreabout the Bank’s approach to this kind of de-velopment, no matter what it is called.

The projects have in common an aspiration(and enabling provisions) for delegating to com-munities (or their representatives) responsibil-ity for taking initiative to plan and implementcertain improvements in infrastructure and/orservices at community level. But “community-driven” means only community-initiated, -im-plemented and -managed, within externally-setparameters. The first point in the Executive Sum-mary makes clear that only participation in “sub-projects” is covered by the study. There is nothingwrong with this, and it can be preferable to moreconventional top-down efforts for local devel-opment, depending on results.

The evaluation of results found that whatwere called CDD projects, meeting certain cri-teria for a community role in activity initiation and

implementation, did not perform much better,and sometimes less well, than conventional proj-ects with similar objectives and environments.This could be because there is nothing inherentlysuperior in CDD-type project, or what werecalled CDD projects were not planned and car-ried out in ways that gave them any advantage.From my experience of over 30 years working onparticipatory development, I would think thelatter explanation is more apt, and indeed, theevaluation team identified many ways in whichthe way the Bank operates “got in the way” of car-rying out the projects included in this study sothat they did not perform noticeably better. Pos-sibly if they had been carried out properly, theymight still have shown no performance advan-tage. But we can’t know that until the CDD con-cept has at least been properly introduced. CanCDD deliver more benefits to the poor? Morecheaply? More reliably? We don’t know and can’tsay because it hasn’t really been properly tried.

The overall hypothesis guiding this evaluationwas, essentially, that CDD projects (as supportedby the Bank) would produce more and betteroutcomes. The evidence assembled and assesseddoes not support the hypothesis; so we are leftwith the null hypothesis. However, as someonewho has worked on participatory development,several times as a consultant for the Bank on thissubject, I must say that I am not surprised, be-cause I have seen the Bank’s efforts, howeverwell-intended, to be continually well behind the“state-of-the-art.” Although there are some waysin which some of the CDD projects have givensuperior results, there is no basis for concludingthat on average, the Bank would get better re-sults by doing more of its projects in what it hasbeen considered and created as a CDD mode.

This said, the converse is also true: therewould be no significant degradation of the Bank’sportfolio by expanding CDD efforts even as cur-rently supported by the Bank. The increasedcost associated with CDD project design andsometimes in implementation can probably bejustified by some subsequent cost reductionsto the government as communities take more fi-nancial responsibility, or by more rapid imple-mentation once the planning has beencompleted, or from better post-project utilization


1 5 4

and maintenance of project services and facili-ties, not all the time, but sometimes. Such aconclusion is based on the evidence provided inthe evaluation. It does not take into account thevarious external costs that Robert Chambers de-lineates very persuasively in his comments on theevaluation. Such broader considerations that gobeyond the terms of reference and the frame-work of this evaluation could justify his sugges-tion that there be a moratorium on CDD effortsuntil the Bank figures out how to pursue themmore effectively.

There is little to find fault with in the waythat the evaluation itself has been done, or in theway that it is reported. The evaluators have usedboth quantitative methods and qualitative de-scription and summarization quite satisfactorily.There is not much to critique on methodologi-cal grounds. However, this does not make theevaluation itself irreproachable, because thestate-of-the-art followed has some shortcomingsthat should be noted. The following commentsrefer to the way in which formal evaluations aredone these days by institutions like OED, ratherthan to the way that this particular evaluation wasdone.

Sample size and statistical significance.One area where the evaluation could have beenmore informative is to have disaggregated theCDD sample more in terms of the kinds of de-velopment work being fostered. This was notdone because that would have reduced sub-sample sizes and impaired the scope and valid-ity of statistical analysis. However, by lumpingdiverse experiences together in one pool, ormaking only a few gross disaggregations, the va-lidity and value of the generalizations are di-luted for the sake of being able to use statisticalmethods and offer assessments of (statistical) sig-nificance. If the analysis had been done onsmaller but more homogeneous subsamples,what could be said with statistical confidencewould have been reduced—but we would knowmore about whether CDD approaches (as theBank has been supporting them) would be moreproductive and sustainable in some sectors thanin others, in some Regions, within certain timeperiods, etc. The OED team made a defensible

decision to analyze internally heterogeneoussamples. But more “less rigorous” analysis couldhave revealed some operationally useful insights,not being “shackled” by statistical analysis con-ventions.

Ontological assumptions. In this analysis, asin most such evaluations these days, there is acertain “reductionism” that obscures more thanit reveals. There is an implicit assumption thatthere is an “essence” of CDD that is the same inall cases that are characterized as CDD, and thatthis quality is rather equally represented in allsuch cases. Then there is a concomitant as-sumption that this “essence” of CDD, whateverit is, has independent, rather than contingent,causal effects. Such assumptions are commonthroughout development studies, and indeedin most social science analyses, which try todraw broad generalizations (about often etherealcharacteristics) rather than stick to more disag-gregated, concretely specified assessments. Thiskind of reliance on abstractions is at the root ofmany of the failures in development efforts, andit is one reason why so much of our social sci-ence is so irrelevant to real-world decision mak-ing and action. This comment is not a critiqueof this evaluation, but rather of the broader en-terprise.

The Bank’s CDD approach has sought to bringmore participation and more flexibility into WorldBank projects. This is commendable and movesthe Bank toward what can be thought of as the“state-of-the-art.” This evaluation documentsthat the Bank’s procedures, staff incentives andorientation, borrower-country predispositionsand capabilities, and still other factors have kept“CDD” projects from achieving the degree ofcommunity assumption of responsibility thatwas anticipated and hoped for, and the amountof improvement in the lives of poor people thatwas expected. What is evaluated in this report is,at most, “partial CDD,” and some of the casestudies suggest this was even “minimal CDD.”When local people say that the project did notaddress their priority needs (see 3.9) or theyare not willing to maintain the facilities or ser-vices beyond life-of-project (4.30), it does not ap-pear that this is an evaluation of truly

ANNEX R

1 5 5

“community-driven” processes. Such processesare directed to meeting priority needs, and ifwhat is created is of value to people, they will findways to maintain the facilities or services some-how, anyhow.

An earlier draft of the report described theBank’s Adaptable Program Loans (APLs) andLearning and Innovation Loans (LILs) as “allow-ing (!) communities greater choice in the selec-tion of activities” and as “providing themopportunities to control investment decisionsand resources during project implementation.”This underscores the extent to which the processis in Bank or borrower government hands. Theseare community-proposed (not community-se-lected) projects, and they are community-man-aged (within often complicated and burdensometerms).Few people would understand this to be“community-driven development.”

The concept of “community” that under-lies this effort is also questionable. Endnote 1 toChapter 1 in the report says: in a World Bank proj-ect, a community is considered a “unified, organicwhole.” This is dubious thinking, suggesting thatthe Bank has learned little from the hundreds ofsocial science assessments of “community” writ-ten over the past half century. These havestressed the pitfalls of internal divisions, con-flicting interests, etc. among any given set ofhouseholds that is delineated on the basis ofresidential area. Even when there are ethnic orother kinds of homogeneity, village residentsoften find personalistic or other bases on whichto factionalize. Some communities will have ahigh degree of solidarity and significant poten-tial for collective action. But this is a variablerather than a defining characteristic. (The Bank’sown efforts to bring some rigor to the conceptof “social capital” have help to clarify this mat-ter, because this concept does not take village sol-idarity for granted, but rather examines differentlevels, kinds, and activation.)

Assuming that the Bank is serious aboutpoverty reduction and empowerment, two ofthe most attractive justifications for its existenceand for its support by member governments,there is much more that could be done to im-

prove prospects for better CDD outcomes. Myown experience with a USAID project introduc-ing participatory irrigation management in SriLanka in the early 1980s has demonstrated thatthere can be significant measurable benefitsfrom truly community-driven development. Theincremental rice production possible in a singleseason, the 1997 dry season, alone covered 4–5times the total cost of the participatory compo-nent, for example (see article by Uphoff and Wi-jayaratna in World Development, November,2000). And such investments in social infra-structure and social capital can be sustainable;the community organizations established in 1981-85 are still functioning well 25 years after end ofproject, and they became the model for a nationalprogram improving management of the wholesector. (On the strategy and implementation ofthis effort, see Uphoff, Learning from Gal Oya,Intermediate Technology Publications, 1996.)The Bank has approached the introduction ofCDD in a rather self-referential manner, trying tolearn mostly from its own experience, which isa limited “slice’ of what is known and what hasbeen done, rather than look beyond its institu-tional boundaries.

One of the most interesting and relevant find-ings of the evaluation concerns the value of rec-ognizing and working with informalorganizations (pp. 39-40, and the very in-structive Annex K), not just with formal organi-zations. We saw this clearly in the Sri Lankaexperience mentioned above. One of the “hy-potheses” we tested and confirmed was that itwould be most effective to begin with informalorganization, creating a demand for formalorganization rather than begin with supplyand then try to create a local demand for it, theusual approach. Our strategy was: work first, or-ganize second. More could be said and donealong these lines, but this last paragraph “flags”this issue as one where the Bank could usefullyfocus some attention.

Norman UphoffCornell University


1 5 6

Paiboon Wattanasiritham The Revised report has taken into accountthe Advisory Panel’s concern about theWorld Bank’s use of the term “CommunityDriven Development” (CDD) and has madechanges which make the report read bet-ter on the whole.

In particular, the reference to the participa-tion of communities in development as either“community-based development” (CBD) or“community-driven development”(CDD) ap-pears more appropriate and more easily under-stood.

Realizing the many difficulties in eval-uating the effectiveness of World Bank sup-port for community development, theevaluation attempt has done well in gath-ering information from several angles,making logical analyses and interpreta-tions, and finally coming up with reason-able and credible conclusions andrecommendations.

The difficulties in evaluating CBD/CDD ef-fectiveness in the context of World Bank pro-grams stem partly, as pointed out in the report,from the fact that the Bank has not , until recently,systematically identified and tracked its portfo-lio of CBD/CDD projects and therefore has lackeda comprehensive understanding of the evolutionand scope of its work on community develop-ment. Further, the Bank has not been sufficientlyclear about the objectives of using CBD/CDDapproaches, criteria for choosing between dif-ferent community development approaches, orhow to measure the results. But evaluatingCBD/CDD effectiveness in World Bank programsis also difficult because, for CBD/CDD, and es-pecially for CDD, to be effective , a complex setof factors have to be at work. Government poli-cies and programs, national and local adminis-trative structures, relevant laws and regulations,attitudes and capacities of officials who can haveimpact on CBD/CDD effectiveness, the eco-nomic, social , cultural and human capacity set-tings of communities in the country, allcontribute to how easy or difficult it is forCBD/CDD programs to be effective, as well ashow long it would take for the cumulative effectsto be realized. World Bank projects or programs,

therefore, may be more or less effective de-pending, to a significant extent, on the nature ofthe programs’ interrelationships with the manyrelevant factors just mentioned.

Utilizing a “systems thinking” approachshould be useful in undertaking such anevaluation as well as in drawing conclu-sions and making recommendations.

A country or society—comprising communi-ties, institutions ,organizations, etc.—is a highlycomplex and dynamic “system.” World Bank pro-grams, therefore, are but a small “part” or “com-ponent” of a very big, complex “whole”, especiallywhen considering that “sustainable poverty re-duction” is the ultimate objective of those WorldBank programs. In this context of “systems think-ing”, it should be added, CBD/CDD is more than“poverty reduction”, and “sustainable povertyreduction” is more than CBD/CDD. Further-more, the differentiation and comparison be-tween CBD/CDD and non-CBD/CDD programscannot be too clear-cut since both, as well asthe many other relevant factors, are very muchinter-related and intertwined in a complex anddynamic manner. Indeed, CBD/CDD and non-CBD/CDD programs can, and perhaps shouldalso, be complementary. Although they may bedifferent in nature and may aim at different out-puts and even outcomes, the ultimate objec-tive, or final impact, should be the same, that is,sustainable development which includes povertyreduction and improvement in people’s qualityof living.

The purpose of evaluation is (or shouldbe) “to learn and improve.” This OED eval-uation should be a useful point of departure,or point of reference, for both learning andimprovement efforts, both of which are mu-tually reinforcing.

Much “learning” can be gained from suchprocesses as “knowledge management” (KM)which in turn may be organized under one ormore, or all, of the following contexts :

• KM within a World Bank program• KM among World Bank programs• KM within a country• KM among countries• KM within the World Bank

ANNEX R

1 5 7

• KM among a number of organizations includ-ing the World Bank

• KM under any other contexts or combinations.

For the World Bank in particular, appropriateknowledge management, or some other learn-ing systems, should be useful for executives andstaffs, not only in understanding and appreciat-ing the true essence of CBD/CDD, but also inbeing able to come up with innovations and / orcreative developments that will be beneficial forall concerned, including the World Bank itself.

As a “development partner” of the itsmember countries (particularly developingcountries), the World Bank may find it con-structive and beneficial, both to the mem-ber countries and to the Bank , to have anassessment study undertaken as to the rel-evant situations and factors that have bear-ing on each country’s potentials for higherdegrees of community empowerment, in-cluding community-driven development.

Such a study would be particularly relevant inthe case where the World Bank is to have a de-velopment lending program in a country, be it aCBD/CDD-related programs or a non-CBD/CDDprogram. The outcome of the study should thenbe useful is helping shape the concept and thedesign of the program in such a way that it is con-ducive to the improvement in community em-powerment or community-driven development(CDD) efforts. For this purpose, it should beborne in mind also that true and effective com-munity empowerment or CDD need to involvea comprehensive, integrated development

agenda of the country, not just the World Bankprogram. In addition, adequate time will beneeded for the many relevant parts and factorsto be in place and become rooted firmly enoughto make the empowerment or CDD sustainable.In this context, therefore, it may be useful for theWorld Bank to be clear from the beginning aboutthe purpose, nature, scope, etc., of its programin a country, particularly whether the programaims at CDD as the main thrust or only a sup-plementary feature, and so on.

The concept of CBD/CDD, especially inthe form of CDD, logically points to theprinciple of “holistic country-driven devel-opment ” (HCDD), about which an in-depthstudy as well as a knowledge managementprocess should be undertaken, which couldlead to a significant change in the way WorldBank programs, especially country pro-grams, are conceptualized, designed, andmanaged.

A number of possible outcomes of such astudy (and / or knowledge management process)can be envisaged. For example, more emphasismay be given to “holistic country-driven devel-opment” programs as opposed to ordinary pro-grams or the so-called “country assistance”programs. In such a “holistic country-drivendevelopment“ (HCDD) program, “community-driven development” (CDD) should automati-cally be a significant part, or even a crucial part,of the total package.

Paiboon Wattanasiritham2 September 2005


1 5 8

1 5 9

IntroductionManagement welcomes a review of the effec-tiveness of Bank support for community-drivendevelopment (CDD)1 and projects that includecommunity participation (covering CBD byOED’s definition).2 Management is encouragedby the evidence that CDD and CBD operationshave in aggregate, higher development outcomeratings than non-CBD and non-CDD operationsand have proven to be an effective tool for clientcountries. Management notes that this review re-inforces findings from previous OED reports in-cluding the OED Review of Social Development3

that community participation contributes tooverall project success and sustainability. Giventhe demand from client governments, thesetypes of operations are expected to remain an im-portant component of the Bank’s assistance toclient countries.

Areas of Agreement. Management concurswith the OED Review that sustainability, moni-toring and evaluation, and local leadership areissues of particular importance to the Bank’swork on community development. Accordingly,for some time management has been devotingincreasing resources to address these issues toimprove the quality of its support for communitydevelopment interventions. These issues arediscussed in more detail below.

Areas of Divergence. Management, however,has questions regarding the relevance, rigor,and clarity of some aspects of the OED Review.In particular, management would like to note:

a. The potential for misunderstanding the Bank’srole in CDD operations and borrowing coun-tries’ ownership of the CDD agenda, as evi-

denced by a growing demand. The OED reviewtalks about “Bank projects” and the “Bank-subproject cycle” and contrasts “Bank proj-ects” with government programs, not notingthat CDD operations are all government pro-grams. The Bank’s role is to provide support.

b. The analysis and findings related to safeguardsand fiduciary aspects of CDD that lead OED torecommend more guidance and oversight.Data from OED’s own reviews of compliancewith safeguard policies have never highlightedCDD or “CBD/CDD” operations as having moreproblems than other types of projects. On thecontrary several other types of projects werehighlighted. Data from the Quality AssuranceGroup (QAG) indicate that CDD operations dobetter in assuring compliance with Bank poli-cies than the average for all other operations,both during preparation and implementation.This is not surprising, given the extensive train-ing programs, thematic reviews, and opera-tional guidance already available to staff.

c. The suggestion that the Bank should do moreup-front analysis and move more slowly on“CDD/CBD” operations; while great care is im-portant in working with client countries onthe preparation of all operations, Managementdoes not see evidence of a need to slow downits processing of CDD operations, given theirbetter quality at entry and better outcomesthat the average for all other projects.

This remaining divergence likely results fromthe scope and methodology of the OED review.Management tracks CDD operations and oper-ations that involve participation broadly, in-cluding community participation. For its review,OED created a set of operations as discussed inAnnex E of the report. The methodology in-

ANNEX S: MANAGEMENT RESPONSE

cludes text searches and OED staff judgment.That set of operations differs significantly fromeither of the sets that Management tracks. Muchof the evidence OED presents for “CDD” con-cerns operations that do not meet the definitionof CDD used broadly, including by Management.4

Further, Management has questions concerningthe methodology used, including the datasources, fieldwork, and surveys (for more detailssee Annex I).

The Approach Paper. As outlined in the Ap-proach Paper,5 the original focus of the reviewwas on CDD. It stated that the objective of thereview was to “assess the relevance, efficacy, ef-ficiency, institutional development impact, andsustainability of the World Bank’s CDD inter-ventions” (para. 6).6 While the Approach Paperclearly stated that the evaluation would also lookat other forms of community participation, thethree primary evaluation questions and the sixsub-questions included in the approach paper fo-cused on CDD.7 In management’s view, a reviewwith a focus on CDD would have been particu-larly useful, given the growth in the CDD port-folio in recent years and the increased corporateattention to CDD over the past five years.8 Man-agement believes that a review of the now sub-stantial set of closed CDD projects (56 operationsat the end of fiscal year 2005) would have beenmore appropriate. When invited by OED to com-ment early in the review process, managementexpressed its concerns with how the ApproachPaper defined the scope of OED’s inquiry andwith the proposed methodology. In retrospect,management should have shared those con-cerns officially in writing with OED and with ex-ecutive directors.

Management CommentsThe management comments that follow refer tothe main issues of agreement and provide an ex-panded set of comments on the evolution ofBank community-driven development opera-tions. As noted above, the OED review highlightsthree areas that, while important for all opera-tions, are particularly pertinent to effective CDDoperations: sustainability, monitoring and evalu-ation, and local leadership. For some time man-

agement has been directing attention to these is-sues in CDD operations. The impetus for the es-tablishment of a Bank-wide CDD Steering Groupand a corporate Anchor unit in FY01 was to drawon global experiences in order to enhance thequality and effectiveness of community-drivendevelopment operations. Client Governmentshave embraced CDD as a valuable approach to de-liver public services to poor people, to strengthenthe capacity of people’s organizations to partic-ipate in development, and to build the assetsand capabilities of poor women and men so asto improve their well-being. Given the heightenedinterest of client governments, management hasinvested significantly in ensuring that ongoing andnew programs benefit from global lessons learnedthrough improved guidance and technical sup-port.

Key Area of Agreement: SustainabilityThe review attests to overall improvements insustainability of the “CBD/CDD” portfolio, al-beit with scope for further improvement (para.4.27). When CDD operations and programs arecarefully designed and well implemented, withclear exit strategies, the likelihood of sustain-ability increases. OED’s Social Development re-view found that “giving the participants theresponsibility for structuring their involvementin the project increases the likelihood of successand sustainability.”9 The Social Developmentevaluation emphasized the value of quality par-ticipation in the success of Bank-supported op-erations and noted that “best practice” projectswere likely to have attributes such as involvementof beneficiaries in project design and imple-mentation and securing of community contri-butions to the project and its future maintenance(Annex D). All are attributes found in most CDDoperations. A combination of top-down and bot-tom-up measures is critical for sustainability ofany type of project. These typically include lineministry budgetary resources and inputs, ap-propriate technical standards, community-ledmaintenance and local government involve-ment.10

Community-Level Sustainability. At the com-munity level, sustainability efforts and project exit


1 6 0

strategies revolve around dynamics of linkagesbetween communities and external support in-stitutions, most notably local governments. De-sign characteristics of CDD operations placeconsiderable emphasis on embedding commu-nity initiatives in permanent institutional frame-works, including local government systems.Evidence from the field suggests that CDD ini-tiatives with these design characteristics aremore likely to be sustainable; for example,schools built through a CDD approach havemore and better teachers than schools builtwithout (Zambia) and child mortality declinedmore in areas that have involved communitiesthan in areas that have not (Bolivia).11

Aid Dependency and Sustainability. Despitethis evidence of trends showing increasing sus-tainability, the OED review mentions lower thanaverage sustainability ratings for “CBD/CDD”operations. This rating covers a number of issuesbeyond sustainability of the service and infra-structure created. Specifically, for several oper-ations rated as non-sustainable, OED noted intheir rationale for the rating the issue of whetherthe implementing agency will be able to continueabsent donor funding. While these issues of aiddependency are valid and important for all donorassistance, they are different from the issue ofsustainability of outputs and outcomes at thecommunity level.

Constant Attention to Sustainability Issues.Overall, what is required is constant attentionduring design and implementation to sustain-ability of community impacts. To this end, Man-agement has augmented its internal qualitycontrol functions and scaled up its technicalguidance through project Quality EnhancementReviews, CDD quality clinics and peer reviewing.

Key Area of Agreement: Monitoringand Evaluation Management agrees with OED that monitoringand evaluation (M&E) is of utmost importance.Management’s approach to monitoring resultsemphasizes the identification and tracking ofthe impact of Bank-financed operations andlearning from approaches that are most suc-

cessful. Given that borrowing governments in-creasingly rely on CDD approaches to addresscommunity empowerment objectives, Manage-ment has for some time been strengtheningmonitoring and evaluation of the CDD portfolio.

Guidance on M&E. Consistent with OED rec-ommendations to promote learning by doing,Management has been encouraging clients im-plementing CDD programs to test out differentapproaches, to build robust systems to monitorimplementation for rapid and operationally rel-evant feedback, and to emulate best practice. Theoperational manuals that guide client imple-mentation give practical guidance for setting upManagement Information Systems to track resultsof decentralized local initiatives. Innovative ben-eficiary assessment approaches collect real timefeedback from those directly involved in com-munity initiatives, so that information about suc-cesses and challenges can be quickly relayed toimplementing agencies.

Evaluating the Impact of CDD Approaches.As OED notes, while real-time information flowis crucial to success, it is also important to eval-uate the impact of CDD approaches. As part ofits efforts to improve impact evaluation acrossthe Bank portfolio, management has placed par-ticular emphasis on CDD. Given the institutionaldevelopment objectives and decentralized im-plementation of CDD approaches, it is particu-larly challenging to evaluate CDD impact.Nonetheless more than 50 impact evaluationshave been carried out for 36 CDD operations (seeAnnex II).12 In addition several evaluations havebeen done that use the most sophisticated tech-niques available, that are more robust than thoseused in the past by essentially any developmentevaluators, including OED. They use randomizedtreatment and control sampling and differencesin difference techniques (for example, the ran-domized studies of the Bolivia Social Fund andthe Indonesia Kecamatan Development Pro-gram). Several additional sophisticated evalua-tions are currently underway. To measure theinstitutional effects of CDD operations, Man-agement has developed and supported extensivequantitative and qualitative techniques for un-

ANNEX S

1 6 1

derstanding changes in social capital, such asthose applied recently in the Philippines, In-donesia, and Thailand.

Cost-Benefit Analysis. In addition to impactevaluation studies, Management also agrees withOED that it is important to do cost-benefit analy-sis of CDD operations—where feasible and at rea-sonable cost. Besides provisional ex-antecost-benefit analysis (such as that carried out inthe Philippines), many operations undertakesome ex-post cost-benefit analyses; some ex-amples include Bank support to the CambodiaSEILA program and the Indonesia KecamatanDevelopment Project. Operational manuals forCDD operations contain guidance on how tocarry out economic analysis. As would be ex-pected when communities are able identify andsupport operations that are their highest prior-ities, the internal rates of return for such oper-ations are high. For example, in the Indonesiacase, two independent cost-benefit analysespoint to internal rates of returns in excess of 60percent for rural roads built using CDD ap-proaches. That said, ex-ante cost-benefit analy-sis of social sector operations is rare (and notrequired by Bank procedures), given the intrin-sic difficulties, and social sector activities sup-ported by CDD operations are no exception.OED is correct in noting that for the majority ofinterventions in client “CBD/CDD” operationssupported by the Bank there was no ex-antecost-benefit analysis of the project itself. OED failsto note, however, that individual sub-projectswere subjected to economic analysis.13

Key Area of Agreement: Local LeadershipManagement agrees with OED on the impor-tance of supporting client countries in workingeffectively with local leadership. Communities arenot homogenous. As management has learnedfrom extensive experience in support of com-munity initiatives, it is critical to ensure buy-infrom diverse social groups within a given com-munity. The role of elites needs to be carefullyconsidered, but practice suggests that they canact benevolently and play a positive leadershiprole, as the OED review points out (para. 3.19,footnote 27). CDD programs, through capacity

development, promote inclusive decision-mak-ing and collective action and strive for the typeof institutional change that underlies participa-tory local governance. These programs are ex-tensively engaged in dealing with the challengeof how to most effectively work with local lead-ership. The degree to which they have succeededis reflected in part by assessments of percep-tions of choice and relevance of investments.Evidence suggests that most benefits accrue topoor people; reviews of beneficiary assessmentsfrom eight countries “were uniform in their find-ing that beneficiaries consistently felt that…proj-ects reflected priority needs of the community.”14

The quality of program design, including therole of community facilitators and local trans-parency and accountability measures, have amajor bearing on a program’s impact on com-munity inclusion and community empowerment.

Evolution of CDD ProgramsThe CDD approach builds on a rich history of in-novations in participatory development, pre-dominantly accumulated from outside the Bank.The composite of CDD programs also reflectlessons from earlier efforts by the Bank in sup-port of community-based development (CBD)initiatives of client countries. There are severalCDD operating principles that are markedly dif-ferent from CBD program mechanisms; thesewere conceived directly in response to perceivedshortcomings in earlier CBD efforts. Of particu-lar relevance are current CDD design features forinstitutional arrangements, the role of local gov-ernment, and broader governance linkages.

The CDD Portfolio Today. The current CDDportfolio comprises iterative and evolving pro-grams that build on lessons from across the port-folio—from prior operations in-country, similaroperations in comparable environments (for ex-ample in post-conflict countries), and opera-tions that address related technical andoperational issues, such as decentralization re-form and public-private partnerships. The CDDportfolio includes a host of long-standing, trans-formative CDD programs which have evolvedover time. In these programs in particular, theareas of sustainability, local leadership challenges,


1 6 2

and M&E have been of critical interest. Man-agement will continue to put resources intolearning in these areas and continue to exercisevigilance in enhancing operational guidance tostaff. Some examples of client operations sup-ported by the Bank that have contributed toand benefited from lessons learned include thefollowing.

Andhra Pradesh (AP) CDD Operations, IndiaThe AP District Poverty Initiatives Project(APDPIP) and AP Rural Poverty Reduction Pro-ject are two statewide community-driven ruralpoverty reduction projects under implementa-tion in Southern India with World Bank supportsince the year 2000. These programs build on theAP Government’s cumulative investments inwomen’s self-help groups over the last 10 years.The APDPIP supports the development of self-managed grass-roots level institutions of poorrural women and their federations; it has sup-ported half a million groups and 800 federationsin 29,000 villages covering 80 percent of all ruralpoor households (6.4 million) over last five years.The key project investments include institutionbuilding, capacity building and local leadershipdevelopment of community organizations, aninvestment risk fund in key livelihood sectors andlivelihood support services for the poor. Theproject uses a learning-by-doing approach andemploys process monitoring and other M&Emechanisms to develop mid-course correctionsand make adjustments in project design andprocedures. The current design features reflectthis evolution.

Scaled-Up Investments in the Poor. An out-come of institutional development and theempowerment and skill enhancement of com-munities has been the willingness of public andprivate sector agencies, including commercialbanks, to scale up their investments in the poor.Annual credit to poor households and householdgroups has increased twelve-fold from $23 mil-lion in 2000 to $276 million in 2005. The totalcredit flow from commercial banks to thesegroups is expected to reach $1 billion by thetime the project closes. World Bank investmenthas proved to be catalytic and mobilization of

other sources of finance has been key to ensur-ing sustainability of investments.

Fiduciary Aspects. The APDPIP project has in-vested in the development of adequate fiduciaryand auditing systems to ensure accountability andtransparency in a program of this scale, includ-ing building up auditing and financial manage-ment capacity among 800 federations of self-helpgroups. These federations have developed ca-pacity to train village-level bookkeepers in fi-nancial management and provide auditingservices. The project has initiated the develop-ment of a rating system jointly with financial in-stitutions to provide performance data on villagegroups to banks and agencies investing in com-munities. The project commissions independentagencies to undertake process monitoring togive feedback on process, institutional and em-powerment aspects.

Zambia Social Investment Fund (ZAMSIF)This fund was conceived as a two-phase pro-gram to support, over a ten-year period, twomain strategic objectives of the Government ofZambia (a) decentralization and empoweringlocal authorities to improve governance and ef-ficiency in service delivery; and (b) increasing ac-cess to basic social services through directpoverty interventions. It followed on two socialrecovery projects (SRP) designed in the late1980s, and is a classic example of a CDD programthat has evolved over time, based on country con-text and lessons learned. The earlier SRP projectsworked directly with communities (and decon-centrated technical staff of central ministries)to deliver quick impacts and open up space foranticipated macro-level reforms. ZAMSIF alsostrove to support the implementation of criticalaspects of the decentralization process and in-troduced a process for strengthening local gov-ernment capacity and performance.

Learning and Design Evolution during Im-plementation. The project devised a process ofgraduation that would align a local government’scapabilities with its responsibilities in each of twoZAMSIF components: a community investmentfund financing community sub-projects and a

ANNEX S

1 6 3

district investment fund financing capacity build-ing and providing funds for district-level sub-projects serving multiple communities managedby local governments. Over time, the projectwas restructured and its development objec-tives modified in the face several structural chal-lenges, including delays by the government inenacting supportive decentralization policies—notably new laws providing for elected local gov-ernments and policies to increase the limitedadministrative, technical and fiscal capacity ofnominated local governments.

Kecamatan Development Project (KDP),IndonesiaThis project is an example of a CDD programbuilt on rigorous social and institutional analy-sis, adapted to country context, which includesmechanisms for poverty targeting and inclusion.In 1996, prior to the Asian economic crisis, alocal level institutions study portrayed the richvariety of organizations and associations at thatlevel capable of planning and managing a broadrange of development projects. It also illustratedthat development resources rarely reached localgroups, and instead were channeled throughpublic sector institutions that crowded out localinitiatives in favor of government and elite con-trolled “user groups.” With the onset of the eco-nomic crisis, economic gains vanished, povertyworsened, and both government policies and in-stitutions were discredited. In this context, arapid response to poverty was required thatwould by-pass weak and ineffective line agencies,establish transparent systems to deliver financialresources to communities to be used for high pri-ority social and infrastructure projects, and de-velop institutional mechanisms at the local levelthat were more inclusive and community driven.

Targeting Financial Resources. To achieve itsobjectives, the project’s financial resources weretargeted to the poorest sub-districts in ruralareas where poverty was most highly concen-trated, and mechanisms were built in to ensurethat poor people, and especially poor women,were incorporated into decision making aboutpriorities and project proposals. Although orig-inally intended to be small, the project was scaled

up to permit financial flows to a large number ofpoor areas. Among the most innovative aspectsof the program were mechanisms for trans-parency (such as access to public records, mediainvolvement, and NGO scrutiny) and to en-courage institutions to respond quickly to pro-posals and be responsive to community needs.The first KDP project covered about 30 percentof the rural sub-districts in the country and it hasbenefited more than 15,000 villages. Ongoingevaluations have determined that KDP invest-ments are more cost-effective than other mech-anisms for delivering similar services.

Learning During Implementation. As an ex-ample of learning by doing, Management en-couraged, supported, and financed (along withDFID) a path-breaking study on monitoring pos-sible corruption (as measured by price versuscost indicators) across different models of KDPinterventions. The study found that increasinggrass roots participation in monitoring village-level KDP interventions altered the method ofpossible corruption (it substantially reduced thetheft of villagers’ wages) but had relatively min-imal effects on the overall level of possible cor-ruption. However, KDP interventions that alsoincluded an announced increased probability ofa government audit substantially reduced thelevel of possible corruption.15 Because there areno other such careful analyses of corruption fornon-CDD operations it is not possible to bench-mark whether the amount of corruption ob-served in the CDD operations is greater or lessthat in other operations.

OED RecommendationsThe following paragraphs provide Management’sspecific comments on the review’s recommen-dations.

Recommendation 1. The Bank shouldprovide operational guidance for the ap-plication of Bank safeguard policies andfiduciary oversight of CBD/CDD projectsand for the strengthening of cost-benefitanalysis and M&E [monitoring and eval-uation] systems; and should commissionan audit of the fiduciary aspects of a rep-


1 6 4

resentative sample of CDD projects forsubmission to the Board within a year.

Operational Guidance on CDD Safeguardsand Fiduciary Oversight. Management notesthat it has been engaged for some time in pro-viding additional operational guidance and fi-duciary oversight and that CDD and “CBD/CDD”operations have benefited as a result. The Qual-ity Assurance and Compliance Unit (QACU), theRural Development anchor, and the CDD an-chor have been conducting specialized trainingprograms on safeguards in CDD operations. Ad-ditionally, the Regional safeguard teams rou-tinely carry out thematic reviews of CDDoperations to assess the level of compliance.The results of these studies have been used fordevelopment of staff guidelines for applicationof safeguard policies in CDD operations. TheCDD anchor will continue to collaborate withQACU and Regional teams in providing opera-tional guidance for the application of Bank safe-guard policies during the preparation andimplementation of CDD operations. There is noevidence from OED’s evaluations of environ-mental and social safeguard compliance, norfrom work by QACU or QAG that indicate thatthose operations perform less well than the restof the World Bank’s portfolio in terms of safe-guard implementation. The “CBD/CDD” opera-tions sampled by OED for this study andreviewed by QAG do better on safeguards and fi-duciary aspects than the average in QAG reviewsfor all operations.16 The OED review’s findingson medical waste do point out difficulties withthe application of the Environmental Assess-ment policy in health projects. However this isan issue for health projects in general and notspecifically for CDD or “CDD/CBD” projects.Similarly the report raises the issue of dams.However, all the dams financed were below theheight that triggers the Operational Policy onDam Safety.

Cost-Benefit Analysis. As indicated above, Man-agement agrees with OED’s recommendationthat it is important to do good economic analy-sis (including but not limited to cost-benefitanalysis) of all operations including CDD oper-

ations and continues to work with implement-ing agencies in borrowing countries to improvesuch analysis. There are fundamental challengesin undertaking ex-ante cost-benefit analysis foroperations that involve decentralized communityactivities. These are principally related to thedifficulties in profiling with any degree of exac-titude in advance the type and proportion of alarge number of dispersed sub-projects. How-ever, all project manuals, the client-produceddocuments that the World Bank appraises, in-clude guidance on how to carry out cost-bene-fit analysis where appropriate. Management notesthat QAG reviews and OED reviewed imple-mentation completion reviews tend to rate Bankperformance in project preparation and super-vision higher for CBD and CDD operations thanfor overall Bank-supported operations.17

Monitoring and Evaluation. Managementnotes that through its results agenda, much ef-fort is being directed at assisting countries in im-proving M&E systems across its portfolio.Concerning impact evaluation of CDD opera-tions, Management notes significant commit-ment to conduct cutting edge research tounderstand the impact of these initiatives and ispleased to note that there are several impactevaluation efforts underway. The Bank’s Devel-opment Economics (DEC) vice presidency in-tends to produce a Policy Research Report inFY07 on CDD operations, once a number of im-pact analyses of sufficient quality have been un-dertaken to provide accurate and operationallyrelevant evidence.

Fiduciary Aspects of CDD Projects. All CDDoperations have audits built into them, mostusing both the borrower’s regular independentaudit system as well as special systems for inde-pendent audits of sub-projects. However, giventhe concerns raised in the OED report Manage-ment will conduct a review of audits of a repre-sentative sample of CDD operations by the endof the year, with independent input into thechoice of the sample, and on the basis of the re-view findings will determine what course of ac-tion to take. That said, Management notes thatIAD’s extensive risk analysis in setting up its

ANNEX S

1 6 5

FY06 work program did not raise CDD as a pri-ority concern for auditing. QAG data indicateno special issue with fiduciary aspects in CDD op-erations; in fact the “CBD/CDD” operations sam-pled by OED tend to score higher on fiduciaryaspects than other operations both in QAG qual-ity at entry and quality of supervision assess-ments.18

Recommendation 2. Future CASs shouldshow how they have analyzed and ad-dressed linkages not only between vari-ous CBD/CDD projects to be undertaken inthe country but also between CBD/CDDand relevant non-CBD/CDD projects. Inparticular, the analysis should addresswhether arrangements for CBD/CDD proj-ect implementation come at the expense oflocal government capacity development.

Results-based CAS Methodology and Use ofExisting Guidelines. Management believes thatits recent guidelines for results-based CASs andthe process of corporate review of CASs andCAS Completion Reports have now establisheda strong basis to learn lessons of past interven-tions and set out the Bank’s proposed inter-ventions within a medium-term resultsframework. All CASs are now required to bebased on a CAS Completion Report that evalu-ates the development impact of past interven-tions and derives lessons for the design of futureinterventions. Moreover, all CASs are required toidentify country-owned development goals thatthe Bank will support. In the CAS, country teamschoose what they view as the most effective mixof instruments to deliver CAS outcomes. In turn,these outcomes are expected to contribute toachieving country’s development goals. In thenew format, teams must adequately demonstratewhy they have chosen a given set of instruments.In addition, CASs go through an extensive cor-porate review process that provides guidance toCAS teams on possible weaknesses and defi-ciencies in analyzing and diagnosing develop-ment challenges and the choice of instrumentsto address these challenges. As part of this review,CAS teams will be asked to explain how opera-tions and instruments, notably including CDD

operations, will contribute to achieving CAS out-comes.

Local Government Capacity Development.One of the mandatory issues of discussion in thecontext of the review of the new generation ofresults-based CASs (along with governance andresults) is capacity development, including localgovernment capacity development. There isgrowing evidence that greater community par-ticipation improves local government capacity,particularly when operations are designed aspart of decentralization efforts. In the mid-1990s,there was a concern that support for decentral-ized initiatives could undermine local govern-ment capacity. As a result of this concern, CDDoperations have evolved to include the explicitobjective to empower communities so that theywork with local government and serve as ac-countability mechanisms to improve local gov-ernance (see examples from Zambia andIndonesia above). As evidenced by an increasingnumber of CASs focused on governance thatrely on combinations of CDD operations anddecentralization support, Management believesthat current guidance meets the objectives setout by OED.

Recommendation 3. For any newCBD/CDD project, the Bank should ana-lyze (using existing processes, such as so-cial assessments) whether it is buildingon indigenously matured initiatives orattempting to begin a CDD program in acountry and then tailor the interventionto local capacity; and the Bank shouldalso selectively undertake rigorous im-pact assessments upon completion of itsongoing CBD/CDD projects to learn forthe future.

Social Assessments. Management agrees thatit is always important to understand the local in-stitutional and social context, though that un-derstanding may not always need to arise froma formal social assessment. Management notesthat social assessments are undertaken for manyprojects and supports their strategic use wherethere are important knowledge gaps. However,


1 6 6

in certain situations such as natural disasters orpost-conflict settings, speed of response is crit-ical and these knowledge gaps would need to befilled in parallel with project implementation.In addition as part of the implementation of thesocial development strategy recently reviewed bythe Bank’s Board of Executive Directors, socialanalysis is increasingly moving upstream to thepolicy and CAS level, which heightens the impactand lowers the cost. On the specific issue ofidentifying whether a country (not the Bank—the Bank supports the client country) is build-ing on an indigenously matured initiative, thatinformation is normally available as part of ap-praisal documentation.

Indigenously Matured Initiatives. It is gen-erally true for all projects that they do betterwhen building on indigenously matured initia-tives and when design reflects local capacity.The key operational question is what type ofintervention to support in low capacity countries,traditional top-down interventions that rely onministries for implementation or projects thatalso build on local community participation andcommunity control over decisions? Manage-ment, therefore, does not interpret OED’s rec-ommendation on tailoring to local capacity tomean that CDD or “CBD/CDD” operationsshould necessarily be the exception in thesesettings. An important aspect of the Bank’s roleis to support innovation in our client countries.

Thus, Management believes that it is importantto offer to support client governments that re-quest assistance with community development,even when the client has little experience inworking with communities. Further, in conflictor recent post-conflict settings where local in-stitutions are very weak—including both for-mal government structures and informalnon-government institutions—Bank efforts tosupport reconstruction and fight poverty maystill best be served by CDD operations and evi-dence suggests that these types of operations dobetter in fragile environments than alternativeapproaches.

Rigorous Impact Assessments. Managementnotes that, as outlined in Annex II, it has sup-ported or facilitated impact assessments for alarge number of client-country CDD operations.Further, through the auspices of its Develop-ment Impact Evaluation (DIME) initiative, it ispromoting more rigorous impact evaluationacross the portfolio. The Office of the ChiefEconomist is particularly committed to increas-ing the number of operations that are subject tocareful impact evaluation and DEC resourcessupport a number of such studies, includingseveral ongoing impact evaluations of CDD op-erations. Management is committed to expand-ing further its evaluation work, especially ofsocial impacts related to enhanced social capi-tal, enhanced local capacity and empowerment.

ANNEX S

1 6 7


1 6 8

Management Action Record

The Bank should provide operational guidance for the application of

Bank safeguard policies and fiduciary oversight of CBD/CDD projects

and for the strengthening of cost-benefit analysis and M&E systems;

and should commission an audit of the fiduciary aspects of a repre-

sentative sample of CDD projects for submission to the Board within

a year.

Future CASs should show how they have analyzed and addressed

linkages not only between various CBD/CDD projects to be undertaken

in the country but also between CBD/CDD and relevant non-CBD/CDD

projects. In particular, the analysis should address whether arrange-

ments for CBD/CDD project implementation come at the expense of

local government capacity development.

Management believes that current operational guidance is adequate.

Concerning the specific sub-set of CDD operations, Management is al-

ready providing operational guidance on safeguard and fiduciary sys-

tems, along with training and special reviews.

Management is committed to a review of its control framework under

IDA14 and considers this as the highest priority use of its resources

in this area. However, Management will conduct a review of the au-

dits of an independently selected representative sample of CDD op-

erations by the end of the year following which it will determine what

course of action to take, including, as needed, updating current fidu-

ciary guidelines for CDD operations

Management concurs with regard to the importance of M&E systems;

however, its commitment for action is related to impact evaluation (see

Recommendation 3 below).

With the mainstreaming of the results-based CAS during FY05, all CASs

are now required to follow a results-based approach. CASs are to be

based on a CAS Completion Report that evaluates the development

impact of past interventions and derives lessons for the design of fu-

ture interventions. The Bank analyses the most effective mix of in-

struments to deliver CAS outcomes and through that analysis

demonstrates why the proposed set of instruments was chosen. CDD

operations need to meet this test. In addition, CASs go through an ex-

tensive corporate review process that provides guidance to CAS teams

on possible weaknesses and deficiencies in analyzing and diagnosing

development challenges and the choice of instruments to address

these challenges. Hence, Management believes that its program of

mainstreaming results-based CASs fulfills this recommendation.

Management finds that many of its most innovative new operations

are designed to combine decentralization initiatives with CDD opera-

tions to improve local governance. These programs are based on the

operational lessons that empowered local communities that are able

to hold local government accountable for service delivery, improve local

government capacity to deliver effective and demand-responsive ser-

vices. Management believes that current guidance and assistance to

staff meet the objectives of the OED recommendation and does not

plan further steps.

OED Recommendation Management Response

ANNEX S

1 6 9

Management Action Record (continued)

For any new CBD/CDD project, the Bank should analyze (using exist-

ing processes, such as social assessments) whether it is building on

indigenously matured initiatives or attempting to begin a CDD program

in a country and then tailor the intervention to local capacity; and the

Bank should also selectively undertake rigorous impact assessments

upon completion of its ongoing CBD/CDD projects to learn for the fu-

ture.

Management agrees that social assessments are valuable tools that

are widely employed in CDD operations. It also agrees that CDD de-

sign will depend on whether the operation is building on an indige-

nously matured initiative or is responding to a request from a client

that does not have such an initiative. Management believes that this

is already the case—the importance of tailoring support for CDD op-

erations to local capacity is already a priority. Therefore, Management

does not plan further follow-up with regard to this recommendation.

(Management notes that CDD is an important approach in post-con-

flict settings where “indigenously matured initiatives” are often ab-

sent and is an important tool in implementing its operational policy on

Development Cooperation and Conflict—OP2.30.)

Management notes that borrower governments supported by the Bank

are already undertaking many impact evaluations of CDD projects. Sev-

eral sophisticated evaluations that meet the highest academic stan-

dards are currently underway. Given the technical challenges of doing

those evaluations well, Management is pleased to note that the Chief

Economist and DEC will lead the effort to conduct a meta-evaluation

of these impact assessments that result. That assessment is planned

for FY07.

OED Recommendation Management Response

Management’s concerns over the conclusionsand recommendations of this review result fromissues around the scope of the OED inquiry.Though Management tracks CDD operationsand those that involve community participation,OED chose to establish its own portfolio of op-erations for this review. As a result of this deci-sion, operations that OED examined, both in itsportfolio review and its specific cases, differ fromthose that Management tracks and guides. Specif-ically, operations OED cites when drawing con-clusions about “CDD” are not CDD operationsin Management’s view. OED based its evalua-tion on a word-search in project documents andrefined the set using OED staff judgment as towhether an operation should be included in theportfolio it examined. Management’s trackingof CDD operations is based on Task Team Lead-ers designating their projects as CDD, subject toreview by Regional Management. Given their ex-tensive, in-the-field experience working withclient governments to identify and appraise op-erations, Management believes that operationalstaff are best placed to identify operations thattake a CDD approach.

Beyond these difficulties of scope and defi-nition, Management also questions OED’s eval-uative methodology, particularly concerning itsfield work. An important illustration is its treat-ment of a Brazil operation in Rio Grande doNorte, which is the only Management-recog-nized CDD operation for which OED conductedfield work. First, there are important questionsregarding the control group in the OED study.The OED selection of control municipalities andcommunity associations was based on their nothaving benefited from the Rural Poverty Allevi-ation Project I (RPAP). However, RPAP is only oneof a series of CDD rural development projects the

Bank supports in Rio Grande do Norte. The con-trol group included municipalities and commu-nity associations that had been involved in thesepreviously Bank-financed CDD-based rural de-velopment projects in Rio Grande do Norte.OED’s “control” communities had specific mu-nicipal councils. Only those municipalities thathad participated in CDD operations had thesemunicipal councils. Second, the household sur-vey methodology, on which most of the con-clusions are based, had an important built-innegative bias concerning the beneficiary com-munities. The beneficiary groups defined byOED included households that were not bene-ficiaries of the operation. The OED study usedtowns and villages to sample beneficiaries andthese generally do not coincide with beneficiaryassociations. Third, the OED review excluded keynon-income measures of poverty (notably, healthbenefits, savings in time because of more ac-cessible and reliable water supply, and reductionto vulnerability to drought) and asset accumu-lation from its analysis of program impact. Italso introduced negative biases in the measureof impacts by excluding from the analysis the kindof activities more directly linked to income gen-eration (the productive subprojects). Fourth,the program OED selected for comparison pur-poses differed fundamentally in design and ob-jectives (top-down, large infrastructureinvestments with no empowerment objective)from the RPAP whose beneficiaries were the sub-ject of OED’s study. Finally, the OED reviewcould have drawn more broadly on the literaturecovering the whole time span of the program.This issue is important because a crucial ele-ment of Brazil CDD program is its evolutionover time, learning from experience.19


1 7 0

Annex S.I: Management Concerns with Scope and Methodology

1. Albania Development Fund2. Armenia Social Investment Fund 3. Armenia Social Investment Fund 2 4. Benin Social Fund5. Bolivia Social Investment Fund 6. Brazil Rural Poverty Reduction Project7. Cambodia Seila/Rural Investment and Local

Governance Project 8. Ethiopia Women’s Development Initiative

Project9. El Salvador Community-Managed Schools

Program (EDUCO)10. Honduras Social Investment Fund11. India Andhra Pradesh District Poverty

Initiative Project12. Indonesia Support for Conflict-Ridden

Areas (SCRAP)13. Indonesia Urban Poverty Project 2 (UPP2)14. Indonesia Kecamatan Development Project

(KDP1, KDP2, KDP3)15. Indonesia KDP and Conflict16. Indonesia Decentralized Agriculture and

Forestry Extension Project (DAFEP)17. Laos Poverty Reduction Fund Project18. Malawi Social Action Fund (MASAF)19. Moldova Social Investment Fund20. Nepal Rural Water Supply Project

21. Nicaragua Emergency Social InvestmentFund (FISE1)

22. Pakistan Aga Khan Rural Support Project(AKRSP-Kwaja Study)

23. Pakistan National Rural Support Program(DEC Study)

24. Panama Rural Poverty and NationalResources Project

25. Peru Social Fund (FONCODES)26. Philippines KALAHI-CIDSS27. Philippines ARMM Social Fund28. Senegal PNIR (Projet National d’Infrastruc-

tures Rurales) 29. Burkina Faso 2nd National Services

Development Project (PNDSA2) andSenegal PSAOP (Programme d’Appui auSecteur Agricole et aux OrganisationsPaysannes)

30. Sierra Leone GoBIFO and IRCBP31. St. Lucia Poverty Reduction Fund32. Thailand Social Investment Fund33. Tanzania Social Action Fund34. Vietnam Northern Mountains Poverty

Reduction Project (NMPRP)35. Vietnam Community Based Rural

Infrastructure Project (CBRIP)36. Zambia Social Investment Fund (ZAMSIF)

ANNEX S

1 7 1

Annex S.II: List of CDD Operations with Impact Evaluations (completed and ongoing)

1 7 3

On August 31, 2005 the Committee on Devel-opment Effectiveness (CODE) reviewed the re-vised draft Management Response to TheEffectiveness of World Bank Support for Com-munity -Based and -Driven Development: AnOED Evaluation as agreed at the first CODEmeeting (June 22, 2005) on this topic. On July27, 2005 a CODE informal roundtable meetingwas held so CODE members may better under-stand some of the issues raised in the OED eval-uation, taking into consideration more recentexperience. Written statements by some mem-bers were circulated before the meeting.

Background. On June 22, 2005 CODE reviewedThe Effectiveness of World Bank Support for Com-munity Development: An OED Evaluation (theReport) and draft Management Response (MR),for which an interim "Green Sheet" was issuedon July 13, 2005. Generally, the Committee per-ceived that community development (CD) op-erations had the potential to effectively helpand empower the poor, but it also took note ofthe report's cautionary findings. At the sametime, several members and speakers had difficultydrawing a clear position on the report and MRgiven the disagreements between OED and Man-agement on some points including the defini-tions used and the scope and methodology of theevaluation, as well as some of the conclusions.Accordingly, as suggested by a member, an in-formal roundtable meeting was organized onJuly 27, 2005 to enable CODE members to gaina better understanding of the issues raised in theOED evaluation, based on current experience,both positive and negative, of community-drivendevelopment (CDD) operations. At the earlierCODE meeting, it was agreed that the MR neededto be revised and further considered by CODE

before completing the discussion on this topic.To remove any ambiguity regarding the scope ofthe review, OED changed the title of the reportto The Effectiveness of World Bank Support forCommunity-Based and –Driven Development.

Revised Draft Management Response (MR).Management concurred with the findings thatCDD and CBD operations achieve better resultsthan other operations that do not involve com-munity participation. Management also agreedwith OED that sustainability, monitoring andevaluation, and local leadership are important is-sues for the Bank's work on CDD, as highlightedin the revised MR. The revised MR elaborated onhow some concerns raised by OED are being ad-dressed in recent CDD operations while com-menting on other issues including: (i) some ofthe analysis concerning safeguards and fiduciaryaspects of CDD; (ii) interpreting the OED rec-ommendation to approach CDD with greatercare; and (iii) potential misunderstanding of theBank's role in CDD operations, which are ownedand implemented by countries and not by theBank. Management also continued to expressconcerns about the definitions used and thescope and methodology of the evaluation. Giventhe lengthy process to review this OED evalua-tion, Management highlighted some key lessonslearned including: (i) importance of discussingthe issue of scope and definition at an earlystage when the Approach Paper of evaluationsare presented; (ii) consider a standing commit-tee to discuss methodological issues; and (iii)need for care in ensuring balanced communi-cation of both OED and Management views.

OED Comments. OED welcomed the revisedMR. OED was happy to note Management's af-

ANNEX T: CHAIRMAN’S SUMMARY: COMMITTEE ON DEVELOPMENT

EFFECTIVENESS (CODE)

firmations that it was already acting on many ofthe issues raised in the evaluation, which it willmonitor and report on results. OED restatedthe importance of a fiduciary audit of a sampleof projects, which should be representative andindependently selected. On the issue of scopeof the evaluation and specific interest in a reviewof only CDD operations, OED stated that it is verydifficult to separate CBD and CDD elements ina project to enable an evaluation of only CDD ini-tiatives. OED expressed confidence in its method-ology and the rigor of its work, which had beenreviewed by internal and external experts.

Overall Conclusions and Next Steps. Fol-lowing extensive discussions, including those atthe earlier CODE and informal roundtable meet-ings, the Committee recognized the importantcontribution of CDD operations to empoweringand helping the poor and it generally supportedscaling-up Bank assistance in response to coun-try demand. At the same time, the Committeenoted the OED report's cautionary findings,which called for greater vigilance in designingCDD operations. The points highlighted in-cluded issues of sustainability and need for exitstrategies; fiduciary matters; environment and so-cial safeguards; monitoring and evaluation; build-ing on indigenously matured initiatives; parallelstructures and local government and communitycapacities; and cost-benefit analysis. The Com-mittee welcomed the revised MR and was gen-erally satisfied with the changes. Members notedthe more positive tone of the MR, although somemembers commented that it could be further ad-justed and also made suggestions to clarify spe-cific aspects, which Management agreed toconsider in the final MR.

Members appreciated Management's accept-ance to undertake a fiduciary review of a repre-sentative sample of CDDs that are selectedindependently. In proceeding with disclosureof the OED evaluation report according to theapproved OED disclosure policy, the Committeeemphasized the importance of balanced com-munication, including the possibility of placingthe summary of CODE discussions upfront in-stead of as annex to the OED report. OED willconsider in a few years the possibility of a more

focused evaluation on CDD operations as pro-posed by some members. Finally, given thelengthy review process for this OED evaluation,Management and OED were requested to ex-plore possible procedural improvements to beconsidered by the Committee in due course.

The following main issues were raised duringthe meeting:

Substantive Issues Raised in the OED Eval-uation. Members commented on some of theissues raised in the OED evaluation, which hadbeen discussed at previous CODE meetings:

a. Sustainability, Aid Dependency, and ExitStrategies. In considering sustainability be-yond the issue of aid dependency, severalspeakers stressed the importance of incorpo-rating exit strategies in the design of CDD proj-ects, and suggested that the MR address thismatter further. Aid dependency was considereda general issue, not specific to CDDs. Man-agement noted that CDD projects includestrategies for exiting communities and that itworks closely with client Governments to learnwhich are most effective. Management alsonoted that OED’s sustainability ratings for theCBD/CDD portfolio review results not only oncommunity-level sustainability but also on thesustainability of funding mechanisms forCBD/CDD approaches, mechanisms that aresusceptible to classic problems of aid de-pendency.

b. Fiduciary Audit. Ensuring appropriate uti-lization of funds, especially in the context oflocal initiatives and systems was consideredcritical in the design of CDD operations. OEDrecommended a fiduciary audit of a repre-sentative sample of CDD operations be un-dertaken. Management assured the Committeethat under the Bank's operational framework,all operations supported by the Bank are sub-ject to audits by independent auditors. It pro-posed to undertake a review of the audits of arepresentative sample of CDD operations se-lected independently; initially, the revised MRproposed a fiduciary review of five large CDDoperations by the end of the year, and on thebasis of the review findings determine whether


1 7 4

to recommend to Internal Auditing Depart-ment (IAD) an audit of CDD operations. Mem-bers welcomed Management agreement toconsider a representative sample of CDD op-erations, which are independently selected.Management noted QAG data that indicatethat CDD does better than other operations onfinancial management.

c. Environment Safeguards. Some memberssought assurance that adequate guidelineswere in place, referring to Annex Q of the OEDreport. A member noted the difficulty of com-pliance in conflict and post-conflict situationswhere country systems may be very weak.Management clarified that Annex Q referred toCBD and CDD projects, and not just CDD op-erations, including issues such as medical wastethat are not normally relevant for CDD oper-ations.

d. Building on Indigenously Matured Ini-tiatives. A few members stated that the Man-agement response should clarify particularcircumstances where Bank support to new ini-tiatives may be merited, given OED's findingsof greater success with indigenously maturedinitiatives. Some members noted that in post-conflict countries and LICUS, there may bemerit in supporting new initiatives. Manage-ment noted that there is evidence that CDDdoes better than other types of operations inthese difficult circumstances.

e. Building Local Government and Com-munity Capacities. Several members high-lighted the importance of strengthening localgovernment and community capacities, andrequested elaboration on how to prevent par-allel structures detracting from this aim. Amember supported the combination of "bot-tom-up" and "top-down" approach to devel-opment, but suggested that the MR clarify howCDD operations should embody the two ap-proaches. Management agreed to clarify furtherin the MR how CDD operations currently workon strengthening community and local andnational government capacity, where relevant.

f. Monitoring and Evaluation (M&E). Not-ing Management and OED agreement on im-portance of M&E, some members soughtassurance that staff is adequately equipped to

strengthen impact evaluation, track quantita-tive and qualitative results, and to supportcountries' M&E capacities. Management notedthat many quantitative and qualitative evalua-tions are already being carried out and agreedthat ensuring government and Bank capacityin this area is important.

g. Cost-Benefit Analysis. While cost-benefitanalysis may not be possible in some cases, afew members stressed the need for stronger in-stitutional commitment in this area. They sug-gested that Management could provide furtherguidance in this area as well as clarify in the MRthe Bank's role given clients' capacity con-straints. Management agreed to address thisissue more fully in the MR, noting that all op-erational manuals for CDD operations containguidance on how to carry out economic analy-sis. Management will continue to monitor to en-sure that guidance remains appropriate and itsuse will be covered under M&E.

Communication and Disclosure. Speakersacknowledged the importance of OED inde-pendence, but also emphasized that there shouldbe balanced communication of all views, in-cluding of Management's response. In disclosingevaluations in accordance with the approvedOED disclosure policy, some speakers proposedthat the summary of CODE discussions be placedat the beginning rather than as annex, particu-larly for potentially controversial reports.

Lessons Learned. Some members noted thatManagement continued to have concerns aboutthe definition, scope, and methodology of theOED evaluation. A member commented thatthere should be Management and OED agree-ment on the objectives of the evaluation up-front, while OED, as an independent evaluationentity, may be best placed to make the final de-cision on methodology. It was pointed out thatthe scope of this evaluation was explicitly out-lined in the Approach Paper. Management notedthat, while it had communicated its difficultieswith the scope and methodology to OED infor-mally, it had erred in not formally commentingin the Approach Paper. In this connection, speak-ers agreed with Management that one key lesson

ANNEX T

1 7 5

emerging from reviewing this OED evaluationwas the need to pay more careful attention to theApproach Paper for each evaluation. They statedthat any significant divergence between Man-agement and OED should be brought to CODE'Sattention at an early stage. A few speakers alsosupported Management's suggestion to explorethe possibility of a standing committee to discussmethodology issues; OED stated that it alreadyhas in place a thorough system for vetting itsmethodology by external experts which could be

shared with CODE members as and when nec-essary. OED independence was considered im-portant, but the need to ensure OED'saccountability was also raised. Some memberssuggested a follow-up evaluation more narrowlyfocused on CDD operations in the near future;OED agreed to consider the possibility in a fewyears time.

Chander Mohan Vasudev,Chairman


1 7 6

Executive Summary1. Management notes that CDD operations are clearly

distinguished by their giving communities control over

decision making and resource allocation.

2. Management notes that it tracks: (1) participa-

tion in operations; and (2) community-driven devel-

opment operations. It finds these two categories as

being more operationally useful than the very diverse

sample that OED has put together for the purpose of

this review.

OED notes: The Bank’s tracking of participation is

not limited to community participation. Since it also

covers participation of all stakeholders at both the proj-

ect and macro levels, it does not permit the separate

tracking of all CBD/CDD nor does it allow for an analy-

sis of participatory issues relevant at the community

level that could help inform the design of future

CBD/CDD operations (in each situation where the

Bank considers fostering community development, it

needs to draw from relevant experience from both

CDD and CBD projects).

3. Management notes that, when discussing weak-

nesses, the table does not compare its set of CDD/CBD

operations with other Bank operations.

4. Management notes that there have been a num-

ber of rigorous evaluations done of operations that in-

volve community participation, which is close to the

definition OED created for this review. For example,

Rawlings, Sherburne-Benz and Van Domelen 2004,

A Cross-Country Analysis of Community Investments,

Washington, D.C, The World Bank, synthesizes six

careful empirical analyses of social funds, each of

which was published in the World Bank Economic Re-

view. In addition, management notes that a large pro-

portion of the most rigorous evaluation work under

way concerns CDD impact. Given the relatively recent

Bank operational support to CDD, there has been

little opportunity for careful before and after evalua-

tion to be completed.

OED notes: The evaluation draws from this docu-

ment (listed in references as World Bank 2003d), the

full title of which is: “Evaluating Social Funds: A Cross-

Country Analysis of Community Investments”; also

known as “Social Funds 2000.” This study is based on

data collected more than 5 years ago and reviews

only the experience of social funds—a specific subset

of CBD/CDD projects.

5. Management notes that Rawlings, Sherburne-

Benz, and van Domelen 2004, cited above by man-

agement, provides rigorous evidence that community

management of investments offers significant po-

tential for cost savings. That report also provides

careful analysis of the benefits of these types of proj-

ects.

OED notes: The study cited by management refers

to social funds only—a specific subset of CBD/CDD

projects—and is based on field work in six countries.

Chapter 11. The “community” in a Bank-supported project

is considered a “unified, organic whole.” Since the

group of people in a “community” live in a particular

area, share a common interest (water users associa-

tions, herders, and the like), and are governed by a

set of norms, its members are assumed to be in the

best position to identify their most pressing needs and

problems. The assumption suggests that there are

common problems that can be solved through com-

munity consensus. While this may be true, it neglects

community members’ differences and power rela-

tionships, the conflicts, and the diversity of interests

that determine day-to-day behavior and that affect

the effectiveness of participatory approaches. The

poor themselves are rarely a homogenous group;

they live in different geographic areas and face different

kinds of deprivations, and each seeks a personalized

way of reducing poverty (Schneider 1999). For more

on community see Annex A.

1 7 7

ENDNOTES

2. Initially, six “ways” of involving communities in

Bank-supported interventions were defined, though

later the Bank’s Social Development Department

adopted a more intuitive fourfold classification (in-

formation sharing, consultation, collaboration, em-

powerment). Although it is not stated, presumably

both categorizations were derived from the work of

Arnstein (1969), who created an eight-rung ladder of

citizen participation.

3. The CDD Anchor in its recently established CDD

database (see Chapter 2), also includes projects that

involve community control in the development and

implementation of their donor-supported develop-

ment plans, as well as projects that focus primarily on

information sharing, consultation, and collaboration.

4. Management notes again that the definition

used by OED is closest to that of participation and that

there are benchmarks for evaluating participation,

specifically those laid out in the Participation Source-

book.

5. Management notes that it tracks: (1) participa-

tion in operations, and(2) community-driven devel-

opment operations. For operational relevance,

management believes that the review would have been

more useful had it been based on either or both of the

categories that management tracks.

6. Management notes that, given the Review’s def-

inition of CBD/CDD, there is a larger set of impact eval-

uation evidence to rely on. For example, Rawlings,

Sherburne-Benz and van Domelen 2004, cited by man-

agement above, reviews a set of high-quality empiri-

cal studies on community participation and its impact.

Chapter 21. The Anchor’s CDD database also includes in-

formation on CBD projects.

2. Over two-thirds of the projects in the sample of

84 devote more than 75 percent of project cost to

CBD/CDD. Since community participation is a process

issue, it has the potential to affect project outcomes

even when it forms a small part of the project cost.

3. Management notes that there is often a detailed

description of participatory processes in the project

appraisal documents. In addition, these processes

are discussed in great detail in project and field

manuals.

Chapter 31. Using the Bank’s primary sector coding.

2. This finding is supported by a Bank study on the

performance of community-level user groups carried

out in the context of three Bank-supported inter-

ventions in India. The study questioned whether user

groups as currently designed and implemented can

achieve long-term sustainability as independent or-

ganizational entities. According to the study, past ex-

perience with various government and NGO initiatives

has habituated user groups to let final responsibility

for the groups’ management rest with an external au-

thority (Alsop and others 2002).

3. Moore (2001) examines the use of the concept

of empowerment in multilateral development agen-

cies, including the World Bank, and finds that while

the notion of empowering the poor was important in

framing the discussion on poverty reduction in these

institutions, the lack of a clear definition could lead

to different interpretations of the concept of em-

powerment. The review finds two distinct implicit

definitions of empowerment within the reports in

these institutions: the underlying proposition of one

definition is that improving the material status of

poor people is empowering because it weakens the

social, economic, and political dependence and pro-

vides the poor with greater freedom and autonomy;

the underlying proposition behind the second defi-

nition is that insofar as empowerment contributes to

collective organization of the poor, it can also con-

tribute to political action and may imply political con-

frontation.

4. It is important to acknowledge that it is much

more difficult to meet and measure the qualitative

goals than the quantitative goals. For example, house-

hold survey data analysis from Benin found that com-

munities’ access to school infrastructure, supplies,

and equipment increased because of the Bank-sup-

ported projects, though this did not necessarily mean

improvement in access to education services, which

is also determined by, among other things, availabil-

ity of teachers. (See Annex P and section on sustain-

ability in Chapter 4 for an explanation.) Moreover,

part of the improvement in capacity is expected to

come about through the participation of the com-

munities in the planning, design, and construction of

the infrastructure. As a result, in the short duration

of a Bank-supported intervention, when achieving

visible results is seen as an indicator of progress on

implementation, progress toward qualitative goals

tends to receive less attention, not only from the


1 7 8

Bank, but also from the country.

Management notes that rigorous cross-country

studies, not cited in this review, found that schools

funded with community inputs have more and better

teachers than those that were funded in the traditional

top-down manner. Further, in the case of Benin, all

schools funded by the project had teachers.

5. “Participation can make development assistance

more effective. But it works best for groups that are

already able to help themselves” (Da Cunha and oth-

ers 1997).

6. The literature and earlier OED evaluations also

consider the dairy experience in India among the few

examples of successful scaling up of sustainable rural

development (Krishna and others 1997; Uphoff and

others 1998; Candler and Kumar 1998).

7. SEWA has had some Ford Foundation support

but no major external assistance. The Orangi Pilot

project received support from several donors, in-

cluding the World Bank (see Uphoff and others 1998;

Krishna and others 1997 for details).

8. Requested and funded by the Aga Khan Foun-

dation.

9. Since in Uttar Pradesh the comparator commu-

nities had no program, whereas in Madhya Pradesh the

comparator was communities with the same activity,

also supported through a participatory approach, but

by the government, it is not surprising that results are

more significant in Uttar Pradesh than in Madhya

Pradesh.

10. Management notes that other reviews have

found quite different and more positive results from

those cited here.

11. “But certain cultural settings are better suited

to local participation and collective action than oth-

ers. Participation works best when it is based in, rather

than in opposition to, existing organizations. In North-

east Brazil, regional tradition and existing social, eco-

nomic, and political structures pose strong challenges

to horizontal social organization and thus to popular

mobilization and participatory development. Although

certain democratizing trends have weakened tradi-

tional authorities, prior organization still poses chal-

lenges, which must be recognized and systematically

addressed in policy planning and in project design and

implementation” (Costa and others 1997).

12. It is worth noting that household data also

show that the poor in project communities in Madhya

Pradesh reported a significantly greater increase in

trust, association life, and mobilization skills than the

relatively better-off (see Annex N).

13. The Bank staff survey found that only 40 per-

cent of the respondents were satisfied or very satis-

fied with their Regional management’s understanding

of the objectives and design of CBD/CDD interven-

tions. This suggests that Bank management may not

completely understand what is needed for CBD/CDD

to succeed. This may also explain why Bank proce-

dures have not yet been fully adapted to CBD/CDD.

14. It is worth noting here that the literature sug-

gests—and earlier OED work on social fund projects

confirms—that having the elite and the local leaders

in the decision-making position does not necessarily

constitute a problem in and of itself. In fact, the lit-

erature shows that the involvement of the elite could

also be helpful in some ways. The important issue is

not so much how to avoid elite domination, but how

best to use the power and energy of the elite to serve

the poor (Narayan 1995). One way of doing so, per-

haps, would be to create appropriate mechanisms to

ensure, before a subproject is funded, “that the ideas

of the leader are also the most important ones for the

community as a whole” (OED 2002b, p. xxvii). As Plat-

teau and Gaspar (2003) argue, so long as the inter-

vention of the elite leads to an improvement in the

situation of the poor, the latter are likely to be thank-

ful to their leaders.

15. Management notes that it did not concur with

the analysis and conclusions of OED’s social funds

evaluation.

16. “The danger is not going far enough, and being

satisfied with any partial progress,” as Ostrom (1999)

puts it, “creating dependent citizens rather than en-

trepreneurial citizens reduces the capacity of citizens

to produce capital.” The costs of development assis-

tance will also inevitably increase—it is not costless

to establish new organizations.” The “stages” theory

also notes that “progression is not taken to be in-

evitable, with outcomes being regression (going back

to the previous stage), stagnation or arrested devel-

opment (remaining at one stage), and extinction (or-

ganizations may fail or terminate)” (Pretty and Ward

2001).

17. A recent Bank study (Alsop and others 2002)

based on fieldwork in three states of India also found

that members’ perceptions of the purpose of user

groups differ from the perceptions of project de-

signers and implementers.

ENDNOTES

1 7 9

18. In his study of 60 villages in the state of Ra-

jasthan in India, Krishna (2001) shows that no matter

how high the level of social capital, it needs to be mar-

shaled strategically and directed toward incentives

available within the broader institutional and envi-

ronment.

19. Management notes that rigorous evaluations on

which Rawling, Sherburne-Benz, and Van Domelen,

2004, cited above, is based show that the bottom two

deciles of the income distribution benefit more than

higher deciles.

OED notes: Of the five social funds for which data

are presented in the study cited by management, four

social funds (Honduras, Nicaragua, Peru, Zambia) had

a neutral or only mildly progressive distribution of ben-

eficiaries at the household level, and the fifth social

fund (Armenia) had a regressive distribution (see

pages 64–65, table 3.3 and figure 3.3 of the manage-

ment-cited study); this lack of progressivity is of con-

cern because social funds aim to reach poor

communities and households using a variety of tar-

geting mechanisms.

20. A recent study by the Institute for Social and

Economic Change in India, which carried out fieldwork

in the States of Karnataka and Uttaranchal, found sig-

nificant elite capture of benefits in participatory wa-

tershed projects (Rajasekhar and others 2004).

21. In the context of irrigation schemes in Tanza-

nia, Koopman and others (2001) find that while

landowners were the main beneficiaries, they did not

shoulder a greater share of the costs. Instead, the net

costs borne by tenants were significantly higher, for

they were not only required to put in as much free

labor as landowners, but also ended up paying higher

rents, because the project increased the value of

land.

22. The expression “participatory spaces” (Corn-

wall 2002) is used here to identify the various types

of community organizations, formal and informal,

created to enable the participation of project benefi-

ciaries in the process of making decisions.

23. The literature also notes reasons for the lack

of participation of the poor. Participation places ad-

ditional demands on community members, which are

likely to be particularly problematic for poorer house-

holds (Pantoja 2000; Garcia and Way 2003). As Baland

and Platteau (2002) point out, the poor often lack in-

centives to take part in collective undertakings, as

these violate their survival constraints. As the authors

explain, poverty tends to shrink the time horizon,

because it forces people to attach considerable im-

portance to their present income opportunities. Con-

sequently, the poor are likely to resist any type of

collective activity that requires them to forgo present

income opportunities—even if it permanently in-

creases future incomes. This argument is echoed by

Weinberger and Jutting’s (2001) quantitative analysis.

24. The literature review carried out for this eval-

uation distinguishes between formal inclusion, which

concerns the extent to which community members are

able to enter decision-making arenas, and substantive

inclusion, which captures the extent to which differ-

ent participants are able to exert influence over de-

cisions. For example, poor people may opt not to

speak up against the views and positions put forward

by more powerful members of the community but,

rather, conform to them (Kolavalli and Kerr 2002,

p. 225). As Linden (1997) points out, keeping a low

profile is an essential survival strategy of poor people.

In the context of patronage, the poor are highly de-

pendent on their leaders, and are hence unwilling to

antagonize them (Kumar 2002).

25. Rajasekhar and others (2004) found that the ac-

tivity preference across the local organizations was to

a large extent influenced by the level of contribu-

tions. The study reports: “The well-off farmers needed

certain type of activities to be undertaken. Under the

framework of beneficiary contribution having impor-

tant say in the activity, the well-off farmers contributed

more and also obtained more from the groups. This

was, thus a case of ‘elite capture,’ which had negative

impact on the processes that the [local organizations]

had to engage in.”

26. The literature also shows that inadequate at-

tention to issues such as inequality within a commu-

nity can also affect targeting outcomes of CDD

interventions, because economic and social hetero-

geneity can increase the risk of elite capture (Elbers

and others 2004) Governments rarely have informa-

tion on the level of inequality in communities.

27. For example, in the context of the Indian Pan-

chayat Raj system in Karnataka, Vyasulu and Vyasulu

(1999) found that women attained high levels of for-

mal inclusion. At the gram panchayat level, well over

40 percent of elected representatives were women.

However, many elected women were surrogates for

husbands and fathers who could not contest because

of the reservation, while others had been put in place


1 8 0

by the wealthy and powerful for their malleability.

28. For example, a socioeconomic impact study was

carried out in connection with the Borgou Pilot proj-

ect in Benin. The OED assessment for the project

questioned the credibility of the ERR rates reported

by that study. In Brazil, a review of an often-quoted im-

pact study by van Zyl and others (2000) revealed that

the study had not compared project communities

against controls. In addition, the study report provided

no information on the number of respondents in the

three different kinds of project communities, and no

indication of how many subprojects had been financed

per community. Further, there are several shortcom-

ings in the cost-benefit analysis reported.

29. That report notes that, “Although social funds

were typically more efficient than other national pro-

grams in term of overhead expenditures, their in-

vestment unit costs tended to be more efficient only

where there was significant input and control by com-

munities.” (World Bank 2003d)

30. Kent and Rimarachin’s (1994) study on public

works in Peru found that community contribution

ranged from 7 to 47 percent, and was on average 20

percent of total project costs. Isham and Kahkonen

(2002) found that community contribution to water

service projects in Sri Lanka was 43 percent of total

construction costs—well above the required 20 per-

cent.

31. Management notes that the background work

for Rawlings, Sherburne-Benz and Van Domelen 2004,

cited above by management, found that community

management of investments offers significant poten-

tial for cost savings, often on the order of 25–40 per-

cent, even after taking into account the full value of

community contributions.

32. Econometric analysis by Kerr and others (2000)

finds that participatory watershed projects in India

performed better than technocratic and top-down

projects. The authors also find that a combination of

participation and sound technical input performed

best of all. Similarly, Kahkonen’s (1999) review of the

literature maintains that community-managed water

and sanitation projects worked better than govern-

ment-managed schemes.

Chapter 41. The capacity of the Bank to support a CBD/CDD

approach is addressed in Chapter 5.

2. The OED assessment of the Matrouh Project

found that the important issue of what impact the on-

going adjudication of Bedouin lands is going to have

on the poor and on rangeland management systems

was ignored. Further, the Egypt country study found

that in the Sohag Project, while the community process

elements were quite well covered, QAG noted that the

larger policy issue of subsidized interest rates was in-

adequately addressed, related fiscal issues were not

well covered, institutions had not been adequately ap-

praised for capacity, and the risk that the project

could undermine the development of alternative

credit sources had not been adequately assessed.

3. Bank management notes in response that the

new national CDD project in Benin (approved by the

Board in October 2005) is fully taking the decentral-

ization framework into account and intends to build

local government’s capacity to implement the CDD ap-

proach.

4. Management notes that communes did not exist

until one year after the Borgou Project became ef-

fective.

5. The majority of the councilors interviewed in Rio

Grande do Norte had not received any training by the

RPAP or the follow-on and ongoing RPRP. From the

NGO focus group session it emerged that a training

event for municipal councilors had been organized in

the state capital in the past. Some NGO representa-

tives criticized this form of training for it only allowed

the participation of few councilors per municipality

and was a one-time event, and not a systematic train-

ing program.

6. It is worth noting that the latest ongoing CDD

projects in Northeast Brazil are attempting to promote

greater integration of existing programs to improve

the impact of public resources available for poverty

reduction. However, it is too early to say how suc-

cessful this effort will be.

7. Three models of implementation are being used

in community-based projects in Philippines: The gov-

ernment agency model has a line ministry or local gov-

ernment as the executing agency for the project. It is

responsible for the design and implementation, pro-

curement, financial management, monitoring, and

evaluation. The project is often carried out by a proj-

ect management unit within the agency, and executed

by local offices of the ministry in conjunction with local

authorities and communities. In the social fund model,

an autonomous agency of government makes grants

directly to beneficiary communities or individuals. In

ENDNOTES

1 8 1

the community-driven development model, the com-

munities themselves decide which projects to un-

dertake and are responsible for their execution. The

Comprehensive Integrated Delivery of Social Services

(CIDSS) component of SEMP 1 and the proposed

SEMP 2 and the proposed “Kalahi” Project are exam-

ples of this model. (Source: “Community-Driven De-

velopment: A Case study of the Philippines.” East Asia

Region Workshop June 12–13 2002.)

8. The rapid multiplication of national NGOs in re-

sponse to the availability of funds from the interna-

tional community is also an issue that has been raised

in the literature (Platteau and Gaspart 2003; Chabal

and Daloz 1999).

9. It is important here to distinguish between

NGOs and grassroots community organizations, which

may have different motivations for working with com-

munities. However, the country studies were not able

to research the role of these organizations.

10. Van de Walle and Johnston (1996) note in the

context of lack of aid coordination, “a conservative es-

timate for a typical African country is that 600 projects

translates into 2,400 quarterly reports a year submit-

ted to different overseeing entities; and more than

1,000 annual missions to appraise, monitor and eval-

uate. Each mission asks to meet with key officials,

and each will ask the government to comment on its

reports. Is it any wonder that the most common com-

plaint voiced by officials interviewed in the case stud-

ies is that aid imposes too many administrative

burdens?”

11. For northeast Brazil, specifically, the van Zyl

study reports that 89 percent of the sample of 3,633

subprojects funded by the RPAP in 1997–98 were

found to be operational in the year 2000.

12. Management notes the evidence presented in

the background research for Rawlings, Sherburne-

Benz and Van Domelen 2004, cited earlier by man-

agement. That work found that “impact evidence

showed that the facilities in which social funds invest

can be at least as sustainable as similar facilities if not

more so. The majority of the infrastructure appeared

to be well constructed and operating adequately and

levels of maintenance were equivalent or better than

comparators.”

13. OED’s Impact Evaluation of Basic Education in

Ghana (OED 2004b) also found that the downside of

community financing of schools is that it leads to dis-

parities in resource availability, because poorer com-

munities have less resources to contribute in com-

parison with richer communities. Further, a review

of community support for Basic Education in Sub-

Saharan Africa also found that support for recurrent

expenditures poses sustainability problems. Accord-

ing to the report, “Schools need a predictable flow of

income to meet their recurrent expenditures, yet

communities are often unable to provide this. This is

especially the case among poor communities, who are

often vulnerable to external shocks, and whose income

is irregular” (Watt 2001, p. 29).

14. See Annex I for an explanation on project-sup-

ported municipal councils in northeast Brazil.

15. It is worth noting that only 18 percent of the

respondents to the Bank Staff Survey agreed with the

statement that community maintenance contribu-

tions are sufficient to sustain infrastructure invest-

ment for Bank-funded CBD/CDD projects (Annex L).

16. The Ethiopia Pastoral Community Develop-

ment Project (fiscal 2003) and the Nigeria Community-

Based Poverty Reduction Project (fiscal 2001) are

examples.

17. Among the infrastructures implemented in a vil-

lage in Kalale commune of Benin’s Borgou region

was an agricultural storage facility financed through

the Borgou Pilot Project. Ten percent of the total proj-

ect cost was to be put into an account for repairs or

other maintenance. When asked whether the funds

needed to be replenished when they were used, the

village leader told OED they did not. When asked if

there was anything left of the 10 percent funds, he also

said no, though it had only been two years since the

subproject had been completed. When asked what the

community would do now if something were to hap-

pen to the storage facility he was not at all worried and

said that they would seek new funds. In this particu-

lar village several Bank projects and other donor in-

terventions had been implemented, so it would be

reasonable for the village committee to assume that

they could just go to another donor when the need

arose.

18. A recent study on the role of local organizations

in water supply and sanitation in the states of Kar-

nataka and Uttaranchal notes that the possibility of ob-

taining resources from the government toward the

maintenance of water supply projects could mean

that local organizations do not have an incentive to

mobilize resources from users (Rajasekhar and

Veerashekhararappa 2004).


1 8 2

19. Further, OED’s social fund evaluation found the

technical quality of social fund infrastructure to be vari-

able across countries and between sectors. The find-

ings of the recent self evaluation of social funds were

similar in that respect (World Bank 2003d).

20. “Ensuring that positions are filled, that staff

report for work, and that they are responsive to all their

clients is a major challenge. …Incentive payments

might encourage professionals to work in remote

areas, but they can be expensive….Even when posi-

tions are filled, staff absence rates can be high” (World

Bank 2004d, pp. 22–23).

21. Thus the ICR for the Cambodia Social Fund ac-

knowledges “sustainability was not set up as one of

the main objectives because of the project’s primary

concern with the rapid and cost effective delivery of

subprojects. Another reason why sustainability con-

cerns were not more integrated into project opera-

tional activities was the assumption that a demand

driven identification process would ensure commu-

nity ownership and that line agencies would take over

operations and maintenance of subprojects.”

22. The case study on Indonesia’s Kecamatan De-

velopment Program for the Shanghai Conference

(May 2004) points out: “Two limitations of the KDP

model are particularly worth underscoring since they

can to some extent be overcome. The first is that

technically difficult activities or activities that involve

recurrent costs are not easily addressed through the

KDP system as it is currently designed. Examples, of

such necessary activities include large-scale health

provision, providing teachers for schools, or any kind

of infrastructure network planning” (World Bank

2004b). If maintaining the services from schools

through CDD interventions is a challenge for In-

donesia, it is an even bigger challenge for countries

like Benin and Nepal.

23. A significant number of focus groups (50 per-

cent of the female focus groups and 40 percent of the

male focus groups) in that state expressed an inter-

est in receiving further training.

Chapter 51. Staff skills within the institution have also become

more diversified, and today Bank staff come from a va-

riety of academic disciplines and varied institutional

experiences.

2. For a CDD intervention covering eight states in

northeast Brazil, construction of small dams was barely

mentioned at appraisal and no effort was made to set

down guidelines for dam safety and to study the cu-

mulative input of a large number of small dams. How-

ever, a Social Development Note (World Bank 2001c)

mentioned that over a thousand such dams were built

under this program.

3. For example, in Nigeria the internal controls

are reported to be weak or inoperative and provide

negligible assurance that the funds are being used en-

tirely for their intended purposes. In Senegal, even

though the scope of NGO involvement has increased

substantially, the CFAA notes that the supervisory ca-

pacity of NGOs’ oversight bodies is very weak. Fidu-

ciary sector work in Pakistan has reported that

compliance with existing internal controls and pro-

cedures have been inadequate and the financial re-

ports and financial accounts have often been untimely

or incomplete.

4. There is considerable evidence in the literature

on mismanagement of aid transfers that occur in class

or caste-based societies (Platteau and Gaspart 2003;

Bardhan 2002; Conning and Kevane 2002).

5. “As compared to other policy areas, performance

in developing countries is in general weakest in pub-

lic sector governance. And within public sector gov-

ernance, it is weakest with respect to transparency,

accountability, and control of corruption. And the

weaknesses are the most pervasive precisely in coun-

tries where stronger institutional capacities are needed

to manage development interventions that will spur

progress toward the MDGs—poor countries” (World

Bank-IMF 2004).

6. At the Fiduciary Forum 04, managers recognized

the need to translate diagnostic into action. “Too often

we do excellent work on Country Financial Account-

ability Assessments and Country Procurement Assess-

ment Reports, and then the country team checks these

off [its ‘to-do’ list] and says, ‘Okay, that’s done,’ when

the real work actually starts afterwards” (Fiduciary

Forum 04, Managing Risk Realizing Opportunity).

7. Internal documents from the midterm review of

the Andhra Pradesh District Poverty Initiatives Project

noted that an independent review of financial and

procurement processes had been completed. The re-

view found that progress had been made, but it also

confirmed some recurring weaknesses. Specifically,

compliance remained inadequate and the processes

in place provided limited assurance that the project

funds had been used for the intended purposes.

ENDNOTES

1 8 3

8. The fieldwork done for the Indonesia Keca-

matan Development Project found that while the ma-

jority of people who lost small amounts of land to

roads and other infrastructure did not seem overly con-

cerned, a few were not happy. Since in virtually all cases

land values rose substantially because of the infra-

structure in question, it would be difficult to argue that

anyone ended up worse off than they were before the

project. However, there is no clarity in the Bank’s re-

settlement policy on whether loss of land requires

compensation, even when land values go up more than

the value of land lost—that is, whether relative loss

or loss comparing the with- and without-project sce-

nario matters. In any event, the OED mission found

the appraisal text on compensation for lost land con-

fusing and likely to put pressure on affected house-

holds not to seek compensation.

9. File:///C:/Documents%20and%20Settings/wb25

1042/Local%20Settings/Temporary%20Internet%20File

s /Content.IE5/1WDNJU3F/300,29,World Bank Staff)

10. “Compartmentalization within major aid or-

ganizations of the expertise and responsibilities to

support administrative reforms, sectoral assistance

programmes and community development projects,

produces fragmented and competing interventions

that do not address and even retard the systemic

changes needed to advance decentralization” (Romeo

2003).

Annex B1. A detailed explanation is presented in the CDD

evaluation discussion paper available at http://www.

worldbank.org/oed/cbdcdd/documents/discussion_

paper.pdf.

Annex C1. From the World Bank community-driven devel-

opment Web site <http://lnweb18.worldbank.org/

ESSD/sdvext.nsf/09ByDocName/ProjectPrepara-

tionImplementationKeyDesignPrinciples> accessed

January 2005.

Annex D1. The projects reviewed were: Albania: Second

Community Works Project (P077297); Cameroon:

Community Development Program Support Project

(P073629); Honduras: Community Based Education

Project (P007397); India: Andhra Pradesh District

Poverty Initiatives Project (P045049); Philippines:

Kalahi - CIDSS Project (P077012); and Yemen: Third

Social Fund for Development Project (P082498).

2. These are undertaken to validate the findings of

project Implementation Completion Reports.

3. The literature on social funds is excluded from

this review, as it has been the subject of a recent OED

evaluation (OED 2002b).

Annex E1. As of September 2004. A detailed explanation of

the methodology for identifying the universe of

CBD/CDD is presented in the document “CDD: A

Study Methodology” available at http://www.world-

bank.org/oed/cbdcdd/documents/discussion_paper.pdf.

This document was shared with workshop partici-

pants and was also posted on the above Web site for

comments. While there was discussion at the workshop

around the methodology adopted for identifying the

universe, no alternative approaches were offered to

identify the universe. Nor did any of the Regions pro-

vide an alternative list of projects.

2. The list of projects identified in the post-2000

period by the word search was compared with the list

produced by the CDD Anchor. Some post-2000 proj-

ects in OED’s universe were absent from the CDD An-

chor’s list and some were additional to the list. All

these projects were reviewed for their appropriateness;

some were dropped from the list of 833 projects that

was available at the design stage and others were

added to give a final total of 847 projects.

Annex F1. A detailed explanation of the methodology for

identifying the sample of CDD is presented in the

document CDD: A Study Methodology: http://www.

worldbank.org/oed/cbdcdd/documents/discussion_

paper. pdf.

Annex G1. OED ratings are based on OED reviews of ICRs,

25 percent of which are subsequently revisited through

OED field assessments. OED ratings are analyzed by

exit year since OED rates each project on only on

exit.

2. OED rates outcome on a six-point scale: 6 =

highly satisfactory, 5 = satisfactory, 4 = moderately sat-

isfactory, 3 = moderately unsatisfactory, 2 = unsatis-

factory and 1 = highly unsatisfactory. These ratings are

presented in most reports, including the Annual Re-


1 8 4

port on Portfolio Performance and Annual Report on

Development Effectiveness, on a two-point scale by

summing the top three ratings (4 to 6) as satisfactory

and summing the bottom three ratings (1-3) as un-

satisfactory. While a small percentage of projects is “not

rated,” to keep the denominator constant for all three

ratings (outcome, sustainability, and institutional de-

velopment impact), calculations for percent satisfac-

tory projects are based on the denominator equaling

all closed investment lending during the specified

period.

3. Post-conflict countries with a CBD/CDD project

exiting during 1994–2003 include Angola, Burundi,

Cambodia, Colombia, Congo, Democratic Republic of,

Djibouti, Eritrea, Ethiopia, Guatemala, Indonesia,

Macedonia, former Yugoslav Republic of, Mozam-

bique, Philippines, Rwanda, Sierra Leone, Sri Lanka,

Tajikistan, Timor-Leste, Uganda, and the West Bank and

Gaza.

4. Regional analysis is restricted to Regions with

more than 25 projects.

5. The non-CBD/CDD portfolio has been per-

forming below CBD/CDD portfolio for both Africa

and Latin America and the Caribbean, however, decline

in the performance of non-CBD/CDD portfolio in the

Latin American and Caribbean Region over the two

phases has been only 2 percentage points, and the in-

crease in the Africa Region portfolio has been 7 per-

centage points.

6. Sectoral analysis is restricted to Sector Boards

with more than 25 CBD/CDD projects.

7. OED rates sustainability on a four-point scale:

4 = highly likely, 3 = likely, 2 = unlikely, 1 = highly

unlikely, plus non-evaluable. Calculations for percent

“likely” or better (includes both “highly likely” and

“likely.”) ratings in this section are based on all closed

projects including projects rated non-evaluable and

uncertain. Excluding “non-evaluable” pushes per-

centages upwards by 7 percentage points for

CBD/CDD projects and by 5 percentage points for

non-CBD/CDD projects in phase 2. The reasons to

keep the non-evaluable in the denominator are

twofold: first, the evaluation wanted to keep the de-

nominator constant for all three OED ratings, namely

outcome, sustainability and Institutional impact; sec-

ond, non-evaluable rating is given to projects largely

because of poor quality of the ICRs, which makes it

difficult for an evaluator to make any concrete as-

sessment on the likely hood of sustainable benefits

from the project. The evaluation wanted to capture

this “negative” aspect as well for all projects—

CBD/CDD and non-CBD/CDD projects.

8. OED rates ID impact on a four-point scale: 4 =

high, 3 = substantial, 2 = modest, and 1 = negligi-

ble. The percentage of projects with “substantial” ID

impact includes both “substantial” and “high” ratings.

Annex I1. In Vietnam, officials picked multiple choices,

unlike the other three countries, where officials picked

one primary choice of central, regional/provincial,

municipal/local, communities, NGOs, other donors,

others, and do not know.

2. This was also raised in interviews with stake-

holders at various levels in Benin.

Annex J1. Note that we are not here talking of the costs of

the investments but of the costs to all players of get-

ting to the point of implementing and then supervising

that investment.

2. It is reasonable to expect that these costs will de-

cline over time.

3. A Study of Rural Hill Potentials and Service De-

livery Systems, by SAPROS and IFAD, April 2002, IFAD.

4. Sample sizes from 110 households up to 154

households.

5. Often there is little detailed evidence to back up

the claims of savings. Making a fair comparison be-

tween provision of infrastructure through different

means is a complex calculation calling for allocation

of a number of fixed costs that are difficult to allocate.

6. The opportunity cost curve of household time

may be concave, with modest amounts of time spent

in meetings having quite small opportunity costs but,

as the time increases, having quite substantial costs

through impact on labor peaks related to the agri-

cultural calendar. There is some anecdotal evidence

that the costs to the poor are greater because they can

least afford the lost labor.

Annex L1. Multivariate analysis indicates that respondents

from the East Asia and the Pacific and Latin America

and Caribbean Regions were more likely to express dis-

satisfaction with coordination within the Bank across

sectors as compared to all other respondents (table

L.5).

ENDNOTES

1 8 5

2. Multivariate analysis indicates that respondents

from the Europe and Central Asia Region, and soci-

ologist/anthropologists were less likely to agree with

the fact as were respondents who had participated in

larger number of CBD/CDD projects. Interestingly,

managers (sector level and above) were more likely

to agree that implementation costs per dollar lent

for CBD/CDD projects are higher than more traditional

projects (table L.5).

3. Multivariate analysis indicates that economist

were more likely to indicate disagreement; and re-

spondents from Europe and Central Asia more likely

to indicate agreement (table L.5).


who had participated in larger number of CBD/CDD

projects were more likely to disagree with this state-

ment (table L.5).


from Africa and Latin America and the Caribbean

were more likely to disagree with the fact that

NGO-supported interventions generally achieve a

better CBD/CDD outcome than Bank interventions

(table L.5).

Annex M1. AgeFIB became inactive in fiscal year 2004; and

the ICR presented it as a CDD intervention, though

from the appraisal document it appeared to be a CBD

project.

2. AgeFIB and PILSA were implemented in other

regions of Benin also.

3. While the first two types of investments were cho-

sen because together they accounted for 60 percent

of investments financed by the RPAP in Rio Grande do

Norte, the second and third type were chosen because

eligible PAC communities had mainly these types of

investments

4. The on-going RPRP had restricted the number

of municipalities targeted under the RPAP because

some were financially able to meet the needs of their

communities. These municipalities could not there-

fore be appropriate comparator for this study.

5. Differences between the comparator community

that had benefited from the reformulated NRDP and

the other two comparator communities were tested

using the same model discussed on page 109, under

“Multivariate Analysis” (specification without inter-

actions). In this model the project dummy repre-

sented the community that had benefited from the

reformulated NRDP. Only two significant differences

were found. Respondents in the community that had

benefited from the reformulated NRDP reported a sig-

nificantly smaller increase in their participation in

political events, and a significantly grater increase in

ownership of medium consumer durables, which was

driven by greater increase in ownership of satellite

dishes.

6. The draft questionnaire is in available on the Web

site http://www.worldbank.org/oed/cbdcdd.

7. Benin was the only country for which weights

were not used due to a lack of information on com-

munity population.

8. The interactive variable for women in project

areas was included for the projects in Benin and Uttar

Pradesh, as these targeted women explicitly (Annex

N).

9. A likelihood ratio test was performed in order

to test the validity of the restricted model and this

was rejected in favor of the unrestricted model, which

includes three separate dummy variables for the

three types of implementation modalities. The three

dummy variables were differently associated with

changes in: (a) in associational life, (b) circle of

friends, (c) access to information and (d) mobiliza-

tion skills.

Annex N1. “CDD empowers poor people (…) Targeted

community-driven approaches devolve control and de-

cision making to poor women and men. This em-

powers them immediately and directly. (…) the speed

and directness with which CDD empowers poor peo-

ple is rarely matched by other institutional frame-

works for poverty reduction. (…) Control over

decisions and resources can also give communities

the opportunity to build social capital (defined as

the ability of individuals to secure benefits as a result

of membership in social networks) by expanding

the depth and range of their networks.” (http://

lnweb18.worldbank.org/ESSD/sdvext.nsf/09ByDoc-

Name/BasicConceptsPrinciplesWhyCDD).

2. As already mentioned in Annex M, the bivariate

analysis for the Brazil project reports only the response

rate of FUMAC communities, which account for 60

percent of respondents in project communities.

3. In Madhya Pradesh, village forest committee

(VFC) or the forest protection committee (FPC) were

created, which comprised of one male and one female


1 8 6

member from each household in the village. In Uttar

Pradesh the project created two distinct committees,

the Site Implementation Committee (SIC) and Water

User Group (WUG), which played distinct role. While

the SIC decided on broader project management is-

sues, the WUG decided on water usage and water

rates of a particular boring around which it was

formed. In Brazil, a Community Association needed

to be legally constituted for the community to take part

in the project. The Community Association was re-

sponsible for selecting the subproject, submitting a

proposal, and if successful for implementing and

maintaining the subproject. In Benin, a Comite de Con-

certation (CC) was set up in each community and

was given primary responsibility for monitoring proj-

ect implementation at the field level.

4. The idea that decision-making forums are neu-

tral and that by entering them people can meet on a

level playing field has been criticized by a number of

scholars because it ignores that differences in the dis-

tribution of power and resources among community

members impinge on the process of collective deci-

sion-making (Leach and others 1999; White 1996;

Molyneux 2002).

5. Attendance at meetings for project selection

was lower in PAC and FUMAC-P communities, where

only 18 and 28 percent of the respondents respectively

attended.

6. In PAC and FUMAC-P communities a much

smaller share of respondents reported speaking at the

meetings for project selection, respectively 10 and

14 percent.

7. In Brazil, a significantly greater number of CA

members reported owning durable goods (car, mo-

torcycle, freezer, and satellite dish) and large animals

(ox and cow) prior to subproject implementation

than did non-members. CA members also reported sig-

nificantly greater mobilization skills and participation

in political and socio-cultural events prior to subpro-

ject implementation than did non-members. In Mad-

hya Pradesh, a significantly greater number of

members of forest committees reported owning land

and large animals (ox and cow) prior to subproject im-

plementation than did non-members. Committee

members were also better educated than non-mem-

bers, and reported greater mobilization skills, social

network and participation in traditional and non-tra-

ditional events prior to subproject implementation

than did non-members. In Uttar Pradesh, SIC mem-

bers were better educated than non-members, and re-

ported greater mobilization skills and social network

prior to subproject implementation than did non-

members. A significantly greater number of SIC mem-

bers also owned oxen prior to subproject

implementation than did non-members.

8. As Kumar and Corbridge (2002) point out, vil-

lage elites are likely to nominate themselves as rep-

resentatives in their role of gatekeepers of

development interventions. In other cases, partici-

patory projects choose to work through village chiefs

or community leaders, for these are seen as legitimate

and appropriate institutions of community repre-

sentation (Kumar and Corbridge 2002; Ribot 1998; Gib-

son and Marks 1995). In addition, communities

generally elect the most prominent members and

those with political connection with power-holders,

as they are believed to be the only ones in the posi-

tion to attract benefits to the community (Linden

1997; Platteau and Gaspart 2003).

9. Around 40 percent of the 84 CBD/CDD projects

in the sample included an extensive dissemination

campaign on project information, while another 40

percent had only some focus on dissemination, and

the remaining 20 percent had no major focus on dis-

semination.

10. The share of respondents in PAC and FUMAC-

P communities who were unaware of the cost of the

Bank-funded subproject was higher than in FUMAC

communities—respectively 92 and 91 percent.

11. Based on her fieldwork in Northeast Brazil,

Tendler (2000) writes that “information … [was] sur-

prisingly low even in the Brazilian programs.”

12. The same is true for respondents in FUMAC-P

communities.

13. The variable that captures changes in respon-

dents’ mobilization skills is a composite variable. See

tables M.9–M.12 for details on how this variable was

created in each of the four project areas.

14. In Brazil, FUMAC-P communities in Brazil are

negatively associated with changes in associational

life, while no significant association was found be-

tween the latter and PAC communities.

15. As defined by the World Bank Social Capital Web

site http://www1.worldbank.org/prem/poverty/scap-

ital/index.htm.

16. The variables that capture changes in trust and

associational life are composite variables. See tables

M.9–M.12 for details on how these two composite

ENDNOTES

1 8 7

variables were created in each of the four project

areas.

17. In Uttar Pradesh the positive association be-

tween the project dummy and changes in trust is

driven by respondent’s greater trust in village mem-

bers and village organizations.

18. In Brazil, although no association was found be-

tween the project dummies and changes in trust, re-

spondents in all three types of project communities

reported a significantly smaller increase in trust in the

municipal government than did respondents in com-

parator communities.

19. The Village-Level Participatory Approach in-

cluded participatory rural appraisal and other part-

nership efforts that enable communities to coordinate

and execute their own rural development project

with assistance from extension agents and financial re-

sources from a variety of programs.

Annex O1. This information is based on an in-depth re-

view of 13 projects from the random sample of 54 tar-

geted projects in the sample of 84—two from fiscal

years 1989–93, five from fiscal 1994–98, and six from

fiscal 1999–2003.

2. In Benin, the Bank’s social funds project (Age-

FIB) targeted poor rural and peri-urban communities,

with a special focus on women and unemployed

youth, and in the Community-based Food Security Pro-

ject (PILSA) pre-identified problem areas were tar-

geted to ensure that assistance would be provided to

the most disadvantaged segments of the rural popu-

lation.

Annex P1. As productive subprojects are normally invest-

ments that interest CA members in particular rather

than communities as whole (though the latter can ben-

efit indirectly form such investments), we only con-

sidered responses from CA members for the analysis

of sustainability of irrigation subprojects.

Annex Q1. The safeguard policies—covering environmen-

tal assessment, natural habitats, pest management, in-

voluntary resettlement, indigenous peoples, forests,

safety of dams, cultural property, projects on inter-

national waterways, and projects in disputed areas—

provide a mechanism for integrating environmental

and social concerns into development decision mak-

ing.

2. The response rate on the questionnaire was 32

percent.

3. PPARs are prepared for selected projects by the

Operations Evaluation Department, whereas ICRs are

prepared for all projects by the Regions.

Annex S1. The standard World Bank definition, available on

the CDD Web site (http://www.worldbank.org/oed/

cbdcdd), states that CDD is an approach that “gives

control over planning decisions and investment re-

sources to community groups and local governments.”

2. The Effectiveness of World Bank Support for

Community-Based and -Driven Development: An

OED Evaluation, July 11, 2005.

3. An OED Review of Social Development in

Bank Activities, Operations Evaluation Department,

World Bank, February 17, 2004.

4. OED notes that this is likely due to the fact that

management’s CDD portfolio covers only projects

approved from FY 00 onwards.

5. Approach Paper: Evaluation of the World

Bank’s Support for Community Driven Development

(CDD) (CODE2003-0052), July 31, 2003.

6. OED notes that the Approach Paper (para. 6) also

said that “Since most of the current CDD projects

have evolved from the Bank’s experience with CBD

projects, the evaluation will review both CBD and

CDD interventions.”

7. The title of the OED report first referred to

CDD, that was changed later to “community devel-

opment,” before changing it again recently to

“CBD/CDD.”

8. OED notes that it found in its investigation that

most projects evaluated contained both CBD and

CDD components and there are very few ‘pure’ CDD

projects. Management notes that this is not surpris-

ing given the fact that OED’s sample was designed to

capture projects with both CBD and CDD compo-

nents (see Annex E of the report).

9. An OED Review of Social Development in Bank

Activities, Operations Evaluation Department, World

Bank, February 17, 2004.

10. Evaluating Social Funds: A Cross-Country

Analysis of Community Investments, L.B. Rawlings, L.

Sherburne-Benz, J. Van Domelen, World Bank, 2004.

11. Ibid.


1 8 8

12. OED notes that the list in Annex II includes im-

pact evaluations which are still ongoing and thus have

not been “carried out” yet.

13. OED notes that it did not find any evidence in

the documentation which it had access to that eco-

nomic analysis was carried out for most individual

sub-projects.

14. Getting an Earful: A Review of Beneficiary

Assessments of Social Funds, D. Owen, J. van Dome-

len, World Bank, 1998.

15. Olken, Benjamin, Monitoring Corruption: Ev-

idence from a Field Experiment in Indonesia, NBER,

November 2004.

16. OED notes that its own reviews do not show

that CDD operations do better on safeguard and fi-

duciary aspects than other operations.

17. OED notes that its database shows no signifi-

cant difference in overall Bank performance between

CBD/CDD operations and overall Bank-supported

operations.

18. OED notes that its own reviews do not show

that CDD operations do better on fiduciary aspects

than other operations. Management notes that QAG

reviews of the sample of operations OED developed

for this report highlight no particular concern about

CBD/CDD operations.

19. OED emphasizes that it has reviewed the

methodology on the project very carefully and stands

by its conclusions.

ENDNOTES

1 8 9

Agarwal, B. 2001. “Participatory Exclusions, Com-munity Forestry, and Gender: An Analysis ofSouth Asia and a Conceptual Framework.”World Development 29(10):1623–48.

———. 2000a. “Conceptualizing Environmen-tal Collective Action: Why Gender Matters.”Cambridge Journal of Economics24(3):283–310.

———. 2000b. “Group Functioning and Com-munity Forestry in South Asia: A Gender Analy-sis and Conceptual Framework.” WIDERWorking Paper 172, UNU/WIDER, Helsinki.

Agrawal, A., and C.C. Gibson. 1999. “Enchant-ment and Disenchantment: The Role of Com-munity in Natural Resource Conservation.”World Development 27(4):629–49.

Alsop, Ruth, Disa Sjoblom, Ceema Namazie, andPawan Patil. 2002. “Community-Level UserGroups in Three World Bank–Aided Projects:Do they Perform as Expected?” Social Devel-opment Papers 40. World Bank Environmen-tally and Socially Sustainable Development(ESSD) Network, Washington, D.C.

Anani, Kofi. 1999. “Sustainable Governance ofLivelihoods in Rural Africa: A Place-Based Re-sponse to Globalism in Africa.” Development42(2):57–63(7).

Arnstein, S.R. 1969. “A Ladder of Citizen Partici-pation” Journal of the American PlanningAssociation 35(4): 216–24.

Arrow, K., R. Solow, P.R. Portney, E.E. Leamer, R.Radner, and H. Schuman. 1993. “Report ofthe NOAA Panel on Contingent Valuation.”Federal Register 58: 4601–14.

Baland, J-M, and J-P Platteau. 2002. “Collective Ac-tion on The Commons: The Role of Inequal-ity.” In J-M Baland, P. Bardhan, and S. Bowles,eds., Inequality, Collective Action and Envi-ronmental Sustainability, papers presented

at the Inequality Cooperation and Sustain-ability Workshop, Santa Fe Institute, Sep-tember 21–23, 2001. <http://discuss.santafe.edu/sustainability/discuss/msgReader$3>accessed March 2005.

Bardhan, P. 2002. “Decentralization of Gover-nance and Development.” Journal of Eco-nomic Perspectives 16(4): 185–205.

Beck, Tony, and Cathy Nesmith. 2001. “Buildingon Poor People’s Capacities: The Case of Com-mon Property Resources in India and WestAfrica.” World Development 29(1):119–33.

Bennett, Lynn, Mike Goldberg, and PamelaHunte. 1996. “Ownership and Sustainability:Lessons on Group Based Financial Servicesfrom South Asia.” Journal of InternationalDevelopment 8(2): 271–88.

Binswanger, H., and S. Aiyar. 2003. “Scaling upCommunity-Driven Development: Theoreti-cal Underpinnings and Program DesignImplications.” World Bank Policy ResearchWorking Paper 3039. Washington D.C.: WorldBank.

Blair, Harry. 2003. “Civil Society and Pro-PoorInitiatives at the Local Level in Bangladesh.”Draft, February 28. Department for Interna-tional Development, UK. <http://idpm.man.ac.uk/cprc/Conference/conferencepapers/Blair%20Harry%20draft%2003.03.03.pdf>accessed March 2005.

———. 2000. “Participation and Accountabilityat the Periphery: Democratic Local Gover-nance in Six Countries.” World Development28(1): 21–39.

Brock, K., A. Cornwall, and J. Gaventa. 2001.“Power, Knowledge and Political Spaces inthe Framing of Poverty Policy.” Institute ofDevelopment Studies Working Paper 143.Brighton, U.K.: IDS. <http://www.ids.ac.uk/

REFERENCES

1 9 1

ids/bookshop/wp/wp143.pdf>accessed March2005.

Candler, W., and N. Kumar. 1998. The Dairy Rev-olution: The Impact of Dairy Developmentin India and the World Bank’s Contribution.Operations Evaluation Department Study Se-ries. Washington, D.C.: World Bank.

Chabal, P,. and J.P. Daloz. 1999. Africa Works: Dis-order as Political Instrument. Oxford: JamesCurrey.

Chambers, R. 1995. “Paradigm Shifts and thePractice of Participatory Research and Devel-opment.” In N. Nelson and S. Wright, eds.,Power and Participatory Development The-ory and Practice. London: Intermediate Tech-nology.

Cleaver, F. 1999. “Paradoxes of Participation:Questioning Participatory Approaches to De-velopment.” Journal of International De-velopment 11(4): 597–612.

———. 1998. “There’s a Right Way to Do It –Informal Arrangements for Local ResourceManagement in Zimbabwe.” Waterlines 16(4).

———. 1997. “Gendered Incentives and Infor-mal Institutions: Women, Men and the Man-agement of Water.” Discussion Paper Series 2,No. 3, Development and Project PlanningCenter. University of Bradford, U.K.

Coady, David, Margaret Grosh, and John Hod-dinott. 2004. Targeting of Transfers in De-veloping Countries Review of Lessons andExperience: Regional and Sectoral Studies.Washington, D.C.: World Bank and IFPRI.

Collinson, M.P. 1982. Farming Systems Researchin Eastern Africa: The Experience of CIMMYTand Some National Agricultural ResearchServices, 1976–81. International Wheat andMaize Improvement Center. Nairobi: CIMMYT.

Conning, J., and M. Kevane. 2002. “Community-Based Targeting Mechanisms for Social SafetyNets: A Critical Review.” World Development30(3): 375–94

Cornwall, A. 2003. “Whose Voices? WhoseChoices? Reflections on Gender and Partici-patory Development.” World Development31(8):1325–42.

———. 2002. “Making Spaces, Changing Places:Situating Participation in Development.” IDSWorking Paper 170. Brighton, U.K.

Costa, Alberto, C.G. Conrad, P. Kottak, andRosane M. Prado. 1997. “The SociopoliticalContext of Participatory Development inNortheastern Brazil.” Human Organization56(2).

da Cunha, Paulo Vieira, and Maria Valeria JunhoPena. 1997. “The Limits and Merits of Partic-ipation.” World Bank Policy Research WorkingPaper 1838. Washington, D.C.

Dahl-Ostergaard, T., D. Moore, V. Ramirez, M.Wenner, and A. Bonde. 2003. “Community-Dri-ven Rural Development: What Have WeLearned?” Inter-American Development BankSustainable Development Department Tech-nical Paper Series. Washington, D.C.<http://www.iadb.org/sds/doc/RUR-COWI-E.pdf> accessed March 2005.

Das Gupta, M., H. Grandvoinnet, and M. Ro-mani. 2000. “State-Community Synergies inDevelopment: Laying the Basis for CollectiveAction.” World Bank Policy Research Work-ing Paper 2439. Washington, D.C.

de Clercq, Francine. 1994. “Putting CommunityParticipation into Development Work: TheDifficult Case of Winterveld.” DevelopmentSouth Africa 11(3).

Desai, V. 1996. “Access to Power and Participa-tion.” Third World Planning Review 18(2):217–42.

Development Committee. 2004. “Global Moni-toring Report 2004. Policies and Actions forAchieving the MDGs and Related Outcomes.”DC2004-0006. Joint Ministerial Committee ofthe Boards of Governors of the Bank and theFund on the Transfer of Real Resources toDeveloping Countries. Washington, D.C.

Devereux, Stephen, and Kay Sharp. Forthcoming.“Trends in Poverty and Destitution in Wollo,Ethiopia. Journal of Development Studies.

Elbers, Chris, Peter F. Lanjouw, Johan A. Mistiaen,Berk Ozler, and Ken Simler. 2004. “On theUnequal Inequality of Poor Communities.”The World Bank Economic Review 18(3).

Einhorn, Jessica. 2001. “The World Bank’s Mis-sion Creep.” Foreign Affairs 80(5): 22.

Estrella, Marisol, and John Gaventa. 1998. “WhoCounts Reality? Participatory Monitoring andEvaluation A Literature Review.” IDS WorkingPaper 70. Brighton, U.K.


1 9 2

Forster, R. 1998. “GTZ’s Experience with Main-streaming Primary Stakeholder Participation.”Paper presented at the International Confer-ence on Mainstreaming and Up-Scaling of Pri-mary Stakeholder Participation—LessonsLearned and Ways Forward. Washington D.C.,November 19–20, 1998.

Garcia, X.M., and S-A. Way. 2003. “WinningSpaces: Participatory Methodologies in RuralProcesses in Mexico.” IDS Working Paper 180.Brighton, U.K.

Gibson, C., and S.A. Marks. 1995. “TransformingRural Hunters into Conservationists: AnAssessment of Community-Based WildlifeManagement Programs in Africa.” World De-velopment 23(6):941–57.

Greiner, L.E. 1972. “Evolution and Revolutionas Organizations Grow.” Harvard BusinessReview 50 (July-August): 37–46.

Grootaert, C. 1998. “Social Capital: The MissingLink?” World Bank Social Capital InitiativeWorking Paper 3. Washington, D.C.

Gueye, B. 1999. “Whither Participation? Experi-ence from Francophone West Africa.” Dry-lands Issue Paper 87. International Institutefor Environment and Development. DrylandProgram<http://www.iied.org/docs/dry-lands/dry_ip87eng.pdf> accessed March2005.

Gugerty, Mary Kay, and Michael Kremer. 2000.“Outside Funding of Community Organiza-tions: Benefiting or Displacing the Poor?” Na-tional Bureau of Economic Research WorkingPaper 7896. Cambridge, MA.

Handy, C. 1985. Understanding Organizations.Harmondsworth: Penguin.

Hentschel, J. 1994. “Does Participation Cost theWorld Bank More? Emerging Evidence.” WorldBank Human Resource Development and Op-erations Policy (HRO) Working Paper 31. Wash-ington, D.C.

IBGE (Instituto Brasileiro de Geografia e Es-tadística). 2003. Perfil dos MunicípiosBrasileiros: Gestão Publica 2001, Rio deJaneiro. <http://www.ibge.gov.br/home/esta-tistica/economia/perfilmunic/2001/munic2001.pdf> accessed March 2005.

Isham, J., and S. Kähkönen. 2002. “InstitutionalDeterminants of the Impact of Community-

Based Water Services: Evidence from Sri Lankaand India.” Economic Development and Cul-tural Change 50(3): 667–91.

Isham, Jonathan, Deepa Narayan, and Lant Pritch-ett. 1995. “Does Participation Improve Per-formance? Establishing Causality withSubjective Data.” The World Bank EconomicReview 9(2):175–200.

Jenkins, R., and A.M. Goetz. 1999. “Accounts andAccountability: Theoretical Implications ofthe Right-to-Information Movement in India.”Third World Quarterly 20(3):603–22.

Jones, Emma, and SPEECH. 2001. “ ‘Of OtherSpaces’ Situating Participatory Practices: ACase Study from South India.” IDS WorkingPaper 137. Brighton Sussex, U.K.

Kähkönen, S. 1999. “Does Social Capital Matterin Water and Sanitation Delivery? A Review ofLiterature.” World Bank Social Capital Initia-tive Working Paper 9. Washington, D.C.

Kent, B.R., and J. Rimarachin. 1994. “Rural Pub-lic Works Construction in the Andes of North-ern Peru.” Third World Planning Review16(4):358–74.

Kerr, J., with G. Pangare, V.L. Pangare, and P.George. 2000. “An Evaluation of Dryland Wa-tershed Development Projects in India.” EPTDDiscussion Paper 68. Washington, D.C.: IFPRI.

Khwaja, A.I. 2003. “A Note on Measuring Em-powerment—An Economist’s Perspective.”Kennedy School of Government Harvard Uni-versity. Paper presented at the Workshop on“Measuring Empowerment: Cross-Discipli-nary Perspectives,” World Bank, Washington,D.C., February 4–5, 2003.

———. 2000. “Can Good Projects Succeed in BadCommunities? Collective Action in PublicGood Provision” Harvard University. Paperpresented at the Annual Bank Conference onDevelopment Economics, May 2, 2001.

Kleemeier, Elizabeth. 2000. “The Impact of Par-ticipation on Sustainability: An Analysis of theMalawi Rural Piped Scheme Program. WorldDevelopment 28(5): 929–44.

Kolavalli, S., and J. Kerr. 2002. “Scaling up Par-ticipatory Watershed Development in India.”Development and Change 33(2): 213–35.

Koopman, J., and R. Kweka, M. Mboya, and S.W.Wangwe. 2001. “Community Participation in

1 9 3

REFERENCES

Traditional Irrigation Scheme RehabilitationProjects in Tanzania: Report of a CollaborativeResearch Project.” Ministry of Agriculture andCooperatives, Irrigation Section, Dar esSalaam, Tanzania.

Korten, D.C. 1980. “Community Organizationand Rural Development: A Learning ProcessApproach.” Public Administration Review40(5): 480–511.

Krishna, A. 2004. “Partnership Between ElectedLocal Governments and Community-BasedOrganizations: Exploring the Scope for Syn-ergy.” Social Development Papers, CommunityDriven Development, Number 52/February.

———. 2003. “Falling into Poverty: The OtherSide of Poverty Reduction.” Special Article,Economic and Political Weekly. HitkariHouse, Mumbai, India.

———. 2002. Active Social Capital. Tracing theRoots of Development and Democracy. NewYork: Columbia University Press

———. 2001. “Moving from the Stock of SocialCapital to the Flow of Benefits: The Role ofAgency.” World Development 29(6): 925–43.

Krishna, A., N. Uphoff, and M. Esman (eds.).1997. Reasons for Hope: Instructive Experi-ences in Rural Development. West Hartford,CT: Kumarian.

Krishna, A., M. Kapila, M. Porwal, and V. Singh.2003. “Falling into Poverty in a High-GrowthState: Escaping Poverty and Becoming Poor inGujarat Villages.” Special Article, Economicand Political Weekly. Hitkari House, Mumbai,India

Krishna, A., M. Kapila, S. Pathak, M. Porwal, K.Singh, and V. Singh. 2004. “Falling into Povertyin Villages of Andhra Pradesh: Why PovertyAvoidance Policies are Needed.” Special Arti-cle, Economic and Political Weekly. HitkariHouse, Mumbai, India

Krishna, A., P. Kristjanson, M. Radeny, and W.Nindo. 2004. “Escaping Poverty and Becom-ing Poor in 20 Kenyan Villages.” Journal ofHuman Development, 5(2): 211–26.

———. 2002. “The Challenges of CommunityParticipation in Forest Development in Nepal.”OED Working Paper Series. Washington, D.C..

Kumar, N. 2003. “Community-Driven Develop-ment: Lessons from the Sahel—An Analytical

Review.” OED Working Paper Series. Wash-ington, D.C.

Kumar, N., N. Saxena, Y. Alagh, and K. Mitra.2000. India: Alleviating Poverty through For-est Development. Evaluation Country CaseStudy Series, Operations Evaluation Depart-ment. Washington, D.C.: World Bank.

Kumar, Sanjay, and Stuart Corbridge. 2002. “Pro-grammed to Fail? Development Projectsand the Politics of Participation.” The Journalof Development Studies 39(2) 830–51.

Leach, M., R. Mearns, and I. Scoones. 1999. “En-vironmental Entitlement: Dynamics and In-stitutions in Community-Based NaturalResource Management.” World Development27(2): 225–47.

Leipziger, Danny, Marianne Fay, Quentin Woden,and Tito Yepes. 2003. “Achieving the Millen-nium Development Goals—A Multi-SectoralApproach Revisited.” World Bank PolicyResearch Working Paper 3163. Washington,D.C.

Linden van der, J. 1997, “On Popular Participa-tion in a Culture of Patronage; Patrons andGrassroots Organization in a Sites and ServicesProject in Hyderabad, Pakistan.” Development9 (1): 81–90.

Lipton, Michael. n.d. “Rural Poverty Reduction:The Neglected Priority.” Work at the PovertyResearch Unit, Sussex University, for the In-ternational Fund for Agriculture Develop-ment’s Report on Rural Poverty.

Manor, J. 1999. The Political Economy of De-mocratic Decentralisation. Directions in De-velopment. Washington D.C.: World Bank.

Mansuri, G., and V. Rao. 2004. “Community-Based and -Driven Development: A Critical Re-view.” World Bank Research Observer 19(1):1–39.

Meinzen-Dick, R.S., K.V. Raju, and A. Gulati. 2002.“What Affects Organization and Collective Ac-tion for Managing Resources? Evidence fromCanal Irrigation Systems in India.” World De-velopment 30(4):649–66.

Minot, N., and B. Baulch. 2004. “Poverty Mappingwith Aggregate Census Data.” MSSD Discus-sion Paper 49. Washington, D.C.: IFPRI.<http://www.ifpri.org/divs/mtid/dp/pa-pers/mssdp49.pdf> accessed March 2005.


1 9 4

Molyneux, M. 2002. “Gender and the Silences ofSocial Capital: Lessons from Latin America.”Development and Change 33(2):167–88.

Moore, Mick. 2001. “Empowerment at Last?”Journal of International Development13(3):321–29.

Moore, Mick, and Anuradha Joshi. 1999. “Be-tween Cant and Corporatism: Creating an En-abling Environment for the Poor.” IDS Bulletin30(4). Brighton, U.K.

Narayan, Deepa. 1995. The Contribution ofPeople’s Participation Evidence from 121Rural Water Supply Projects. Environmen-tally Sustainable Development OccasionalPaper No. 1. Washington, D.C.: World Bank.

Narayan, Deepa, and Lant Pritchett. 1997. “Centsand Sociability: Household Income and SocialCapital in Rural Tanzania.” World Bank PolicyResearch Working Paper 1796. Washington,D.C.

Nelson N., and S. Wright (eds.). 1995. Powerand Participatory Development Theory andPractice. London: Intermediate Technology.

North, D.C. 1991. Institutions, InstitutionalChange and Economic Performance. Politi-cal Economy of Institutions and Decisions.Cambridge, U.K.: Cambridge University Press.

Oakley, Peter, and others. 1991. Projects withPeople: The Practice of Participation in RuralDevelopment. Geneva: ILO.

OED (Operations Evaluation Department, WorldBank). 2004a. Monitoring and Evaluation:Some Tools, Methods and Approaches. Wash-ington, D.C.: World Bank.

———. 2004b. Books, Building and LearningOutcomes: An Impact Evaluation of WorldBank Support to Basic Education in Ghana.Operations Evaluation Department Study Se-ries. Washington, D.C.: World Bank.

———. 2003c. An OED Review of Social Devel-opment in Bank Activities. Operations Eval-uation Department. Washington, D.C.

———. 2002a. IDA’s Partnership for PovertyReduction: An Independent Evaluation ofFiscal Years 1994–2000. Operations EvaluationDepartment. Washington, D.C.: World Bank.

———. 2002b. Social Funds: Assessing Effec-tiveness. Operations Evaluation DepartmentStudy Series. Washington, D.C.: World Bank.

———. 2002c. The Next Ascent: An Evalua-tion of the Aga Khan Rural Support Pro-gram, Pakistan. Operations Evaluation De-partment Study Series. Washington, D.C.:World Bank.

———. 2001. “The National Agricultural Re-search Project, the National Agricultural Ex-tension Project, the National LivestockServices Project, the Agricultural Diversifica-tion Project, and the Agricultural Sector In-vestment Project “(sector overview). ProjectPerformance Assessment Report. Report No.22439, Operations Evaluation Department.Washington, D.C.

———. 1999a. Nongovernmental Organiza-tions in World Bank-Supported Projects. Op-erations Evaluation Department. Report No.19061. Washington, D.C.: World Bank.

———. 1999b. Agricultural Extension Projectsin Kenya—An Impact Evaluation. ReportNo. 19523. Washington, D.C.: World Bank.

———. 1998. The World Bank’s Experience withPost-Conflict Reconstruction. Operations Eval-uation Department Study Series. Washing-ton, D.C.: World Bank.

Ostrom, E. 1999. “Social Capital: A Fad or a Fun-damental Concept?” In P. Dasgupta and I. Ser-ageldin, eds., Social Capital: A MultifacetedPerspective. Washington, D.C.: World Bank.

Overseas Development Institute. 1996. “The Im-pact of NGO Development Projects.” BriefingPaper 2/96. <http://www.odi.org.uk/publica-tions/briefing/2_96.html> accessed March2005.

Pantoja, E. 2000. “Exploring the Concept of SocialCapital and its Relevance for Community-BasedDevelopment: The Case of Coal Mining Areasin Orissa, India.” World Bank Social Capital Ini-tiative Working Paper 18. Washington D.C.

Picciotto, Robert. 1992. “Participatory Develop-ment Myths and Dilemmas.” Participatory De-velopment Working Paper 930. WashingtonD.C.: World Bank.

Pimbert, M., V. Bainbridge, S. Foerster, K. Pasteur,G. Pratt, and I.Y. Arroyo. 2000. “Transform-ing Bureaucracies: Institutionalizing Partici-pation and People Centered Processes inNatural Resource Management: An AnnotatedBibliography.” London: IIED. <http://

1 9 5

REFERENCES

www.iied.org/docs/sarl/AnnotBib20001.pdf>accessed March 2005.

Platteau, Jean-Philippe. 2003. “Institutional andDistributional Aspects of Sustainability inCommunity-Driven Development.” Paper pre-pared for the Fifth Biennial World Bank Con-ference on Evaluation and Development,Evaluating Development Effectiveness: Chal-lenges and the Way Forward. Washington D.C.,July 5–16, 2003.

Platteau, J-P, and F. Gaspart. 2003. “The Risk ofResource Misappropriation in Community-Driven Development.” World Development31(10):1687–1703.

Pretty, Jules. 1998. “Sustainable Agricultural In-tensification: Farmer Participation, Social Cap-ital and Technology Design.” Paper Preparedfor the World Bank’s Rural Week, Washington.D.C.

Pretty, Jules, and Hugh Ward. 2001. “Social Cap-ital and the Environment.” World Develop-ment 29(2) 209-227.

Pritchett, Lant, and Michael Woolcock. 2004.“Solutions When the Solution is the Problem:Arraying the Disarray in Development.” WorldDevelopment 32(2): 191–212.

Pronk, Jan P. 2003. “Aid as a Catalyst: A Rejoinder.”Development and Change 34(3):383–400.

Rajasekhar, D., and Veerashekhararappa. 2004.“Role of Local Organizations in Water Supplyand Sanitation Sector.” A Study in Karnatakaand Uttaranchal States, India. Institute forSocial and Economic Change. Bangalore,India.

Rajasekhar, D., D.V. Gopalappa, and MadhushreeSekhar. 2004. “Role of Local Organizations inWatershed Development.” A Study in Kar-nataka and Uttaranchal States, India. Insti-tute for Social and Economic Change.Bangalore, India.

Rao, V., and A.M. Ibáñez. 2003. “The Social Im-pact of Social Funds in Jamaica: A Mixed-Methods Analysis of Participation, Targeting,and Collective Action in Community-DrivenDevelopment.” World Bank Policy ResearchWorking Paper 2970. Washington, D.C.

Ravallion, Martin, and Gaurav Datt. 1996. “India’sCheckered History in the Fight againstPoverty: Are There Lessons for the Future?”

Economic and Political Weekly 31 (SpecialNumber):2479–86.

Renata, Elad L. and Jack E. Houston 2002. “Sea-sonal Labor Constraints and Intra-householdDynamics in the Female Fields of SouthernCameroon.” Agricultural Economics 27:23–32.

Ribot, J.C. 2001. “Integral Local Development: ‘Ac-commodating Multiple Interests’ through En-trustment and Accountable Representation.”International Journal of Agricultural Re-sources, Governance and Ecology 1(3/4):327–50.

———. 1998. “Decentralization, Participationand Accountability in Sahelian Forestry: LegalInstruments of Political-Administrative Con-trol.” Institute of International Studies. Berke-ley Workshop on Environmental Politics.Working Paper WP00-5-Ribot. <http://repos-itories.cdlib.org/iis/bwep/WP00-5-Ribot> ac-cessed March 2005.

———. 1995. “From Exclusion to Participation:Turning Senegal’s Forestry Policy around?”World Development 23(9):1587–99.

Romeo, Leonardo G. 2003. “The Role of Exter-nal Assistance in Supporting DecentralizationReform.” Public Administration and Devel-opment 23: 89–96.

SAPROS & IFAD. 2002. Nepal—A Study of RuralHill Potentials and Service Delivery Systems.Sapros:Kathmandu, Nepal.

SAPROS & World Bank. 2000. Delivery of RuralDevelopment Services, A Comparative As-sessment of Alternative Institutional Arrange-ments with a Focus on the Terai. Sapros:Kathmandu, Nepal.

Schady, Norbert R. 2000. “Picking the Poor: In-dicators for Geographic Targeting in Peru.”World Bank Policy Research Working Paper2477. Washington, D.C.

Schneider, H. 1999. “Participatory Governance forPoverty Reduction.” Journal of InternationalDevelopment 11(4):521–34.

Schou, Arild. 2002. “Popular Participation in De-centralised Service Delivery and Social Funds.”NIBR Working Paper 124. Oslo.

Schroeder, Larry. 2000. “Social Funds andLocal Government: The Case of Malawi.”Public Administration and Development20:423–38.


1 9 6

Schultz, Theodore W. 1964. Transforming Tra-ditional Agriculture. New Haven, CT: YaleUniversity Press.

Sen, Amartya. 1999. Development as Freedom.New York: Alfred A Knopf.

Serrano, R. 1996. “Who Knows What’s Best forthe Poor? Demand-Driven Policies and RuralPoverty in Northeast Brazil.” Thesis, MIT.

Shepherd, A. 1998. “Participatory Environmen-tal Management: Contradiction of Process,Project and Bureaucracy in the HimalayanFoothills.” In J. Blackburn and J. Holland,eds., Who Changes? Institutionalizing Par-ticipation in Development. London: Inter-mediate Technology.

Tendler, Judith. 2000. “Why Are Social Funds SoPopular?” In Shahid Yusuf, Simon Evenett,and Weiping Wu, eds., Local Dynamics in anEra of Globalization: 21st Century Catalystsfor Development. Oxford, U.K.: Oxford Uni-versity Press.

———. 1997. Good Government in the Tropics.Baltimore and London: The John HopkinsUniversity Press.

———. 1989. “What Ever Happened to PovertyAlleviation?” World Development 17(7):1033–44.

Thompson, John. 1995. “Participatory Ap-proaches in Government Bureaucracies: Fa-cilitating the Process of Institutional Change.”World Development 23(9):1521–54.

Tripp, Aili Mari. 2001. “Non-Formal Institutions,Informal Economics, and the Politics ofInclusion.” Discussion Paper No. 2001/108.World Institute for Development EconomicsResearch/United Nations University. NewYork.

Turton, C., and J. Farrington. 1998. “EnhancingRural Livelihoods through Participatory Wa-tershed Development in India.” Natural Re-source Perspectives 34, London: ODI.

UNDP (United Nations Development Program).2004. Human Development Report 2004: Cul-tural Liberty in Today’s Diverse World.<http://hdr.undp.org/reports/global/2004/>accessed March 2005

Uphoff, N., and C.M. Wijayaratna. 2000. “Demon-strated Benefits from Social Capital: The Pro-ductivity of Farmer Organizations in Gal Oya,

Sri Lanka.” World Development 28(11):1875–90.

Uphoff, N., M. Esman, and A. Krishna. 1998.Reasons for Success. Learning from Instruc-tive Experiences in Rural Development.Bloomfield, CT: Kumarian.

van de Walle, N., and T.A. Johnston. 1996. “Im-proving Aid to Africa.” Policy Essay No. 21.Baltimore: Overseas Development Council.

van der Linden, J. 1997. “On Popular Participa-tion in a Culture of Patronage; Patrons andGrassroots Organization in a Sites and ServicesProject in Hyderabad, Pakistan.” Environ-ment and Urbanization 9(1):81–90.

van Donge, Jankees. 2003. Nurtured from aboveand Growing from the Roots: Social Fundsand Decentralization in Zambia andMalawi. The Hague: Institute of Social Stud-ies. <http://www.devstud.org.uk/publica-tions/papers/conf03/dsaconf03vandonge.pdf>accessed March 2005.

van Zyl, J., L. Sonn, and A. Costa. 2000. “Decen-tralized Rural Development, Enhanced Com-munity Participation and Local GovernmentPerformance: Evidence from the NortheastBrazil.” Unpublished paper.

Vyasulu, P., and V. Vyasulu. 1999. “Women in Pan-chayati Raj: Grassroots Democracy in India. Ex-perience from Malgudi.” Paper presentedat the Meeting on Women and Political Par-ticipation: 21st Century Challenges, UNDP,New Delhi, March 24–26, 1999. http://mag-net.undp.org/events/gender/india/VYA-SULU3.htm

Wagel, S., and P. Shah. 2003. “Case Study 2—Porto Alegre, Brazil: Participatory Approachesin Budgeting and Public ExpenditureManagement.” Social Development Note 71,World Bank. <http://www.worldbank.org/participation/sdn/snd71.pdf> accessed March2005.

Walter, Ian, Rafael Del Cid, Fidel Ordoñez, andFlorencia Rodríguez. 1999. “Ex-Post Evaluationof the Honduran Social Investment Fund.”Report for the World Bank. <http://wbln0018.worldbank.org/HDNet/HDdocs.nsf/socialfunds/49898ac982b7c7de852568e2005d75ce/$FILE/FHISexecsummary.pdf> accessedMarch 2005.

1 9 7

REFERENCES

Wassenich, Paul, and Katherine Whiteside. 2004.“CDD Impact Assessments Study: OptimizingEvaluation Design Under Constraints.” SocialDevelopment Papers Number 51, February.

Watt, Patrick 2001. “Community Support forBasic Education in Sub-Saharan Africa.” WorldBank Africa Region Human DevelopmentWorking Paper Series. Washington, D.C.

Weinberger, K., and J.P. Jutting. 2001. “Women’sParticipation in Local Organizations: Condi-tions and Constraints.” World Development29(8):1391–1404.

White, S.C. 1996. “Depoliticising Development:the Uses and Abuses of Participation.” De-velopment in Practice 6(1):6–15.

Wilks, Alex. 2001. “Overstretched and Under-loved: World Bank Faces Strategy Decisions.”Bretton Woods Project. Washington, D.C.

Williams, G. 2003. “Toward a Re-politicisation ofDevelopment: Political Capabilities and Spaceof Empowerment.” Paper presented at theconference Participation: From Tyranny toTransition? Exploring New Approaches to Par-ticipatory Development,” University of Man-chester, Manchester, February 27–28,

World Bank. 2004a. “Community-Driven Devel-opment in Local Government Capacity Build-ing Projects: Emerging Approaches in Africa.”Social Development Notes No. 86 (Novem-ber). Washington, D.C.

———. 2004b. “Case Study Summaries for Re-ducing Poverty Sustaining Growth, ScalingUp Poverty Reduction.” A Global LearningProcess and Conference, Shanghai, May 25–27,2004.

———. 2004c. “Negotiating Social Change inRural Nepal: Crosscutting Gender, Caste andEthnic Dimensions of Empowerment and So-cial Inclusion. Measuring Empowerment andSocial Inclusion Study: Phase I Report, SocialDevelopment.” Draft.

———. 2004d. World Development Report 2004:.Making Services Work for Poor People. Wash-ington, D.C.

———. 2004e. “Community-Driven Developmentin Urban Upgrading.” Social DevelopmentNotes No. 85 (February). Washington, D.C.

———. 2003a. “The Role of Selectivity in En-hancing IDA Effectiveness: A Progress Report.”

IDA/R2003-0199. Washington, D.C.<http://s i teresources .worldbank.org/ IDA/Re-sources/MTRselectivityERS.pdf> accessedMarch 2005.

———. 2003b. PRSP Sourcebook. Washington,D.C.

———. 2003c. “Review of Bank Supervision ofCDD Operations in the EAP Region” (Draft).EAP, VPU. <http://lnweb18.worldbank.org/ESSD/sdvext.nsf/09ByDocName/ReviewOfBankSupervisionOfCDOperationsInTheEAPRegion-RevisedDraft/$FILE/EAPCD-SupervisionRevisedRptMar2003.pdf> ac-cessed March 2005.

———. 2003d. “Evaluating Social Funds: A Cross-Country Analysis of Community Investments.”Regional and Sectoral Studies. Washington,D.C.

———. 2003e. “Community-Driven Recon-struction as an Instrument in War-to-PeaceTransitions.” CPR Working Paper No. 7, August.Social Development Department ESSD Net-work. <http://lnweb18.worldbank.org/ESSD/sdvext.nsf/60ByDocName/Community-dri-venReconstructionasanInstrumentinWar-to-PeaceTransitions/$FILE/SDP43-CPRWP07Aug21.pdf> accessed March 2005.

———. 2002a. “A Tale of Two Projects: Commu-nity-Based Reconstruction in East Timor andRwanda.” Social Funds Innovation Update2(4). Human Development Network-SocialProtection Unit, Social Funds Thematic Group.Washington, D.C.

———. 2002b. Empowerment and Poverty Re-duction: A Sourcebook. <http://sitere-sources.worldbank.org/INTEMPOWERMENT/Resources/486312-1095094954594/draft.pdf>accessed March 2005.

———. 2002c. “Sleeping on our Own Mats: AnIntroductory Guide to Community BasedMonitoring and Evaluation.” Africa Region,Rural Development II. Washington, D.C.

———. 2001a. “Decentralization and Gover-nance: Does Decentralization Improve PublicService Delivery?” PREM Note Number 55.Washington, D.C.

———. 2001b. “World Bank Strategic Frame-work” January 24. Washington, D.C.<http://siteresources.worldbank.org/EXTABOU-


1 9 8

1 9 9

TUS/Resources/strategic.pdf> accessed March2005.

———. 2001c. “Empowering the Poor throughDecentralization: Brazil Rural Poverty Allevi-ation Program.” Social Development NoteNo. 51. Washington, D.C.

———. 2000. “The Community Driven De-velopment Approach in the Africa Region:A Vision of Poverty Reduction Through Em-powerment.” Washington, D.C.

———. 1996. The World Bank ParticipationSourcebook. Environmentally Sustainable De-velopment Network. Washington, D.C.

———. 1995. “Designing Community Based De-velopment” (by Deepa Narayan). Environ-ment Department Paper No. 007. ParticipationSeries. Washington, D.C.

World Bank and IMF 2004. “Global MonitoringReport 2004: Policies and Actions for Achiev-

ing the MDGs and Related Outcomes.” Back-ground Paper, Development Committee (JointMinisterial Committee of the Boards of Gov-ernors of the Bank and the Fund) on theTransfer of Real Resources to DevelopingCountries. Washington, D.C.

Wunsch, James S. 2001. “Decentralization, LocalGovernance and ‘Recentralization’ in Africa.”Public Administration and Development 21:277-88.

Yukio, Ikemoto. 2001. “Poverty Alleviation Poli-cies and Ethnic Minority People in Vietnam.”Institute of Oriental Culture, University ofTokyo. <http://www.st-edmunds.cam.ac.uk/vhi/sen/papers/ikemoto.pdf> accessed March2005.

REFERENCES

Date post:	02-Jun-2020
Category:	Documents
Upload:	others
View:	4 times
Download:	0 times