Power, Politics, and Performance

Community Participation in South African

Public Works Programs

Michelle AdatoJohn HoddinottLawrence Haddad

RESEARCHREPORT 143INTERNATIONAL FOOD POLICY RESEARCH INSTITUTEWASHINGTON, D.C.

Copyright © 2005 International Food Policy Research Institute. All rights reserved.Sections of this material may be reproduced for personal and not-for-profit usewithout the express written permission of but with acknowledgment to IFPRI. Toreproduce the material contained herein for profit or commercial use requires expresswritten permission. To obtain permission, contact the Communications Division<ifpri-copyright@cgiar.org>.

International Food Policy Research Institute2033 K Street, NWWashington, DC 20006-1002 USATelephone +1-202-862-5600www.ifpri.org

Library of Congress Cataloging-in-Publication Data

Adato, Michelle.Power, politics, and performance : community participation in South African public

works programs / Michelle Adato, John Hoddinott, and Lawrence James Haddad.p. cm. — (Research report ; 143)

Includes bibliographical references.ISBN 0-89629-147-2 (alk. paper)1. Public works—South Africa. 2. Community development—South Africa.

I. Hoddinott, John. II. Haddad, Lawrence James. III. Title. IV. Research report(International Food Policy Research Institute) ; 143.HD4350.5.A33 2005363.60968—dc22 2005027717

Contents

List of Tables iv

List of Figures v

Glossary vi

Foreword viii

Acknowledgments ix

Summary x

1. Introduction 1

2. Participatory Development and the Origins of Community-Based Public Works in South Africa 4

3. Research Methods 13

4. The Impact of Participation on Project Outcomes: Evidence from the Full Sample 20

5. Community Participation in the Western Cape Projects 38

6. Social, Political, and Institutional Challenges to Participation 54

7. Conclusions 67

References 70

iii

Tables

3.1 Public works programs in Western Cape province included in the study 14

3.2 Description of data collection methods 15

3.3 Data collected at the project-level, 101 Western Cape public works projects 16

3.4 Summary profile of the eight in-depth case studies selected from the 101 projects 17

4.1 Project characteristics 27

4.2 Mean project performance by degree of community participation 28

4.3 Probit estimates of the determinants of de facto participation 31

4.4 Reduced form determinants of project performance 32

4.5 Parameter estimates for the first stage residual in the determinants of project performance 33

4.6 The impact of participation on attainment of project performance 34

4.7 The impact of participation on the distribution of project benefits 35

4.8 The impact of participation on project cost-effectiveness 36

5.1 Quantitative indicators for the eight case study projects 39

5.2 Comparison of de facto participation in the case study and full project database 40

5.3 Types of participation where broad community involvement occurred 43

5.4 Responsibilities of role-players in case-study projects 51

6.1 Factors limiting the roles of the project steering committees (PSCs), case-study projects 55

iv

Figures

3.1 Western Cape province: Magisterial districts of the case studies 18

v

Glossary

ABLIC Accreditation Board for Labour-Intensive Construction

ACDP African Christian Democratic Party

ANC African National Congress

CAG Clean and Green Programme

CBO Community-based organization

CBPWP Community-Based Public Works Programme

CEP Community Employment Programme

CLO Community Liaison Officer

CNC Cape Nature Conservation

COSATU Congress of South African Trade Unions

CSS Central Statistical Services

DBSA Development Bank of Southern Africa

DP Democratic Party

DPW Department of Public Works (national)

DTPW Department of Transport and Public Works (Western Cape)

DWAF Department of Water Affairs and Forestry

FF Freedom Front

FWCP Fynbos Water Conservation Programme

FWWP Fynbos Working for Water Programme

GEAR Growth, Employment and Redistribution Programme

IDT Independent Development Trust

IFPRI International Food Policy Research Institute

NCC National Co-ordinating Committee of the NEF

NEF National Economic Forum

NGO Nongovernmental organization

NP National Party

NPWP National Public Works Programme

vi

OHS October Household Survey

PAC Pan African Congress

PDC Provincial Development Council

PIR Poverty and Inequality Report

PSC Project Steering Committee

PSLSD SALDRU/World Bank Project on Statistics for Living Standardsand Development

RDP Reconstruction and Development Programme

SAIRR South African Institute of Race Relations

SALDRU Southern Africa Labour and Development Research Unit

SANCO South African National Civics Organisation

SMME Small, micro, and medium enterprises

WCEDF Western Cape Economic Development Forum

WWP Working for Water Programme

GLOSSARY vii

Foreword

Community-driven development is indelible in the development landscape. It is increas-ingly visible in the policy design of many governments, nongovernmental organizations,and multilateral institutions and features in important debates involving democracy,

governance, institutions, and decentralization. As this research report points out, this has philo-sophical and instrumentalist underpinnings, with participation as both means and end. Participa-tory or community-driven development is advocated on the basis that, among other advantages,it can reduce information problems for development planners and beneficiaries, increase theresources available to poor people, and strengthen the capacity for collective action amongpoor and other marginalized societal groups.

While these arguments are persuasive, there is a need to scrutinize the benefits of partici-pation more closely, along with the complex operational problems inherent in participa-tory approaches. The unique contribution of this study—which derives from IFPRI’s researchprograms on targeted safety net programs, policy processes, and governance—is threefold.First, it examines an innovation in public works programs designed to achieve multiple de-velopment objectives, from job creation to community empowerment; second, it uses a unique,project-level quantitative dataset and econometric analysis to uncover hard evidence of theeffects of different forms of participation on key outcomes of a safety net program; and, third,it uses rich case study data to explain a range of ways in which participation can affect the out-comes of public works programs. The report also explores the complexities of institutionalarrangements and policy processes involving government, the private sector, and community-based organizations representing a diverse and often conflicting set of values, identities, andinterests. Such an understanding of the policy process is key to understanding why policiessucceed or fail in achieving the expected results.

The study finds that community participation does lead to improved project outcomes, butit is not an easy road to travel. The authors point out that participation does not have to be allor nothing, and its best forms are likely to vary under different conditions. Different modali-ties exist that can capture local preferences, achieve accountability and transparency, and buildcapacity and local empowerment, while at the same time delivering the quality infrastructureneeded by the poor. However, while the research has shown convergence between diverseprogram objectives, policymakers will also have to resolve trade-offs by weighing economicand social priorities. Furthermore, the research shows that policy processes do not end withpolicymaking—decisions must be supported by solid commitments and institutional arrange-ments that enable agreements to be carried out and monitored.

While many of the findings of this study are specific to the South African context, manyof the issues confronted are common to the challenges of participation, community-driven de-velopment, scaling up, and policy processes encountered across the globe. The report providesresearch methods and insights for researchers, policymakers, and development practitionersseeking a better understanding of the relationship between governance and poverty alleviation.

Joachim von BraunDirector General, IFPRI

viii

Acknowledgments

T he authors thank Dudley Horner, Ridwaan Haywood, Jaqui Goldin, Boniswa DebiHene, and Faldie Esau for their important roles in carrying out the research; Tim Besleyfor assistance in developing the model in chapter 4; John Pender and two anonymous

reviewers for comments on an earlier draft of the report; Patricia Biermayr-Jenzano and NellyRose Tioco for production assistance; and the U.K. Department for International Developmentand the Rockefeller Foundation for funding the research. We also express our deep appreciationto the many people in government, NGOs, the private sector, community-based organizations,and the research communities for contributing their time and insights to this research.

ix

Summary

Following South Africa’s democratic transition in 1994, the new government launchedpublic works programs that shared four objectives: to create jobs to respond to ex-tremely high levels of poverty and unemployment; to build or rehabilitate infrastruc-

ture in poor, underserved areas, or improve the natural environment; to provide job trainingthat would enable workers to find postproject employment; and finally to build the capacityof communities to control their own development processes through community participationin public works projects.

The long national struggle for democracy, combined with exposure to democracy in for-eign institutions and in trade unions and civic associations within South Africa, infused thecountry’s new leadership and many of its constituencies with a broad democratic agenda—from “one person, one vote” to control of local development projects by elected community-based organizations. At the same time, South Africa’s history of political conflict, includingextreme racial and class polarization and violent political strife, meant that attempts to createdemocratic institutions would not be easily or wholly achieved. A national public works pro-gram, which distributed resources to reduce poverty while building capacity and promotingparticipation by local institutions, would be no exception: it involved actors who had only re-cently emerged from the apartheid system, under which they had experienced a vast political,economic, social, and cultural divide.

Through a study of seven public works programs implemented in Western Cape province,this report examines the benefits and challenges of pursuing community participation, togetherwith the effects of participation on meeting the other objectives of the programs. Althoughaspects of South Africa’s experience are unique to its political economy, the study’s findingsreveal insights, dilemmas, and possibilities of considerable relevance in the wider context ofparticipatory or “community-driven” development programs, which have increasingly becomeintegral to the development agenda throughout the world.

The extent to which public works programs achieve their objectives depends on institu-tions at several levels and their performance based on access to information, the incentivesthey face, and the requirements of a changing political, social, and economic environment.Thus this study considers key institutions, their agents, and their interrelationships in the im-plementation of public works projects. This coverage includes the intended and actual responsi-bilities of community-based committees; the advantages and disadvantages of community par-ticipation in different aspects of project selection, design, and management; the effects ofcommunity participation on project performance; and the factors that prevent community-basedcommittees from acting in the capacity intended in policy documents. Key to understandingthese constraints is the relationship of community organizations to other project-level actorsin government and the private sector, and the changes in professional identity, institutionalpractices, and local politics necessary for communities to receive the type of opportunitiesenvisioned.

The report integrates quantitative and qualitative data from a survey of 101 public worksprojects with in-depth qualitative data from eight project-level case studies. The quantitative

x

analysis examines the outcomes of different forms of participation on job creation, labor in-tensity, efficiency in transferring income to the poor, and other variables. It finds that, evenafter accounting for the endogeneity of participation, de facto participation has a statisticallysignificant, positive effect on the share of the project budget spent on labor, the log number ofdays of work created, and the log number of training days undertaken. It increases women’sshare of employment and is associated with a reduction in the ratio of the project wage to localunskilled wages. It also reduces the cost of creating employment and of transferring incometo the poor. The magnitudes of these impacts are sizeable and robust to a variety of modelspecifications and the inclusion of other covariates.

The case studies indicate the ways in which different forms of participation added value toprojects by enabling beneficiary communities to influence the choices of priority assets andproject design features, which increased safety and convenience and affected the number ofjobs created in the short and long terms. They also demonstrated the importance of regularcommunication with communities to build a sense of local ownership, increasing cooperationand the maintenance of assets. The case studies also show how community participation in-troduces politics, conflict, and lengthier decisionmaking processes—as well as how lack ofparticipation can result in even more conflict and delays further down the line, when timebecomes more expensive.

The study finds a wide gap between ideas for community-driven projects embodied in na-tional government programs and the beliefs and practices of professional public and privateproviders at provincial and local levels. Although consensus on the importance of communityparticipation has been voiced, there is also profound ambivalence about it, as well as widelydifferent ideas about what it means and where it is appropriate. Community-based committeeswere involved in some way in almost all the projects, but their roles were often limited to com-munity and worker liaison functions. Private-sector and local government project managersexcluded community participation from management tasks because of their views on effi-ciency and specialization; their lack of identification with the programs’ capacity-building andempowerment objectives, which were set at the national level; and ambiguities about the rolesof different stakeholders. Community committee members often did not have sufficient skillsand were not trained.

One policy implication might be to reduce the role of communities to opportunities forcommunication and some consultation. Local government, if downwardly accountable to poorconstituents, could manage public works projects. An alternative policy implication is that gov-ernment should increase its commitment to improving structures and processes for commu-nity participation for the value that it adds. Participation does not have to be all or nothing,and its best forms are likely to vary under different circumstances. Where trade-offs betweenthe achievement of different program objectives are required, these must be weighed in light ofnational and local priorities. If participation and local empowerment remain important objec-tives, policymakers and program designers must creatively explore institutional arrangementsand methods for increasing local capacities. But new policy proposals alone are unlikely toeffect significant change. This would require involving influential government, private-sector,and community stakeholders in the process; identifying common ground; and securing thecommitment necessary for agreements to be carried out and monitored. Politics, conflicts ofinterest, struggles over resources, and processes of consultation and consensus-building arepart of the landscape of community-driven development. If participatory development is to re-main on South Africa’s development agenda, all actors must commit to realizing this objective,including generating sufficient resources, creativity, and patience to see the process through.

SUMMARY xi

C H A P T E R 1

Introduction

You see we consultants are used to the conventional way of working. We design, there’sa contractor that is responsible, they are the communities or the beneficiaries—theyknow nothing what is happening to the funds . . . They are just asked to work and getsalaries at the end of the day and leave. Now this situation is a new situation that needsboth parties, communities and the consultants, to make a joint effort to educate eachother about the new process because the whole process is new to us. The whole processis new also to the community because they were not involved before. They were notparticipating, so they want to participate.

Consulting engineer for the Thembaletu roads project

Between 1994 and 1995, six new public works programs were initiated by the new SouthAfrican national government, mainly administered by newly elected and transformingprovincial level governments. A seventh program was continued that had begun prior

to the 1994 election.1 These programs shared four objectives: First, to create jobs for thepoor and unemployed; second, to build or rehabilitate infrastructure, or to improve the natu-ral environment; third, to provide job training that would enable workers to find post-projectemployment; fourth, to build the capacity of communities to exert more control over their owndevelopment processes, through strengthening local institutions and community participationin public works projects. The nature and objectives of public works programs in South Africawere shaped by three corresponding considerations: the high level of unemployment in SouthAfrica;2 the backlog in infrastructure such as roads, water, and sanitation systems in blackrural and urban areas; and the new government’s development philosophy that stressed sus-tainability and democracy. Through a study of these seven programs as implemented in SouthAfrica’s Western Cape province, this report examines the benefits and challenges of attempt-ing to achieve the fourth objective—community participation—and the effects of participationon the programs’ ability to meet their other objectives.

1The six programs were: The Clean and Green (CAG) and the Community Based Public Works (CBPWP), de-veloped and financed by the National Department of Public Works (DPW); the Community Employment Pro-gramme (CBPWP/CEP), part of the CBPWP administered by the Independent Development Trust (IDT); theTransport Projects (Trans), funded by the national Department of Transport; the Working for Water Programme(WWP) of the national Department of Water Affairs and Forestry (DWAF). The seventh program was the inter-sectoral nongovernmental National Economic Forum (NEF) program begun in 1993.

2According to Klasen (1997) these were 30 percent of working age South Africans; 53 percent in the poorest20 percent of households.

1

Public works are widely used through-out Asia, Africa, and Latin America forpoverty alleviation, social insurance, infra-structure development at various levels oflabor intensity, or most commonly, all threeobjectives (Subbarao et al. 1996; Subbarao1997; Deolalikar 1995; von Braun, Teklu,and Webb 1992; von Braun 1995). SouthAfrica’s programs share these objectives butadd those of sustainable job creation, insti-tutional capacity building, and communityempowerment. The South African programsare unusual in that they envisioned long-term poverty alleviation—rather than a short-term horizon most often held by publicworks and other safety-net programs—byattempting to generate skills that enhancedpeople’s ability to secure employment afterthe project has ended. The emphasis oncommunity participation was part of thislong-term horizon, the intention of whichwas to develop community skills in organi-zation and management to enable commu-nities to access new development resourcesand engage broader political processes.

Because community participation wascentral to many aspects of the new gov-ernment’s Reconstruction and DevelopmentProgramme (RDP), our study of publicworks in the Western Cape province pro-vides one window onto the broad develop-mental changes and challenges faced bySouth Africa since its transition to democ-racy in 1994. At the same time, althoughaspects of the South Africa experience areunique to its political economy and to theapplication of a participatory approach topublic works programs, this experience hasrevealed insights, dilemmas, and possibil-ities relevant to the wider context of partic-ipatory or “community-driven” developmentprojects, which have become an increas-ingly important part of the developmentlandscape.

The extent to which labor-intensive pub-lic works programs achieve their objectivesdepends on institutions at several levels, andhow these institutions perform in the con-text of the information they have access to,

the incentives they face, and the changesrequired by new program objectives and anew political dispensation. Based on a studyof seven public works programs in WesternCape province, this report thus looks at keyinstitutions, their role-players and their re-lationships in the implementation of publicworks projects. The primary emphasis ison people in communities and community-based project steering committees (PSCs).

This report examines the intended andactual responsibilities of community-basedcommittees; the advantages and disadvan-tages of community participation in differ-ent aspects of project selection, design, andmanagement; outcomes of community par-ticipation on project performance; and thefactors that have constrained the ability ofcommunity-based committees to play therole envisioned for them in policy docu-ments. Key to understanding these con-straints is the relationship of community or-ganizations to other project-level actors ingovernment and the private sector, and thechanges in professional identity, institutionalpractices, and local politics necessary if com-munities are to receive the type of opportu-nities envisioned. The research also looks atoutcomes, exploring how participation canbe an efficient means to achieving the pro-grams’ other material objectives. It revealedtensions and trade-offs of different levels andtypes of participation. Participation does nothave to be all or nothing, and good programdesign choices adapt to different desires,capacities and objectives. Where there aretrade-offs between the achievement of dif-ferent program objectives, these must beweighed in light of nationally and locallydefined priorities.

The rest of the report is organized in thefollowing manner. Chapter 2 provides abackground to the research, discussing sev-eral inter-related aspects: some key ideas thatinfuse much of the international literatureon participation in development, the politi-cal and economic context in South Africathat explains how community-based publicworks got on the policy agenda, and a de-

2 CHAPTER 1

scriptive overview of the seven nationalpublic works programs that are included inthe research. It also provides the politicaland economic context in Western Cape prov-ince, which helps to explain the course of im-plementation of the public works programs.Chapter 3 describes the research design andmethods used in the quantitative and quali-tative research. Chapter 4 focuses on oneof the core objectives of our work: to assesswhether community participation affectsproject performance. To do so, we first setout a conceptual model to capture some ofthe analytics of participation. We use thisto help frame an econometric analysis ofproject outcomes. Chapter 5 then introduceseight case studies, which examine closelywhat community participation actually lookslike in practice, and help to explain the re-sults in Chapter 4. Chapter 5 first considerscertain key quantitative variables in the casestudies, including levels of participation andperformance outcomes of concern in thequantitative analysis in order to show how

the case studies compare to the overall data-set. Qualitative data are then used to explainwith depth the different forms that commu-nity participation took in these projects, andcompare them to potential roles for inclu-sive community-wide forums and for rep-resentative community-based structures. Wediscuss the advantages and drawbacks ofthese different types of participation, fromforms of communication to participation inproject design and management, as well aswhen and why people might not want toparticipate. Continuing to use the case studydata, Chapter 6 then analyzes the social,political, and institutional constraints onmeaningful participation, including beliefsand practices of government and the privatesector with regard to project management,specialization and efficiency, political con-flict, and other issues. Chapter 7 draws someconclusions about the need for the reclarifi-cation or redefinition of program objectivesand about what would need to take place inorder to meet these objectives.

INTRODUCTION 3

C H A P T E R 2

Participatory Development andthe Origins of Community-BasedPublic Works in South Africa

T his chapter begins with an overview of the meaning of “participatory development,”alternatively referred to in the literature as “community participation,” “community-driven development,” and other terms. It then explains the historical, political, and eco-

nomic context in which community-based public works programs were conceived in SouthAfrica, including the conditions that led to a call for public employment programs and, in par-ticular, those controlled by communities. It also details the institutional context for the devel-opment of the seven programs included in the study, and describes each of these programs.Finally, it discusses the conditions in the Western Cape province of South Africa, where weconducted our research on public works programs.

Participatory development, where local people are engaged in some active way in develop-ment planning and implementation, has a long history and a respected place among devel-opment intellectuals, policymakers, and practitioners throughout the “first” and “third”worlds.3 Participatory development, however, means very different things to different people:from “local people doing what planners wanted” (Guijt 1991) to the “education for participa-tion” approach of Paulo Freire where “men and women will analyze and critically interprettheir world and their problems, and will be able to acquire the skills necessary to respond tothem in a cooperative and democratic way” (OEF 1986 in McDonald 1995). It can involvenotions of individual empowerment as well as influencing wider processes of political change,and even redefining the term “development” itself.

Support for participation has instrumentalist, philosophical, and political underpinnings.The instrumentalist foundation involves a recognition that top-down, technocratic forms ofdevelopment imposed on diverse local realities often result in failure; that local people bestunderstand their own needs; and that involving local people can be cost-effective in terms ofreduced capital costs and increased involvement in operation and maintenance. The philo-sophical-political foundation involves the belief that poor people should be empowered andshould have more command over their lives (Chambers 1995); and that they should be em-powered “to determine choices in life and to influence the direction of change” (Moser1989, 1815).

3For reviews of a wide range of philosophies and practices of participatory development, see Nelson and Wright(1995) and Scoones and Thompson (1994). For additional perspectives see Rahman (1993), Long and Villarreal(1993), and Rahnema (1992).

4

In order to see how ideas about partici-patory development made their way intoSouth African development policy, becom-ing a centerpiece of public works programs,it is important to understand the historicalpolitical and economic conditions that char-acterized the country in the period leadingup to and following the birth of the newgovernment. In 1994, South Africa’s vastblack majority emerged from almost a cen-tury of racial oppression, segregation, andengineered poverty, codified for four decadesas the system of apartheid. The long nationalpolitical struggle for democracy, coupledwith a period of exposure to democratic in-stitutions both outside South Africa andinside the country through trade unions andcivic associations, infused the new SouthAfrican leadership (primarily in the AfricanNational Congress Alliance) and many oftheir constituencies with a broad democraticagenda—from “one-person one-vote” tocontrol of local development projects byelected community-based organizations. Afocus on democracy, participation, and citi-zen empowerment infuses the African Na-tional Congress’s (ANC)—and soon afterthe first government’s—Reconstruction andDevelopment Programme (RDP):

[T]his programme must become apeople-driven process. Our people,with their aspirations and collectivedetermination, are our most importantresource. The RDP is focused on peo-ple’s immediate needs and it relies, inturn, on their energies. Irrespective ofrace or sex, or whether they are rural orurban, rich or poor, the people of SouthAfrica must shape their own future.Development is not about the deliveryof goods to a passive citizenry. It isabout active involvement and growingempowerment (RDP 1994, 5).

Sectoral social movements and commu-nity-based organizations, formed in opposi-tion to apartheid, had a considerable historyof participatory democracy in their decision-

making structures. These were seen in theRDP as an important asset. The governmentpledged that

[S]tructured consultation processes atall levels of government will be intro-duced to ensure participation in policy-making and planning, as well as projectimplementation. The empowerment ofinstitutions of civil society is a funda-mental aim of the Government’s ap-proach to building national consensus. . . To facilitate effective involvement,the Government will introduce pro-grammes that will enhance the capacityof community organisations (RDP1994, 41).

RDP forums, to be broadly representative,were to be formed in each communityto contribute to developing planning andimplementation.

At the same time, South Africa’s his-tory of political conflict, including extremeracial and class polarization, violent politi-cal strife, and brutality characterizing thestate and spilling over into the citizenry,meant that any attempts at creating demo-cratic institutions working toward a nationalconsensus and common good, were impor-tant but elusive goals—not to be achievedeasily or wholly. A national public worksprogram distributing resources to reducepoverty while building capacity and promot-ing participation by local institutions wouldbe no exception. These programs would in-volve the same range of political and eco-nomic actors (government, private sectorcompanies, poor communities, trade unions,and others) who had struggled with eachother under the old apartheid system andhad just recently emerged from that system.

The National Public Works Programme(NPWP) was also conceived and operation-alized in an institutional context defined bySouth Africa’s unique political-economichistory. A strong trade union movement(born largely in the 1970s and progressivelygaining economic and later political power)

PARTICIPATORY DEVELOPMENT AND THE ORIGINS OF COMMUNITY-BASED PUBLIC WORKS IN SOUTH AFRICA 5

and liberal elements within the private sec-tor in 1992 formed a negotiating forum tojointly formulate strategies for economicdevelopment. In 1993, a task team withinthe National Economic Forum (NEF) wasformed to develop a proposal for a NationalPublic Works Programme (NPWP).

The focus on public works can be under-stood within the context of the high povertyand unemployment rates inherited by thenew government. Using October HouseholdSurvey (OHS) 1995 data and the broad def-inition of unemployment (the number ofpeople aged 16–64 who are not working butwould like to work and are either activelyseeking work or are too discouraged to con-tinue looking), the Poverty and InequalityReport (PIR) (May et al. 1998) generatesan estimate of 30.3 percent for all adults.Klasen (1997) also uses this broad defini-tion of unemployment for his analysis usingthe SALDRU 1993 data (PSLSD 1994). Heestimated unemployment at 29.9 percent forthe entire country.

Disaggregated by race, unemploymentunder the broad definition was estimatedfor “African” at 38.3 percent and for“Coloured” (the main racial group in theWestern Cape province where our study tookplace) at 20 percent. Poverty and unemploy-ment went hand in hand as they normallydo, but in South Africa it is exacerbated byrace:

apartheid has left the majority of thepopulation in poverty, with little accessto employment, education, health orother basic services. It enforced a spa-tially unsustainable residential pattern,leaving a majority of the poor in eco-nomically marginal areas of the coun-try, dependent on transfers and what-ever incomes they could eke out of theavailable resource base (Klasen 1997).

Klasen (1997) estimated a poverty rate ofbetween 23.7 and 56.7 percent, dependingon the poverty line used. May et al. (1998)used 1995 National Income and Expendi-

ture Survey data to derive a poverty rate of49.9 percent, with a rate of 61 percent forAfricans and 38.2 percent for the “Col-oured” population.

Facing high unemployment and povertylevels and a newly elected democratic gov-ernment pledged to alleviate these prob-lems, a wide range of institutions called forgovernment and the private sector to makejob creation a central priority. Between 1994and 1998, a number of institutions and pol-icy documents called for public works pro-grams as part of a strategy for employmentcreation. These included the NEF (NEF1994a, 1994b), the RDP (RDP 1994), theCongress of South African Trade Unions(COSATU), the Urban Foundation (UrbanFoundation 1994), the government’s Growth,Employment and Redistribution (GEAR)macroeconomic policy (Ministry of Finance1996), the Presidential Commission on theLabour Market (Department of Labour1996); the Poverty and Inequality Report(May et al. 1998); and the Department ofLabour’s Employment Strategy Framework(Department of Labour 1998).

The NEF team tasked with developingthe NPWP had the mandate to develop anational program that would not be justtemporary “make work” programs but, aspart of the Reconstruction and DevelopmentProgramme (RDP), they were to be partici-patory and result in sustainable poverty all-evation. The NPWP took its cues from theparticipatory development focus of the RDP,of which it was a part. One of the NPWP’sfour objectives is to “build the capacity ofcommunities to manage their own affairs”(NEF 1994a). A flagship program of theNPWP was the Community-Based PublicWorks Programme (CBPWP):

The CBPWP (Community-Based Pub-lic Works Programme) believes thatprojects should have strong communityparticipation. The community shouldcontrol all processes leading to the ulti-mate establishment of the asset. Thismeans that even before the process of

6 CHAPTER 2

institution building occurs, the totalcommunity should have adequately andactively participated in all the aspectsof the second phase i.e., project con-ception and prioritization. Further, thismeans that the community through itsrepresentative community structure,should make the decisions about whatshould be constructed, how it should bedesigned and constructed, who shouldwork on the project, as well as the ratesand system of employment (DPW1996, 38).

Along the project cycle, community par-ticipation can take place at different stages.This cycle includes selecting project type;many aspects of project design, from choos-ing labor versus capital intensity to func-tionality for whom; project implementation,including hiring contractors; selecting, hir-ing, and supervising workers; supervision ofconstruction; and maintenance and man-agement of assets. Bringing communitymembers into these roles would necessarilyrequire skills training and capacity develop-ment. Each stage also involves the activa-tion of a range of different material andpolitical interests.

There are also different types of partici-pation that can be employed at each of thesestages. Paul (1987) distinguishes four typesof participation: (1) information sharing;(2) consultation; (3) decisionmaking; and(4) initiating action. Moving from (1) to (4),communities tend to gain and exercisemore power and control. The intention ofthe NPWP was that participation would ide-ally involve up to and including the fourthtype but definitely the third. In practice,the type of participation varied greatly inthe 101 projects in the Western Cape andamong the projects in the case studies, withmost involving (1) and (2), and some in-volving (3) or (4).

Nelson and Wright (1995) make anotherimportant distinction, between participationas a means and participation as an end. Thefirst uses participation to accomplish the

aims of a project more efficiently or effec-tively; the second involves a community orgroup setting up a process to control its owndevelopment: “Both types imply very dif-ferent power relationships between membersof a community as well as between them andthe state and agency institutions. Simply put,the extent of empowerment and involve-ment of the local population is more limitedin the first approach than in the second.”

In South African public works programs,the focus was strongly on participation asan end, as community capacity building wasone of the main objectives. Participation wassupposed to be maximized, and communitiesempowered, without a great deal of consid-eration given to the efficiency implicationsof this choice. However, as this report willshow, participation can also be a means toan efficient outcome, and ultimately this hasto be considered in the South African con-text, because good quality infrastructure at areasonable cost is also a political and eco-nomic requisite. Although the two can becomplementary in practice, our research re-vealed a potential tension. The challenge is tofind practices that enable the maximizationof both types of goals, but where there aretrade-offs, policymakers and citizens shouldmake these explicit within policy debates.

The Socioeconomic andPolitical Context in Western Cape ProvinceAlthough community-driven developmentwas an ANC priority at the national level,implementation had to be carried out at theprovincial and local levels. It is thus equallyessential to understand the economic andpolitical context in the Western Cape prov-ince. Labor-intensive public works programsin the Western Cape are inserted into a set ofgovernment and civil society institutions atthe provincial, regional, and local levels. Inthe period following the transition to democ-racy, these institutions were in a process ofchange, yet built upon inherited structuresand racially defined relationships.

PARTICIPATORY DEVELOPMENT AND THE ORIGINS OF COMMUNITY-BASED PUBLIC WORKS IN SOUTH AFRICA 7

The Western Cape is characterized byeconomic and political conditions, as wellas a politicized racial profile, that standsout from the rest of South Africa. In thelate 1990s the province contained approx-imately 9.7 percent of the total SouthAfrican population and was highly urban-ized at 89 percent. The province is domi-nated by the Cape Metropolitan Regionwhich contained nearly 70 percent of thepopulation of whom 98 percent were classi-fied as urban. The legacy of apartheid withthe application of stringent influx control,coupled to the implementation of the col-ored labor preference policy, constrained thedistribution of the African populace of whomthe vast majority (over 76 percent) are con-tained within the metropolitan area. WesternCape is predominantly colored and white,with a substantial African minority, largelyurban, with better than national averagelevels of per capita incomes and other indi-cators of well-being (with the notable ex-ception of self-reported illnesses, which tendto be positively correlated with income andeducation levels) including access to healthcare and other infrastructure. On the aver-age, Western Cape is among the most well-off of all the nine provinces in South Africa.Nevertheless, there is a substantial povertyproblem as indicated by the relatively highlevels of child stunting, the prevalence oftuberculosis where the rate of 703 per100,000 is three times higher than the na-tional rate, and an above the national aver-age percent of housing from shacks. Someof the 42 districts in the province have ex-tremely high levels of poverty (Adato et al.1999).

The 1996 census reported 299,114 asunemployed in the province which gives abroad rate of 17.9 percent (the lowest in anyof the nine provinces, but still high enoughto generate concern and a call for a share ofa national public works program). This esti-mate is very similar to those estimates gen-erated by the October Household Surveyseries from 1994–1997 as well as with the

1993 PSLSD data and a number generatedfor the RDP in 1994 (Horner 1994).

Historically, social segregation in theprovince was not as intensely enforced inthe Western Cape as it was elsewhere, andin the first part of the century, people ofcolor had some limited rights. The NationalParty took power in 1948 in all districts out-side the metropole, removing electoral rightsof colored people, Africans, and Indians(Hellmann 1949; Horrell 1971). Variousforms of limited representation for coloredpeople throughout the sixties and seventieseventually meant some improvements intheir material conditions, with the Africanpopulation left behind. However, services forcolored people remained particularly bad inthe rural areas, where poverty was highest.

In 1980 the Nationalist Party Govern-ment devised its system of tricameral gov-ernment which downgraded the second tierof government, provincial councils, to man-agerial rather than elected governmental or-gans, consolidating power at the center withan elaborate system of local authorities basedon racially defined criteria (SAIRR 1985).It neglected, however, to provide many ofthese organs at the third tier of governmentwith the necessary financial support tofunction efficiently and effectively. In 1986the Provincial Government Act abolished the(white, elected) provincial councils, em-powering the state president to appoint pro-vincial administrators and executive com-mittees. In the Cape, the state presidentconfirmed the re-appointment of the previ-ous (nationalist) administrator and includedcolored people in the five-man executivecommittee (SAIRR 1987).

Opposition to apartheid also swelled inthe 1980s. In 1983 the newly formed UnitedDemocratic Front (UDF) drew strong sup-port in those areas that were later demar-cated as constituting the Province of theWestern Cape. This broadly based move-ment was to play an influential part in thepolitical life of the province in the 1980s untilit was disbanded in August 1991 (SAIRR

8 CHAPTER 2

1992). In 1994, as the ANC won 62.6 percentof votes nationally to the National Party’s20.4 percent and the Democratic Party’s 1.7percent, the newly demarcated Western Capeprovince remained the one province withNational Party control (SAIRR 1995). Inthe election for the provincial legislature theNP won 55 percent of the 42 seats (23), theANC 33 percent (14), and the DemocraticParty (DP) three seats, with one seat eachto smaller parties. The NP took seven of theseats on the provincial executive councilincluding the premiership, with the ANCbeing allocated the other four seats, includ-ing the portfolios of health; economic affairs;environment and tourism; and road trans-port and public works (SAIRR 1995). Thisassignment of the public works portfolio toan ANC ministry was one important factorplaying into political tensions that emergedin the operationalization of public worksprograms at the provincial and local levels.In 1997 the ANC lost control of this min-istry when the party withdrew from the pro-vincial executive committee.

Local government elections in WesternCape province scheduled for November1995 were delayed in certain areas becauseof a National Party challenge to demarca-tion of boundaries in the metropole and ruralareas. In those areas where voting occurredfor 886 seats, the ANC secured 34.54 per-cent and the NP 32.96 percent while inde-pendents and ratepayers associations securedthe bulk (27.2 percent) of the rest (SAIRR1996). Deferred elections in the metropoleand certain other areas were held on May 29,1996. The final outcome of the election forthe 1,446 local government seats in the prov-ince was: NP 43 percent; ANC 34.2 percent;independents and ratepayers 18.3 percent;with the DP securing 22 seats, the AfricanChristian Democratic Party (ACDP), 5; thePAC, 4; and the Freedom Front (FF), 3(SAIRR 1997).

This picture of the Western Cape’s polit-ical development helps to explain politicaltensions that would confound post-apartheid

development efforts, including community-based public works. Historical tensionsbetween colored and African communities,based on greater economic and political ben-efits awarded the former, remained at thelocal level, while also helping to explainpolitical support for the National Party, par-ticularly in rural areas. Community-drivendevelopment was ANC national policy, buthad to be implemented by largely NP offi-cials at the provincial and local levels.Despite support of an ANC public worksminister for the first three years, the imple-mentation of these community-driven pro-grams were largely the responsibility of NPadministrators who did not generally sharethe ANC objectives. There was also a racialas well as political dimension to the rela-tionships, tensions, biases, and incentives ofthese different actors—with most of theofficials at the national level African, mostof the officials at the provincial level white,and most officials at the local level whiteor colored. However, race and politics ishighly correlated, with most African officials(at national, provincial, and local levels)members of the ANC, and most white offi-cials (at provincial and local levels) mem-bers of the NP, and colored officials at thelocal level mainly NP affiliated. The effectsof these relationships will be further ex-plored in later chapters.

Public Works Programs inWestern Cape ProvinceGiven the enormous need for basic infra-structure in poor communities and the newgovernment’s commitment to addressing thisneed, many infrastructure programs wereinitiated in and around 1994. In choosingwhich provincial programs to include in thisresearch, we were guided by two criteria.First, we selected programs that had multi-ple objectives. This excluded programs thathad a unitary or strong primary focus on in-frastructure creation, with a relatively minorrole given to employment creation and/or

PARTICIPATORY DEVELOPMENT AND THE ORIGINS OF COMMUNITY-BASED PUBLIC WORKS IN SOUTH AFRICA 9

community participation. We employed thedistinction derived by I. T. Transport Ltd.(1997) from the international literature,which distinguishes between “Public WorksProgrammes” and “low-cost infrastructureprogrammes” that have different objectives.Low-cost infrastructure programs

seek to reduce the cost of infrastructureconstruction by using appropriate tech-nology which includes a high labourcontent on grounds of technical feasi-bility and economic efficiency, withemployment creation as a by-product.These programme [sic] are not strictlyPWPs as they start from a totally dif-ferent basis. A PWP starts with a budgetwhich may be more or less flexible,and a brief to create employment oralleviate poverty. It then looks for suit-able projects (I. T. Transport Ltd. 1997).

However, the programs included in thestudy also constructed infrastructure or im-proved the environment in ways that werenot simply “make-work” projects, but ratherundertook the construction of assets or ac-tivities identified as high priority for the com-munities involved. They also provided jobtraining, and were concerned with commu-nity participation, though to differing de-

grees. Seven programs met these criteriaand are described below (some key vari-ables for each program are found in the nextchapter in Table 3.1).4 As noted in the pre-vious section, each of the public works pro-grams conceived at the national level had aprovincial level department and programresponsible for program administration. In-frastructure development and job creationprograms in Western Cape province werelocated primarily within the Department ofTransport and Public Works, the nationalDepartment of Water Affairs and Forestry,and the Department of Housing.5 The sevenprograms included in the study are describedbelow.

The National Economic Forum JobCreation ProgrammeBecause it began prior to 1994, The NEFjob creation program was conceived andimplemented by institutions of civil societywith participation of the old government.However, it was the first of the new genera-tion of public works programs that beganto take a broad development perspective,focussing on sustainability and multiple ob-jectives of job creation, quality assets, skillstraining, and institutional capacity building.

10 CHAPTER 2

4One of the seven programs we have included, Public Works Programme Transport, could be seen as a low-costinfrastructure program. We have decided to include it because, managed by the provincial Department of PublicWorks and Transport (DTPW), it was considered part of the same group of RDP programs that fell under the newPublic Works Programme Directorate and thus followed parallel principles to the other programs we have included.These principles included increased labor-intensity, training, promotion of small, micro, and medium enterprises(SMMEs), and community participation. We also felt it was important to include because it was a program in-tended to experiment with integrating the PWP (the Western Cape version of the NPWP) principles into normal(line function) roads projects of DTPW (our primary institutional focus), and thus had far-reaching implications.

Several programs were excluded by our selection criteria because they fit the description of low-cost infra-structure programs. The Municipal Infrastructure Programme (water, sanitation, roads, refuse, electricity, andhealth facilities), the Integrated Service Land Projects (providing similar bulk service, as well as schools, policestations, and other community facilities); the Clinic Upgrading and Building Programme; and the Culture ofLearning Programme (schools) were considered low-cost infrastructure programs rather than public works pro-grams. Although job creation was stated as one objective of all these programs, it was not at the top of the list,formally or in practice, and thus labor-intensity was only encouraged and not required.

5The responsibility for the physical construction of assets for the Departments of Health (clinics and hospitals)and Education (schools) was also transferred to the Department of Transport and Public Works. Construction per-formed for these two departments did not meet the criteria used for inclusion in the research.

PARTICIPATORY DEVELOPMENT AND THE ORIGINS OF COMMUNITY-BASED PUBLIC WORKS IN SOUTH AFRICA 11

In August 1993, the NEF launched itsjob-creation program and identified theDevelopment Bank of South Africa (DBSA)to monitor implementation of the projects.A number of regional economic forumshad been established by local stakeholders(SAIRR 1994, 440–441), which in theWestern Cape became the Western CapeEconomic Development Forum (WCEDF),composed of representatives from the tradeunions, civic organizations, and private sec-tor. These regional forums were involved inappraising and approving applications forNEF funding for public works projects intheir areas. Approved applications weresubmitted by the regions for final evaluationby the Accreditation Board for Labour-intensive Construction (ABLIC) appointedby the National Co-ordinating Committee(NCC) of the NEF.

The Community-BasedPublic Works Programmeand the CommunityEmployment ProgrammePart of the national government’s NPWP,the Community-Based Public Works Pro-gramme (CBPWP) aims at poverty allevia-tion through job creation, skills training,delivery of needed assets, and capacitybuilding. It was seen as a critical safety netprogram for vulnerable groups during thetransition period. The main target groupswere the rural poor, youth, and disabled peo-ple. It also aimed at building the capacity ofcivil society to engage with developmentissues, giving opportunities for community-based organizations to manage developmentprojects, and NGOs to manage deliveryand provide training (DPW 1997, 8). TheCommunity Employment Programme (CEP)is part of the CBPWP, administered by alarge national NGO (the Independent De-velopment Trust [IDT]).

The Clean and GreenProgramme (CAG)This program was funded by the nationalgovernment as a “Presidential Lead Project”

and administered by the Provincial Depart-ment of Transport and Public Works. Itwas conceived to clean up the environment,improve the quality of life and tourism po-tential, and alleviate poverty through jobcreation, human resource development, andincome generation strategies. The programhad a provision for local authority matchingto ensure compliance with the requirementsof the RDP fund. For the sustainability ofthe project, institutions (particularly localauthorities) were to make provisions for therecurrent cost associated with the provisionof a sustainable service. The projects werealso to include a concerted educational cam-paign to enhance a people-driven and sus-tainable process.

Pilot Project ProgrammeAlso part of the CBPWP, this program wasinitiated by the National Department of Pub-lic Works in conjunction with provincial de-partments of public works, and consisted of12 projects spread across the nine prov-inces. The purpose of these projects was to“demonstrate to other government depart-ments how reorientation of expenditure oninfrastructure projects may be achieved.” In-formation gained through these pilot projectswas to be (1) disseminated to government de-partments at national, provincial, and locallevels to further the re-orientation process;and (2) used to develop guidelines and pro-vide technical information (CSIR 1997). Keyfeatures of the projects are labor-intensivedesigns, tender documentation, skills train-ing, community participation and liaison,labor management systems, and task-basedpayment.

Public Works Programme TransportThis was part of a fund allocated from theRDP fund to the National Department ofTransport in July 1995 (DTPW 1996). Thiswas divided up among the provinces, andadministered by the provincial departments(in the Western Cape, the DTPW). Thesewere managed by consultants, with partici-pation of community committees.

12 CHAPTER 2

The Working for Water ProgrammeThe Working for Water Programme (WWP)is a job creation and environmental programof the national Department of Water Affairsand Forestry (DWAF 1996, 3). A high profileprogram of the national government and per-ceived to be particularly successful, it hasreceived very high levels of funding and ex-panded rapidly. DWAF calculated in 1998that it had grown to 240 projects with over42,000 workers nationally (DWAF 1998, 6).

The program involves cutting down “in-vading alien plants,” that is, nonindigenousspecies such as wattle, pine, and others thatconsume large quantities of water. The pro-gram thus has two main benefits: job cre-ation, and increasing water availability fordomestic and commercial use, significantto this water-poor country. Like the NPWPprograms, the WWP includes skills trainingand education, support for SMMEs, andlocal institutional capacity building as partof its mandate. It also encourages secondaryindustries, and targets women, youth, andthe disabled as beneficiaries. In the WesternCape, the program was called the FynbosWater Conservation Programme and laterthe Fynbos Working for Water Programme(FWWP).

Program-Level Differencesand OutcomesThe differences between these programs—and the government departments responsiblefor conceiving and implementing them at

the national and provincial levels—wouldbe expected to account for some of the dif-ferences in community participation foundacross the project dataset and case studies.The WCEDF would be expected to behighly participatory because it was under-taken prior to the establishment of legiti-mate government structures and was largelygoverned by civil society institutions. Com-munity participation was an important cri-teria for project funding. In practice, how-ever, as the first program of its kind tryingto realize multiple objectives, it had manypreoccupations and the operationalizationof participation was less explicitly thoughtthrough and emphasized than in theCBPWP initiatives. The three programsfalling under the CBPWP (CBPWP, CEP,and Pilot) would be expected to have thehighest level of participation, particularlythe CEP where the IDT required sole com-munity control and did not involve localgovernment. The Transport projects werenot part of the NPWP and could be expectedto be less participatory, as the national De-partment of Transport was less preoccupiedby participation than the national DPW, andthe provincial DTPW had more independ-ent authority to implement these than theydid the CBPWP. Finally, the FWWP had somany objectives that participation hadgreater competition than in the NPWP/CBPWP projects. This program variable isexplored further in our analysis as we lookat project-level outcomes.

C H A P T E R 3

Research Methods

Public works programs contain public works projects. As explained in Chapter 2, we se-lected seven programs for inclusion in the study. We then took a census of all projectswithin the programs. Table 3.1 presents each of the seven programs, the institution

responsible for project administration, the number of Western Cape projects under each pro-gram, the number of these categorized as “rural,” and the types of infrastructure or environ-mental project activity included within each program.

Methods of Data CollectionA number of qualitative and quantitative data collection methods were used in this study.6

Table 3.2 describes each of these methods, the level at which data were collected (project, dis-trict, local, provincial, and national), and the informants included in the research, from projectworkers to national policymakers.

Database of 101 Public Works Projects in Seven ProgramsUsing project documents and mail-in questionnaires with follow-up telephone calls and visits,quantitative and qualitative data were collected on approximately 45 variables for each proj-ect, within the categories as outlined in Table 3.3.

The data were collected in the following manner. Initially, all program level documentswere identified for each of the seven programs (for example, monthly reports, final projectclose-out reports, project review summaries). However, these were new programs that hadfocused more on getting off the ground than monitoring. It was soon determined that many ofthese documents either: (1) contained data taken from project applications and did not reflectactual data collected during project implementation; (2) were incomplete, existing for someprojects and not others, and/or containing certain pieces of data for some projects and notothers; or (3) contained data that were of questionable origin or contradictory. Thus, in orderto get accurate data, a project-level questionnaire was designed and administered to imple-menting agents for each project. Project records were mainly used to fill out the question-naire wherever they existed, though some qualitative information categories (for example,identification of different agents) had to come directly through interviews with implement-ing agents, which we triangulated with other information wherever possible. In manycases, the implementing agent did not have records and visits had to be made to various

6An earlier IFPRI research report by Kerr (2002) also combined quantitative and qualitative methods to look ateffects of participatory development projects and different institutional arrangements.

13

program and project administrators ormanagers, consultants, contractors, and ac-countants who helped us to track down datathrough internal project paperwork. In thecase of the Independent Development Trust

projects, however, the program managerswould not allow the research team access toproject facilitators, and so certain categoriesof information are missing for a number ofthese projects.7

14 CHAPTER 3

7The IDT was a large NGO with a large public sector funding base. Because it was not a public sector programwe did not have direct access through our contacts in government, who became committed to the process and rec-ognized the benefits of the research. The IDT in contrast appeared to take a defensive posture toward being thesubject of research, and despite directives from their funders in the government to cooperate with the research,management decided midway through the research to refuse.

Table 3.1 Public works programs in Western Cape province included in the study

Number outside CapeAdministering Number of Metro/Winelands

Program institution projects (rural)a Types of infrastructure

Clean and Green (CAG) Provincial Department of 10 1 Cleaning (2), greening, alien Transport and Public vegetation clearing (7), Works (DTPW) parking area

Community Based Public DTPW 18 10 Community center, roads, Works (CBPWP) stormwater drainage, sanitation,

water supply

Community Employment Independent 22 21 Community center, roads, Programme (CBPWP/CEP) Development Trust stormwater drainage, sanitation,

(IDT) school, creche, clinic, greening,roads and stormwater

Fynbos Water Conservation Department of Water 14 11 Alien vegetation clearingProject (FWCP) also known Affairs and Forestry as the Fynbos Working for (DWAF)/Cape Water Project (FWWP) Nature Conservation

Pilot Projects (Pilot) Department of Public 2 0 Roads and stormwaterWorks/DTPW

Transport Projects (Trans) DTPW 6 4 Roads, roads and stormwater

National Economic Forum/ WCEDF/DBSA 29 19 Community center, roads, Western Cape Economic stormwater drainage, sanitation, Development Forum water supply, cleanup, recreation(WCEDF/NEF) grounds, roads and stormwater,

multiple services, bridge

aIn South Africa, there is no easy distinction between rural and urban, because so many localities have characteristics often associated withboth. Previous government definitions of urban based on the existence of a local authority is not useful given the restructuring of local gov-ernment in urban and rural areas. We thus instead use an (imperfect but useful) distinction between locations in or on the outskirts of CapeTown, Paarl and Stellenbosch (the Cape Metro/Winelands regions), and those elsewhere in the province.

Project Case StudiesEight8 in-depth case studies were conductedin the second phase of the research. The proj-ects were selected purposively so as to gener-

ate variation in type of institutional arrange-ment, rural/urban location, and infrastructuretype. We selected a subset of candidate proj-ects in numbers roughly proportional to the

RESEARCH METHODS 15

8We actually captured events at a total of nine projects because the Kahayelitsha project comprised two separateconstruction projects. We describe it as one project because the governing structure, processes, community, and

Table 3.2 Description of data collection methods

Source of information

Focus HouseholdProject Semi- groups/ surveys from

Information records and Project Worker structured PRA secondaryabout the . . . documents questionnaires questionnaires interviews workshops sources

District October householdsurveys and the1991 census

Project Various documents Mail-in with from 101 projects telephone and

visit follow-ups

Local government 12 individual officials and two group

interviews

Community-based 13 individualproject steering and nine group committees. interviewscommunity-based organization (CBO) or other community-based key instruments

Consultants, Eight individual contractors, and and one group emerging interviewscontractors

Other project staff, supervisors, trailers

Former project 193 questionnaires Eight workers in seven projects workshops

National and provincial 36 people policymakers, interviewedgovernment program managers, and policy advisors

frequency of the values of the stratifyingvariables. From the subset of candidateprojects we made a final project selectionbased on criteria such as geographicalspread; formal roles of local government,CBOs, NGOs, and consultants;9 existenceof second-round effects;10 existence of sub-

contractors; types of payment systems; andinterest generated among policymakers.Table 3.4 presents a profile of each of thecase studies selected, including their pro-gram, magisterial district, and type of assetor activity. It also includes some basic indi-cators for these projects, including employ-

16 CHAPTER 3

infrastructure were the same, though the timing and workers were different (the projects started on the same date,but one ran much longer in part due to a conflict-related delay).

9This was intended to capture different levels and types of participation. Although we selected on de jure ratherthan de facto participation, we ended up with de facto participation at the following levels: one case study eachin the categories highest, very high, high, and no participation; two projects with medium participation (count-ing Khayletisha as two), and two medium-low participation. Table 5.1 indicates which projects fall into thesecategories.

10Second-round effects are indirect effects or “spin-offs” from the project, beyond their immediate transfer of in-come for workers. Examples include development of skills that could help workers secure future employment, acommunity hall that is used as a skills training workshop, or an enhanced ability of community members to man-age development activities.

Table 3.3 Data collected at the project-level, 101 Western Cape public works projects

Category of information Type of data collected

Project location Town, residence of workers, magisterial district, development region; rural/urban

Durations Actual and projected; dates

Institutional arrangements Administering institutions; applicants; implementing agents; community-basedorganization (CBO) roles; identity of consultants and contractors

Assets and activities Primary activities and project components

Costs Projected and actual; sources of funding (tiers of government and private)

Employment generated Projected and actual work days, for men and women

Wages Wage rates, initial and final, skilled and unskilled; comparative rural sector wagerates

Payment systems Daily wage or task-based

Labor disputes Existence or absence

Training Costs; number of days; content; training institution

Small/medium enterprises Existence or absence

Second-round effects Existence or absence; types

Maintenance arrangements Responsible institution

Comments Any other points of interest

ment generated; employment for women;whether they paid daily wages or gavetask-based payments; labor stoppages; andwhether there were second-round effects.Figure 3.1 shows a corresponding map of

the magisterial districts where the case-study projects are located.

For seven11 of the eight project casestudies, a survey was conducted among for-mer workers to determine their employment

RESEARCH METHODS 17

11In the project at Kylemore, a post-project survey could not be conducted because most of the original projectworkers were still working on the project. The FWWP projects were different from the other programs in thatthey were ongoing, so workers at this newer project were still on their contract.

Table 3.4 Summary profile of the eight in-depth case studies selected from the 101 projects

Second-Worker Percent of Method round

Asset days workdays of Labor effectProject Program Districta generated generated to women payment stoppage reportedd

Langa CAG Goodwood Clean-up andgreening 11,600 60 Wage No Yes

Khayelitshab Pilot Mitchells Plain Roads and 12,289 18 Task Yes Yes4C stormwater

drainage

Khayelitshab Pilot Mitchells Plain Roads and 29,462 5 Task Yes Yes3A stormwater

drainage

Lutzville CBPWP Vredendal Community 2,490 12 Wage No Yescenter

Thembaletu Trans George Stormwater 16,500 40 Wage Yes Yesand roads

Clanwilliam CBPWP Clanwilliam Stormwater 2,981 10 Wage Yes Yesdrainage

Murraysburg CBPWP Murraysburg Water 4,364 0 Wage Yes Yesreticulation

Kylemorec FWWP Stellenbosch Removal of Wage/task Yes Noalien vegetation

Stellenbosch FWWP Stellenbosch Removal of 47,285 44 Wage/task Yes Yesalien vegetation

aSee Figure 3.1.bThe Khayelitsha Pilot projects were two separate contracts in two areas, with data kept separately, so for the purposes of our database they aretreated separately. However, because they had the same consultants, contractors, and community steering committees, for the purposes of ourcase study they are treated as one.cWorker day figures are not available for Kylemore because at the time the database was constructed, the Stellenbosch project included Kyle-more (prior to our case studies it was split into a separate project with separate managers and committees).cNote that the Fynbos Working for Water projects are ongoing, and thus do not have end dates. For data collection on these projects, we chose a15-month period from initial start-up until the end of January 1997, when most of the other public works projects were complete.dSecond-round effects are indirect economic or social impacts of a project; for example, the building of a community center that is later usedas a skills-training facility.

histories pre and post public works projectsto determine if participation in the projectsassisted them in improving their access toemployment following the project. A surveyof the impact of the projects on the povertystatus of former workers was not under-taken due to resource constraints, thoughthe nonexistence of baseline data wouldhave limited the value of this exercise had itbeen undertaken. Given this stage in thedevelopment of public works programs inSouth Africa—the completion of a firstround of programs and the start of planningexpanded programs building on lessons fromthe first round—we believed it most pro-ductive to focus on program and project-level variables and issues to be resolved.

In designing and selecting the partici-pants for the case studies and institutionalstudy, a “stakeholder” approach was used.“Stakeholder” analysis is defined in severalways, all of which we considered in our iden-

tification and selection of interviewees andfocus group participants. In program evalu-ation, stakeholders are defined as “peoplewhose lives are affected by the programmeand people whose decisions can affect thefuture of the programme” (Greene 1988).Selection should be informed by diversityand representativeness (Guba and Lincoln1981), and in a utilization-focused approach,“stakeholders are people who have a stake—a vested interest—in evaluation findings. . . decision-makers and information userswho have questions about a programme”(Patton 1986).

Using the conceptual framework we de-velop in more detail in the next chapter, wecan identify the stakeholders as financiers,providers, and beneficiaries. Financiers werethe national government officials, includingpolicymakers and program managers, andto a lesser extent provincial governmentofficials (in a dual role, mainly serving as a

18 CHAPTER 3

Figure 3.1 Western Cape province: Magisterial districts of the case studies

Oudtshoorn

Knysna

Uniondale

Prince Albert

Beaufort West

MosselBay

George

Ceres

Laingsburg

Ladismith

RiversdaleSwellendam

Montagu

Caledon

Clanwilliam

Bredasdorp

Pidetberg

Malmesbury

Paarl Wor

ceste

r

Vanrhynsdorp

Vredendal

Other districts

Magisterial districts where public works case study projects were located

Murraysburg

1

1. Bellville2. Calitzdorp3. Cape4. Goodwood5. Heidelberg CP6. Hermanus7. Hopefield8. Kuils River9. Mitchells Plain

10. Mooreesburg

11. Robertson12. Simons Town13. Somerset West14. Stellenbosch15. Strand16. Tulbagh17. Vredenburg18. Wellington19. Wynter

2

34

5

6

7

8

9

10

11

12

131415

16

17

18

19

provider). Providers compose the largestgroup of stakeholders. These included:community-based project steering commit-tee (PSC) representatives; community-basedorganization representatives; local govern-ment councilors, municipality officials, andstaff. There were also agents of providers:including consulting engineers and archi-tects (sometimes crossing the line into be-ing providers themselves), contractors, thecommunity liaison officer (CLO, in a dualrole—also serving as agent of beneficiaries),and supervisors. Beneficiaries included proj-ect workers, other community members, andsubcontractors. Agents of beneficiaries in-cluded worker committees (sometimes in adual role as providers, with their role limitedto decisionmaking on labor issues), tradeunion representatives, and the CLO. Not allprojects had the exact same role-players butall had variations of the above categories.

For each case study, these individualswere identified and in-depth, semi-structuredinterviews were conducted based on a struc-tured interview guide. Interviews were con-ducted with either all of the members of aparticular group (if there were between oneto five people involved, such as the consult-ants or contractors), or where a particularrole-player involved multiple people (forexample, the PSC, local government), a se-lection of representatives were interviewed,based on several factors. For example, onlylocal government officials identified ashaving been involved in the project insome capacity, directly or as an interestedobserver, were interviewed. In the case ofPSCs, approximately 60–90 percent of themembers were interviewed, depending ontheir availability.

Research workshops were held with for-mer project workers, using a combinationof focus groups and participatory appraisal(PRA) methods that use visual exercisesto engage participants and stimulate theirthinking. In selecting the former projectworkers to participate in the workshops, weinvited all workers in the three projects that

had under 30 workers in total. In the fiveprojects with over 100 workers, we used astratified random sample from a list of allproject workers, proportional to the numberof men and women workers. Approximately70 workers were selected per project to en-sure that at least 30 showed up to the work-shop (between 30 and 50 showed up). Al-though we held workshops on the weekendto minimize a bias toward those who wereunemployed (that is, less successful in find-ing a job after the project), we are aware thatin the three large urban projects it may havebeen disproportionately unemployed peoplewho came to the workshops. This was be-cause of the incentive of a small “trans-portation” fee and meal offered (in theseurban areas many people would be unlikelyto show up on a weekend without this in-centive), and the fact that when people hearabout a workshop related to the public worksproject, many come in the hope that it maylead to a job (regardless of the fact that thepurpose of the workshop is explained inthe invitation).

In selecting former project workers forthe survey, we also used a random sampleproportional to the number of men andwomen on the projects, but a different sam-ple from that used for workshop invitations.For the projects with under 30 people, allworkers were surveyed. For the projectswith over 100, a sample of 30 percent wasused. In addition, at the two projects usingsubcontractors, five subcontractors per proj-ect were also surveyed.

Finally, the study also included an analy-sis of the policy process surrounding thesepublic works programs at the national andprovincial levels, as well as political econ-omy issues. This required key informantinterviews with a wide range of nationaland provincial policymakers, program man-agers, policy advisors, and trade unions andNGO representatives. Altogether over 80 keyinformant interviews were carried out forthe study (see Adato et al. 1999 for a full de-scription of study methods and findings).

RESEARCH METHODS 19

C H A P T E R 4

The Impact of Participation on ProjectOutcomes: Evidence from the Full Sample

Our research is concerned with participation as both an end and a means, and we em-ploy a particular analytical approach to studying the latter. Having set out the back-ground for this study as well as the methods used to collect our data, we now turn to

one of the core objectives of our work: to assess whether community participation affectsproject performance. To do so we first set out a conceptual model to capture some of the an-alytics of participation. We then use this model to help frame an econometric analysis of projectoutcomes.

We begin by identifying the actors, making links between community participation, socialcapital, and trust as well as noting the importance of defining precisely what is meant bycommunity participation. We then set out a simple model that identifies the main trade-offs involved in how different modes of delivery vary in the costs of intervening and the designof objectives.

Some Analytics of Participation:Financiers, Providers, and BeneficiariesConsider a world comprised of three groups: financiers, providers, and beneficiaries.12 Theprimary role of the financier is to provide funds. Multilateral and bilateral donors, ministriesof finance, and nongovernmental organizations (NGOs) are all examples of financiers. Asnoted in Chapter 2, in the case of the Western Cape, the financiers for these public works pro-grams were largely from national government. Given the discussion in Chapter 2, it is reason-able to assume that the financier is interested in reducing poverty but has only a limited budgetto do so. The main role of providers is to implement interventions. Providers may be line min-istries, autonomous government agencies, private firms, NGOs, or community-based organi-zations (CBO). Here, we are thinking of “communities” as being a group of individuals withina geographically defined area who collectively implement an intervention with financial back-ing from the financier. (The following chapters provide specific examples of these.) Benefi-ciaries comprise the communities, households, and individuals who are the intended recipientsof program benefits, the poor. Note that the roles of these three groups are not always strictlydelineated. Depending on context, financiers, providers, or beneficiaries may initiate, design,

12A similar version of this model with a longer literature review appeared in earlier drafts of this report and inHoddinott (2002).

20

and/or implement the intervention. Commu-nities and beneficiaries can be co-financiersas well as beneficiaries of interventions.

Poverty alleviation projects typicallyhave multiple objectives or outcomes valuedby the actors involved in the intervention.In the Western Cape, objectives included jobcreation; the building and rehabilitation ofinfrastructure, or improvements to the natu-ral environment; the provision of job train-ing that would enable workers to find post-project employment; and capacity buildingof communities. Without loss of generality,we will develop our model on the basis of aproject that is characterized by a pair of ob-jectives (z1, z2). For example, suppose thatthe anti-poverty intervention aims to reducepoverty in both the short and long term. Inthis context, z1 would be the level of cur-rent consumption of intended beneficiarieswhile z2 is the extent to which the programtries to create human capital and eliminatelong-term poverty.

There is no reason to expect that all ac-tors will have identical preferences overprogram objectives. In the case of publicworks, for example, actors may differ in theweights given to job creation and training.Given a pair of realized objectives, we de-note the outcome that accrues to the pooras B(z1, z2). We assume that B(., .) is in-creasing in both arguments and that bene-ficiaries do not pay any of the costs ofpoverty reduction; z1 and z2 are measuredso that more of both of them is consideredto be a good thing.

Providers—in the case we will considerhere, community-based organizations—andfinanciers—here government—have theirown objectives in addition to caring aboutthe poor. For the government this is

G(z1, z2) = βg(z1, z) + (1 – β)B(z1, z2)

(1)

where g(⋅, ⋅) represents any “private” bene-fit that the government receives from hav-ing the program designed in a particular way.For example, the government might wish todeliver benefits to particular ethnic groups.It could also represent differences in dis-count rates that imply different weightingof long- and short-term poverty allevia-tion benefits. The parameter β denotes theweight given to the government’s versus thepoor’s payoff; where β = 0, the governmentand the poor have identical preferences. Thecommunity’s preferences is denoted by

R(z1, z2) = αr(z1, z) + (1 – α)B(z1, z2),

(2)

where r(⋅, ⋅) denotes the “private” payoff ofthe community organization and α is theweight that it attaches to its own preferencerelative to that of the poor beneficiaries.

We now consider what would happen ifthe government managed the poverty reduc-tion program, that is it is both financier andimplementer. Any government expendituresnot allocated to the project can be spent onsome other valuable activity whose price isnormalized at one. Thus, the government’sobjective is to choose (z1, z2) to maximizeG(z1, z2) – C(z1, z2). Let the optimal valuesof this be, (z1

G, z2G ). Thus, the benefit to the

poor is BG(z1G, z2

G ). The community’s pay-off is R(z1

G, z2G ).

Alternatively, the government could con-tract provision to the community whose costfunction is denoted by c(z1, z2). We assumethat the community enjoys an absolute ad-vantage in production of both goals so thatC(z1, z2) > c (z1, z2). We also assume that

∂c(z1, z2) ∂C(z1, z2)———— < ————–

∂zi ∂zi

for all (z1, z2), i = (1, 2).13 Given this costadvantage, a Pareto improvement is, in

THE IMPACT OF PARTICIPATION ON PROJECT OUTCOMES 21

13Hoddinott (2002) provides a detailed set of examples showing the different ways in which communities have acost advantage.

principle, possible from decentralizing theprogram to have some kind of communityinvolvement. This is because c(z1

G, z2G ) <

C(z1G, z2

G ). Thus, the government couldpay the community organization a transferof t = c (z1

G, z2G ) to undertake the project on

its behalf, thereby saving money. Note,however, that solution is not incentive com-patible unless the government has somedirect way of controlling the community or-ganization’s inclination to change the pro-gram’s objectives ex post. This is becausepreferences over project objectives may dif-fer if G(z1, z2) ≠ R(z1, z2). Define

{z1C( y), z2

C( y)} = z1≥0,

argz2≥0max {R(z1, z2) : C(z1, z2) = y}

(3)

Thus, if it were given a transfer of c(z1G, z2

G )to undertake the project, the communityorganization would be {z1(c(z1

G, z2G )),

z2(c(zz1G, z2

G ))}. This would be the solutionunder community management if it were notpossible for the government to write somekind of contract that restrained the commu-nity’s behavior. Thus, we are assuming anextreme form of contractual incompletenessin the model, a reasonable assumption whenthe precise objectives of poverty alleviationprograms are very hard to describe ex ante.

It is interesting to ask when the gov-ernment would wish to decentralize man-agement of poor support to the communityorganization. Let

yG = arg max{G(z1C( y), z2

C( y)) – y},(4)

as the optimal poverty alleviation budgetto grant to a community given that the re-source allocation decision will be made atthe community level. Then the governmentwill prefer to have a community organizationmanage a poverty alleviation project if

G(z1C(yG), z2

C(yG)) – yG > G(z1G, z2

G )– C(z1

G, z2G ). (5)

The left-hand side is the payoff of thegovernment if it gives a budget of yG tothe community and the right-hand side is thepayoff to the government under pure gov-ernment provision. It is easy to see thatthe likelihood of community involvementis highest where (1) government and com-munity preferences are more congruent and(2) the absolute cost advantage of the com-munity (net of government’s monitoringcosts) is largest.

Such cost advantages could stem frommultiple sources. There may be interventionswhere knowledge of local conditions is es-pecially important and where the cost of theacquisition of such knowledge by outsidersis high. Second, community participationmay reduce the likelihood of moral hazardor adverse selection problems. In the con-text of public works projects where a dailywage payment is made, involvement by thecommunity in the hiring of labor may in-crease the likelihood that those most in needof work and/or are more likely to show upreceive employment, and those less in needand/or with a propensity to shirk, are ex-cluded. Third, communities may have abili-ties to lower costs that are not available tooutsiders. For example, communities couldencourage participants to accept lowerwages than those officially sanctioned so asto generate savings that can be put towardother objectives such as training. However,it does not necessarily follow that all costsare lowered when beneficiaries have in-creased involvement in the design and im-plementation of interventions. Abraham andPlatteau (2001, 11) document the impor-tance placed on ensuring unanimity andconsensus in self-help projects in a Nairobislum, resulting in “considerable efforts andtime spent in lengthy discussions at meet-ings and assemblies as well as in the pro-tracted mediation procedures to settle inter-personal conflicts.” Khwaja (2001) finds thatcommunity participation in what is describedas “technical decisions”—such as projectcapacity—resulted in projects requiring

22 CHAPTER 4

greater maintenance. If all costs are paidby an external financier and the budget con-straints are not hard (that is, strict limits oncosts are not imposed), communities may nothave an incentive to control costs where theycan capture additional resources that flowfrom the financier. Other providers mayoffer cost advantages in other areas. Forexample, liaising with a myriad number ofcommunities will carry higher administra-tive costs to the financier than contractingwith a single provider. A large NGO orprivate firm may be able to benefit fromeconomies of scale, or superior manage-ment skills.

The community organization also has tobe willing to undertake management of theproject—it is not reasonable to assume thatprojects can be foisted on an unwilling or-ganization. This requires that

R(z1C(yG ), z2

C(yG )) ≥ R(z1G, z2

G ). (6)

Now consider the well-being of the poor.Most of the discussion of poverty reductiontends to assume that the community organi-zations are more in tune with the prefer-ences of the beneficiaries. If the communitycares solely about the beneficiaries, thenwhenever community management is goodfor the poor, it will be chosen by the com-munity. However, if there is an agency prob-lem, in the sense that the well-being of thepoor and the community organization arenot fully in tune, there is no guarantee thatthis will be the case.

Example: Suppose that the only differ-ence in preferences is which group to targetresources on. Thus, let b (zi) be utility ofmembers of group i when the aim of the anti-poverty program is to get them to an incomeof zi. We assume that b (zi) = log (zi). Thereare two groups and let λ be the share of type1’s in the population. The overall benefit in-dicator of the poor is λb(z1) – (1 – λ)log(z2).The government and the community organi-zation differ in the weight that they attach tothe well-being of each group. Thus,

G(z1, z2) = β log(z1) + (1 – β)log(z2)(7)

and

R(z1, z2) = α log(z1) + (1 – α)log(z2)(8)

where α ≥ λ > β. This says that the govern-ment favors group 2 when it designs theprogram. We assume that there is a trans-action cost ci(Ci) for the community organ-ization (government) to reach group i, andthe initial (pre-transfer) income for group iis the same and fixed at y. Then the cost ofachieving the objective is

λz1 + (1 – λ)z2 – y + C1λ + C2(1 – λ) ≡ λ1z1 + (1 – λ)z2 + Γ (9)

if the government manages the project and

λz1 + (1 – λ)z2 – y – c1λ1 + c2(1 – λ) ≡ λ1z1 + (1 – λ)z2 + γ (10)

if the community organization does. It is noweasy to check that

β (1 – β)(z1

G, z2G) = {—, ———}.

λ (1 – λ)

It is also easy to check that

z1C( y), z2

C( y) α (1 – α)

= {— ( y – γ), ——— ( y – γ)}λ (1 – λ)

and that yG = 1 + γ. So in this case, the un-constrained community optimum and theconstrained optimum yield the same allo-cation. The community organization spendsmore on the group that it favors relativelyto the government. The two conditions forcommunity participation to be optimal are

α 1 – αβ log(—) + (1 – β) log(———) – 1 – γλ (1 – λ)

β 1 – β> β log(—) + (1 – β) log(———) – 1 – Γλ (1 – λ)

THE IMPACT OF PARTICIPATION ON PROJECT OUTCOMES 23

for the government and

α 1 – αα log (—) + (1 – α) log (———)γ (1 – λ)β 1 – β> α log (—) + (1 – α) log (———).λ (1 – λ)

The latter is clearly satisfied. The former willbe satisfied when Γ is much larger than γ andα is closer to β. Whether the poor’s benefitgoes up or down depends upon whether

α 1 – αλ log (—) + (1 – λ) log (———)λ (1 – λ)β 1 – β> λ log (—) + (1 – λ) log (———).λ (1 – λ)

This will tend to be the case if α iscloser to λ than is β, that is, there is lessof an agency problem with communityorganizations.

There are two additional issues thatshould be noted. In this three-entity context,participation in antipoverty interventions isa process whereby financiers enter into acontractual arrangement with beneficiarieswho become involved in aspects of provi-sion, and possibly finance. A feature that wewill return to at several points is the factthat, as in all contractual relations, it is im-possible to set out in advance all possiblecontingencies that will arise. In such cir-cumstances, establishing who has decision-making power has a direct bearing on whatis meant by community participation. Specif-ically, one can distinguish between formal(de jure) and real (de facto) authority. For-mal authority is the right to decide; real au-thority is the effective control over decisions

(Aghion and Tirole 1997). The importanceof this distinction lies in the possibility thatin the absence of delegation of de factodecisionmaking power, potential benefi-ciaries may be reluctant to act because ofconcerns that they will be overruled sub-sequently. Alternatively, central authoritiesmay subsequently renege on commitmentsand the threat of this generates a hold-upproblem (providers are unwilling to act outof concern that decisions they make will besubsequently over-ruled).

A second issue is that this conceptualframework treats communities as homoge-neous entities.14 Yet as one of our key in-formants observed

There’s very few community leaders in this country who will really let theircommunities work because they arealso engulfed in this great bitterness ofthe past and they’re always infightingabout this opinion and that opinion.And that really tears communities apartbecause they never get down to sayingthat these are the issues that we need towork on. Because I mean after all youcannot work in conflict, you cannotwork in difference of opinions espe-cially if you take on projects. You haveto work as a team to let that projectwork, be it to put a local clinic there or be it to build a couple of houses orwater. You will always find the com-munities who are successful in SouthAfrica are those communities whobasically stick together as a unit. Theymight have their difference of opinion,but their difference of opinion have not

24 CHAPTER 4

14Purely on theoretical grounds, this is dubious given Arrow’s (1951) well-known impossibility theorem. Notetoo that financiers and other providers may not be homogeneous either. The donor acting as a financier might en-compass a project leader who is interested in maximizing prestige; middle managers anxious to protect theirbudgets; and lower level functionaries interested in promotion. Further, each actor may be accountable to a num-ber of constituencies. For example, an engineer within a government public works department who is responsiblefor providing drinking water to a locality may be answerable to four different groups. These could include hersuperiors within the bureaucracy, a professional organization that accredits and monitors her professional conduct,local politicians seeking to maximize the benefits flowing to their constituents, and the intended beneficiaries ofthe intervention.

driven them apart (interview, C. B.,C.B. Associates).

Community participation requires thatthere must exist mechanisms for the discus-sion of common issues, for the resolutionof differences in opinions, for the allocationof costs and benefits associated with groupdecisions, and for the efficient and effectivemonitoring of actions taken by a few onbehalf of the many. In the next chapters,specific examples are provided of how thiscame about in the provision of public worksin the Western Cape. Here we note that theefficacy of community participation maybe a function of community characteristicssuch as the size of the group involved, thedegree of inequality, and the outside optionsavailable to members (see Baland and Plat-teau (1997) and Dayton-Johnson (2000) fora discussion). To this can be added the so-cial capital literature’s emphasis on theimportance of trust across individuals. Trust,built up by repeated interactions, can helpproblems of information asymmetries (dif-ferences in information held by differentparties) to be solved thereby allowing self-enforcing agreements to be reached (Knackand Keefer 1997). Common to both theliterature on collective action and socialcapital is the view that polarization inhibitsthe construction of relationships of trust.Easterly and Levine (1997, 1205–06) write,“polarized communities will be prone tocompetitive rent-seeking by the differentgroups and have difficulty agreeing onpublic goods like infrastructure . . . therebyimpede agreement about the provision ofpublic goods.”

Concerns regarding the determinants ofeffective community action raise two issues,one practical and the other analytical. Thepractical issue is that financiers may be un-willing, or unable, to assess the ability ofcommunities to engage in collective action,nor may they be willing in practice to invest

heavily in the development of such capacityin the context of a particular developmentproject.15 Financiers may decide instead thatcommunities should self-select into partici-patory projects, perhaps signaling their seri-ousness by making an up-front contributionof money or other resources. Indeed, suchactions are often seen as the hallmark ofmany successful rural development projects(Uphoff, Esman, and Krishna 1998). Thismay mean that better organized, less frac-tionalized communities may benefit dis-proportionately from increased emphasison participation. The analytical issue is thatif one accepts the notion that successfulcommunity interventions are those in whichcommunities self-select into the projects,this self-selection needs to be taken into ac-count. It might be the case that localities inwhich community participation thrives arethose in which project outcomes are morelikely to be favorable in any case.

To summarize, our conceptual frame-work identifies three actors: financiers, pro-viders, and beneficiaries. These multipleactors may have divergent preferences thatlead them to attach different weights to themultiple objectives found in most anti-poverty interventions. One benefit of bene-ficiary participation lies in the prospect ofreducing the cost of providing antipovertyinterventions. In essence, these are benefitsthat are derived from improved implemen-tation. This is likely to occur where knowl-edge of local conditions is especially im-portant, where moral hazard or adverseselection concerns play a role, where verifi-cation of actions is needed. Further, com-munities may have ways of lowering coststhat are not available to outsiders. A furtherset of benefits lie in terms of improved de-sign. Beneficiary participation offers thepotential for the selection of project objec-tives that more closely reflect the prefer-ences of the population that they are designedto serve.

THE IMPACT OF PARTICIPATION ON PROJECT OUTCOMES 25

15See Parker and Skytta (2000) for a discussion of this issue in the context of water provision in rural Mali.

However, realization of these benefits iscontingent on two characteristics. The firstis the ability of communities to engage incollective actions. In fractionalized commu-nities where trust is weak (perhaps becauseof little prior history of collective action ora history of political conflict such as in theSouth African case), there is a risk thatcommunity participation may result in thecapture of benefits by local elites to the detri-ment of the poor, or that conflict may stallprogress. The second characteristic is theimportance of delegating decisionmakingauthority. Failing to do so—allowing de jurebut not de facto participation—may result inbeneficiaries being reluctant to act becauseof concerns that they will be subsequentlyoverruled.

Data OverviewTable 4.1 provides some descriptive statistics.Different types of public works activities—the construction of buildings and roads andbridges and removing alien vegetation andgarbage—will have differences in require-ments for materials and for specialist inputssuch as engineering design. Accordingly,Table 4.1 groups the assets being constructedas part of these programs into three broadcategories: community buildings such ascenters, schools, and clinics; basic infra-structure activities such as roads, storm sew-ers, sanitation sewers, and water reticulation;and other activities such as the removal ofalien vegetation and general “cleaning andgreening.”

The average project operated for aboutnine months, with 20 percent being com-pleted in less than four months and 33 per-cent operating for one year or longer. Onlyone project operated for more than twoyears.16 While projects typically went over-budget, it was only in the case of the con-struction of community buildings thatthese cost-overruns were significantly high.

Materials-intensive projects such as theconstruction of community buildings devotea lower share of their budgets to labor andcreate fewer jobs. Infrastructure develop-ment and community building projects typ-ically employed fewer women than otherpublic works projects.

As noted above, assessing the impactof community participation and project out-comes requires a precise definition of whatis meant by these concepts. In particular, itis important to distinguish between de jureauthority, in the sense of “whose name is onthe contract,” and de facto authority, in thesense of “who actually is responsible forplanning and implementing the project.”Our measure of de facto authority is in-formed by Paul’s (1987) continuum of par-ticipation from information sharing to con-sulation to decisionmaking to the initiationof action as applied to qualitative informa-tion collected from project managers andadministrators as well as other relevantparties. Projects were divided into four cat-egories: (1) the community based organi-zation (CBO) is solely responsible for allaspects of the project, including design,overseeing the contractors, setting wages,selecting workers, controlling the bank ac-counts, and so on (32 percent of projects);(2) the CBO, together with another im-plementing actor, jointly participates indecisionmaking over some or all aspects ofthe project, including design, overseeing thecontractors, setting wages, selecting workers,controlling bank accounts, and so on (23percent of all projects); (3) the CBO assistsin selecting workers, mediates disputes, andliaises with the community but is not a de-cision maker; (31 percent of projects); and(4) cases where the community has little orno involvement in the project (15 percent ofprojects).

Our data are also rich in information onproject outcomes. These can be divided into

26 CHAPTER 4

16The exception is the projects in the FWWP, which were continuous. We took a fixed period, from project startto the end of the period of quantitative research (a 12- to 15-month period) for the purpose of data collection.

several categories. First, we would like toknow if community participation enhancedthe attainment of program objectives to cre-ate employment and training opportunities.Three outcomes capture this: the share ofthe project budget allocated to labor; thelog of the number of days of work created;and the log of the number of training daysundertaken. Second, we would like to knowwho in the community captures these bene-fits and this is measured in two ways: viathe ratio of the daily project wage to thelocal unskilled wage (consistent with the lit-erature on self-targeting of public works, alower ratio is indicative of improved target-ing toward the poor); and the percentage ofemployment that goes to women. Lastly, wehave information on the cost-effectivenessof community participation. This includescost overruns (computed as the ratio of costoverruns to projected costs as submitted inthe project proposal) relative to projectedcosts, the log cost in rands of creating oneday of employment (calculated by dividing

the number of days of employment gener-ated by the project by its total cost); and thecost to the government of transferring onerand to the poor. This variable is the re-ciprocal of the benefit stream generated bythe project divided by the government ex-penditure on it. The benefit stream consistsof transfer benefits to workers net of whatthey would have earned in the project’s ab-sence plus non-transfer benefits capturedby the poor.17 A low value indicates thatthe project is cost-efficient in delivering re-sources to the poor.

Table 4.2 provides mean values of theseproject outcomes, disaggregated by varyingdegrees of community participation.18 Thetop panel classifies projects as to whether ornot there is any participation. Communityparticipation appears to be associated withan increased share of project budgets beingallocated to labor, and with greater amountsof job creation and training. However, thesedifferences are not statistically significant.Community participation is associated with

THE IMPACT OF PARTICIPATION ON PROJECT OUTCOMES 27

17There are numerous details associated with calculating this figure, including estimating the size of leakages tothe nonpoor, the probability of obtaining work in the absence of the project, worker’s wages on the project,workers’ wages in the absence of the project, the probability of finding non-project work while working on theproject, and the level of nontransfer benefits. These are documented in Haddad and Adato (2002/03).

18Median values are broadly similar to mean values except for actual costs and number of training days.

Table 4.1 Project characteristics

Constructs Constructscommunity basic Other All

Project characteristics buildings infrastructure projects projects

Duration (days) 334 236 329 294Project cost (’000s rands)a 1,084 1,458 1,322 1,299Ratio: Cost overruns to planned project costs 30.6 11.3 11.4 18.7Percent projects over budget 65.6 61.5 66.7 64.3Project budget share spent on labor 30.7 34.8 70.4 43.3Ratio: Project wage to local unskilled wage 71.1 81.3 85.0 79.0Number of person years of work created 33.7 27.1 141.8 60.8Number of person years of training undertaken 3.8 1.8 5.9 3.6Percent jobs taken up by women 13.9 20.1 40.2 23.7Sample size 32 39 27 98

aAlmost all projects (with the exception of four) were completed between early 1995 and the end of 1997. Theexchange rate was R3.5 = $1.00 in January 1995; R4.8 = $1.00 in December 1997.

Table 4.2 Mean project performance by degree of community participation

Attaining program objectives Distribution of benefits Cost-effectiveness

(1) (2) (3) (1) (2) (1) (2) (3)Mean ratio: Mean ratio: Mean

Mean Mean log Project Cost Mean log cost ofproject number of Mean log wage to Mean overruns to cost of transferringbudget days of number of the local percentage planned creating one rand

Degree of community share spent work training unskilled employment project one day to a poorparticipation on labor created days wages to women costs of work person

No participation 0.37 7.94 5.49 0.85 8.25 13.4 5.12 11.10Any participation 0.44 8.45 5.94 0.78 25.94 19.2 4.41 7.19

F test on differences in means 0.76 1.25 0.76 0.90 5.80** 0.17 8.75** 3.93**

No participation 0.37 7.94 5.49 0.85 8.25 13.4 5.12 11.10Community advises but does not decide 0.52 9.61 6.63 0.82 33.84 4.9 4.32 6.47Community is joint decisionmaker 0.45 7.69 5.09 0.85 21.31 2.1 4.51 8.15Community is sole decisionmaker 0.35 7.90 5.82 0.70 21.93 41.0 4.40 6.92

F test on differences in means 3.03** 14.25** 5.01** 2.59* 3.78** 4.71** 3.31** 1.70

Note: * significant at the 10 percent level; ** significant at the 5 percent level.

improved targeting to the poor, as proxied bythe ratio of project wages to local unskilledwages as well as a markedly increased shareof employment going to women. While proj-ects with at least some degree of commu-nity participation have relatively higher costoverruns, this difference is not statisticallysignificant and projects with communityparticipation appear to have lower costs ofcreating work and transferring resources tothe poor.

The bottom panel of Table 4.2 assesseswhether the extent of community participa-tion affects project performance. While theF statistic indicates that we can reject thenull that mean outcomes are equal acrossdiffering degrees of participation, the pat-tern of these differences is not uniform. Forexample, job creation and training appearto be highest in projects where the commu-nity has an advisory or liaison role but nodecisionmaking authority whereas projectwages are relatively lowest where the com-munity has sole decisionmaking authority.Projects with the highest cost overruns,relative to projected budget are those wherecommunities have sole decisionmakingpower, next highest where there is no com-munity participation, and lowest where thereis joint decisionmaking. That all said, thereare a number of reasons why we might notwant to put too much weight on these find-ings. First, they do not account for otherfactors—for example, project characteristicssuch as size and type of asset created—thatmight also affect these outcomes. Second,they do not take into account two issuesalluded to in our conceptual framework: theprocesses by which projects were situated inparticular localities (also referred to as en-dogenous program placement [Rosenzweigand Wolpin 1986 and Pitt, Rosenzweig, andGibbons 1993]) and the possible endogene-ity of community participation.

Project performance may also be affectedby locality characteristics. For example,labor costs might be higher in areas withhigh local wages and low levels of unem-ployment. To account for this, we will draw

on data that describe the 34 districts in whichthe projects are located. These are taken fromthe 1995 October Household Survey (OHS),conducted by the Government of SouthAfrica’s Central Statistics Service. TheOHS collected detailed household data onjobs, wages, and employment status; educa-tion and demographic data; information onaspects of living standards such as housingquality and access to infrastructure (water,electricity, telephones, transport, and healthfacilities); and crime and safety. In ourmultivariate analysis, these household leveldata are aggregated into district means.

Results

Estimation IssuesAn estimable model that captures the rela-tionship between community participationand project performance can be written as

Iij = β ⋅ CPij + γ ′ ⋅ Pij + η′ ⋅ Li + eij (1)

where: Iij is the performance indicator ofproject j located in locality i; CPij capturesthe extent of community participation in theproject; Pij is a vector of other project char-acteristics; Li is a vector of locality charac-teristics; β, γ and η are parameters to be es-timated; eij is an error term; and vectors arewritten in bold.

As noted above, there are two potentialproblems associated with attempting to ob-tain consistent estimates of β. First, as notedearlier, governments choose to contract withcommunity-based organizations and theseorganizations choose to accept these con-tracts. Factors that affect this choice couldalso affect project performance. This im-plies that E(CPij eij) ≠ 0 and therefore thatestimates of β are biased and inconsistent.Second, government may choose to locateprojects on the basis of locality characteris-tics, for example citing infrastructure proj-ects in places with poor infrastructure. Thisimplies that E(Pij eij) ≠ 0 and that estimatesof both γ and β are biased and inconsistent.

THE IMPACT OF PARTICIPATION ON PROJECT OUTCOMES 29

One way of addressing these problemsis to think of participation as an endogenous“treatment.” That is, we estimate equation (1)as a treatment effects model using Heck-man’s (1979) two-step consistent estimator.This takes the following form:

Iij = β ⋅ CPij + γγ ′′ ⋅ Pij + ηη′ ⋅ Li + eij

where CPij, the endogenous dummy vari-able is assumed to reflect an unobservablelatent variable CPij* which itself is deter-mined by:

CPij* = ττ ⋅ wij + νij

where wij are covariates that affect partici-pation, ττ are their associated parameters, νijis a disturbance term and the relationshipbetween CPij and CPij* is given by:

CPij = 1, if CPij* > 0= 0, otherwise

and where eij and νij are bivariate normal.Amongst others, Maddala (1983, 120–22)shows that consistent estimates of β can beobtained by first estimating the determinantsof treatment (here, community participation).From this probit, the hazard (λ) or inverseMill’s ratio is calculated and then inserted asan additional regressor. This gives us:

Iij = β ⋅ CPij + γγ ′ ⋅ Pij + ηη′ ⋅ Li+ ωλij + eij

(2)

Estimating (2) requires that we identifycovariates (instruments) that plausibly af-fect participation but do not directly affectproject outcomes. We use two such covari-ates as instruments for participation.

The first draws on insights found inEasterly and Levine (1997), Mauro (1995),and Knack and Keefer (1997). Easterly andLevine’s note that “an assortment of politi-cal economy models suggest that polarizedcommunities will be prone to competitiverent-seeking by the different groups and havedifficulty agreeing on public goods like infra-

structure . . . ethnic diversity may increasepolarization and thereby impede agreementabout the provision of public goods (East-erly and Levine 1997, 1205–06). Addition-ally, where groups have a history of limitedinteraction—something certainly true of racerelations in South Africa—one would expectthat levels of trust across groups (which arebuilt up by repeated interactions) would belower. As Knack and Keefer (1997) empha-size, trust can be thought to help solve prob-lems of information asymmetries thereby“allowing self-enforcing agreements to bereached.”

This suggests that racial fractionaliza-tion would be a natural determinant ofthe likelihood of community participation.Specifically, we construct an index of ethnicfractionalization that takes the form: 1 –Σ I

i=1(ni / N)2 , i = 1, . . . , I; where ni is thenumber of people in the ith group, N isthe total population, and I is the number ofgroups. By construction, it can range invalue from 0, complete homogeneity—to 1,complete heterogeneity. We use the OHSdata at the district level on the percentageof individuals from different racial groups(White, Colored, African, Asian) to con-struct the index. These data were collectedprior to the implementation of these pro-grams so there should not be any reversecausation from project implementation toracial fractionalization.

The second covariate follows from theobservation found in the literature on com-munity-based development that stresses thatpast experience with collective action en-hances the capacity of people to take on newparticipatory projects. One covariate thatcaptures this is the share of the vote that theAfrican National Congress (ANC, the lead-ing political party in the struggle for major-ity rule) received in local elections held inNovember 1995 and March 1996—electionsthat occurred prior to the implementation ofthese projects. The political struggle formajority rule in South Africa had a strong“grassroots” community-level participatorycomponent and communities where this was

30 CHAPTER 4

most marked were communities that sup-ported the ANC. As such, these commu-nities may be able to engage in a differentform of collective action.19 Further, therewere elections held for approximately 136local councils; and it appears that for at leastone-third of our sample, our unit of analy-sis, “the community” maps directly onto alocal election. For the remainder, the “localcouncil” refers to a geographical entity largerthan “the community” but smaller than adistrict.

To see whether these characteristics af-fect the likelihood of de facto communityparticipation, we estimated a probit where

the dependent variable equals 1 if there is anycommunity participation (that is, cases wherethe community is the sole decisionmaker,joint decisionmaker, or plays an advisoryrole), zero otherwise. These results are re-ported in Table 4.3. Racial fractionalizationand ANC voting share in local electionshave a statistically significant effect on thelikelihood of participation at the 1 percentand 12 percent confidence levels, respec-tively. A chi squared test does not accept thenull hypothesis that these three correlates ofparticipation are jointly zero at the 2 percentconfidence level. The sign on racial frac-tionalization is negative, consistent with the

THE IMPACT OF PARTICIPATION ON PROJECT OUTCOMES 31

19While other political parties such as Pan African Congress (PAC) were involved in these political efforts, theyreceived little electoral support in the local elections.

Table 4.3 Probit estimates of the determinants of de facto participation

Dependent variable equals one wherecommunity, de facto, participates

Variable in project decisionmaking

Log projected duration of project (days) –1.601(4.63)**

Project constructs community buildings –3.934(4.17)**

Project constructs basic infrastructure –3.191(6.04)**

Log average wage in district 6.624(3.86)**

District unemployment rate 0.386(3.86)**

District poverty (P0) rate –5.366(1.92)*

Percentage of households reporting that they feel safe or very safe –0.348(1.47)

ANC voting share in local council election 4.302(1.53)

(Significant at the 12% level)Index of racial fractionalization –47.03

(2.73)**Constant 47.26

(4.01)**Chi squared on joint significance of instruments 7.61**

(Significant at the 2% level)

Notes: Covariates that serve as instruments for participation are in italics. Absolute values of z statistics are inparentheses. Standard errors account for clustering at the magisterial district level. * Significant at the10 percent level; ** significant at the 5 percent level. Magisterial district dummy variables for Metro-politan Cape Town and Winelands included but not reported. Sample size is 98.

hypothesis that increased diversity makesthe provision of public goods more prob-lematic. The sign on ANC voting share ispositive.

In addition to showing that these co-variates are correlated with participation, wealso need to demonstrate that they are un-correlated with project performance. We doso in two ways.

First, we estimate reduced form deter-minants of the eight project outcomes wehave been considering. In these regressions,we exclude community participation, includethe “instruments” racial fractionalization andANC vote share, and test to see if these co-variates are individually and jointly signifi-cant. These results are reported in Table 4.4.Neither covariate has a statistically signifi-cant effect on these project outcomes. Sec-ond, we construct a “pseudo-Hausman” test.We estimate a linear probability model withde facto community participation as the de-

pendent variable and use the same covariatesas in Table 4.4 as regressors. We recover theresiduals and include these in a linear re-gression where the dependent variables arethe measures of project performance listedin Table 4.5 and the regressors are (exoge-nous) de facto community participation aswell as project and locality characteristics.20

In only one case, cost overruns, is an instru-ment statistically significant at the 15 per-cent level or better and so we drop thismeasure from the remainder of our analysis.Collectively, we conclude from Tables 4.3,4.4, and 4.5 that our instruments are corre-lated with participation but uncorrelated withthe remaining seven performance measures.

In addition to addressing the concern ofendogenous participation above, we notedthat endogenous project placement is also apotential concern. However, there are sev-eral reasons why it is unlikely to be prob-lematic for our work here.

32 CHAPTER 4

20We also constructed the over-identification test as set out in Wooldridge (2002, 123) but did not obtain resultsany different from those described here.

Table 4.4 Reduced form determinants of project performance

Attaining program objectives Distribution of benefits Cost-effectiveness

Ratio:Log Project Ratio: Cost Cost of

Project number Log wage to overruns to Log cost transferringbudget of days number of the local Percentage planned of creating one rand to

share spent of work training unskilled employment project one day a poorVariable on labor created days wages to women costs of work person

ANC vote share 0.172 0.475 –0.029 –0.060 –16.15 –0.152 –0.562 –5.45(1.05) (0.65) (0.03) (0.15) (0.91) (0.26) (1.32) (0.45)

Racial fractionalization 0.180 –1.105 –1.176 0.303 56.48 –1.323 –0.392 –12.99(0.67) (0.72) (0.88) (1.13) (0.97) (1.92)* (0.51) (1.63)

F test on joint 1.22 0.31 0.87 0.69 0.64 2.52 0.93 1.37significance of racial fractionalization and vote share

Notes: Additional variables included but not reported are log projected project duration, type of infrastructure built, district mean wages, un-employment, poverty (P0) rate, percent households feeling safe or very safe, and location dummies. Absolute values of t statistics arein parentheses. Standard errors account for clustering at the magisterial district level. * Significant at the 10 percent level; ** signifi-cant at the 5 percent level. Sample size is 98.

First although each program had its ownstated set of location criteria,21 none of theprograms explicitly state that they use socio-economic data to determine which areasshould be prioritized and only two pro-grams (encompassing 7 percent of our sam-ple) mentioned infrastructure needs as atargeting requirement. Second, no localityreceived funding for more than one projectwithin a program but programs paid noattention to what other development projectfunding the locality received. Third, inseparate work, Adato and Haddad (2002)systematically examine the relationship be-tween locality and project selection, andcharacteristics of the districts in which theseprojects are sited. They find no evidence ofany systematic relationship between these.All these observations suggest that projectswere allocated to localities in a somewhatunsystematic fashion and that, therefore,concerns regarding nonrandom placementof interventions in selected communities arenot warranted here.

Estimation StrategyGiven the discussion above, we adopt the fol-lowing estimation strategy. We will estimatethe determinants of seven different measuresof project performance: the share of the proj-

ect budget allocated to labor; the log of thenumber of days of work created; the log ofthe number of training days undertaken; theratio of the daily project wage to the localunskilled wage; the percentage of employ-ment that goes to women; the log cost inrands of creating one day of employment;and the cost to the government of transfer-ring one rand to the poor.

We control for project characteristicsthat might affect these measures: projectedduration and type of asset being created(construction of community buildings, con-struction of basic infrastructure). We willcontrol for locality characteristics that mighthave an independent effect on project out-comes: log mean local wages for comparablesemiskilled work; the local unemploymentrate; the locality’s poverty rate; and the per-centage of households reporting that theyfeel safe or very safe. Lastly, we includeregional dummy variables (regions are geo-graphical entities smaller than a provincebut larger than a district) to capture regionalfixed effects. With these covariates, we willestimate models (1) and (2).

Basic ResultsEstimates of (1) and (2) are reported inTables 4.6, 4.7, and 4.8. We have grouped

THE IMPACT OF PARTICIPATION ON PROJECT OUTCOMES 33

21See Adato and Haddad (2002).

Table 4.5 Parameter estimates for the first stage residual in the determinants of project performance

Attaining program objectives Distribution of benefits Cost-effectiveness

Ratio:Log Project Ratio: Cost Cost of

Project number Log wage to overruns to Log cost transferringbudget of days number of the local Percentage planned of creating one rand to

share spent of work training unskilled employment project one day a poorVariable on labor created days wages to women costs of work person

Residual 0.109 –0.471 –0.389 0.194 74.21 –1.132 –0.258 –11.11(0.29) (0.35) (0.29) (0.72) (1.40) (1.16) (0.24) (0.98)

Notes: Additional variables included but not reported are de facto community participation, log projected project duration, type of infra-structure built, district mean wages, unemployment, poverty (P0) rate, percent households feeling safe or very safe, and location dum-mies. Absolute values of t statistics are in parentheses. Standard errors account for clustering at the magisterial district level. * Signif-icant at the 10 percent level; ** significant at the 5 percent level. Sample size is 98.

these tables according to type of perform-ance measure being considered. Table 4.6looks at the impact of participation on theattainment of project objectives. Table 4.7looks at the measure of the distribution ofproject benefits and Table 4.8 examines as-pects of cost-effectiveness.

Treating participation as exogenous, par-ticipation has a statistically significant, pos-itive effect on the project budget share spenton labor and the log number of trainingdays undertaken. Treating participation asendogenous, participation has a statisticallysignificant, positive effect on the projectbudget share spent on labor, the log numberof days of work created and the log num-

ber of training days undertaken. It increasesthe percentage of employment that goes towomen and is associated with a reduction inthe ratio of the project wage to local un-skilled wages. It reduces the cost of creatingemployment and reduces the cost of trans-ferring income to the poor.

Robustness Checks and Additional ResultsA concern we had in developing these re-sults was that our measure of communityparticipation was picking up the impact ofsome other locality characteristic. For thisreason, in preliminary work we added a widerange of locality characteristics to these spec-

34 CHAPTER 4

Table 4.6 The impact of participation on attainment of project performance

Project budget share Log number of Log number ofspent on labor days of work created training days

Variable (1) (2) (1) (2) (1) (2)

De facto participation 0.069 0.160 0.702 1.180 0.625 1.251(2.19)** (1.71)* (1.13) (2.53)** (0.76) (1.85)*

Project characteristicsLog planned duration –0.029 –0.023 1.450 1.535 1.239 1.295

(0.74) (0.74) (8.15)** (10.10)** (5.11)** (6.09)**Constructs community buildings –0.425 –0.422 –0.277 –0.265 0.457 0.512

(7.48)** (8.63)** (0.62) (1.08) (1.65) (1.51)Constructs basic infrastructure –0.382 –0.381 –0.429 –0.410 –0.425 –0.392

(6.58)** (7.68)** (1.64) (1.66)* (2.37)** (1.15)

Locality characteristicsUnemployment rate 0.002 0.001 –0.026 –0.029 –0.029 –0.036

(0.36) (0.20) (1.14) (1.07) (0.77) (0.97)Log average wages –0.176 –0.179 –0.761 –0.775 1.289 1.279

(2.15)** (1.77)* (1.99)* (1.54) (1.61) (1.89)*District poverty (P0) rate –0.463 –0.448 –1.744 –1.660 2.933 2.893

(2.86)** (1.93)* (1.61) (1.43) (1.26) (1.75)*Percentage of households reporting –0.002 –0.001 –0.044 –0.038 –0.023 –0.018

that they feel safe or very safe (1.16) (0.44) (3.44)** (3.00)** (2.00)* (1.14)Inverse Mill’s Ratio –0.089 –0.470 –0.558

(1.33) (1.42) (1.21)Chi squared statistic, all regressors 112.63** 189.58** 92.51**R2 0.508 0.649

Notes: (1) is basic specification; (2) is a “treatments” regression with community participation treated as endogenous. Absolute values of t sta-tistics are in parentheses for specification (1). Absolute values of z statistics are in parentheses for specification (2). Standard errorsare robust to cluster effects at the magisterial district level. * Significant at the 10 percent level; ** significant at the 5 percent level.Magisterial district dummy variables for Metropolitan Cape Town and Winelands are included but not reported. Sample size is 98 forbudget share and days work created, 86 for days training created.

ifications. These characteristics included: av-erage household size; proportion of femaleheaded households; the percent of individ-uals that have completed standard 5 andstandard 10 of schooling (standard 10 is theequivalent of completing high school);mean per capita incomes; the standard devi-ation of per capita incomes; the standarddeviation of male and female wages; qualityof infrastructure; the proportion of house-holds who report that they are unable to feedtheir children; the district rate of unemploy-ment; the district rate of long-term (greaterthan one year) unemployment; proportionof adults by occupational class; housing

quality (size, building materials, sanitation);levels of home ownership; access to water,electricity, telephones, transport, and healthfacilities; and reported levels of crime. Noneof these variables had explanatory power,thus they are not used in the results reported.We experimented with different controlsfor project location; again doing so had nomeaningful impact on the results reportedhere. Finally, we included a set of dummyvariables that reflected our subjective as-sessment of the quality of the project leveldata we had obtained. These had no statisti-cal significance and their inclusion does notsubstantively alter our results.

THE IMPACT OF PARTICIPATION ON PROJECT OUTCOMES 35

Table 4.7 The impact of participation on the distribution of project benefits

Ratio: Project wage to Percentage employmentthe local unskilled wages to women

Variable (1) (2) (1) (2)

De facto participation –0.131 –0.209 14.42 17.79(1.45) (2.19)** (2.29)** (1.69)*

Project characteristicsLog planned duration –0.011 –0.017 3.894 4.144

(0.20) (0.56) (1.42) (1.21)Constructs community buildings –0.149 –0.151 –24.895 –24.808

(5.69)** (3.01)** (9.07)** (4.51)**Constructs basic infrastructure –0.037 –0.041 –20.319 –20.185

(0.75) (0.80) (5.79)** (3.63)**

Locality characteristicsUnemployment rate –0.001 –0.0004 –0.868 –0.890

(0.17) (0.07) (1.36) (1.48)Log average wages –0.602 –0.599 –2.131 –2.227

(9.80)** (5.81)** (0.17) (0.20)District poverty (P0) rate –0.383 –0.398 –3.900 –3.308

(2.59)** (1.67)* (0.08) (0.13)Percentage of households reporting –0.005 –0.006 –0.402 –0.364

that they feel safe or very safe (2.32)** (2.36)** (3.77)** (1.27)Inverse Mill’s Ratio 0.077 –3.323

(1.12) (0.43)Chi squared statistic, all regressors 80.96** 58.64**R2 0.415 0.330

Notes: (1) is basic specification; (2) is a “treatments” regression with community participation treated as en-dogenous. Absolute values of t statistics are in parentheses for specification (1). Absolute values of z sta-tistics are in parentheses for specification (2). Standard errors are robust to cluster effects at the magis-terial district level. * Significant at the 10 percent level; ** significant at the 5 percent level. Magisterialdistrict dummy variables for Metropolitan Cape Town and Winelands are included but not reported. Sam-ple size is 98.

We experimented with the inclusion ofother variables that we believed might af-fect community participation. These in-cluded measures of economic stratification(such as the standard deviation of earnings;levels of unemployment; percentage of indi-viduals in different occupations; levels andseverity of poverty); political fractionaliza-tion, derived in the same manner as the indexof ethnic fractionalization and drawing oninformation on the shares of votes obtainedby different political parties and measures ofcommunity access to infrastructure (such asdistance to various facilities; access to tele-phones); and other measures of political ac-

tivity, such as voter turnout and the identityof the party that controlled the local council.None of these variables had explanatorypower in the regressions used to predict theprobability of participation and hence theirexclusion does not affect our results.

ConclusionThis chapter has examined the relationshipbetween different types of community partic-ipation and the performance of public worksprojects. We develop some simple analyticsthat are used to motivate the empirical analy-sis of the impact of participation on the effi-

36 CHAPTER 4

Table 4.8 The impact of participation on project cost-effectiveness

Log cost of creating Cost of transferring oneone day of work rand to a poor person

Variable (1) (2) (1) (2)

De facto participation –0.438 –0.909 –3.841 –8.045(1.43) (2.78)** (1.56) (2.66)**

Project characteristicsLog planned duration 0.117 0.082 –0.210 –0.522

(1.52) (0.77) (0.16) (0.53)Constructs community buildings 0.714 0.702 3.380 3.272

(3.13)** (4.10)** (2.79)** (2.06)**Constructs basic infrastructure 0.927 0.908 3.165 3.000

(6.27)** (5.24)** (3.93)** (1.87)*

Locality characteristicsUnemployment rate 0.026 0.030 –0.418 –0.390

(1.74)* (1.58) (1.74) (2.24)**Log average wages 0.806 0.819 0.267 0.386

(1.96)* (2.32)** (0.09) (0.12)District poverty (P0) rate 1.037 0.955 –1.418 –2.155

(1.12) (1.18) (0.55) (0.29)Percentage of households reporting 0.017 0.011 0.004 –0.043

that they feel safe or very safe (3.91)** (1.28) (0.11) (0.53)Inverse Mill’s Ratio 0.464 4.141

(2.10)** (2.00)**Chi squared statistic, all regressors 93.58** 42.34**R2 0.455 0.224

Notes: (1) is basic specification; (2) is a “treatments” regression with community participation treated as en-dogenous. Absolute values of t statistics are in parentheses for specification (1). Absolute values of zstatistics are in parentheses for specification (2). Standard errors are robust to cluster effects at the mag-isterial district level. * Significant at the 10 percent level; ** significant at the 5 percent level. Magisterialdistrict dummy variables for Metropolitan Cape Town and Winelands are included but not reported.Sample size is 98.

cacy of public works interventions in theWestern Cape. The empirical implicationsof our model are straightforward: (1) weshould expect to see an empirical differencebetween programs that are communitymanaged and those that are purely govern-ment run; (2) there is no direct link betweenthe cost of delivering benefits in a programand whether a program is community run.We could easily find community managedprograms that have more or less cost perunit in equilibrium because of changes inobjectives. Formally testing these impli-cations is, however, challenging. For areasonably large number of interventions,we need information about the extent ofcommunity participation, the level of ben-efits to the community, the distribution ofbenefits within the community, and indica-tors of cost effectiveness. Further, such data,while necessary, are not sufficient. Projectperformance might also be affected by lo-cality characteristics and so we need addi-

tional data that can control for these. Lastly,the fact that in our model financiers chooseto work with community-based organiza-tions implies that we cannot necessarilyassume that community participation isexogenous and so our data must also be richenough to take this potential endogeneityinto account.

We find that de facto participation has astatistically significant, positive effect on theproject budget share spent on labor, the lognumber of days of work created, and thelog number of training days undertaken. It in-creases the percentage of employment thatgoes to women and is associated with a re-duction in the ratio of the project wage tolocal unskilled wages. It reduces the cost ofcreating employment and reduces the costof transferring income to the poor. The mag-nitudes of these impacts are sizeable. Thesefinding are robust to a variety of modelspecifications and the inclusion of othercovariates.

THE IMPACT OF PARTICIPATION ON PROJECT OUTCOMES 37

C H A P T E R 5

Community Participation in theWestern Cape Projects

While the econometric work in the previous chapter shows the effects that commu-nity participation had on a range of project outcomes, it does not tell us why sucheffects were found. To get inside the “black box” of community participation, this

chapter uses the case study analysis to examine the forms of community participation in theeight case study projects, both intended and actual, and discusses their implications—the ben-efits and consequences of different levels and types of participation.

The Case Study ProjectsThis section provides an overview of the eight project-level case studies. These case studiesinvolved in-depth investigations of the institutional arrangements, actors involved, and pro-cesses that occurred in the course of planning and implementation of each project. Althoughextensive qualitative data collection is what distinguishes the case studies from the rest of thepublic works dataset, as a subset of the 101 projects quantitative data were also collected.

Table 5.1 provides quantitative characteristics and performance of each of the eight proj-ects, comparing this performance to the average for the different asset categories. As noted inChapter 3, the case studies were selected to be roughly proportional to the full datasetwith respect to institutional arrangements, rural/urban location, and infrastructure type. Boththe selection on institutional arrangements and the secondary level of selection that consideredin more detail the role of government, NGOs, CBOs, and consultants were intended to cap-ture variation in de jure and de facto participation.

With respect to de facto participation outcomes, there were three case studies with veryhigh to high levels of participation (Lutzville—the highest, Thembalethu, and Clanwilliam);two cases of medium-level participation (Khayelitsha and Langa);22 two cases of medium-lowparticipation (Stellenbosch and Kylemore); and one case of no participation (Murraysburg).Table 5.2 shows this case study stratification compared to that of the total project database, in-dicating a roughly reasonable reflection of the different categories, with the category of no par-ticipation under-represented. However, it is important to point out again that the case studiesare not intended to be statistically representative of the full dataset, but rather to capture vari-ation that provides insights into a range of issues and conditions, taking a close look at theprocesses behind each of these outcomes.

22As explained in Chapter 3, the Khayelitsha project was actually two projects and thus we actually studied threecases of medium-level participation.

38

Table 5.1 Quantitative indicators for the eight case study projects

Distribution of Attainment of project performance project benefits Project cost-effectiveness

Ratio:Project Number Project Cost of

budget share Person of person wage to Log cost of transferringForm of spent on years years of local Percentage creating one rand to

community Duration labor of work training unskilled employment one day a poorProject name Project activity participation (days) (percent) created undertaken wages to women of work person

Lutzville Community center (CBPWP) Highest 120 24 11.1 0.7 1.00 11.6 4.73 12.2

All projects constructing community centers 339 30.2 34.7 3.8 0.71 14.4 4.57 8.5

Khayelitsha 4C Roads and stormwater drainage (Pilot) Medium 330 14.3 54.6 3.9 0.54 18.0 5.27 6.7

Khayelitsha 3A Roads and stormwater drainage (Pilot) Medium 520 11.9 130.9 5.1 0.54 4.9 5.49 7.5

Thembaletu Stormwater and roads (Trans) Very high 390 28.2 73.3 4.2 1.26 40.0 5.91 8.0

Clanwilliam Stormwater drainage (CBPWP) High 180 20.1 13.2 0.5 0.71 10.1 5.11 8.2

Murraysburg Water reticulation (CBPWP) None 240 17.3 19.4 0 1.04 0 5.31 7.9

All basic infrastructure (road, storm sewers, 236 34.8 27.1 1.9 0.81 20.1 4.75 8.5sanitation sewers, and water reticulation)

Langa Clean and green (CAG) Medium 330 52.9 51.6 44.3 0.42 60.0 3.80 3.7

Stellenboscha Remove alien vegetation (FWCP) Medium 390 92.2 210.1 2.2 0.69 43.9 3.61 2.5

Other projects (removal of alien vegetation and 329 70.4 141.7 5.9 0.85 40.2 4.00 5.2general cleaning and greening)

aData are not available for Kylemore because at the time the database was constructed, the Stellenbosch project included Kylemore (prior to our case studies it was split into a separate project with sep-arate managers and committees).

With respect to the main performancevariables considered in Chapter 4, the Lutz-ville project with the highest level of defacto participation performed close to theaverage for all projects in its class of infra-structure (community halls) with respect tolabor intensity (budget share spent on labor),employment for women, and efficiency increating work and transferring income tothe poor. The Thembalethu project, with avery high level of participation, performedclose to the average in its class of infra-structure (roads, storm sewers, and waterreticulation) with respect to labor intensity,and efficiency in creating work and trans-ferring income to the poor. However, it per-formed at twice the average participation ingiving employment to women. The Clan-william project with a high level of partici-pation performed far below the average withrespect to labor intensity23 and employmentof women, but at the average for efficiencyin creating work and transferring incometo the poor. The Khayelitsha project witha medium level of de facto participationperformed far below average with respect tolabor intensity; at the average for employingwomen in one of the two projects (the onecompleted on time) but far below in the otherproject; and at the average for efficiency increating work and transferring income to thepoor. Langa, the other project with mediumlevel participation, performed below average(for “cleaning and greening” projects) inlabor intensity; above average in employing

women; and at around average in efficiencyin creating work and transferring incometo the poor. The Stellenbosch/Kylemoreprojects performed above average in laborintensity; at average in employing womenand efficiency in creating work; but some-what below average in transferring incometo the poor. Although there were some vari-ations from the average (some of which isexplained in the following two chapters), onthe whole these project outcomes are reason-ably reflective of outcomes across the widerdatabase.

What CommunityParticipation Looks LikeFunds for public works projects were allo-cated to government or nongovernmentproviders representing a geographic entity,usually a local level political and adminis-trative unit (under the system in place at thetime the program was initiated). Programfinanciers in the central government requiredthe participation of a community-based or-ganization. In the South African contextthis meant a civic association, RDP Forum,or project committee composed of commu-nity members. These were nongovernmental,though Project Steering Committees (PSCs)were sometimes composed of both localgovernment officials and other communitymembers. These CBOs were supposed to berepresentative of the broader community. Inmost (though not all) cases they were elected

40 CHAPTER 5

23One of the stakeholders commented that the engineers on the project did not present any options for labor-intensive construction.

Table 5.2 Comparison of de facto participation in the case studyand full project database

Case studies All projectsDe facto Case studies (percent (percentparticipation (no.) of total) of total)

Highest-high 3 37.5 44.44Medium-low 3 37.5 24.24None 1 12.5 31.31

bodies. Community participation in the pub-lic works projects then had two main dimen-sions: (1) participation of the community atlarge through mass meetings involving con-sultation or decisionmaking; and (2) par-ticipation of representative members of thecommunity in a CBO, often in the form of aPSC. (PSC is used here to represent all typesof community-based project committees/organizations, which varied in structure andname between programs and projects).

While roles given to the community atlarge or to a CBO/PSC were sometimesclearly differentiated, the meaning of theterm “community” is often blurred. Whenprogram stakeholders refer to “the commu-nity” they are sometimes referring to com-munity residents in general, sometimes to thePSC, other times imprecisely to both. “Com-munity” is a problematic term in several re-spects. Bozzoli (1987) problematizes com-munity in the South African context, placingit within its apartheid history. “Community”was long used to signal race by the apartheidgovernment, but also became part of the dis-course of resistance movements. “Commu-nity” signals inclusion and solidarity, but issimultaneously exclusionary. Reference to“the community” can obscure divisions ofrace, class, gender, political affiliation, andother differences. As Lund (no date, refer-ring to Bozzoli 1987) points out, commu-nity is not a static variable: “actually exist-ing residential areas will at different times,and in the face of different threats, and undercertain conditions, organize collectively, witha real sense of community, but always ex-cluding some groups and including others.”The question of who in the community isrepresented or excluded by a PSC or localgovernment is discussed in Chapter 6.

All the role-players in the eight projectsstudied, including financiers, providers (in

government, the private sector, NGOs, andcommunity groups), and beneficiaries agreedthat community participation in some formis essential to the functioning of public worksprojects. In contrast to the past when thestate engaged in top-down planning andimplementation, the community was seenby all as an indispensable player, althoughthere were vastly different perceptions of itsappropriate roles—from the view that com-munities should be able to run projects ontheir own and that even where this mightlead to financial disaster this is still an im-portant learning experience (interview, for-mer NGO staff, Langa project), to otherswho felt that participation does not havemuch place once construction is underway(interviews, contractor, construction firmon Khayelitsha project; official, DTPW). Thereality of community participation in his ex-perience was explained by the assistant cityengineer for Khayelitsha (the implementingagent/provider for the Khayelitsha Pilotprojects):

I believe that we have to involve thecommunity. It’s the way that this com-munity works . . . so one has to respectthat and take cognizance of it and ac-commodate it. And if you don’t do it inany case, if you think it’s going to takelonger because you’ve got to consult acommittee, you’re making a big mis-take. Because if they don’t want you todo it, you won’t do it—you won’t getthe job going. They24 will just comearound and make a lot of trouble. Andif I were them I’d most probably do itmyself, as well. You know, I’m sittingin a shack down there, and there mightbe a job opportunity here, and no-oneconsults me or my representative. Wehave to walk the extra mile, we have to

COMMUNITY PARTICIPATION IN THE WESTERN CAPE PROJECTS 41

24Interpretations of “they” is a frequent theme in this research. The idea of African, colored, and white commu-nities as “different” or “other”—views based on historical racial relationships, lack of familiarity and under-standing, and stereotyped discourse—shaped responses to these programs, responses to the “other,” and theirreceptiveness to participation of the “other.” This point is elaborated, with examples, in Chapter 6.

go there, and get them on board. Andwe do that, or we try to do that, in anycase . . . it works generally. People25

tend to see the whole community partic-ipation process I think also in the wronglight. They feel these guys are interfer-ing in the process and they don’t knowanything about it. Yes, they don’tknow anything about it—but theywant to know about it (interview, engi-neer, Lingulethu West City Council).

The tradition of participation in urbancommunities—not just in civic organiza-tions but also groups involved with servicedelivery—can be traced through the his-tory of activism and community consultationthrough civic organizations that became animportant feature of the urban landscape inthe mid-1980s. The organizations often en-gaged in both antiapartheid activism andservice delivery in communities where thestate was not delivering services (Seekings1988; Lodge 1991). However, expectationsof community consultation are not con-fined to urban environments. In the smallmostly Afrikaaner and “coloured” town ofLutzville in the rural northwest corner of theprovince, the town clerk explained thatcommunity participation even at the designstage is “very important because if the com-munity is not satisfied with the design andyou spend R3 million, it won’t be a satis-factory building. That, one must clearlyunderstand.” (interview acting town clerk,Lutzville).

Although there is increasing recogni-tion of the importance or at least the in-evitability of participation, in practice pro-

viders in Western Cape provincial and localgovernments continued to view infrastruc-ture delivery as a matter of technical expert-ise and were slow to hand over real respon-sibility to community members. The findingsof our database of 101 projects show thatalthough all programs had an emphasis oncommunity participation in program design,CBOs ran approximately 30 percent of theprojects.26

Types of Participationfrom the Case StudiesThrough the case studies, we identified anumber of areas as actual and potential areasin which projects did or can benefit fromcommunity participation—from infrastruc-ture selection to project design and manage-ment. Table 5.3 shows how each projectactually performed in relation to types ofparticipation that could potentially have in-volved the community as a whole27—wherethe wider community/beneficiaries (not justa PSC) take on some aspects of the providerrole. The gray boxes indicate that participa-tion did take place in this area. The check-ered boxes indicate that there were mixedresponses, with either significant differencesof opinion or a circumstance not possible tosimplify with a positive or negative answer.The white boxes indicate that participationdid not take place. The FWWP projects have“n.a.” in some categories where the designof the program made the question not appli-cable. Lutzville had the greatest degree ofbroad community/beneficiary involvement,with the community as a whole involved inselecting the asset; electing a CBO/provider;

42 CHAPTER 5

25 “People” here refers to other white professionals in government and the private sector. The informant speakingis also a member of this group, reflecting critically on these relationships.

26A national evaluation commissioned by DPW of the CBPWP ranked Western Cape among the three lowest per-forming provinces (out of the nine) in terms of participation, using primarily the criteria involvement in choiceof project, planning, financial decisionmaking, and knowledge of maintenance (Everatt et al. 1997). The CASEevaluation sampled only three projects from the province and thus is not representative. However, it does appearto be consistent with the findings from our dataset.

27This of course does not imply that virtually everyone in the community participates, but that a communityforum enabling broad participation took place. There will always be people who either do not want or are not ableto participate for a variety of reasons. This issue of choosing not to participate is taken up later in this chapter.

Table 5.3 Types of participation where broad community involvement occurred

Level of participation Lutzville Langa Khayelitsha Thembalethu Clanwilliam Murraysburg Kylemore Stellenbosch

Community helped n.a. n.a.select asset

Community felt n.a. n.a.asset a good choice

Had an elected PSC or CBO involved

Community selected workers

Community contributed n.a. n.a.to project design

Adequate communi-cation with community during project

Community had sense of ownership

Notes: PSC indicates Project Steering Committee; CBO, community-based organization; n.a. indicates that, due to program design, the ques-tion was not applicable.

Broad community participation did not occur.Broad community participation occurred.Results were mixed, with either significant differences of opinion or complex circumstances not able to be characterized as

simply positive or negative.

receiving adequate communication fromproviders; and reporting a sense of projectand asset ownership. Murraysburg had thelowest degree of broad community partici-pation, where the wider community agreedafter the fact that the asset was a goodchoice, but no further communication tookplace between providers and beneficiaries.

Selection of InfrastructureAs suggested earlier, community partici-pation can potentially secure a better dis-tribution of benefits. Broad community/beneficiary involvement in the selection ofinfrastructure has three main advantages:(a) assets are chosen that are in line withcommunity priorities, (b) access to those as-sets can be more equitable, and (c) greater

awareness of the cost trade-offs of differentchoices is generated among the community.

The NPWP states that communitiesshould be involved with choosing the typeof asset on which to spend public worksfunds. It also recognizes that governments(in the role of financier or provider) haveadditional or different considerations to theneeds of communities/beneficiaries in prior-itizing projects, and if priorities conflict,“the needs of the community must be rec-onciled with government planning” (DPW1994, 10). In four of the 18 CBPWP proj-ects, community facilitators contracted byDTPW28 found that local government hadchosen projects that were not the highestpriority for the community, and followinga process of consultation, the project type

COMMUNITY PARTICIPATION IN THE WESTERN CAPE PROJECTS 43

28In the framework of Chapter 4, community facilitators were agents of the financiers, helping to assure thatproviders more closely represented their priorities, in this case greater community participation.

was changed (interview, C. B. Associates).29

Community involvement can also help cor-rect for bias that local government mayhave with respect to one constituency. How-ever, a better outcome could also beachieved if a community decisionmakingprocess is facilitated by local government.Local government may have information thatcommunities are unaware of; for example,plans may be underway or funding alreadyallocated to certain kinds of infrastructure,making another type of asset a better choice,or there may be environmental or health im-plications associated with one type of assetcompared with another. Of 101 projects inthe Western Cape, 22 were community halls.It is not possible to know whether these werereally the highest infrastructure need of thecommunities that requested them. Adato andHaddad (2002) show how consultant engi-neers, many of whom favored building com-munity halls, positioned themselves to getaccess to project contracts. Access to goodinformation is an important factor in facili-tating successful community participation.

Choices of infrastructure have equityimplications often reflecting power relation-ships in the communities. A decision tobuild the stormwater system in Clanwilliammeant that the “coloured” community ofCedarville received the benefits, rather thanthe African community of “Khayelitsha 2,”the new squatter area for which an alter-native project was initially planned for ablu-tion blocks30 until it was later changed. Theablution block project was abandoned be-cause people were not living on the sitesyet, because this could have been funded byhousing subsidies, and because “the com-munity” decided the higher priority was thestorm water project (interview, C. B. Asso-ciates). It may be that the ablution blockproject would not have been a good choice

(when these were eventually built peopledid not move to the site). However, this casecautions against seeing “the community”as one entity, as residents of Khayelitsha 2and Cedarville had different interests. Theformer may not want to use their housingsubsidies for a project otherwise paid for,may have had other ideas for infrastructurethey wanted, and did not benefit from thechoice of the stormwater project. Commu-nity participation can embody and exacerbateinequities depending on who participates,and is thus not a guaranteed protectionagainst one part of the community capturingthe benefits.

One role of provincial department facil-itators (as agents for financiers or providers)can be to monitor this process to increaseits inclusiveness. An inclusive participationprocess for choosing the type of asset to bebuilt can minimize the possibility of havingthe benefits captured by a less poor part ofthe community, or at least bring these choicesout into the open. This can also help to min-imize conflict later on: in Khayelitsha, thedecision by the municipality to build roadsin one part of the section and not another(meaning also that jobs were not createdthere), later led to the disruption of theproject. Bringing community leaders into adecisionmaking process, where the facts oflimited funds and difficult choices are dis-cussed, could avoid this kind of problem—“could” rather than “would,” because theneed for resources in poor communities canoverpower negotiated agreements (as in thecase of workers who agree to a wage rateand then strike afterwards).

Understanding Cost ImplicationsCommunity involvement in the choice of in-frastructure also means people will be awarethat choices also involve costs. In Murrays-

44 CHAPTER 5

29Note that all programs had varying boundaries around the type of project that could be chosen, with FWWP themost limited, the CAG limited to environmental clean-up, and the Pilot and Transport projects limited to roads.The most flexible were the infrastructure projects of the CBPWP, CEP, and WCEDF.

30Ablution blocks are normally buildings with toilets and running water, sometimes sinks and baths or showers.

burg, project workers (from the community)said they were having difficulty paying thenew water bills that came with householdtaps (installed by the project) instead of com-munal taps. As one of the Murraysburg localcouncillors explained it, there was inade-quate communication on the need to pay forwater: “The only thing they went for is toget more water, but they didn’t realize thatthey had to pay more for the water” (inter-view). Although some responses indicatedthat people would prefer the taps but neededto understand the costs, one comment at theworkshop was that

there are people who are looking backto the old street taps because now thebig accounts are coming to them evenwith lawyers letters. And there are peo-ple who are saying: “why should I payfor water that comes from God?” andall these types of things. When the tapswere installed people were happy butnow that they see what they have to payfor it they are not longer so happy (proj-ect worker, Murraysburg workshop).

As Table 5.3 indicates, lack of participa-tion does not mean that the community dis-agrees with the choice. In four of the sixcases, the PSC and workers felt that the assetwas a good choice (though not necessarilythe best) and in the remaining two no dis-satisfaction was found. The two FWWPprojects were by definition alien vegetationclearing and thus did not involve any proj-ect choice.31 However, in both Kylemoreand Stellenbosch, workers did express thatthey were not consulted on the type of proj-ect and that basic infrastructure and hous-ing were higher community needs. Althoughalien vegetation clearing does increase waterresources, this does not directly affect theworkers’ communities, so it is not surprising

that they would see infrastructure as a higherpriority. However, they are not consideringthe scale of job creation, and it is very pos-sible that with this information, they mightchoose alien vegetation clearing. This againimplies the importance of good information(for example, on relative job creation andskill development potential of different in-frastructure choices) communicated by pro-viders to beneficiaries if the latter are to par-ticipate in selecting between asset types.

Under the re-aligned CBPWP that DPWintroduced in 1998, community-based steer-ing committees have co-responsibility forproject identification, along with local gov-ernment. Unlike project management andother implementation issues, this is an areawhere the PSC can facilitate the participa-tion of the wider community.

Community RepresentationThe main way that the community partici-pates in projects is through a project steeringcommittee (PSC), the mechanism throughwhich the community becomes the provider.Thus the degree to which the committee isrepresentative of the broader community be-comes an important proxy for communityparticipation in an ongoing capacity. In fiveof the eight cases, projects had steering com-mittees that were elected by the community—either pre-existing community organiza-tions such as street committees or RDP fo-rums, or project steering committeesformed for this project but chosen throughthese existing structures. More fragmentedcommunities may result in less representa-tive steering committees, particularly wheremechanisms do not exist to resolve conflicts.This was clearly the case in Thembalethu.The legitimacy of Thembalethu’s commit-tee, where community members had beenappointed from among the leadership ofthe controversial Thembalethu Development

COMMUNITY PARTICIPATION IN THE WESTERN CAPE PROJECTS 45

31The program only involves one type of project and takes place far outside of communities where the relevantspecies are found. The equity issues arise only in determining where to locate the project in terms of which com-munities will obtain employment.

Trust, was disputed and this is reflected inthe many conflicts that arose in the courseof the project, with the committee some-times at the center of the conflict rather thanmediating.

Where communities are politically highlyfragmented, like in Thembalethu, Clan-william, and Murraysburg, it is particularlyimportant that project committees are elected.As a key role player from the Clanwilliammunicipality put it, “what is very importantis the committee is representative of all therole players. In that way, half of the problemis solved. If this process is not in order thereis problems” (interview).

Community Input into Project DesignThis falls into stage 2 and 3 of Paul’s (1987)categorization of types of participation. InSouth Africa, this was a highly contested areawith regard to the benefits of communityinput/decisionmaking. Project design in-volves many—though not solely—technicaldecisions: in a road project, for example, itwould include the type of road built; thematerials and machines used; the shape andsize of the road and other features; the typeof training. Two CBPWP directors in theprovince, one with extensive experiencefacilitating community participation, saidthat communities tend not to be that inter-ested in project design, generally preferringto leave that to those with technical training.Using participatory research methods in thestudy workshops, project workers listed alltasks in their project and then chose the roleplayers they thought should be responsible

for those tasks. Out of seven workshops, inonly one did workers mention project de-sign as a task in the project (much lesssomething that workers or the PSC shouldbe responsible for). It is important to notethat these were project workers/communitymembers; PSC members would have beenmore likely to mention design.

At the same time, participation does in-volve opportunity costs, and it is possiblethat communities will opt out of participat-ing in a given area. Participation takes timeand interest, and not everyone will prioritizethis use of their time or have this interest.32

In several of the case studies, individuals onthe PSC tended to be those who were onother committees and played other leader-ship roles in the community—those individ-uals for whom participation in developmentor political processes was a priority. Theseindividuals also tended to be more educatedand skilled, and thus more likely to haveaccess to employment and alternative usesfor their time—contradicting the opportunitycost argument. Nevertheless, the fact thatmany people do not come to meetings orvolunteer to join the PSC implies an oppor-tunity cost, financial or otherwise. In caseswhere PSC members are remunerated fortheir time—some projects recognized thecost of participation—there is of coursemore interest in joining—presenting greateropportunities for conflict over these posi-tions. But in these South African communi-ties with very high unemployment and fewresources to access, the opportunity costsof participation are relatively low33 and theinterest in monitoring the use of resources

46 CHAPTER 5

32Writing on the issue of infrastructure maintenance, Ostrom, Shroeder, and Wynne (1990) discuss opportunitycosts in terms of the investment of resource inputs, particularly time and labor, that must be exceeded by the per-ceived benefits.

33In industrialized countries with high employment and more abundant resources, there is less interest in partic-ipating, and the chances of getting a mass turnout to a meeting about a road project is low. The city engineer forthe Khayelitsha project observed this in commenting on the strong interest in participation and large meetingturnouts in Khayelitsha and the contrast with his own white, middle class community: “It’s not the way that wework. I mean if [my town of] Somerset West wants to build a park [next] to my house, I believe they know whatthey’re doing and they’re doing it the right way, and they’ve planned it properly—so I don’t worry about it toomuch” (interview, engineer, Lingelethu West Development Council).

is high—this is evidenced by the large turn-outs to mass meetings.

Taking into account varied levels of in-terest, there are several reasons for gettingcommunity input during the project designphase. It is likely that as people becomefamiliar with the implications of designchoices they would be more interested inparticipating for these reasons:34

1. Problems can be resolved early avoid-ing disruption later. Involvement upfront means that if there are problemsraised that need to be resolved, this oc-curs before the project is underway andchanges become more costly. A com-munity unhappy with how they see aproject progressing may disrupt it mid-stream, as they did in the Khayelitshaproject.

2. Design aspects may be more appropri-ate to community needs and preferences.Beneficiaries can express design pref-erences, for example, with respect toallocation of space in community build-ings. In Lutzville, the committee helpedto decide on the design of a creche,kitchen, and meeting hall. In Langa, thecommunity decided where they wantedto start with the greening, choosing theentrance to Langa to create a good firstimpression, and to not favor one part ofthe community over another. However,this choice did not benefit the poorestpart of the community, as would an al-ternative idea to build a small park nearthe hostels.

3. Local knowledge can enhance safety,convenience, and efficiency. Commu-nity members can contribute ideas thatmake a design better (for example,safer) for the community. In Them-balethu, residents wanted speed bumpsand taxi pull-offs, pointing out that theabsence of these features created safetyhazards for children and adults (given

how fast taxis and other traffic traveledon the long straightaway). To controlspeed, the South African National CivicsOrganization (SANCO) representativehad also proposed a zig-zagging designrather than a long straightaway. Thesesuggestions were not adopted by theproviders, for reasons related to costand jurisdictions.

4. Access to relevant inside informationabout the community can be provided.In Khayelitsha, the community raisedthe point that a brick road would bevandalized by people taking bricks outof the road for use as building materials.This was a point engineers had not real-ized on their own. In this case the choiceinvolved a trade-off in labor use as abrick road would have created morejobs. Still the community chose paving.

5. Community priorities can be reflectedand trade-offs understood. People canmake decisions involving benefits andtrade-offs that affect their communities.The consultants at Thembalethu calleda mass meeting where they asked whattype of road they should build. Thecommunity chose bricks instead ofpaving, because more jobs would becreated. It also would mean that brickmaking skills would be left in the com-munity, enabling new job opportunitiesin the housing construction industry.The Regional Services Council and taxidrivers argued against using bricks, andthe consulting engineers told the com-munity that a brick road would be 35percent more expensive but would cre-ate more jobs. The community chosemore jobs (interview, consultant engi-neer, Thembalethu project). The abilityto participate in these kinds of decisionsin an informed way depends on therange of choices presented—which willbe shaped by technical, economic, and

COMMUNITY PARTICIPATION IN THE WESTERN CAPE PROJECTS 47

34Participation in design choices can occur through the PSC or through a mass meeting where the PSC solicitssuggestions, as occurred at Thembalethu.

political considerations and interests—and the kind of education that is builtinto the process. Explaining how he hasseen the process work in many commu-nity projects in the past, one DTPW of-ficial suggested that communities couldbenefit more than they currently do ifthey did participate in design issues:

The supply of materials, they’re not tooconcerned about that. . . . And to methat’s perhaps a shortcoming because in any building construction work . . .money generally goes for materials.And if you look at the 60% of it—andthe change in those things—that iswhere you actually create more work. . . Let’s just say a brick [from the fac-tory] costs you 50 cents and a brickmade on site costs you 55 cents—itmay cost you more, you get less bricks,you get less building for it, but 80% ofthat 55 cents can be money earned bythose participating in brick-makingwhereas the entire 50 cents really goesout of the community. And you knowthose kind of understandings cannot beseen easy because that decision is longbefore the bricks arrive. . . . Peoplemust be made aware of where the re-sources are, how those resources can be changed (interview).

Some kinds of projects or infrastructureare more conducive to community input intothe design than others. In the two FWWPprojects, there is little room for communityparticipation in project design, as the alienvegetation removal is standardized. Waterreticulation projects do not lend themselvesto as many choices as a road or a commu-nity hall, although there are some. In Clan-william, no design options were discussedwith communities or the PSC, althoughthere is a wide variation in the labor inten-sity of different drainage and pipe laying de-signs (interview, official, DPW). This againpoints to the importance of good informa-

tion if communities are to make informedchoices for or against design options thatinvolve trading off job creation for otherbenefits.

The Murraysburg and Khayelitsha casesserve to illustrate where community partici-pation in design has to be approached care-fully. In both cases, community membersexpressed their desire for design inputs inthe form of extensions to the project thatwere not within the budget. If this occurs,technical consultants or local authorities canclarify the limits, and if a design change isfinancially infeasible, it will not be done.However, a discussion around the issue canenable communities to decide whether theyare willing to trade off one kind of benefitfor another, and leaves people with a betterunderstanding of why a choice has beenmade. This educational process is beneficialto the community, and to the project by re-ducing the chance of disgruntlement andprotest later (as occurred in Khayelitsha andThembalethu).

One option raised by engineers was theuse of standardized designs, to save costs andensure that certain criteria are met with re-gards to quality, labor intensity, and cost. Ifthis route is taken, communities can still haveinput through including community repre-sentatives in forums where design optionsare discussed. Context-specific changeswould need to fit local needs.

Communication and Community EducationAnother way in which beneficiaries can beseen to participate is through periodic com-munication from and with providers (Paul’s[1987] first stage of participation). In thecase studies (and most of the other projectsin the seven programs), once the project isunderway, mass meetings are no longer heldto receive community input, except underunusual circumstances that significantly anddirectly affect beneficiaries—for example, anunforseen need to remove shacks. Instead,community participation takes place from

48 CHAPTER 5

that point on through the PSC—the com-munity organization becomes the provider,alone or in partnership with local govern-ment and/or the private sector. In the re-search workshops with former projectworkers, when they refer to involving “thecommunity,” they generally were referringto CBOs such as street committees, SANCO,and PSCs.

Despite Paul’s (1987) designation ofinformation sharing as a first stage of par-ticipation, it is also arguably not participa-tion if there are no mechanisms for partici-pants to influence actions. Nevertheless, thecase study projects revealed ways in whichperiodic reports given to beneficiaries wereimportant—on progress of the project, rel-evant announcements such as upcomingshack removals or safety hazards, and finan-cial reports. In Thembalethu, workers saidthey would have liked an early meeting withthe consultants to discuss wages, amount offunds to contractors, type of work tasks, andcontracts, and that there needs to be linesof communication so there is no breakdownin understanding between community andcontractor (Thembalethu Workshop). InKhayelitsha and Thembalethu, workers andPSC members expressed dissatisfactionover the absence of financial reporting fromthe managing engineers. Aware that a largesum of money has been given by the gov-ernment for their benefit, they want to knowhow it is being used. Sudden infusions ofproject money can breed distrust, and peo-ple may assume that someone is benefitingunduly if they do not know how funds arebeing spent. A simple financial report givenat intervals can help communities feel moreinvolved and confident. If periodic reportsare not given, the provider can at a mini-mum present a final report. As the head ofSANCO in Thembalethu explained it,

you must know what the professionalconsultant engineer is getting, whatthe quantity surveyor is getting, what thearchitect is getting you see. You mustknow that at least most of the moneygoes back to the community. If there’sany money perhaps left, it can also startto do another project you see. That’s athing that should happen because [other-wise] we’ll never know. Because it canperhaps be that . . . you’ve got a prob-lem with the project. [Or] perhaps theproject did go smooth and you neverhad any problems (interview).

In Thembalethu, community membersand the engineers alike commented on theproblem of communication breakdowns. Theengineers/providers did not explain to thecommunity why the budget increased by R2million. There is no evidence that those fundswere not spent legitimately on the project,but beneficiaries were bound to be suspi-cious when no explanation was offered. Thefact that no final report was given left long-standing bad feelings in the community to-ward the consultants and the provincialgovernment.35 This makes it harder foroutsiders to subsequently work with thiscommunity.

Communication with beneficiaries is notjust important for its intrinsic democraticvalue, but it also has financial implications.Greater participation can help to foster asense of ownership, valuable not only forthe social benefits of instilled pride, but alsohelping to prevent theft, vandalism, workstoppages, and other conflicts. Of the eightcase study projects, a sense of communityownership was exhibited to some extent atKhayelitsha, Thembalethu, Clanwilliam,and Lutzville. While communication helpsto promote ownership, ownership appears

COMMUNITY PARTICIPATION IN THE WESTERN CAPE PROJECTS 49

35The consultants and a DTPW official did come awhile after the project finished to try to report on the financesin response to a request, but were unable to report because of changing memberships of organizations and dis-putes in the community about which was the appropriate group to report to.

to be more closely related to higher levelsof participation. Lutzville, for example, hadthe highest levels of participation and peo-ple said that the community “put their wholeheart and soul in it” because of a sense thatit is “our hall” (interview, PSC head). Asidefrom instilling pride, a sense of local owner-ship has instrumental benefits such as in-creasing cooperation, improving mainte-nance, and reducing vandalism.

In the Langa environmental project in-volving township clean-up and greening,workers were supposed to be the link withthe wider community through their StreetCommittees, but such participation did notreally take place as envisioned. The NGOprovider did not have sufficient time to runthe project and carry out the kind of linkswith the community, primarily around envi-ronmental education, it had hoped for. Oneof the project coordinators felt this was alost opportunity, because

six months later the dirt was right backthere again. And that just showed wedidn’t address the core, we didn’t talkto the people who were leaving thegarbage there in the first place. Andthat takes time and time take money(interview, Tsoga representative).

In Khayelitsha, a good community con-sultation process occurred through the RDPforums in the early stage of the project, butlater distrust became extremely high duemainly to conflict around the task systemand other labor-management issues, and apoor working relationship between the PSCand the contractor. Now, according to thePSC, “in a recent community meeting thepeople said that [this contractor] was nevergoing to be taken for any jobs in theircommunity as it has caused problems andhas left the community in conflict” (inter-view, PSC).

Murraysburg had no community partic-ipation and no community steering com-mittee. One councilor commented that suchprojects should do more to empower com-munities by bringing them through the stepsof a development project, explaining thetypes of decisions involved and considera-tions involved in making them. At the re-search workshops with former project work-ers, they were taken through an exercise ofthis type, asked to identify who played whichroles on the project and debate who they feltshould play these roles in the future. Thelevel of interest, energy, and heated argu-ments generated through these workshopexercises revealed their value as a learningprocess. The benefits may extend beyondthis, making projects run more smoothly ifworkers and community members betterunderstand the rationale for decisions aroundwages, why some people should get jobsbefore others, or why certain role-playersand not others hold certain types of decision-making power. The conceptual issues theyintroduce and the opportunity for debatingthem could be introduced through pre-projectworkshops (rather than as an evaluative tool,as we used them).

The Role of the ProjectSteering CommitteesWhile most of the issues discussed abovepertain to forms of participation for com-munities at large or for smaller representa-tive groups such as PSCs or CBOs, thesesmaller groups are the focal point of mostparticipation, particularly those forms thatinvolve capacity building and the assump-tion of most provider roles. This arrangementassumes that this small group is indeed rep-resentative of the beneficiaries—in the vastmajority of cases this group was elected—and on this basis it is brought into the role of(partial) provider.36 It is important to stress,

50 CHAPTER 5

36The fact that the PSC has been elected does not necessarily guarantee that it will always act in the interest ofthe community, however. Problems that occur where it is not are discussed in Chapter 6.

in referring back to the model in Chapter 4,that the PSC rarely takes on the full role ofprovider, but rather is in this role partially,along with other role-players such as localgovernment, consultants, and contractors.

In the 101 Western Cape projects studied,the PSCs’ roles varied widely, from Paul’s(1987) lowest to highest levels of participa-tion. Since in these South African programsparticipation was an end as well as a means,the nature of participation in these projectsis important. The analysis in Chapter 4considered whether communities were jointor sole decisionmakers, advisors but notdecisionmakers, or had no participation. Thecase studies further disaggregate these roles,looking more closely at what community-based organizations actually did or did not doin the projects, and the effects of these roles.

Table 5.4 shows the responsibilities heldby PSCs in the eight case studies, indicated

by grey shading. The responsibilities of theother role players are also indicated.

The pattern that emerges from thesestudies is that, with the exception of Lutz-ville, community-based PSCs were not in-volved with technical managerial tasks. In-stead, their roles have been primarily relatedto labor and community liaison responsibil-ities, including worker selection, communi-cation, and conflict resolution. This is wheretheir skills are strongest and they are mostconfident. It is also the area that they tend toperceive as having the most direct impacton the community. As one of the DTPWPWP directors observed:

When it comes to the physical con-struction work, that is when peoplewant to participate, for the basic needthat—a job is needed and they wouldlike to ensure that people that they

COMMUNITY PARTICIPATION IN THE WESTERN CAPE PROJECTS 51

Table 5.4 Responsibilities of role-players in case-study projects

Responsibility Lutzville Langa Khayelitsha Thembalethu Clanwilliam Murraysburg Kylemore Stellenbosch

Implementing agent NGO LG With LG CNC CNCconsultants

Selecting project NGO LG Consultants LG LG CNC CNC

Selecting contractor n.a. LG Consultants LG LG n.a. n.a.

Selecting workers Consultant/ Project CNCcontractor manager

Community liaison/ LGconflict resolution

Represents workers/conflict resolution

Overseeing contractor n.a. LG, Consultants Consultant n.a. n.a.consultants

Managing finances NGO LG, Consultants LG LG Project Project consultants manager manager

Maintenance LG LG LG LG LG n.a n.a

Notes: LG indicates local government; NGO, nongovernmental organization; CNC, Cape Nature Conservation (the implementing agent forthe Department of Water Affairs and Forestry); n.a. indicates that, due to program design, the question was not applicable.

Responsibility held by another agent, as indicated.Responsibility held by the Project Steering Committee.

know within the community, let’s say a welder for burglar bars to a door, theywant to make sure that it’s done by thelocal guy and not by somebody outside.And in doing that they ignore the pro-cess of competition amongst variousservice providers . . . it takes the bur-den of another mouth or few mouths to feed off the general community, be-cause that’s what’s happening in com-munities . . . they help each other . . .So their concern is what’s going to hap-pen to the money, where’s the moneygoing to? (interview).

Our research workshop findings supportthis view. When asked who they thoughtshould select the workers, former projectworkers in five of the eight projects desig-nated only community members (PSC orthe “community”) for this role, and in theother three projects this task was to be sharedby the community and contractor or fore-man. This is also how professionals in thecase studies viewed appropriate roles forthe PSC. The committee is seen mainly inthe role of community liaison: it is impor-tant for enabling access to work in commu-nities; for communicating with the com-munity about relocation, safety, and otherissues with direct impact on inhabitants ofthe area; and for mediating conflict withinthe community; between the community andcontractor; and between the project workersand the contractor. These are roles that areobviously of benefit to the contractors, whileinvolvement in technical management issuesis less desirable to them.

As seen in Table 5.4, in five of the casestudy projects community committees (andsometimes the community as a whole) wereinvolved in the worker selection process.This lets elected community members onthe committee deal with the contentiousequity issues that arise in choosing whogets jobs and who does not, in communitieswhere there are far fewer jobs than peoplewho need them, where whole families areunemployed and many people have not

worked for a very long time. There is thepotential for patronage, though this did notarise as an issue in the case studies. It isimportant to note, however, that some rep-resentatives from the Congress of SouthAfrican Trade Unions (COSATU) object tothis role for community members, becausethese union leaders believe this arrangementdrives divisions into communities by makingcommunity members into employers (objec-tions were even greater to participation indecisions regarding firing; see Adato andHaddad 2002). This is one of the dilemmasposed by the dual identity of communitymembers as providers and beneficiaries.

Adato and Haddad (2002) discuss howPSC involvement in selecting workers per-mits considerations that do not usually enterinto targeting processes for antipovertyprograms—for example, the criterion ofbeing “active community members” (whohave dedicated time to meetings, activities,and organizations) was used in the Langa andKhayelitsha projects—and also introduceslocal perceptions of poverty and the mostneedy. These criteria do not necessarily re-sult in the best targeting of the poor, andmay have other disadvantages in terms ofelite capture or underemployment of womenif women have less opportunity to participatein community activities due to time con-strains or social biases. But using local cri-teria can also improve upon a self-targetingprocess by choosing those the communityfeels are most needy or deserving fromamong those who will accept the low wage(that is, in theory the poorest of the poor),where demand for the jobs is higher thanthose available—almost always the case inSouth Africa (Adato and Haddad 2002).

The quantitative data in Chapter 4 foundthat de facto participation increased theamount of employment days for women;however, infrastructure development andcommunity building projects typically em-ployed fewer women than other public worksprojects. The case study research helped tointerpret these findings. First, we found anumber of biases against employing women,

52 CHAPTER 5

one of the main ones being the perceptionthat construction was not women’s work.Several of our case studies showed quite abit of bias against hiring women. The Lutz-ville project, with the highest levels of com-munity participation, employed only threewomen on the community hall project—ascleaners. However, biases were held in dif-ferent cases by community members, localgovernment, and contractors, so communityparticipation does not necessarily affect this,and the quantitative results suggest that com-munity run projects perform better in thisarea. Our research found that the main fac-tor influencing women’s employment wasthe emphasis given on targeting women atthe program level, though this was still notsufficient—these directives were followedmore closely in some programs than others(Adato et al. 1999). With respect to commit-tee participation, several of the case studieshad women members but they were consid-erably underrepresented compared to men.

Lutzville was the one case study wherethe project was entirely community-run:there were weekly community-wide meet-ings, and a community-based PSC managedthe project with no local government in-

volvement. The DTPW provided some as-sistance with facilitation, and the PSC man-aged the contractor. The main feature ofLutzville West that stands out among theother case study projects was its high degreeof social capital37—it was a small town withtightly knit social networks and little appar-ent political conflict (inter-party or intra-party), and strongly influenced by its churchcommittee and its community, church, andproject leader (all the same person).38 Thesefactors seem to explain why there was noreported conflict on the project (see Chap-ter 6 for a discussion of the significance ofconflict), and that community participationwas effective. There were no labor disputes,time or cost overruns, or theft, and negli-gible vandalism.

Although beneficiaries and most privatesector and government providers gave thehighest value to the worker-selection/com-munity liaison functions of the community-based provider, there are benefits to bederived from community participation inother areas. Evidence from the case studiesreveals that given the opportunity and skillsto participate in other areas, communitymembers are interested in doing so.

COMMUNITY PARTICIPATION IN THE WESTERN CAPE PROJECTS 53

37Isham (2002) finds that social cohesion positively affects participation and project outcomes.

38Certainly the leadership of this individual was important to the smooth running of the project. However, we donot attribute participation outcomes primarily to “leadership” as the case study projects had widely variable leader-ship that does not appear to be very correlated with participation outcomes.

C H A P T E R 6

Social, Political, and InstitutionalChallenges to Participation

In this chapter we use the case study material to describe and analyze the extent and natureof the social, political, and institutional challenges posed to participatory development, aswell as consequences of increasing or limiting participation. In particular we focus on the

project steering committee (PSC), negotiating with other agents that either support or chal-lenge their role as (partial) provider.

Table 6.1 describes eight reasons that were found to explain why PSCs were not able toplay more substantive roles in the case study projects.

The Views of Professionals on Efficiencyand Technical SpecializationThe first three factors in Table 6.1 share a common underlying theme, related to the tensionbetween participation and professional views on technical specialization.39 In four of the eightprojects, participation was limited by the consultant’s, contractor’s, or local government’sbelief that committees did not have the technical ability to assume greater responsibility.Government and private sector project managers are accustomed to planning and managingprojects on their own, and are concerned primarily with bringing projects in at high qualityand within budget. These role-players do not have incentives for sharing the provider role withcommunity members or their representatives. The perception of most is that community par-ticipation will increase time and costs, and reduce quality. In terms of financial and profes-sional rewards, their perceived incentives mitigate against participation. The exception tothis is when failure to share this role would disrupt the construction process, but the effect ofthis absence is not apparent until after the project is underway. In the competitive process oftendering, there is no incentive for allowing generous costs for training community-basedcommittees and allowing them to make mistakes. Contracts have fixed prices and the con-tractor bears the cost overruns if they occur. There were no new incentives developed in theseprograms to encourage a new way of working.

The main incentive for local government and private sector providers to share their rolewith communities is that central government financiers required this before awarding publicworks funds. This was specified in all the programs studied. With the IDT CBPWP, the cen-tral government/financiers required CBOs to be the sole formal implement agent/provider, and

39The impact of these views and professional discourses more broadly on institutional transformation in the con-text of these public works programs is analyzed in more detail in Adato (2000).

54

this was a powerful factor explaining whythere was de facto participation in theseprojects. With respect to the other six pro-grams, where there could be joint providers,financiers did not give very specific criteriawith respect to the nature of communityparticipation and the type of training givento support it, nor were directives for partici-pation monitored. Specifying these criteriaand monitoring their implementation are nec-essary because once professional providersobtained funds, the incentives to promoteparticipation, particularly higher levels ofparticipation, were removed. Communitiesmay, of course, also decide to limit theirown participation to specific areas due toopportunity costs or their own perceptionsof where their abilities to contribute lie.

A central issue, reflecting long-standingdebates on democracy, is whether com-munity members have the qualifications tocarry out the specialized, requisite respon-sibilities. In these projects, the questionsrevolved around whether PSC membershad the skills to carry out planning, book-keeping, purchasing, and managing trans-port, personnel, and other tasks. Actual levelsof education and skills varied across theprojects, with higher capacity at Clan-william, Murraysburg (in the CBO thatwas excluded), Lutzville, Kylemore, andStellenbosch, and less in Khayelitsha andThembalethu.40 However, no formal or on-the-job training was provided to build theseskills. At Khayelitsha, computer courseswere offered on the weekends, but not well

SOCIAL, POLITICAL, AND INSTITUTIONAL CHALLENGES TO PARTICIPATION 55

40There were too many different people involved in the Street Committees at Langa to make a judgment.

Table 6.1 Factors limiting the roles of the project steering committees (PSCs), case-study projects

Factors limitingPSC roles Lutzville Langa Khayelitsha Thembalethu Clanwilliam Murraysburg Kylemore Stellenbosch

Consultant/govt. viewson efficiency and tech-nical specialization

Community empower-ment not prioritized

Lack of technical capacity within com-mittee; no training

Local politics and conflict

Local govt. seen as legitimate representative

PSC commitment reduced over time

Lack of clarity as to what function should be

Note: Shading indicates the presence of factors limiting the roles of the PSCs.

attended.41 Special weekend training work-shops were designed for the PSC, but wereonly given when the project was close tocompletion.

The objectives of central governmentfinanciers to promote community partici-pation and empowerment—that communitymembers should develop new skills and gainmanagement experience—was not shared bymany professional providers on the proj-ects. Where participation was appreciated, itwas with a different vision. For example,when asked if the PSC assisted the projectin any way, the contractor at Khayelitsharesponded:

Right at the beginning of the contractthere were a lot of houses that werebuilt in the road reserve and we gothold of the Project Steering Committeeand said listen we can’t build a roadunless the people move out and relo-cate, and there they did a fair amountof work. Because their job wouldn’t goon unless they did that. So they didtheir job there but up to a point, thenthe job would just naturally run inward[toward the contractor]. And yes wecalled them for problems, but it wasmore just to do the administrative partof it because . . . they didn’t really addvalue to the contract. I don’t think. You need them for the community sake . . . the thing runs it’s like a bigvicious wheel, it just runs by itself.Then the Project Steering Committee,they can’t really change the directionthat this whole thing is running(interview).

The contractor was also critical of thePSC, questioning their commitment to theproject, their failure to really learn and com-

municate the principles of labor-intensiveconstruction, their dwindling attendance atmeetings, and their failure to show up forcomputer training classes. The vast gulf be-tween perceptions of the contractor and thePSC at Khayelitsha is illustrated in the con-trast between the contractor’s commentsabove, and those of PSC members, who ex-pressed that the contractor “never wanted togive people a sense of ownership and re-sponsibility towards the project and theprocess” and later that the “contractor isclearly taking some part of our responsibili-ties away.” This was in fact one explana-tion why their enthusiasm for the projectwaned and why some members left thecommittee (interview, Khayelitsha PSC).Their perspective on appropriate roles forthe committee also contrasts with that of thecontractor:

We should have been part of monitor-ing the overall aspects of the project—things like ordering, buying the mate-rials and tools . . . even the CLOs[Community Liaison Officers] werenever involved in such operations. Wedo not even know as the PSC howmuch was allocated to this project—wedo not know how much was actuallyexpected in this project, also if there isany over-spending . . . all these are thedecisions that were taken in the forumswhere the officials meet without us . . .I don’t think even our local councilorswere part of such meetings (interview,Khayelitsha PSC).

This influenced the committee’s viewnot only of the contractor, but also thecommittee’s opinion of provincial govern-ment, which was the department responsiblefor the project (though central government

56 CHAPTER 6

41We do not know the reasons, because the PSC did not mention these classes, and after the contractor did, wewere not able to re-contact the PSC to inquire. It would be worthwhile to understand why, given the committee’sexpressed desire for more managerial inclusion: whether it was insufficient interest, quality of the classes, timeor logistical constraints, or other factors.

was the primary financier; as explained inChapter 2, the funds passed through theprovince). One member remarked that “thepoliticians/leadership usually say that theygovern according to the needs and views ofthe people, but it is them who turn aroundand not implement their own policies aboutcommunities taking part in deciding what itfeels is good for them” (interview, Khaye-litsha PSC). This is a case where financiersshare interests with beneficiaries but profes-sional providers did not—at least with re-spect to the empowerment goal. The interestin building a good quality road was sharedamong all parties.

Similar divides between the perspectivesof community committee members and pro-fessionals were revealed at Thembalethu. Inthis case committee members, including twowho were also employed as office workers,were not trained in using the computers orbookkeeping, and were not involved in fi-nancial issues nor kept informed on substan-tive issues. One of the partner consultantsexplained the divide that needs to bridged inthe following terms:

You see we consultants are used to theconventional way of working. We de-sign, there’s a contractor that is respon-sible, they are the communities or thebeneficiaries—they know nothing aboutwhat is happening to the funds . . . Theyare just asked to work and get salariesat the end of the day and leave. Nowthis situation is a new situation thatneeds both parties, communities andthe consultants, to make a joint effort toeducate each other about the new pro-cess because the whole process is newto us. The whole process is new also tothe community because they were notinvolved before. They were not partici-pating, so they want to participate. Atthe same time we consultants, we’re notused to communities getting involvedin the administration of the project. . . .[We’re] not used to accountability andthe community wanted some account-

ability from us, especially with finance(interview, Manong Associates).

Asked whether he felt some of the questionsthe community members asked were validones, he replied:

Yes, absolutely. That’s why I said if Ihave to be given such a project again inthe future I’d run it differently from theway we ran that one . . . In the begin-ning put the proper administrationmechanism to train them, let them runwith the whole thing. Make sure theykeep the books, but at the same timethe financial part of it we have a propercontrol of it, you see. They know whatis happening with each and every cent,but at the same time their access to fi-nance doesn’t become that busy becauseit’s difficult if you keep books, finance,you can draw money and all thesethings. There is a way of getting aroundthese things you know. But I will makeit a point that they know it’s transparent,they know themselves what is happen-ing to the money, but they don’t keepthe money (interview).

Community representatives were eventuallyremoved from the project following a con-flict where the contractor was held hostageover a dispute involving bonuses. This wasa case with a very high degree of politicalconflict in the community (more detail isgiven below), and a wide gulf between theperspectives of the consultant engineers andcommunity members on the PSC. The Lutz-ville case discussed in Chapter 5, on the otherhand, shows that community-based organi-zations can play a productive role in man-aging projects without strife and with goodperformance outcomes.

There are two ways of responding to theproblem that community committee mem-bers may not have technical skills to con-tribute to project management. The first isthat they thus should not be involved inmanagement in the interest of efficiency (as

SOCIAL, POLITICAL, AND INSTITUTIONAL CHALLENGES TO PARTICIPATION 57

suggested by some stakeholders in the casestudies). An alternative approach is to seeparticipatory development projects as oppor-tunities to develop local capacities throughtraining and oversight. In the case of SouthAfrican public works projects, policy-makers (with an implicit democratically ex-pressed mandate from their constituents)made community participation an objective,an end as well as a means. If it remains sopolicymakers must live with the tensionsinherent in the process and continue to ex-plore creative methods to achieve multipleobjectives. One such method, for example,is that suggested by the Thembalethu con-sultant in the quote above. The extent towhich there is a tension between empower-ment of communities and technical effi-ciency varies on a case by case basis. Butpolicymakers need to make the prioritiza-tion of objectives clear to the professionaland community-based providers, so eachparty knows what to expect and what is ex-pected of them.

It is not only the private sector that hashad difficulty adjusting to the idea of com-munity participation. Local government wasoften the main alternative to the commu-nity-based provider, or the main partner in ashared arrangement. In Murraysburg thelocal government did not involve a CBO atall until the proxy financier (DTPW) threat-ened to withhold funds, and even then itcontinued to resist bringing in the CBO. Inthis case, the CBO had some leadershipwith a high standard of education and couldhave taken on management roles (they re-ported having managed a R700,000 projectfor the Department of Health). The localCBO explained the situation as follows:

In as far as the CBPWP is concerned itwas made clear that the municipalityhad to involve a community organiza-tion and [we are] the only local com-munity organization. [The municipality]informed [DTPW] that there is a com-munity organization and so the depart-ment approached us. They came up

with a contract and we had to sign itand say whether we approved or not.We did not sign the contract becausewe disagreed on a few things. When weread through the contract we found thatwe . . . had to be involved from thestart but we were only involved whenthe project was already in the middle.We found that they only wanted oursignatures and they were doing thewhole thing. . . . [For example] R74,000was allocated for training. So we saidwe will not sign the contract unlessthey could prove that the R74,000was spent for the right purpose. . . .We did not sign the contract becausewe wanted answers as to how themonies were spent and they could notcome up with the answers . . . they re-ally did not want to come out straightand say we still have the money orthis money has been given. So we saidhow can we sign because we will berubber-stamping something of whichwe have no knowledge (interview,Murraysburg CBO).

The organization attributed this to

the old idea that still persists that [thegovernment] decides for the people andthey just have to say thank you, andthey don’t ask questions. . . . The mu-nicipality is also seeing [the CBO] as athreat because if they are going to give[the CBO] more of a say they will nothave the ability to do their own thingwithout the people having a say. . . .They still have that old attitude thatwe decide for you and at the end ofthe day you just accept (interview,Murraysburg CBO).

Local government officials on the otherhand saw no reason to involve the CBO:“Even today I can not see that with [theCBO] involved, that anything could haverun better than the way it was” (interview,local official).

58 CHAPTER 6

The case studies also revealed how par-ticipation can also work well, and the bene-fits that come from this. In Lutzville, theLutzville West Development Trust (a CBO)was the provider/PSC, with its leader theproject manager. The PSC contracted anarchitect (again a partial provider) but man-aged the construction, labor force, and fi-nances on its own, without any involvementof the municipality and with occasional ad-vice from consultants. The community hallis of a high quality standard, and it came inwithin the budget (a small overrun was madeup through interest). It was the first time thecommunity had managed a project, and com-mittee members said they are “still shakenup about this. [We] can’t believe that we didthis.” Their confidence had increased, andthey were “waiting anxiously for the next[project]” (interview, Lutzville PSC). Therewere expressions of similar feelings of ac-complishment and new capacity (thoughmixed with more frustrations) from PSCmembers at Khayelitsha, Thembalethu, andClanwilliam. These are described in thelarger report upon which this research re-port is based (Adato et al. 1999).

Participation, Politics,and ConflictCommunity participation is also complicatedby the introduction of local politics into proj-ects, which often brings conflict and some-times expensive delays. The politicizationof projects means that a process that is in-tended to be inclusive and serve the interestsof beneficiaries, often excludes some part ofa community and serves the interests of someintended beneficiaries more than others. Insome cases national level politics enteredthe projects, particularly when a project wasknown as an RDP project and communitymembers who were not working on the proj-

ect felt a sense of ownership. In an incidentin the Stellenbosch case study, workers hadjoined a national work stay-away, but thenrequested and were allowed to work on aweekend day to make up their wages. Othercommunity members felt this ran contraryto the purpose of the strike and arrived at theproject, insisting that the work be stopped.

More often, however, political conflict inprojects reflected local and provincial-levelpolitics. Political tensions are the reality ofworking in communities where resourcesare very scarce, where there are conflicts ofinterest, and where a history of conflictbetween communities and the state, and be-tween workers and managers, re-emerges innew forms in development projects.42 Onelong-time Western Cape activist, who be-came a specialist in community facilitationand worked in public works projects, de-scribed the Hout Bay CBPWP communityhall project, where

they wanted to politicize each and everything. After all, that is a community-based project. The process designed tolet everybody get involved in it. At theend the workers did shady work, thestandard was low, they didn’t listen tothe foreman. Everything just went hay-wire . . . the whole idea was that thecommunity was going to empowerthemselves. . . . And we had endlessproblems with them. We had moneyproblems, that they wanted more money,they wanted to change the design. . . .I can give you a million examples . . .There is so much friction in some ofthese communities and everybody isafraid to talk about it. Everybody stillbelieves that the struggle was a greatunifier, but they don’t understand we’renot in the struggle anymore, we’re inreality now and people have difference

SOCIAL, POLITICAL, AND INSTITUTIONAL CHALLENGES TO PARTICIPATION 59

42See Adato (1996) for a discussion of the challenges of South Africans moving from “strategies of opposi-tion” in the apartheid and transitional era to “strategies of development” in a post-apartheid political and eco-nomic context.

of opinions. . . . It’s over resources, it’s over power43 (interview, C. B.Associates).

According to another consultant, therewere many political groups involved andsome of those that felt excluded ended upvandalizing the community hall. This isanother reason to facilitate an inclusiveprocess, although such facilitation was at-tempted in this case and does not alwayssolve the problems.

Local politics affected projects in dif-ferent ways in all eight cases studied, but hada particularly strong impact in four. Localpolitics involved conflict between politicalparties, between CBOs and local govern-ment, and among CBOs. The nature of theproblem affecting the PSCs were mainlytwo: (1) Projects were used by political ri-vals to gain political advantage, where onegroup attempted to exclude another to avoidsharing the credit for infrastructure delivery.Once a group no longer perceived a politicalgain, it lost its incentive to support the proj-ect; (2) Local politics were carried inside theprojects, causing disruptions which under-mined the effectiveness of the PSC and theauthority of its members. The case studiesprovide insight into these processes and theirimplications for community-based projects.

In Thembalethu, where communitymembers were on a joint PSC with consul-tants and the contractor, political conflictreached a point where the consultants saidthey eventually found the need to exclude thecommittee in order to get the job done. Oneconsultant said that the community commit-tee members had their own agendas, weretoo influenced by local political struggles,changed their minds too often, and did notstick to agreements. In one incident, theycalled a strike that workers did not want,and in another participated in holding the

contractor hostage over the issue of bonuses.Thembalethu was a particularly difficultcommunity to work in, with a highly frac-tious political environment, and many po-litical groupings vying for influence (weidentified five, there may have been more).The community committee members wereappointed from a trust with links to a sepa-rate engineering firm, and not elected by thecommunity. In this sense, it was a particu-larly complicated and atypical project, butnevertheless illustrates the type of prob-lems that can be encountered in community-based projects.

Clanwilliam and Murraysburg weretowns with a common political configura-tion in rural Western Cape, with NationalParty (NP) and African National Congress(ANC) members sitting together on a coun-cil controlled by the NP, and a community-based organization officially neutral but moresympathetic to the ANC. In Clanwilliam,when the RDP Forum took on a manage-ment role, local government pulled back,though in the end they were impressed:“everyone was sort of standing on the side-lines to see whether the thing would workout. Everyone wanted to see where a mis-take was going to be made, but in the endthey came to us and congratulated us” (in-terview, local official, PSC). The municipal-ity resisted involving the RDP Forum in theadministration of the project, until a greaterrole for the forum was negotiated with theassistance of provincial facilitators. Commu-nity members and local government officials,though they were not in a happy alliance,did both participate substantively in imple-mentation and performed their tasks well.

In Murraysburg, in addition to the munic-ipality seeing delivery as its own job, therewas also a strained political relationship be-tween itself and the local CBO, which re-sulted in the latter dropping out of the proj-

60 CHAPTER 6

43There were of course differences of opinion in the anti-apartheid movement as well. But this speaker’s point isthat that the conflicts with authority and between political interests that became so familiar in the anti-apartheidera were being brought into projects, where beneficiaries of the projects would stand to lose.

ect altogether. The CBO’s perception wasthat “some people in the municipality arepaying too much attention to politics atthe expense of the community” (interview,Murraysburg CBO). Referring to these twoprojects but also others in the province, oneof the provincial facilitators observed that

We need to be able to de-politicize theprojects and that is always a big prob-lem in our country. How do you de-politicize anything? [A project is] agreat political football in the ranks of a political party and also you know try-ing to score points on the oppositionand that has also made it bloody diffi-cult in some cases where NationalParty and ANC people have also fought(interview, C. B. Associates).

Both issues identified here—views ontechnical expertise and local politics—indicate the important role played by nationalgovernment financiers and their provincialcounterparts in setting and monitoring pro-visions for community participation. In alleight case study projects involvement of aCBO was a requirement. In Clanwilliam,where local government did not want to in-volve RDP forum members in monitoringof the funds, the provincial DTPW made adifference:

Here it was again the issue of whether[community] people could be trustedand whether [we] had the experience.They have a very low opinion of us.For this to change we just had to putour foot down and with the help of[facilitation consultants hired by DTPW]we gave them the option that eitheryou give us a say in the administrationof the funds or we will find a way toadminister the funds ourselves. Themunicipality wanted to administer the

funds and only they will have full sayin this, but this was then sorted out(interview, Clanwilliam PSC).

DTPW was less successful in Murrays-burg. There, the CBO made the point thatit was the provincial department’s role tomake sure the requirements for communityparticipation were met: “unfortunately for usthe people who are supplying the moneyare also negligent because they are not fol-lowing up” (interview, Murraysburg CBO).DTPW tried to insist on involvement ofthe CBO by sending out its facilitators andwithholding funds, but eventually turned thefunds over after the municipality alreadyspent the money and built the asset (usingHousing Department funds that were ear-marked for a different project and requiredreimbursement). Returning to the model inChapter 4, these cases of politics and con-flict illustrate where a provider or particularprovider arrangement is not working in thebest interest of beneficiaries. Instead, the fin-ancier more closely served these interestsand intervened to do so.

The case studies also suggest that acommunity with a high degree of politicalfragmentation will introduce a higher de-gree of conflict to a project, whereas a po-litically homogenous community will seea smoother participatory process. In somecases this political conflict took on a racialas well as political dimension; in others theconflict was inter-party or intra-party. InClanwilliam and Murraysburg, politicaltensions between the ANC and NP led tocompetition and conflict. Lutzville West onthe other hand was a politically and raciallyhomogenous, tight knit community with alarge stock of social capital at the outset,and little if any conflict. This project hadone of the highest levels of participationamong the 101 projects and was efficientlymanaged and reasonably labor intensive.44

SOCIAL, POLITICAL, AND INSTITUTIONAL CHALLENGES TO PARTICIPATION 61

44As noted in Chapter 5, in comparison to other community hall projects, Lutzville performed better than theaverage in efficiency in transferring money to the poor, and just below average with respect to labor intensity.

These conditions cannot easily be influencedby an external intervention, though media-tion or authority given to a legitimate, down-wardly accountable local government mightbe able to soften or reduce the impact of pre-existing conflicts or distrust.

However, fractionalization and conflictdo not necessarily prevent participation—itmay even increase it where rivals seek in-fluence (this may have positive or negativeimpacts on the project). In Clanwilliam therewas eventually a high degree of participa-tion; in Murraysburg there was none. Therewas also a high level of conflict (intra-partyrather than inter-party) in the Thembalethuand Khayelitsha projects, but a high andmedium level (respectively) of participation.However, the outcomes of conflict might beotherwise problematic, as they were in thecase studies, as well as the Hout Bay projectdescribed above.

Participation and ProjectCompletion TimeThere is a generalized belief that participa-tion increases the time of project implemen-tation. In five of our eight case study projects,at least some informants commented thatcommunity participation adds time to proj-ects. This is usually true up front—decisionsmade by community members or commit-tees about project design, hiring, and otherissues are likely to take longer than thosemade by government officials or consult-ants, because of the time required for com-munity members to understand the issues,and the pressure for consensus. It is alsolikely that politics and conflict will causedelays. In the Khayelitsha and Thembalethuprojects, cost overruns were high, in partattributed to community conflict. Differentconclusions were drawn from these experi-ences. The lesson one official in the provin-cial government came away with was thatcommunity participation was too expensiveand there should be less of it. Similarly, thecontractor at Thembalethu felt that com-

munity participation takes too long, and thatfrom his standpoint conventional contractsare easier to manage.

Yet while increased time would nor-mally be associated with higher costs, thequantitative analysis in Chapter 4 foundhigher de facto participation to be more cost-effective. The case study research providessome insight into why this might be thecase. One part of this explanation revealedin the case studies is that participation earlyon avoids costly conflict and delays later.In the Clanwilliam project, a key municipalofficial felt that the strike there could havebeen averted if there had been more com-munity involvement in the project. AnotherDTPW program director warned that tryingto avoid conflict or save time by sideliningcommunity members is unlikely to succeed,since

If you don’t have community participa-tion you run a large risk of having yourproject blocked. To get a project doneon time you must get acceptance of theproject by community leadership, pro-vide assurance that labor will comefrom community, but also agreementthat some can come from outside. Youmay spend a bit more up front in get-ting this participation, but if you don’tyou may end up with project stoppedor having to bring people from outsidewhich costs more (interview).

In the case studies, we also found thatthe pressure to get projects off the groundquickly is at odds with the time needed toadequately discuss issues with communitymembers and project workers before con-struction starts. This may increase thechances of a slipshod participation processthat goes awry. In the CBPWP, delays inobtaining funds and other obstacles at theearly stages meant that once projects beganthere was a rush to finish them within theremaining time allotted. Also, where consul-tants are the providers, there are usually high

62 CHAPTER 6

capital costs and overheads, and set tenderprices that cannot be adjusted upwards, lead-ing to pressure to not extend the time frame.From an engineer’s point of view, a logisticsprocess has a “critical path” and there are in-centives not to veer from this (interview, of-ficial, DTPW)—another reason to deal withproblems early, before that path is embarkedupon. From the beneficiaries point of view,however, there is no reason to rush—in fact,a longer project means more employment—and consensus is considered important.These competing incentives must be me-diated, particularly during the constructionphase (if workers slow down to extend theproject, which does occur). In politics de-bates can continue indefinitely, and theredoes need to be a process for facilitatingtimely decisions, within a work plan.

There are also political decisions inher-ent in what type of participation to pursue.Under Paul’s (1987) four types of partici-pation, time can be expected to increasemoving from type (1) to (4). Determiningthe relative priority of efficiency and em-powerment objectives is a policy process.Politics, conflicts of interest, struggles overresources and drawn-out processes of con-sultation, consensus and even new consen-sus post-conflict are part of the landscapeof community-driven development. Train-ing, facilitation (by local government orothers), and patience are all required. If com-munity empowerment remains an objective,then all role-players must take the realiza-tion of this objective seriously through allo-cating the necessary training, time, and pa-tience to see the process through. Our casestudies and quantitative analysis indicatethat there is not necessarily a trade-off be-tween participation and efficiency.

Local Government and RepresentationOne basis for resistance to participation onthe part of some providers in local govern-ment and the private sector was the belief

that local government was the representa-tive of “the community”—they did not seethe need to involve a CBO. In Clanwilliam,one local councillor asked:

if [the CBO] got involved, what roledid the Council have to play? Becausewe were chosen by the community.There was a clash of interest in my wayof thinking. [The CBO] got involvedon behalf of the community; we got in-volved because the community madetheir crosses behind our names. So whywould the one do a better job than theother one? (interview).

The respective roles of local governmentand CBOs and the relationship betweenthem is a salient issue in debates about de-centralization and participation. Ribot (2001)persuasively argues that local associations(such as CBOs) do not necessarily representthe public nor are they systematically ac-countable, and that elected local governmentis a legitimate representative of local peo-ple. Different interest groups hold varyingdegrees of power, and representative gov-ernance levels the playing field. However,he also recognizes the less than ideal op-erationalization of decentralization to date.Among other problems: powers are usuallyplaced with upwardly accountable agentsor with private groups who are not system-atically accountable at the local level; andthere are many ways that local elite or polit-ical parties are able to capture the electoralprocess. Ribot also posits certain factorsthat influence how accountable local gov-ernment might be; for example, the ways inwhich they are embedded in social relationsin their communities and the level of civicdedication.

All of these points have considerable rel-evance to the South African experiment withcommunity participation. There are manyarguments why local government rather thanCBOs should be the main provider in publicworks projects—democratic accountability;

SOCIAL, POLITICAL, AND INSTITUTIONAL CHALLENGES TO PARTICIPATION 63

access to resources, and capacity with re-spect to the complicated task of infrastruc-ture delivery. Note that local government waspart of the joint provider arrangement in themajority of the 101 projects. However, thereare also several factors that weigh in favorof a strong role for CBOs in the SouthAfrican context. First, most of these CBOswere elected. In the context of powerfuldiscourses of democracy leading up to andfollowing the transition to a democraticstate (Adato 1996), it is less likely that acommunity committee with access to re-sources could simply be appointed. Second,when many of these programs started, a le-gitimate local government was not in place.But even when local elections did take place,the observation that local governments canend up being upwardly rather than down-wardly accountable, controlled by elite orpolitical parties had relevance in WesternCape province. In many of the communitieswhere our research took place, local gov-ernments were embedded in social and po-litical relations—in particular racial, class,and political configurations left over fromapartheid—in a way that significantly re-duced their downward accountability. Fi-nally, in several of our case study com-munities, local authorities represented dualcommunities of different race and class, andwere more directly accountable to a non-poor constituency. There is also the consid-eration of the RDP, that aimed to build theskills and capacities of local people throughtheir direct involvement (not through rep-resentation) in development processes.

Nevertheless, in a more accountableframework, the arguments are still strongfor elected local authorities to play the roleof provider in public works projects. Ribot(2001) also makes a strong case for the im-portant role that local associations can playin fostering accountability through moni-toring, informing, and lobbying. Capacityfor political action can be built this way (al-though the acquisition of certain skills ismore questionable). The argument for localgovernment responsibility won out in the

subsequent CBPWP in South Africa, whereauthority for implementation was placedwith local government. As the country’sdemocracy grows older, local governmentshould become more accountable than itwas in the immediate post-apartheid envi-ronment. Still, the history of civil societyparticipation at the local level may still re-sult in a significant role for CBOs, formallyor informally.

Reduction of CBO CommitmentIn five of the seven case study projectswhere community-based committees wereinvolved, the commitment of members tothe job was reduced over the course of theproject. In four of the projects, committeemembers explained that their interest dimin-ished when their power was undermined.However, other stakeholders observed thatcommittee members got busy and over-extended, especially those with skills andcommitment to many community activities.Their skills make them likely to have otherjobs as well as other community engage-ments (a factor at Clanwilliam and Kyle-more). Committee members, particularlywomen, may have problems attendingmeetings in the evening (although this wasnot raised by any committee members). Ifcommittee participation wanes or membersdrop out, implementing agents and consul-tants should try to learn why; what are theconstraints they face in terms of time orforegone opportunities? What dissatisfactionmight they have with the process? Compen-sation for serving on committees can also beconsidered. This involves risks, because itincreases the chance that people will try toget on the committee for money rather thancommitment to the project, and the likeli-hood of conflict as people vie for member-ship in the context of high unemployment.In Lutzville, the project was carried offsuccessfully and with sustained commitmentwithout any compensation. In Khayelitsha,attendance problems were solved by paying

64 CHAPTER 6

members a stipend to attend meetings—probably a better option than paying a flatsalary because the amount is small and thestakes lower. Also, if a PSC member is es-sentially an employee of the provider (in-cluding possibly itself), its independenceas a worker representative is potentiallycomprised.

Lack of Clarity in theRole of the PSCGiven the joint provider institutional ar-rangements of most of these public worksprograms, and the ambivalence of manyproject stakeholders about the role of CBOsin project management, there was a signifi-cant degree of confusion in many of theprojects as to what the role of the PSC wassupposed to be. For example, in Khayelitshait emerged that the PSC’s main de factorole was as community liaison and worker-management liaison, but this was also therole of the two elected Community LiaisonOfficers (CLOs), as well as a role occasion-ally and most effectively played by the localgovernment.

The two Fynbos Working for Water casestudy projects provide particularly colorfulexamples of this problem. The national levelfinancier, the Department of Water Affairsand Forestry (DWAF) specified that all proj-ects would have Facilitating Committeescomposed of local government, an RDPForum, the Workers Committee, and some-times other CBOs. However, in the two casestudy projects the composition of the com-mittee was different. At Kylemore it wascomposed of assorted community memberswhom the project manager requested to join,skilled people who were active in commu-nity affairs. In Stellenbosch, it consisted oflocal councilors and initially RDP forummembers. In both cases (1) the roles were notclearly defined; (2) there were conflictingperceptions of these roles by the committeemembers themselves and project manage-ment; and (3) their participation dwindled,in part because of (1) and (2) above. There

were also other reasons, for example, inKylemore the members had too many othercommitments.

DWAF envisioned the steering commit-tee as assisting with “recruitment, appoint-ments, promotions; disputes between thecommunity and Fynbos Water ConservationProgramme; discipline, facilitation of com-munity tendering, and promoting socialactivities” (SALDRU/IFPRI project survey).Other functions were seen as “advising uson social questions in the community” andthis and discipline were the committee rolesseen as most useful by project implementers(interview, DWAF official). Steering com-mittees were functioning to greater or lesserdegrees across the FWWP projects, but in thetwo projects in our study, they were hardlyfunctioning at all. The committees were moreactive in the early stages, but this had wanedfor the reasons cited here. In both cases, thecommittees did not want to be brought inonly to deal with labor conflicts. They didnot want to be seen by workers as just adisciplinary body (reported at Kylemore) andfelt that they did not know enough about theproject and working conditions to feel com-fortable in that role (reported at Stellen-bosch). In Kylemore, they also felt uncom-fortable in a social worker role:

We can advise that the social workermust work there or the supervisor musttry to do that or so on, but we cannotbe physically involved and I think theythought that the steering committeecould take a lot of their responsibilitywhich is not working now. . . . So theyshould have thought about the thingsthat are going to happen in the projectand work out protocols, how we’regoing to handle some of these issues(interview, Facilitating Committee,Kylemore).

Increasingly, the committee felt that themanager, who tended to want to control mostaspects of the project herself, saw the com-mittee as an interloper, eventually excluding

SOCIAL, POLITICAL, AND INSTITUTIONAL CHALLENGES TO PARTICIPATION 65

it almost altogether. In Stellenbosch, the com-mittee was even less clear about its role.

Let me tell you, we don’t actual knowwhat is the whole objective of theSteering Committee. If we don’t havefunctions, so we don’t know what’sour function. We cannot tell you whatpeople want from the steering commit-tee (interview, Facilitating Committee,Stellenbosch).

Committee members felt that they shouldbe monitoring the project and reporting backto the community, but were too uninvolved

to know what to report back. Subsequently,the committee played no role, and did notknow why it was no longer called to meet-ings. There was a need to revisit and re-define the role of the steering committeesin these projects, which DWAF did under-take. Given that the WWP was operatingat an increasingly large scale, introducingmany new economic and social develop-ment measures rapidly, it was particularlyimportant that experimental institutionalarrangements were monitored, so that other-wise good program ideas were not under-mined by problems going undetected andunsolved.

66 CHAPTER 6

C H A P T E R 7

Conclusions

For several decades, development theory and practice has increasingly called for greaterparticipation of local people and community organizations in the initiation, design, andimplementation of development initiatives that directly affect their communities. This

call has philosophical, political, and instrumentalist underpinnings. There has been an in-creasing recognition that top-down, technocratic forms of development imposed on diverselocal realities often result in failure; that local people best understand their own needs andwhat is likely to work and not work; that involving local people can be cost-effective in termsof reduced capital costs, and increased involvement in operation and maintenance, and thusgreater sustainability; and finally, that poor people should have more power to direct thecourse of their own economic and social development.

These ideas have not been lost on South Africa. In fact, the last point infuses much SouthAfrican development policy. In terms of different levels of participation—information sharing,consultation, decisionmaking, and initiating action—the NPWP aimed to achieve all of these.In fact, by making community capacity building one of its central objectives, the NPWP sawparticipation in part as an “end” in itself. But it can and, from a political economy perspective,must simultaneously be a “means” of efficient infrastructure delivery if it is to survive asa policy. Although the goals of participation as an end and as a means to other ends can becomplementary, there are various tensions in this relationship.

The case studies pointed to ways in which participation, through wide community forums orthrough CBOs, added value to projects through enabling beneficiary communities to influencethe choices of priority assets and by contributing to project design features that can increasesafety and convenience, and affect the number of jobs created in the short and long term. Reg-ular communication with communities can increase a sense of local ownership with variouspositive spin-offs such as more cooperation and better maintenance. Yet community partici-pation also introduces politics, conflict, and lengthier decisionmaking processes, increasingthe time required up front. On the other hand, forgoing participation can result in even moreconflict and time required further down the line when time becomes more expensive.

The quantitative work demonstrates that even after accounting for the endogeneity ofparticipation, de facto participation has a statistically significant, positive effect on the projectbudget share spent on labor, the log number of days of work created, and the log numberof training days undertaken. It increases the percentage of employment that goes to womenand is associated with a reduction in the ratio of the project wage to local unskilled wages. Itreduces the cost of creating employment and reduces the cost of transferring income to thepoor. The magnitudes of these impacts are sizeable and they are robust to a variety of modelspecifications and the inclusion of other covariates.

67

Despite these benefits from participa-tion, the qualitative research found a widegap between ideas for community-drivenprojects embodied in national governmentpolicies and programs and the beliefs andpractices of professional providers at theprovincial and local levels, in governmentand the private sector. Although there is avoiced consensus on the importance ofcommunity participation, there is profoundambivalence about it, as well as widely dif-ferent ideas about what it means and whereit is appropriate.

Community-based PSCs were involvedin some way in almost all projects but theirroles tended to be limited to community andworker liaison functions. These liaison func-tions are important to the community anduseful for contractors, but far more limitedthan the vision of the NPWP, where the com-munity “through its representative commu-nity structure, should make the decisionsabout what should be constructed, how itshould be designed and constructed, whoshould work on the project, as well as therates and system of employment (DPW1996, 38). In many projects, consultants orlocal government excluded PSCs from par-ticipation in management tasks because ofprofessional views about efficiency and spe-cialization, and a lack of identification withthe program objectives of capacity buildingand empowerment. PSC members in someprojects did not have sufficient skills, andwere not trained. There were other problemsthat limited participation too: local politics,dwindling commitment of PSC members,and unclear role assignments.

There are two ways of responding tothese findings. One is to substantially re-duce the role of communities, providing op-portunities for communication and in some

areas consultation. Local government, if it islegitimate and downwardly accountable topoor constituents/public works beneficiaries,could be the sole provider. Alternatively,government could increase its commitmentto improving structures and processes forcommunity participation for the value thatit adds. In the same way that Ribot (2001,342) suggests that the “ ‘democratic’ decen-tralization experiment has not yet happened,”it can be argued that the community partici-pation experiment, in the way envisionedby the RDP and NPWP, has hardly occurred.Certainly the first phase of community-basedpublic works programs involved severegrowing pains, and subsequent participa-tory programs have still not solved all theseproblems.

Community participation in its pureforms envisioned by the government in 1994may not necessarily be the most appropri-ate path under newer political structuresand relationships. Participation does nothave to be all or nothing, and its best formsare likely to vary under different circum-stances.45 There are a range of modalitiesfor capturing local preferences and throughwhich to achieve accountability, trans-parency, capacity building, and local em-powerment, while delivering quality infra-structure needed by the poor. Debates aboutprioritization of objectives and how toachieve them must take place among policy-makers and their constituencies, and thendecisions reflected in program design. Ifparticipation and local empowerment re-main important objectives—and it is likelythat they will if differently conceived—policymakers and program designers mustexplore creatively institutional arrangementsand methods for increasing skills and ca-pacities. This might include a primary role

68 CHAPTER 7

45Pritchett and Woolcock (2004) argue that the quest for “one solution” is the problem. While they mainly pointto the problems with uniform imported technocratic and bureaucratic systems, they suggest that calls for partici-pation and empowerment be regarded with analytical complexity as well, advocating for local solutions to localproblems. They also note that tensions between the interests and incentives of all involved can be seen as an op-portunity to enhance creative solutions.

for legitimate downwardly accountablelocal authorities (as has been the intentionfor new community-based public worksprograms). But new policy proposals aloneare unlikely to bring about significantchange. This would require that all impor-tant stakeholders be brought into the pro-cess, common ground found, and a commit-ment generated to carry out and monitoragreements. This would also require provid-ing the time and training necessary to oper-ationalize agreed roles.

The Lutzville community hall projectwhere a CBO managed all aspects of a highlysuccessful project—meeting cost, quality,training, and capacity building objectives—

is an example of the feasibility of capablecommunity-based project management. Onthe other hand, not all communities will havethe social and political conditions of Lutz-ville West, so the road will not always be sosmooth. Nevertheless, politics, conflictsof interest, struggles over resources, andprocesses of consultation and consensus-building are part of the landscape of com-munity-based development. If participatorydevelopment remains part of South Africa’sdevelopment agenda, then all role-playersmust take the realization of this objectiveseriously through providing the necessaryresources, creativity, and patience to see theprocess through.

CONCLUSIONS 69

References

Abraham, A., and J. P. Platteau. 2001. Participatory development in the presence of endogenous com-munity imperfections. Paper presented at the 13th Annual Bank Conference on DevelopmentEconomics, World Bank, Washington D.C.

Adato, M. 1996. Democratic process, mediated models and the reconstitution of meaning in democraticorganizations: Trade unions and cooperatives in South Africa. Ph.D. dissertation, Cornell Uni-versity, New York.

———. 2000. From infrastructure construction to social reconstruction: Transforming institutionalidentity through South African public works programs. Washington, D.C.: International FoodPolicy Research Institute. Photocopy.

Adato, M., and L. Haddad. 2002. Targeting poverty through community-based public works programmes:Experience from South Africa. Journal of Development Studies 38 (3): 1–36.

Adato, M., L. Haddad, D. Horner, N. Ravjee, and R. Haywood. 1999. From works to public works: Theperformance of public works in Western Cape Province, South Africa. Final Report. Washing-ton, D.C.: International Food Policy Research Institute.

Aghion, P., and J. Tirole. 1997. Formal and real authority in organizations. Journal of Political Economy105 (1): 1–29.

Arrow, K. 1951. Social choice and individual values. New Haven, Conn., U.S.A.: Cowles Foundation.

Baland, J.-M., and J.-P. Platteau. 1997. Wealth inequality and efficiency in the commons: The unregu-lated case. Oxford Economic Papers 49: 451–482.

Bozzoli, B. ed. 1987. Class, community and conflict: South African perspectives. Johannesburg: RavanPress.

Chambers, R. 1995. Paradigm shifts and the practice of participatory research and development. In Powerand participatory development, ed. N. Nelson and S. Wright. London: Intermediate TechnologyPublications.

CSIR (Council for Scientific and Industrial Research). 1997. Third quarterly report on case studies ofNPWP pilot projects. Pretoria: Division of Roads and Transport Technology, CSIR.

Dayton-Johnson, J. 2000. Determinants of collective action on the local commons: A model with evi-dence from Mexico. Journal of Development Economics 62: 181–208.

Deolalikar, A. 1995. Special employment programs and poverty alleviation. Asian Development Review13 (2): 50–73.

Department of Labour. 1996. Creating jobs, fighting poverty: An employment strategy framework. Gov-ernment Gazette No. 19040. Pretoria: Department of Labour.

———. 1998. Restructuring the South Africa labour market. Labour Commission Report. Pretoria: De-partment of Labour.

DPW (Department of Public Works). 1994. The National Public Works Programme. What is it? Howdoes it work? A guide for communities, trade unions, government departments, contractorsand NGOs. Pretoria: DPW.

70

———. 1996. Project management system, Vols. 1–3, May. Pretoria: DPW.

———. 1997. Public works: Towards the 21st century. White Paper. Pretoria: DPW.

DTPW (Department of Transport and Public Works). 1996. Provincial progress report. Cape Town:DTPW.

DWAF (Department of Water Affairs and Forestry). 1996. Annual report 1995/96. Pretoria: DWAF.

———. 1998. Annual report 1997/98. Pretoria: DWAF.

Easterly, W., and R. Levine. 1997. Africa’s growth tragedy: Policies and ethnic divisions. QuarterlyJournal of Economics 112: 1203–1250.

Everatt, D., M. Ntsime, L. Stevens, L. Morrison, R. Govender, C. Relf, D. Sahle, S. Selatile, andJ. Walaza. 1997. An experiment in empowerment: An evaluation of the Community Based PublicWorks Programme. Johannesburg: Community Agency for Social Enquiry/Internation LabourOrganisation.

Greene, J. G. 1988. The youth employment program evaluation. Evaluation Review 12 (2): 91–116.

Guba, E., and Y. S. Lincoln. 1981. Effective evaluation. San Francisco, Calif.: Jossey-Bass.

Guijt, I. 1991. Perspectives on participation: Views from Africa. London: International Institute for En-vironment and Development.

Haddad, L., and M. Adato. 2002/03. “Maximizing Benefit Transfers to the Poor: Evidence from SouthAfrican Employment Programmes,” International Labour Review 141 (3).

Heckman, J. 1979. Sample selection bias as a specification error. Econometrica 47 (1): 153–163.

Hellmann, E., ed. 1949. Handbook on race relations in South Africa. Cape Town: Oxford UniversityPress.

Hoddinott, J. 2002. “Participation and poverty reduction: An analytical framework and overview of theissues,” Journal of African Economies 11: 144–166.

Horner, D. 1994. Job creation in the Western Cape: An approach to the Reconstruction and Develop-ment Programme. Cape Town: Southern Africa Labour and Development Research Unit, Uni-versity of Capetown.

Horrell, M. 1971. Action, reaction and counter-action. Johannesburg: South African Institute of RaceRelations.

I. T. Transport Ltd. 1997. International perspectives on public works programmes. Policy ResearchPaper Series. Pretoria: Department of Public Works.

Isham, J. 2002. Can investments in social capital improve local development and environmental out-comes? A cost–benefit framework to assess the policy options. In Social capital and economicdevelopment: Well-being in developing countries, J. Isham, T. Kelly, and S. Ramaswamy, eds.Cheltenham, U.K., and Northampton, Mass., U.S.A.: Edward Elgar.

Kerr, J. with G. Pangare and V. L. Pangare. 2002. Watershed development projects in India: An evalua-tion. Research Report 127. Washington, D.C.: International Food Policy Research Institute.

Khwaja, A. 2001. Can good projects succeed in bad communities? Collective action in the Himalayas.Paper presented at the 13th Annual Bank Conference on Development Economics, World Bank,Washington, D.C.

Klasen, S. 1997. Poverty, inequality and deprivation in South Africa: An analysis of the 1993 SALDRUsurvey. Social Indicators Research 41: 51–94.

Knack, S., and P. Keefer. 1997. Does social capital have an economic payoff? A cross-country investi-gation. Quarterly Journal of Economics 112: 1251–1288.

Lodge, T. 1991. Ideology and people’s power. In All, here and now: Black politics in South Africa in the1980s, ed. T. Lodge and B. Nasson with S. Mufson, K. Shubane, and N. Sithole. South AfricaUPDATE Series, Ford Foundation.

REFERENCES 71

Long, N., and M. Villarreal. 1993. Exploring development interfaces: From the transfer of knowledge tothe transformation of meaning. In Beyond the impasse: New directions in development theory,ed. F. J. Schuurman. Atlantic Highlands, N.J., U.S.A.: Zed Books.

Lund, F. No date. Draft paper on community care. University of Kwazulu-Natal, Durban, South Africa.

Maddala, G. S. 1983. Limited dependent and qualitative variables in econometrics. Cambridge, U.K.:Cambridge University Press.

Mauro, P. 1995. Corruption and growth. Quarterly Journal of Economics 110 (2): 681–712.

May, J., J. Govender, D. Budlender, R. Mokate, C. Rogerson, A. Stavrou, and N. Wilkins. 1998. Povertyand inequality in South Africa. Durban, South Africa: Praxis Publishing.

McDonald, L. 1995. NGOs and the problematic discourse of participation: Cases from Costa Rica. InDebating development discourse: Institutional and popular perspectives, ed. D. B. Moore andG. J. Schmitz. New York: Macmillan.

Ministry of Finance. 1996. Growth, employment and redistribution: A macroeconomic strategy. Min-istry of Finance, Pretoria, South Africa.

Moser, C. 1989. Gender planning in the Third World: Meeting practical and strategic gender needs.World Development 17 (11).

NEF (National Economic Forum) Technical Committee on a Public Works Programme. 1994a. A frame-work for implementation of a National Public Works Programme adopted by the NEF. Pre-investment investigation. Final report. Pretoria: NEF.

NEF (National Economic Forum). 1994b. Pre-investment investigation. Report of the Technical FocusGroup. Pretoria: NEF.

Nelson, N., and S. Wright. 1995. Participation and power. In Power and participatory development, ed.N. Nelson and S. Wright. London: Intermediate Technology Publications.

OEF International. 1986. Documento del Programa Educación para Participación (PEP) para Centro-america. San Jose, Costa Rica: OEF International.

Ostrom, E., L. Schroeder, and S. Wynne. 1990. Institutional incentives and rural infrastructure sustain-ability. Burlington, Vt.: Associates in Rural Development. Decentralization: Finance & Man-agement Project.

Parker, R., and T. Skytta. 2000. Rural water projects: Lessons from OED evaluations. World Bank Op-erations Evaluation Department, World Bank, Washington, D.C.

Patton, M. Q. 1986. Utilization-focussed evaluation. 2nd ed. Beverly Hills, Calif., U.S.A.: SagePublications.

Paul, S. 1987. Community participation in development projects. World Bank Discussion Paper No. 6,World Bank, Washington, D.C.

Pitt, M., M. Rosenzweig, and D. Gibbons. 1993. The determinants and consequences of the placementof government programs in Indonesia. World Bank Economic Review 7 (3): 319–348.

Pretty, J. N., and I. Scoones. 1995. Institutionalizing adaptive planning and local level concerns: Look-ing to the future. In Power and participatory development, ed. N. Nelson and S. Wright. London:Intermediate Technology Publications.

Pritchett, L., and M. Woolcock. 2004. Solutions when the solution is the problem: Arraying the disarrayin development. World Development 32 (2): 191–212.

PSLSD (Project for Statistics on Living Standards and Development). 1994. South Africans rich andpoor. Baseline household statistics. South African Labour and Development Research Unit,University of Cape Town, Cape Town.

Rahman, M. D. Anisur. 1993. People’s self development: Perspectives on participatory action research.Atlantic Highlands, N.J., U.S.A.: Zed Books.

72 REFERENCES

Rahnema, M. 1992. Participation. In The Development Dictionary: A guide to knowledge as power, ed.W. Sachs. Atlantic Highlands, N.J., U.S.A.: Zed Books.

RDP (Reconstruction and Development Programme). 1994. White paper on reconstruction and develop-ment: Government strategy for fundamental transformation. RDP white paper discussion doc-ument. Pretoria: Reconstruction and Development Programme.

Ribot, J. 2001. Integral local development: “accommodating multiple interests” through entrustmentand accountable representation. International Journal of Agricultural Resources, Governanceand Ecology 1 (3–4).

Rosenzweig, M., and Wolpin, K. 1986. Evaluating the effects of optimally distributed public programs.American Economic Review 76: 470–487.

SAIRR (South African Institute of Race Relations). 1985. Race relations survey 1984. Johannesburg:SAIRR.

———. 1987. Race relations survey 1986. Johannesburg: SAIRR.

———. 1992. Race relations survey 1991/92. Johannesburg: SAIRR.

———. 1994. Race relations survey 1993/94. Johannesburg: SAIRR.

———. 1995. Race relations survey 1994/95. Johannesburg: SAIRR.

———. 1996. Race relations survey 1995/96. Johannesburg: SAIRR.

———. 1997. Race relations survey 1996/97. Johannesburg: SAIRR.

Scoones, I., and J. Thompson, ed. 1994. Beyond farmer first: Rural people’s knowledge, agriculturalresearch and extension practice. London: Intermediate Technology Publications.

Seekings, J. 1988. The origins of political mobilisation in the PWV Townships, 1980–84. In Popularstruggles in South Africa, ed. W. Cobbett and R. Cohen. Trenton, N.J., U.S.A.: Africa WorldPress.

Subbarao, K. 1997. Public works as an anti-poverty program: An overview of cross-country experience.American Journal of Agricultural Economics 79 (May): 678–683.

Subbarao, K., A. Bonnerjee, J. Braithwaite, S. Carvalho, K. Ezemenari, C. Graham, and A. Thompson.1996. Social assistance and poverty-targeted programs: Sourcebook. Washington, D.C.: Povertyand Social Department, World Bank.

Uphoff, N., M. Esman, and A. Krishna. 1998. Reasons for success: Learning from instructive experiencesin rural development. West Hartford, Conn., U.S.A.: Kumarian Press.

Urban Foundation 1994. Public works programmes in South Africa: A key element in employment pol-icy? An executive summary. Johannesburg: Urban Foundation.

von Braun, J. 1995. Employment for poverty reduction and food security: Concept, research issues, andoverview. In Employment for poverty reduction and food security, ed. J. von Braun. Washing-ton, D.C.: International Food Policy Research Institute.

von Braun, J., T. Teklu, and P. Webb. 1992. Labour-intensive public works for food security in Africa:Past experience and future potential. International Labour Review 131 (1).

Wooldridge, J. 2002. Econometric analysis of cross section and panel data. Cambridge, Mass., U.S.A.:MIT Press.

REFERENCES 73