D I R E C T I O N S I N D E V E L O P M E N T D I R E C T I O N S I N D E V E L O P M E N T 35032 Empowerment in Practice From Analysis to Implementation RUTH ALSOP, METTE BERTELSEN, AND JEREMY HOLLAND DIRECTIONS IN DEVELOPMENT Empowerment in Practice From Analysis to Implementation Ruth Alsop Mette Frost Bertelsen Jeremy Holland THE WORLD BANK © 2006 The International Bank for Reconstruction and Development/THE WORLD BANK 1818 H Street, NW Washington, DC 20433 USA Telephone 202-473-1000 Internet www.worldbank.org E-mail feedback@worldbank.org All rights reserved First printing 1 2 3 4 09 08 07 06 The findings, interpretations, and conclusions expressed here are those of the authors and do not necessarily reflect the views of the Board of Executive Directors of the World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of the World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The World Bank encour- ages dissemination of its work and will normally grant permission promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; telephone 978-750-8400, fax 978-750-4470, www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, World Bank, at the above address, by fax 202-522-2422, or e-mail pubrights@worldbank.org. Cover photos: Jane Harper ISBN-10: 0-8213-6450-2 ISBN-13: 978-0-8213-6450-5 eISBN: 0-8213-6451-0 DOI: 10.1596/978-0-8213-6450-5 Library of Congress Cataloging-in-Publication Data Alsop, Ruth. Empowerment in practice : from analysis to implementation / Ruth Alsop, Mette Bertelsen, Jeremy Holland. p. cm. -- (Directions in development) Includes bibliographical references and index. ISBN-13: 978-0-8213-6450-5 ISBN-10: 0-8213-6450-2 1. Community development. 2. Social planning. 3. Economic development--Citizen participation. I. Bertelsen, Mette, 1975­ II. Holland, Jeremy, 1966­ III. Title. IV. Directions in development (Washington, D.C.) HN49.C6A375 2006 307.1'4--dc22 2005044731 Contents Preface x Acknowledgment xii About the Authors xiii Acronyms and Abbreviations xiv 1 Introduction 1 PART ONE: UNDERSTANDING AND APPLYING THE CONCEPT OF EMPOWERMENT 2 Empowerment: An Analytic Framework 9 Framework Overview 10 Agency 11 Opportunity Structure 13 Interaction between Agency and Opportunity Structure: Explaining Degrees of Empowerment 15 Where Empowerment Takes Place: Domains and Levels 19 Domains 19 Levels 21 The Framework Summarized 21 Empowerment and Development Outcomes 22 Understanding Empowerment: Summary 25 3 Methodological Issues in Measuring Empowerment 29 Challenges to Measurement 30 Toward a Methodology for Measuring Empowerment 32 Identifying Empowerment Indicators 33 Sequencing Methods and Data 37 iii iv CONTENTS Multidimensionality and Interactive Variables: Summarizing Methodological Issues 42 4 Using the Empowerment Framework: Five Country Studies Summarized 45 Design and Structure of the Studies 45 Application of the Framework 46 Brazil: The Participatory Budget Initiative 49 Ethiopia: The Women's Development Initiatives Project 50 Honduras: The Community-Based Education Project 52 Indonesia: The Kecamatan Development Project 53 Nepal: The Rural Water Supply and Sanitation Project 55 Conclusion: A Robust but Flexible Framework 56 5 Applying the Empowerment Framework to Operations 57 Framing Program and Project Interventions 58 Empowerment--Means or End? 59 Investing in Agency and Opportunity Structure 60 Project-Level Monitoring Systems 64 Policy Formulation and Implementation 71 Policy-Level Monitoring Systems 74 Operationalizing the Empowerment Framework: A Summary with Questions 79 6 Operationalizing a Difficult Concept: Summarizing Empowerment Analysis and Action 83 Structure and Agency: Prime Concepts 83 Using the Empowerment Framework in Operations and Analysis 85 The Missing Link: Need for a Coherent Approach to Empowerment 87 PART TWO: MEASURING AND EVALUATING EMPOWERMENT: FIVE COUNTRY STUDIES Preface to the Country Studies 91 7 Evaluating Empowerment: Participatory Budgeting in Brazilian Municipalities 95 The Participatory Budgeting Initiative 96 Research Framework 97 Empowerment Impact of PB: The Matched-Pair Analysis 98 Selection of Pairs 98 CONTENTS v The Matched Pairs 100 Citizen Engagement in Budgeting 102 Full PB: Binding Participation 104 Partial PB: Consultative Participation, Even When Contentious 106 State-Controlled Participation 107 Civil Society Capacity: Agency 108 Agency, Opportunity Structure, and Empowerment 110 Poverty Outcomes: Estimating the Effects of Participatory Budgeting 112 Methodology and Data Set 112 Effects on Growth, Poverty, and Inequality 115 Conclusion 120 8 Measuring Women's Empowerment in Ethiopia: The Women's Development Initiatives Project 125 Gender Inequality and the WDIP Project 125 Conceptual Framework of the Study 127 Research Design and Methodology 127 Instruments and Sampling 127 Indicators 129 Empowerment Outcomes and Determinants 129 State Domain 133 Market Domain 137 Social Domain 141 WDIP and Women's Empowerment 144 Impact in the Market Domain 146 Impact in the Social Domain 147 Conclusion 149 9 Impact Evaluation of the Honduras Community-Based Education Project 152 Conceptual Framework of the Study 153 Research Design and Methodology 154 Direct Indicators of Empowerment 155 Indirect Indicators of Empowerment: Agency 157 Indirect Indicators of Empowerment: Opportunity Structure 158 School Council Empowerment Outcomes 160 Empowerment of Excluded Community Members 165 Women 167 Indigenous People 167 The Poorest 168 Conclusion 169 vi CONTENTS 10 Empowerment and Local-Level Conflict Mediation in Indonesia 172 Conceptual Framework and Research Hypothesis 173 Research Design and Methodology 175 Assessing Empowerment: An Indonesian Case Study 177 State-Owned Resources in Kecamatan Badegan, Kabupatan Ponorogo, East Java 178 KDP Management in Kecamatan Proppo, Kabupatan Pamekasan, East Java 179 Land Conflicts in Mangarrai, Nusa Tengarra Timor 180 KDP in Kecamatan Pasean, Kabupatan Pamekasan, East Java 181 Other Cases 182 Analysis and Implications 183 Countervailing Power and Adaptive Conflict Management 183 Procedures for Countervailing Power 184 Disaggregating Conflict Management 185 KDP's Impact on Empowerment: A Typology 186 Conclusion 189 11 Negotiating Social Change: Gender, Caste, and Ethnic Dimensions of Empowerment and Social Inclusion in Rural Nepal 193 Conceptual Framework of the Study 194 Defining Empowerment and Social Inclusion 195 Complementary Dimensions of Social Change 196 Research Design and Methodology 197 The Empowerment Index 198 The Social Inclusion Index 199 The Composite Empowerment and Inclusion Index 199 The Women's Empowerment and Inclusion Index 200 Analysis: Who Was Empowered and Who Was Included? 202 Hierarchies Based on Caste and Ethnicity 203 Cross-Cutting Hierarchies of Gender and Caste-Ethnicity 206 Shifting Patterns in Women's Empowerment by Caste and Ethnicity 208 Cross-Cutting Caste, Ethnic, and Gender Dimensions of WEI 210 What Determines Empowerment and Social Inclusion? 212 Conclusion 215 CONTENTS vii Boxes 2.1 The Effect of Education on Other Assets 13 2.2 Persistence of Informal Institutions in Ethiopia 15 2.3 Assets, Opportunity Structure, and Effective Political Choice 16 2.4 Poverty-Reducing Effects of Access to Information in Ethiopia 25 5.1 Empowerment as an End in Guinea 61 5.2 Empowerment in a Watershed Development Project in India 62 5.3 Women in India: Enhancing Assets and Opportunity Structures 66 5.4 Empowerment for Poverty Reduction in Ethiopia 60 5.5 Purpose and Functions of Poverty Monitoring Systems 77 5.6 Organizational Requirements of Poverty Monitoring Systems 78 7.1 Assessing the Development Benefits of Participatory Budgeting: Dimensions and Indicators 113 8.1 Illustrative Indicators Used in Three Domains of Women's Lives 130 10.1 The Madurese Tradition of Carok 182 11.1 Indicators Used for the Empowerment Index (EMI) 200 11.2 Indicators Used for the Social Inclusion Index (SII) 201 11.3 Indicators Used for the Women's Empowerment and Inclusion Index (WEI) 204 Figures 2.1 The Relationship between Outcomes and Correlates of Empowerment 10 5.1 Schematic Representation of Nested Monitoring and Learning System 69 7.1 Brazil Study: Frame and Sequence of Analysis 99 7.2 Changes in Agency and Opportunity Structure in 10 Brazilian Cities, 1997­2000 111 8.1 Conceptual Framework for WDIP Evaluation 128 11.1 Composite Empowerment and Inclusion Index (CEI) Levels, by Gender and Caste and Ethnicity 202 11.2 Woman's Empowerment and Inclusion Index (WEI) Levels, by Caste and Ethnicity 203 11.3 Composite Empowerment and Inclusion Index (CEI) Levels, by Caste and Ethnicity 205 11.4 Composite Empowerment and Inclusion Index (CEI) Levels, by Gender 207 viii CONTENTS 11.5 Community Empowerment Index (EMI) Gender Gap by Caste and Ethnicity: Result of Multivariate Analysis 208 11.6 Composite Empowerment and Inclusion Index (CEI) Levels for Women, by Caste and Ethnicity 209 11.7 Influence of Caste and Ethnicity on Composite Empowerment and Inclusion Index (CEI) Levels, by Type of Analysis 213 Tables 2.1 Summary of the Analytic Framework 23 3.1 Indirect Indicators of Empowerment: Asset Endowment 33 3.2 Indirect Indicators of Empowerment: Operation of Institutions 35 3.3 Direct Indicators of Empowerment 36 4.1 Definitions of Empowerment Used in the Country Studies 47 4.2 Case Study by Analytic Focus 48 5.1 Actions to Develop Assets and Opportunity Structures in the Madhya Pradesh Poverty Initiatives Project 65 5.2 Summary of a Project's Monitoring and Learning System 70 5.3 Data Availability and Critical Actions for Poverty Reduction in Ethiopia 73 7.1 Citizen Participation in Budgeting in 10 Brazilian Cities, 1997­2000 103 7.2 Synthesis of Participation in Budgeting, 1997­2000 104 7.3 Civil Society before and after 1997­2000 108 7.4 PB Municipalities Compared with Other Municipalities 116 7.5 Estimating the Effects of Participatory Budgeting on Multiple Dimensions of Development Performance 118 8.1 Explaining Women's Happiness 132 8.2 Explaining Women's Political Engagement 134 8.3 Women and Justice: Proportion of Institutions Considered "Fair" in Different Circumstances 135 8.4 Experience and Awareness of Legal Protection against Sexual Violence 136 8.5 Experience and Awareness of the Right to Choose One's Spouse 137 8.6 Explaining Women's Economic Empowerment 140 8.7 Women's Associational Life: Participation in Groups among Rural and Urban Women 144 8.8 Women's Influence on Community Decisions 145 CONTENTS ix 8.9 Social Outcomes: WDIP Beneficiaries and Nonbeneficiaries 148 8.10 Number of People Beneficiaries Can Rely on for Loans 149 9.1 Empowerment of School Councils 160 9.2 Level of School Council Empowerment, by Region 161 9.3 Multiple Regression against School Council Empowerment Index 162 9.4 Stepwise Regression against School Council Empowerment Index, by Number of Variables Entered 163 9.5 Empowerment of Excluded Commuity Members, by Region 166 10.1 Routines for Managing Local Conflict 187 11.1 Multiple Regression Analysis Examining Influence of Caste and Ethnicity on Women's Empowerment and Inclusion 212 Appendixes 1 Selected Examples of Approaches to Measuring Empowerment 219 2 Understanding the Concept of Power 230 3 Measuring and Monitoring Empowerment in World Bank Projects 243 4 Indicators Used in the Five Country Studies 257 5 Empowerment Themes and Strategies in Selected PRSPs 277 6 Draft National Survey Empowerment Module 284 Bibliography 341 Index 353 Preface People in all walks of life often have difficult choices to make--but for some the range of options is more restricted than for others. In remote villages and urban communities, many women, men, and children have limited choices available to them, resulting in a daily struggle to simply survive. The poor in particular have limited employment opportunities, little voice in decision making over locally available resources, often lack basic services, have lim- ited recourse to state-sponsored systems of justice, and are rarely able to exercise the right to hold their representatives accountable. These people suffer from inequality in terms of the power they have to change their lives and escape poverty. Some, such as women and excluded ethnic or social groups, are even more disempowered than others. The World Bank, along with development partners in donor agencies, gov- ernments, and within civil society, recognizes that poverty reduction involves understanding and addressing the often complex processes that limit people's capacity to make life-changing choices. It is necessary but insufficient for poverty alleviation to deliver services, to develop infrastructure, and raise income levels. Effective and sustainable development means designing poli- cies and interventions that both build the stocks of assets people have at their disposal and ensure that the "rules of the game" operate to allow the transformation of these assets into poverty reducing benefits. The model for understanding and operationalizing an empowerment approach to development presented in this volume is simple. The book translates a long-standing academic discourse on structure and agency into an actionable framework that can, in practice, help change power relations and in turn reduce poverty. Using the concepts of asset-based agency and institution-based opportunity structure, this empowerment framework can be applied across dimensions of both action and analysis.As such it provides an instrument to help policy makers and practitioners do their business better. As an analytic tool the framework also aids in tracking levels of empowerment and assessing progress made toward that end, as well as offering guidance on how to embed this analysis into broader poverty- monitoring systems. x PREFACE xi We are delighted to be able to share a document that draws upon both theory and applied experience to translate the abstract notion of empow- erment into development practice. Ruth Alsop Mette Frost Bertelsen Jeremy Holland Acknowledgments This book is the outcome of a collaborative effort between the Empowerment Team of the World Bank and five country teams. We would like to express sin- cere thanks to country coordinators and authors of the draft country stud- ies -- Lynn Bennett (Nepal), Arianna Legovini (Ethiopia), Mike Walton (Brazil), Mike Woolcock (Indonesia), and Emanuela di Gropello and Nina Heinsohn (Honduras). These coordinators were ably supported by teams who provided disciplinary and country expertise. These included Kishor Gajural, KimArmstrong, and Sandra Houser (Nepal); the Ethiopian Economic Association (Ethiopia); Gianpaolo Baiocchi, Shubham Chaudhuri, Patrick Heller, and the Centro de Assessoria e Estudios Urbanos (Brazil); Patrick Barron, Leni Dharmawan, Claire Smith, Rachael Diprose, LutfiAshari,Adam Satu, and Saifullah Barwani (Indonesia); and ESAConsultores (Honduras). Further background materials for the appendixes were contributed by Anirudh Krishna (appendix 1), Fruzsina Csaszar, Joy Moncrieffe, and Duncan Holtom (appendix 2), Jacob Young (appendix 3), and Simon Brook (appen- dix 6). We are grateful to all the above for the use of their material. We are also very grateful to the following peer reviewers for their time, insights, and contributions: Regina Birner, Paul Francis, Markus Goldstein, and Howard White. Valuable additional comments were provided by mem- bers of the Empowerment Team, by colleagues in the World Bank's Poverty Reduction Group, and by a network of colleagues within and beyond the World Bank: Tara Bedi, Rob Chase, Luc Christiaensen, Anis Dani, Ishac Diwan, Rosalind Eyben, Osvaldo Feinstein, Estanislao Gatitua-Mario, Nina Heinsohn, Reidar Kvam, Sarah Ladbury, Joy Moncrieffe, Caroline Moser, Veronica Nyhan-Jones, Patti Petesch, Uri Raich, Abigail Somma, Carrie Turk, Michael Walton, and Susan Wong. We must also thank PRMPR man- agement for support and guidance. Technical support was provided by Alex Arenas, Oykiao Koo Tze Mew, Nelly Obias, and Cecile Wodon. Finally, many thanks to Cathy Sunshine, for her excellent work in editing the book despite being presented with a very tight deadline. This book and the team's work on measuring empowerment have been funded by the Finish-Norwegian Trust Fund in the World Bank. xii About the Authors RuthAlsop is the Empowerment Team Leader in the Poverty Reduction unit of the World Bank. In a career spanning over 20 years she has worked in UK academia, for British NGOs, the Ford Foundation, the International Food Policy Research Institute and bilateral donor organizations. She has published widely, with journal articles and books focusing on the organi- zational, institutional, and political aspects of development interventions. Mette Frost Bertelsen has been with the Empowerment Team since 2004 as a Junior Professional Officer financed by DANIDA. She previously worked with a civil society organization in Nicaragua on developing a system for mon- itoring democratic processes and has also undertaken research on power relations and civil society participation in Poverty Reduction Strategy Papers. Jeremy Holland lectures in Social Development at the Centre for Development Studies, University of Wales Swansea, working with the World Bank's Empow- erment Team in 2004­5. During the past 10 years he has managed research and development projects in the Caribbean, Africa, and the Middle East. His published work focuses on participation, institutions, and poverty dynamics in development and on combining methods in development research. xiii Acronyms and Abbreviations AECO Asociación Educativa Comunitaria BCN Brahmans, Chhetris, and Newars CDD community-driven development CEI Composite Empowerment and Inclusion Index EMI Empowerment Index HIV/AIDS Human Immunodeficiency Virus/Acquired Immune Deficiency Syndrome INEA Institute for Adult Education KDP Kecamatan Development Project LSMS Living Standards Measurement Survey M&E monitoring and evaluation MESI Measuring Empowerment and Social Inclusion MID monitoring, information flow, and decision making NGO nongovernmental organization PB participatory budgeting Proheco Honduras Community-Based Education Project PRSP Poverty Reduction Strategy Paper PT Partido dos Trabalhadores (Workers' Party) RWDEP Rural Women's Development and Empowerment Project RWSS Rural Water Supply and Sanitation Project SDPRP Sustainable Development and Poverty Reduction Program SHG self-help group SII Social Inclusion Index THP traditional harmful practices WDIP Women's Development Initiatives Project WDO Women's Development Office WEI Women's Empowerment and Inclusion Index Note: All dollar amounts are U.S. dollars. U.S. equivalents for birr are based on the exchange rate in effect as of August 7, 2005. xiv 1 Introduction Empowerment has become a familiar term within many development agen- cies.1 Empowerment objectives also appear with increasing frequency in policy documents issued by governments, especially strategy papers deal- ing with poverty reduction. By 2005 more than 1,800 projects in the World Bank's lending portfolio mentioned empowerment in their project docu- mentation.2 Yet there are many different interpretations of what empow- erment means, analytically and operationally, and these interpretations are often inconsistent even within one organization. In addition, limited mate- rial is available on the practicalities of measuring, tracking, and evaluat- ing progress made toward empowerment (see appendix 1 for a summary of measurement efforts). This volume offers one way of understanding the concept, using a simple framework that can be laid across both analytic and operational work to identify issues, focus discussion, and prioritize practical entry points for promoting and tracking empowerment. In this volume, empowerment is defined as "the process of enhancing an individual's or group's capacity to make purposive choices and to trans- form those choices into desired actions and outcomes." Using the concepts of asset-based agency and institution-based opportunity structure, the framework presented suggests that investments and interventions can empower people by focusing on the dynamic and iterative relationship between agency and structure. In short, it is hypothesized that interven- tions to improve agency and enhance opportunity structures can increase people's capacity to make effective choices, and that this in turn can bring about other development outcomes. The framework has emerged from two years of work involving conceptual reviews, five country studies, and efforts to test the approach within parts of four other World Bank country pro- grams. The authors believe that the conceptual work, combined with coun- try activities centered on in situ measurement and application of empowerment analysis, gives a theoretical and applied robustness to the framework and subsequent suggestions for its use.3 Thisdocumentisdividedintotwoparts.Part1explainshowtheempower- ment framework can be used for understanding, measuring, monitoring, 1 2 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION and operationalizing empowerment policy and practice. Part 2 presents summaries of each of the five country studies, using them to discuss how the empowerment framework can be applied in very different country and sector contexts and what lessons can be learned from these test cases.4 Empowerment is clearly something that development agencies of all types are taking seriously. As policy, analysis, and interventions increas- ingly use the term, there is an urgent need to clearly conceptualize and track empowerment. But why is empowerment important? In their recent book Growth and Empowerment, Stern, Dethier, and Rogers (2005) propose a strategy for development that demands a dual approach: building a cli- mate that encourages investment and growth while at the same time empowering poor people to participate in that growth. Empowerment is suggested both as a goal in itself and as a driver of development. This view resonates with a body of work linking empowerment and broader development goals. These works support the proposition that efforts to change power relations, giving people an equal capacity to make effective choices, have both intrinsic and instrumental worth. Arguments for the intrinsic value of empowerment are found in the philosophical underpinnings of literature on democratization and decentralization (Dollar and Kraay 2002; Kaufmann, Kraay, and Mastruzzi 2004; Moore and Putzel 1999); on nonmonetary aspects of poverty reduction (Bourguignon, Ferreira, and Menendez 2003; Sen 1999a); and on human rights approaches to devel- opment (Alsop 2005; DFID 2000; Eyben 2003; Moser and Norton 2001). The instrumental arguments for empowerment are simple. Empowerment asso- ciates positively with achievements in other spheres of development, includ- ing growth, poverty alleviation, and the realization of human rights, as the discussion of participatory budgeting in chapter 7 suggests (see also Lokshin and Ravallion 2005; Varshney 2005). While this book can offer only a lim- ited empirical basis for the positive association between empowerment and development outcomes, it does add to the body of work supporting the existence of such a relationship. Perhaps more importantly, it also pro- vides a framework for future research to test the association and to prior- itize practical interventions seeking to empower individuals and groups. Whether empowerment is seen as a means to poverty reduction or as an end in itself, a basic question arises: Who is to be empowered? Uncertainty in this regard often causes problems when moving beyond rhetoric and into operationalization.As a relational concept, empowerment often means redress- ing imbalances of power between those who have it and those who do not. This can imply that empowerment is a zero-sum game--that is, one person or group gains power at the expense of another. Unfortunately, while this does not have to be the case, it is often taken to be so, and actions to empower certain groups or individuals can meet with resistance. Efforts to empower may be undermined at all levels. Government staff can find it difficult to INTRODUCTION 3 hand over power and resources to elected leaders; elected leaders or those implementing development interventions are not always comfortable when power to hold them accountable is vested in citizens; elite groups used to controlling local resources and decision making can feel that their power, authority, and prestige are undermined along with their ability to capture development benefits. While such resistance does not always occur, it is a factor that policy makers and practitioners must consider when designing and monitoring empowerment strategies and interventions. Understanding and accounting for poor people's weak "bargaining posi- tion" is critical to the success of policies and actions for poverty alleviation. Those in the most unequal bargaining positions, that is, those unable to make effective choices, are usually among the poorest--typically indigenous peo- ple, women, or certain other groups such as the Dalits in India or pastoral- ists in Ethiopia.Amass of evidence shows that disenfranchised people, who are often defined by their social attributes, are usually the poorest in income, consumption, and welfare terms. They are also subject to greater vulnera- bility, that is, they are less able to withstand stress or shocks. Power relations therefore need to be taken seriously if these people are to make their way out of poverty. However, this touches upon sensitive cul- tural issues, and thus empowerment may mean changing deeply ingrained beliefs and practices. For example, as reported recently in the Financial Times of London, "Azhar Ramadan, a legislator from Baghdad and part of a women's alliance lobbying constitutional drafters, is opposed to any men- tion of tribes in the constitution. Tribal customs, she says, contain such anachronisms as nahi--in which a paternal cousin can claim the right to marry her, and kill anyone else who does so--or fasi settlements, in which the transgressing tribes offer their daughters to aggrieved tribes, who can either marry them or take them as servant girls. It also takes an indulgent view on `honor killings'--when families kill their wives or daughters, and their lovers, for having extramarital sex" (August 13­14, 2005). This is an extreme example of women's disempowerment, but demonstrates that tra- ditional norms and beliefs can be highly inequitable, reducing opportuni- ties for people to make effective choices. The term empowerment is commonly used to indicate both an outcome, in which a person or group enjoys a state of empowerment, and a process, an action that moves a group or person from a lower to a higher state of empow- erment. While not of central importance to the discussion here, this distinc- tion can be helpful to potential users of the framework suggested in this book. For example, in a project monitoring system, different indicators would be used for process and outcome. In an education project, process indicators might track the establishment of school management commit- tees as well as information or training activities that build parents' capac- ities to use these opportunities to participate. For the same project, outcome 4 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION indicators would measure the degree of empowerment achieved, perhaps with indicators such as levels of participation in school committees by par- ents and the effect of parental participation on decision making.5 In addi- tion, the distinction between process and outcome indicators will be critical to statistical analysis and will become even more important as analytic techniques become more complex. Of course, indicators and interventions appropriate to one sector are unlikely to be appropriate to another. Also, because the concept of empow- erment implies a changing relationship between individuals or groups, the values placed on indicators will vary depending upon the countries and subnational contexts within which these relationships unfold. The frame- work developed in this book addresses the problem of generating com- parative analysis by looking at three main domains that affect people's lives--the state, market, and society. Therefore, while both indicators and their values may vary, the generic framework allows for discussion of empowerment within the same domains in different locations. The frame- work also recognizes that a person or group can experience different degrees of empowerment in different contexts. For example, a poor man is likely to experience very different degrees of empowerment within his house- hold or family, within his community, or within higher-level administra- tive jurisdictions. That is, his degree of empowerment varies according to the level at which he is operating. This recognition is important both for measuring empowerment and for considering what an intervention should focus upon. Developing a framework that can be used across different contexts is not easy--and some may argue that it is not even desirable. However, the real- ity facing development organizations today is that the usefulness of a base, generic analytic and operational framework outweighs such criticisms. In this volume, therefore, the authors have sought to balance an acute con- sciousness of diversity with the need to provide a robust and simple frame- work that can be adapted and selectively used for different purposes. The recognition of diversity underpinned the selection of five different cultural and sectoral cases through which to develop and test the empowerment framework. Drawing on the country team studies, the volume summarizes and develops these cases to show how the framework can be applied for interventions with very different objectives and conditions. Following this introduction, part 1 of this volume begins by introducing an analytic framework for empowerment in chapter 2, describing its ele- ments and outlining its theoretical influences. Chapter 3 discusses a mixed- methods approach to gathering, recording, and analyzing both qualitative and quantitative empowerment information. Chapter 4 uses the five country cases to show how the framework can be applied to analysis and measure- ment in different contexts. Chapter 5 discusses application of the frame- INTRODUCTION 5 work in project and policy operations, and chapter 6 closes part 1 with an overview of the use of this approach in understanding and operationalizing empowerment. Part 2 consists of five chapters, each one summarizing a coun- try case study. This volume represents an effort to present an easily accessible framework to readers, especially those for whom empowerment remains a puzzling development concern, conceptually and in application. While rooted in aca- demic discourse, the discussion in the following pages focuses on presenta- tion and illustration of a framework that can be used for development practice rather than on theoretical discussion of the framework's conceptual under- pinnings. This volume is complemented by two other products. One is a basic, case-based learning module on understanding and applying empowerment. The other is the base questionnaire and a set of accompanying guidelines on the collection, recording, and analysis of empowerment data.6 Notes 1. Following the publication of World Development Report 2000­2001, the World Bank launched Empowerment and Poverty Reduction: A Sourcebook (Narayan 2002), representing the Bank's first attempt to define the concept of empowerment and articulate areas of applica- tion. The present volume builds upon and furthers the ideas laid out in the sourcebook. 2. Appendix 3 summarizes a small sample of World Bank project documents that pursue empowerment as a central project objective, to illustrate how empowerment is presently measured and evaluated in Bank projects. 3. Country cases were managed by Lynn Bennett (Nepal), Arianna Legovini (Ethiopia), Mike Walton (Brazil), Mike Woolcock (Indonesia), and Emanuela di Gropello and Nina Heinsohn (Honduras). These task managers worked in collaboration with a team of international and local consultants: Kishor Gajural, KimArmstrong, and Sandra Houser (Nepal); the Ethiopian Economic Association (Ethiopia); Gianpaolo Baiocchi, Shubham Chaudhuri, Patrick Heller, and the Centro deAssessoria e Estudos Urbanos (Brazil); Patrick Barron, Leni Dharmawan, Claire Smith, Rachael Diprose, Lutfi Ashari, Adam Satu, and Saifullah Barwani (Indonesia); and ESA Consultores (Honduras). For additional information please visit www.worldbank.org/empowerment/. 4. Each case study provides insight into the association between empowerment interven- tions, empowerment outcomes, and broader development outcomes. However, the primary objective of the case studies was to develop and test indicators and instruments for understand- ing and tracking empowerment. The evidence on outcomes and causal links should there- fore be treated only as illustrative of the kind of analysis that might be undertaken, and the results should be seen as indicative of potential causal links. 5. Similarly, in development interventions, differentiating between outcomes, such as degrees of empowerment, and the means or processes required to achieve those outcomes, ensures clarity of purpose during design and implementation. 6. The learning module and questionnaire with guidelines are available through the World Bank's empowerment Web site (www.worldbank.org/empowerment/). PART ONE Understanding and Applying the Concept of Empowerment 7 2 Empowerment: An Analytic Framework This chapter provides an analytic framework for understanding and mea- suring empowerment and for framing action to further empowerment of individuals or groups. The framework introduced here draws on a long sociological tradition of the analysis of power, discussed at greater length in appendix 2. The framework focuses on the dynamic and iterative rela- tionship between structure and agency and underpins the five country studies presented in part 2 of this book. Demonstrating a key message of this publication--that diversity in application is critical--these studies illus- trate the adaptation and use of the framework in specific country and devel- opment contexts. Although the framework is primarily influenced by sociological theory, it accommodates theoretical insights from a range of disciplinary traditions that have tackled aspects of the structure-agency relationship (see Narayan 2005 for papers relating to multidisciplinary perspectives on measuring empowerment). For institutional economists, institutions are part of the solu- tion to reducing transaction costs and moving toward a more complete market by realigning incentives and providing sanctions for rational actors (Atkinson and Stiglitz 1980). "New" political economists and behavioral economists diverge from simplistic notions of rational choice in recognizing the influ- ence of politically motivated and subjectively formed interests of those with greater bargaining strength (Besley 2004; North 1990). Political scientists have similarly examined the role and influence of vested interests in decision making and the implications of these interests for "progressive" institutional change (P. Evans 2004; Hall and Taylor 1996). Additional theoretical work on agency includes contributions from a range of disciplines across a range of subject matter. In particular, work on both social capital--that is, the trust and reciprocity underpinning effective social relations (Bourdieu 1986; Coleman 1990; Putnam 1993)--and adult education, on the psychological challenges and the role of motivation for learning and behavioral change, have been useful in shaping the empowerment approach suggested in this volume (Crocker andAlgina 1986; Nunnally and Bernstein 1978). This chapter first presents an overview of the analytic framework, exam- ining the concepts of agency and opportunity structure. It then considers 9 10 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION the degrees of empowerment that can result from the interaction of oppor- tunity structure and agency and outlines how this can be used to assess empowerment and factors associated with it in different domains of life. The chapter concludes with a discussion of links between empowerment and other development outcomes. Framework Overview The framework draws heavily on discussions of power in social theory lit- erature. Since Hobbes's explication of the social contract and state-citizen relations, many other social and political scientists have explored power as a concept that extends beyond the realm of the state to involve relations between individuals or groups (see appendix 2).1 Social theorists such as Giddens (1984) emphasize the relationship between agency and structure. The analytic framework used in this book treats empowerment as contin- gent upon this relationship. Empowerment is defined as a group's or individual's capacity to make effective choices, that is, to make choices and then to transform those choices into desired actions and outcomes. As has already been suggested and as figure 2.1 illustrates, this capacity is primarily influenced by two sets of inter- related factors: agency and opportunity structure. Agency is defined as an actor's or group's ability to make purposeful choices--that is, the actor is able to envisage and purposively choose options. But agency cannot be treated as synonymous with empowerment. Even when people have the capacity to choose options, they may not be able to use that agency effec- tively. They are constrained by their opportunity structure, defined as those aspects of the institutional context within which actors operate that influ- ence their ability to transform agency into action. By establishing the "rules Figure 2.1. The Relationship between Outcomes and Correlates of Empowerment Agency Iterative Degree of Development relationship empowerment outcomes Opportunity structure Source: Authors. EMPOWERMENT: AN ANALYTIC FRAMEWORK 11 of the game" for the exercise of agency, institutional contexts determine, to a greater or lesser extent, the effectiveness of agency. To complicate matters further, these rules can also influence the accrual of stocks of assets and determine the value of benefits that flow from these assets. Working together, these factors give rise to different degrees of empow- erment and are assumed to have mutually reinforcing effects on develop- ment outcomes.2 Figure 2.1 indicates that agency and opportunity structure associate with the degree of empowerment a person or group experiences.3 It also suggests a relationship between empowerment and development outcomes.4 Agency Agency is defined as an actor's or group's ability to make purposeful choices--that is, the actor is able to envisage and purposively choose options. In terms of both measurement of and action to enhance empowerment, a person or group's agency can be largely predicted by their asset endowment. Assets are the stocks of resources that equip actors to use economic, social, and political opportunities, to be productive, and to protect themselves from shocks (Moser 1998; Swift 1989). In addition to the question of which assets are most important in understanding and tracking empowerment, two other points are discussed below: first, the range of assets considered in this frame- work, and second, the interaction between different kinds of assets. The assets requiring measurement (for monitoring or analysis) or con- sideration in operational work include psychological, informational, orga- nizational, material, social, financial, and human assets. Some are easier to measure than others. For example, undertaking a survey to gather infor- mation on human assets such as skills or literacy, or designing an inter- vention to enhance them, is less problematic than doing the same for assets such as social capital. Even more difficult is the measurement or enhance- ment of psychological assets, such as the capacity to envision.5 Yet all these assets are critical in this treatment of empowerment, in which power is understood to result from a combination of resources and rules (Giddens 1984). Although they are complicated and often ignored, psychological assets are particularly crucial both to interventions and to measurement of asset-based agency.Actors need a raised level of consciousness if they are to translate their assets into choices--that is, to become "agents."6 Women, in particular, are often locked into a cultural framework in which they perceive their disem- powerment to be right and proper (Kabeer 1999). In a recent survey of women in Ethiopia, for example, large percentages agreed that a husband is justified in beating his wife for the following reasons: burning food (65 percent), argu- ing with him (61 percent), leaving the house without telling him (56 percent), 12 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION neglecting the children (65 percent), and refusing sexual relations (51 percent) (Central Statistical Authority 2001). The capacity of these women to choose a different way of living requires a change in their psychological assets-- which, in this case, is prerequisite to a change in the social institutions gov- erning the right of a husband to beat his wife. Similarly, many outsiders would find it hard to understand why a group of low-caste "toddy tap- pers" in Bihar, India accept that the richest man in the village incites the police to beat them periodically.7 These are but two examples of people's acceptance of low levels of empowerment, with similar patterns to be found in most cultures. As the statistics show, this passive concurrence with the right of others to practice coercive forms of power through violence is deeply ingrained in the psyche of many individuals and groups. Psychological assets, therefore, are particularly important, but often unrecognized, assets in development terms. People make choices that are characterized by "adaptive preferences," or narrow practical aspirations that shape how they conceive of their life possibilities. To illustrate this point, Becker (1995) observes that women and minority groups frequently under- invest in their own human capital because they have been brought up to believe that they cannot do certain things that other people can do. As a result, actors and social groups internalize their second class status in ways that cause them to make choices that perpetuate their disempowered status (see also Nussbaum 2000). Furthermore, actors with low levels of psychological assets are also less likely to make choices that can build or strengthen the other assets that form the basis of their agency. Understanding the interaction among assets and the effect of this interaction on agency presents methodological and analytic challenges. It is clear that the endowment of a single asset, such as owner- ship of land or capacity to aspire, can affect a person's or group's ability to make meaningful choices. A person's or group's command over one asset can also, however, affect their endowment of another asset. For example, education (a human asset) often gives an actor greater access to information (itself an asset) and at times improves his or her capacity to envision alter- native options (a psychological asset).8 Similarly, for groups of people, col- lective savings (a financial asset) can give access to enhanced productive assets. In these cases, more than one asset contributes to the capacity to make meaningful choices. Box 2.1 provides a further example of the inter- play between different assets and how this serves to enhance agency. While figure 2.1 initially suggests a relatively simple model of relations between causes and effects, the factors contributing to agency are often inter- active. Assets interact with each other, and opportunity structure can influ- ence the accumulation and use of asset stocks. This has implications for data collection and--not least--analysis, implying a need for information on the range of asset endowments and for analysis that tests the effects of one asset on another as well as their association with empowerment outcomes. EMPOWERMENT: AN ANALYTIC FRAMEWORK 13 Box 2.1. The Effect of Education on Other Assets The Institute for Adult Education (INEA) in Mexico provides literacy training and basic education to disadvantaged young adults who have not attended or have dropped out of the formal school system. Students' testimony indicates that enrollment in INEA programs has not only improved their education levels but has also provided them with other skills and assets. They state that being able to read and write has enhanced their self-confidence, and that, as a result, they are less hesitant to voice opinions and speak in public. INEA courses also give students access to information. Women learn, for example, that domestic violence is an infringement of their rights and that they are entitled to seek help or redress. Coupled with increased self- confidence, such information might lead an INEA student to take action to stop abuse. INEA schools also provide an arena for interaction that contributes to a community's level of social capital. Studying together, students learn to trust each other and develop friendships and networks of support. Source: Heinsohn 2004. Opportunity Structure An actor may be able to choose options, but the effective realization of those choices will largely depend upon the institutional context within which the actor lives and works. The opportunity structure comprises these insti- tutions that govern people's behavior and that influence the success or fail- ure of the choices that they make.9 Institutions are the "rules of the game" devised by societies to shape and constrain human interaction and individual choices (North 1990).10 Institutions can be formal or informal. Formal institutions include the sets of rules, laws, and regulatory frameworks that govern the operation of polit- ical processes, public services, private organizations, and markets. Informal institutions include the "unofficial" rules that structure incentives and govern relationships within organizations such as bureaucracies, firms, or industries, as well as the informal cultural practices, value systems, and norms of behavior that operate in households or among social groups or communi- ties. In practice, changes such as shifting power relations involve an inter- action between formal and informal institutions and can, as the last part of this section demonstrates, result in tension as these changes play out. Utilitarian, preference-based approaches to analysis of choice assume that choice is a straightforward process in which actors make rational selections 14 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION designed to maximize utility. This "rational choice" approach is questioned, however, by those who recognize that choice is constrained by social circum- stance or social rules, with implications for individual and group expression of agency as discussed above. Kabeer, in discussing women's empowerment, cites Bourdieu's (1977) notion of doxa, meaning those aspects of tradition and culture that have become so habitual as to be "naturalized." These tradi- tions are rooted in deeply entrenched cultural institutions: The passage from "doxa" to discourse, a more critical consciousness, only becomes possible when competing ways of "being and doing" become avail- able as material and cultural possibilities, so that "common sense" propo- sitions of culture begin to lose their "naturalised" character, revealing the underlying arbitrariness of the given social order. (Kabeer 1999, 441) Increases in agency, through the accumulation of assets such as educa- tion, information, psychological consciousness, and income or consumption wealth, have been shown to associate with changes in traditional informal institutions such as untouchability in India and Nepal and gender-based inequalities in many countries. However, the relationship does not always hold, and in practice it may be extremely hard for disempowered people or groups to use agency effectively. When discussing agency, the distinction between habitual choice and reasoned choice, therefore, becomes impor- tant. Habitual choices made within routines and customs may be com- fortable, but rarely confer new or higher levels of agency. Formal institutions touch the lives of most people. Common examples include a country's legal framework, tax regulations, and local governance rules, such as what constitutes a quorum in a local committee or how pas- ture land is managed. As with informal institutions, measurement efforts need to track not only the existence of these institutions, but the way they work in practice. In India, for example, constitutional amendments in 1992 reserved seats for women representatives, but in many places these have been less than entirely successful because social norms governing women's public behavior undermine their capacity to operate as political leaders.11 Changes in legislation, such as these constitutional amendments and sub- sequent state acts, often predate changes in practice, so that efforts to mon- itor outcomes must track changes in practice. Examples of tension between formal and informal institutions are partic- ularly apparent in the legal sphere, where policies and laws are often weakly enforced and in many cases provide contradictory or incomplete coverage in their protection of marginalized people. For example, while Ethiopia's constitution prohibits violations such as female genital mutilation, wife battering, domestic violence, and sexual harassment, the penal code con- tains no provisions for adjudicating them, and existing laws are often EMPOWERMENT: AN ANALYTIC FRAMEWORK 15 Box 2.2. Persistence of Informal Institutions in Ethiopia Most rural people in Ethiopia continue to apply customary laws to their economic and social relationships. This is most apparent, and perhaps most damaging, in the ways in which customary conflict resolution mechanisms and the civil courts are legally integrated. While in theory this integration was meant to enable citizens to retain their ethnic and religious identities, in practice it has reinforced damaging attitudes and customs toward women. Article 34(7) of the Constitution reserves the option to adjudicate dis- putes related to personal matters in accordance with religious or customary laws, rather than under the civil code, if the parties to the disputes agree. In practice, personal disputes, particularly between men and women, are frequently directed to traditional adjudication mechanisms by the choice of men, without the consent of women. In Muslim areas, if a husband goes to the Sharia court first to institute divorce proceedings, the wife often does not have recourse to the civil court (World Bank 2004b). Focus group dis- cussions among Orthodox Christians in Addis Ketama reported that if there is a conflict between husband and wife, the case is first handled by a traditional court, and noted that even if a formal court was approached directly, the case would be passed to traditional courts (Legovini 2004). Source: Alsop and Kurey 2004. applied by judges in a manner reflecting social norms rather than women's rights (World Bank 2004b). Box 2.2 provides further evidence (also from Ethiopia) showing that the creation and presence of formal institutions does not always mean they operate effectively. It is in understanding, track- ing, and addressing the interplay of formal and informal institutions that change toward empowerment can be better effected. Interaction between Agency and Opportunity Structure: Explaining Degrees of Empowerment The analytic framework introduced here goes beyond income-based and util- itarian approaches to poverty in which "real incomes are presumed to trans- late unproblematically into well-being via utilitarian consumption choices" (Evans 2002, 57). Empowerment is based on tackling the differences in capa- bilities that deny actors the capacity to make transforming choices.As box 2.3 illustrates, it is a dynamic process through which the interaction of agency and opportunity structure has the potential to improve the capacity of individ- uals or groups to make effective choices. This concept has similarities to Sen's 16 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Box 2.3. Assets, Opportunity Structure, and Effective Political Choice A study of local government in India finds that elected representatives who are landless participate in the meetings of local elected bodies, or panchayats, to a significantly lesser extent than those who own some land. As the landless are dependent economically, they are less likely to raise dissenting opinions against their potential employers in the village. Education and access to information also significantly associate with par- ticipation among elected representatives. Every additional year of educa- tion, on average, raises a representative's participation by more than two and a half percentage points. A representative who has 10 years of educa- tion scores on average 27 percentage points higher on this scale than one who has no formal education. Similarly, more access to information is associated with greater participation among representatives by almost 3 percentage points, on average, for each additional source of information that they consult. Respondents in the study stated that individual benefits from the pan- chayat could be accessed only by people who had a relationship with the family of the sarpanch, or panchayat president. Such relationships were based on performing frequent labor for the sarpanch and his or her kin, purchasing goods from shops owned by them, and voting in their favor. An elected representative, or wardpanch, from a low-status tribe said that he had no powers, but that he and other wardpanches "have to go along with whatever the patidars [the caste group of the sarpanch] decide in the panchayat," as many of the wardpanches are dependent on the patidars for labor and therefore for their livelihoods. The people who feel that they are excluded from the individual benefits of the panchayat say they lack the awareness of what to do to change the situation and do not know to whom they should turn outside the panchayat. Source: Adapted from Alsop, Krishna, and Sjoblom 2001. notion of expanding human capabilities and freedoms by focusing on people's ability to "enhance the substantive choices they have" (1997, 1999a). Prerequisite to empowerment is an opportunity structure that allows people to translate their asset base into effective agency, through more equitable rules and expanded entitlements. For example, an individual's human assets might be improved through completion of secondary education, while at the same time new opportunities for citizen participation in budget alloca- tions might open up through the institutionalization of budget planning pro- cessesatthelocallevel.Usingthenewskills,confidence,andknowledgegained through formal education, and taking advantage of the opportunities opened EMPOWERMENT: AN ANALYTIC FRAMEWORK 17 up in the planning process, that person may be empowered to effectively participate in local-level decision making. The iterative relationship between structure and agency has implica- tions for the way that empowerment is measured. When the analytic frame- work is applied, the measurement of assets and institutions provides indirect indicators of empowerment. Direct indicators of empowerment, however, are extremely difficult to find in any national sample survey or poverty monitoring system. They are more common within project monitoring sys- tems, but while attempts have been made to track empowerment outcomes, the indicators used are often rather limited (see appendix 1). Prioritizing indicators used by these systems and embedding them within the discourse on power suggests that three direct measures are important for measuring or tracking empowerment: 1. Whether an opportunity to make a choice exists (existence of choice). 2. Whether a person or group actually uses the opportunity to choose (use of choice). 3. Whether the choice brings about the desired result (achievement of choice). To illustrate, if a policy or project design team were trying to assess the degree of political empowerment of women, it would need to gather infor- mation on (a) whether opportunities for political participation exist, such as whether elections are held; and if so, (b) whether women attempt to vote; and (c) whether they actually vote. If the same team were using the infor- mation to design an intervention to politically empower women, they would then need to ensure that the structures and processes of the intervention were such that these three ends were achieved and monitored. For several reasons, including the geographic, social, or economic posi- tioning of a person or group, the opportunity to make a desired choice may not exist. To take the case of a rural woman in the hills of Nepal who wants to send her daughter to primary school, if a school does not physically exist within walking distance, she has no option. It may not matter that the formal opportunity structure--in this case a policy on education for all-- exists. If the asset of a local school is not present, the opportunity for that woman to make a choice does not exist. The use of choice involves measuring whether a person or group takes advantage of an opportunity to choose. If a school exists, does the Nepalese woman choose to send her daughter there? She may or she may not. If she does not, an analysis of the reason would involve documenting the inter- play between her assets and her opportunity structure. She may choose not to send her daughter to school because her financial assets are insuffi- cient: for example, she cannot afford shoes the child would need to walk to school or she cannot pay the bribe the teacher levies. She may also not use the opportunity because her mother-in-law, with whom she lives, strongly 18 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION feels that girls need no education and that the child is of more use herding the goats. Here there is interplay between informal institutions and eco- nomic assets. Or the reason that the woman does not send her daughter to school may simply be that her husband will beat her if she does--another informal rule coming into play. As these examples show, the framework of agency and opportunity struc- ture helps in understanding which issues need to be addressed in operational work and which factors need tracking in monitoring systems. There is one further dimension of the capacity to use choice that needs to be considered. This is whether the use of choice is direct or indirect. The example of the Nepal- ese woman illustrates a direct use of choice. However, in many situations peo- ple may choose to indirectly use opportunities to express choice--for example, by accepting the legitimacy of an elected representative to engage on their behalf. This could occur in a local-level budgeting or planning exercise where the costs of direct use of choice are too high for individual citizens to bear. The achievement of choice is a measure of how far a person or group is able to achieve the desired outcome. If the woman in Nepal has the option to send her daughter to school, and if she makes the choice to do so, does her daughter actually attend school? If she does not, the analytic framework calls for assessing whether there is something in the opportunity structure that affects the achievement of choice. For example, it might be that this is a low-caste girl and only Brahman and Chhetri (high-caste) children are allowed to attend the school, so the girl was sent home on her first day. The failure to achieve this choice could also relate to the assets of the girl or her house- hold. She may not have the requisite skills to attend a class on her age level, or she may not continue attending because the school requires her to wear a uniform or shoes and her household cannot afford to buy them for her. While these degrees of empowerment capture individuals' or groups' capacity to make effective choice, it is too simplistic to automatically treat these three degrees of empowerment as a continuum, with the final degree-- the achievement of choice--considered as the most desirable degree of empowerment. The subject matter of choice has to be considered in rela- tion to each degree. If, for example, a person in a well-functioning democ- racy chooses not to go to a local council meeting because his or her elected representative is considered effective, this should not be considered to indi- cate a lesser degree of empowerment than if the person participated directly in the meeting. The person has chosen an indirect route of participation-- one that many citizens in established democracies use. In this case, the operation of the formal institutions that influence the person's choice to use an opportunity (the second degree of empowerment) are effective to the point that they result in indirect use of opportunity. Each degree, and the two key groups of factors associated with that degree, have therefore to be considered on their own merit. EMPOWERMENT: AN ANALYTIC FRAMEWORK 19 Where Empowerment Takes Place: Domains and Levels Using agency and opportunity structure to frame analysis of empower- ment is helpful, but leads to two further questions. First, does a person's capacity to make effective choices vary according to what he or she is doing? Second, does empowerment vary according to the level at which a person is acting? The answer to both questions is yes. To illustrate, an Indian woman may experience one degree of empowerment when she is trying to exercise choice over domestic resources within the household, and a different form of empowerment when in a bank trying to access a loan. Her experiences will also differ according to whether she is trying to operate in her village, at a market or office located at a distance from her village, or in a capital city.12 These added complexities in the measurement of empowerment are dealt with by conceptualizing three different domains and three different levels of actors' lives.13 This conceptualization is important to an analytic framework that has to span the multiple political, social, and economic conditions found in dif- ferent countries. It is also important, as discussed in chapter 5, to an oper- ational approach that must identify the most effective points of intervention for empowerment. As the following discussion illustrates, conceptualiza- tion of empowerment and its determinants as operating in three domains and at three levels can help to refine tracking, analysis, and operations in any country. Domains The framework identifies three domains: · State, in which a person is a civic actor · Market, in which a person is an economic actor · Society, in which a person is a social actor. The three domains are further divided into eight subdomains: · The domain of state is divided into the subdomains of justice, politics, and public service delivery. · The domain of market is divided into the subdomains of labor, goods, and private services (for both production and consumption). · The domain of society is divided into the subdomains of intra-household and intra-community.14 These should be treated as opportunities to explore relations within the household and within the community. In certain con- texts it may be necessary to refine or add to these subdomains. For exam- ple, an extended family, ethnic group, or caste group may be critical subdomains in some cultures. 20 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION In each of these subdomains, the individual or collective actor experi- ences a certain degree of empowerment. This is likely to vary between people or groups and will also vary according to whether an actor is in the position of a provider (supply side) or client (demand side). In the state domain, for example, different citizens and their organizations may experience very different degrees of empowerment in terms of access- ing justice, participating in politics, or accessing social services. In India, a well-educated, high-caste man with good social connections would likely experience a higher degree of empowerment in all three state subdomains than his low-caste, illiterate counterpart. In a perfect or "complete" market domain, everyone plays by a set of equitable and transparent rules with highly efficient outcomes for the par- ties involved in a transaction (Rajan 2004). However, differences in control over resources and information, a lack of contract enforcement, or the abil- ity to distort or control market prices through monopolistic or monopson- istic practices can result in highly inefficient and unequal outcomes. Hence, a purchaser may enjoy a marketing monopoly or control price informa- tion, thus forcing producers to accept below-market prices. Conversely, a single supplier or producer (or collective of producers) may be able to dom- inate and set prices above the level that could be achieved within a com- petitive market. In the social domain, social norms will combine with local implemen- tation of formal institutions to affect the choices available to individuals and social groups. A son in an Indian household, for example, is likely to experience a higher degree of empowerment than a daughter; yet, in her community, a high-caste daughter would experience a higher degree of empowerment than the daughter of a low-caste household. While no prior assumption can be made about how empowerment in any one domain or subdomain relates to empowerment in some other domain or subdomain, the degree of empowerment in one domain may well correlate with a similar degree of empowerment in another domain. For example, an individual who is severely disempowered in one domain, say the market, may also be simultaneously disempowered in another domain. Equal market opportunity might be denied to this person, and relations with the state might be repressive or exploitative. In this way, institutional constraints on agency can be mapped onto the empowerment framework's domains. For example, in each domain one can ask questions about the influence of institutions on agency: · In the state domain, do political processes allow people a political voice? Are the judiciary and the police force independent and able to enforce contracts and apply sanctions? Has the state created the regulatory and policy environment for accessible and high-quality service provision or accountability of those in charge of state budgets? EMPOWERMENT: AN ANALYTIC FRAMEWORK 21 · In the market domain, do regulations and systems of property rights allow access to markets, land, labor opportunities, and credit? · In the social domain, to what extent is the agency of individuals and social groups constrained by power-based institutions of social hierar- chies, patriarchy, or exploitative patron-client relations? Levels People experience domains and subdomains at different levels--macro, intermediary, and local. For ease of analysis, a level is defined as an admin- istrative boundary. These levels are common to most countries. For exam- ple, in Ethiopia, the macro level would correspond to the federal, the intermediary to the woreda, and the micro to the kebele or village. In Nepal, the macro level could correspond to the national, the intermediary to the administrative boundaries of a district, and the local to the jurisdiction of a village development committee. In India, where the vast size of the coun- try means that states are extremely important administrative units, the macro level could correspond to the state, the intermediary to the district, and the local to the village. In India it may also be necessary to add a supra- macro, federal level. Another feature that can be generalized across countries is the distance of administrative boundaries from the individual or group. The local level will consist of the immediate vicinity of a person's everyday life. This is likely to be an area contiguous with that person's residence. The intermediary level will consist of a vicinity that is familiar to the person but that is not visited on an everyday basis. This is likely to be the level between the res- idential and national levels. The macro level will consist of a vicinity that is farthest away from the individual, usually the national level. Acertain degree of empowerment at one level does not necessarily reflect the same degree of empowerment at other levels.As research demonstrates, individuals or communities empowered at the local level are not necessar- ily empowered at the intermediary or macro level (Fox 1996; Moore 2001).15 The Framework Summarized In sum, empowerment can be assessed within different domains of a person's life (the state, the market, society) and at different levels (macro, intermedi- ary, and local). Each domain can be further divided into subdomains, which will indicate where and in what areas of their lives actors are empowered. At the intersection of the domains and levels, a person can experience dif- ferent degrees of empowerment, addressing the issues of whether and to what extent the person is empowered. Two clusters of interdependent fac- tors associate with the different degrees of empowerment an individual or group experiences: the agency of the actor and the opportunity structure 22 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO INTERPRETATION within which that actor operates.Analysis of agency and opportunity struc- ture helps explain why an actor is empowered or not, and to what degree. Table 2.1 presents a summary matrix. As the examples above demon- strate, data do not have to be collected, analysis undertaken, or action ini- tiated for all the domains and levels that are included in the table. Nor will a given intervention necessarily address them all. Rather, as chapter 4 will demonstrate, the contents of the framework can be selectively applied in dif- ferent contexts. The number of domains and levels considered depends on the nature and objectives of the development intervention, the purpose of the measurement exercise, and the identities of the key stakeholders, whether suppliers or clients. Empowerment and Development Outcomes As discussed in the introduction, this analytic framework for empowerment recognizes the intrinsic value of empowerment, but also emphasizes its instru- mental role in improving development outcomes. Clearly, demonstrating a causal link between empowerment and development outcomes would pro- vide strong incentives for fostering empowerment in interventions ranging from the policy level to individual projects. Though not always adequately addressed in policy or practice, the associ- ation between social identity, the degree of power a person or group experi- ences,andpovertyoutcomesnowhasalegitimateplaceindevelopmentpolicy making and analysis (Du Toit 2003; Eyben 2004; Moore 2001). The World Bank first gave organizational recognition to this association with the publication of World Development Report 2000/2001, and followed this with Empowerment and Poverty Reduction: A Sourcebook (Narayan 2002). The first, launching the World Bank's strategy to attack poverty, emphasized three main themes, of which empowerment was one. It articulated the link between empowerment and development outcomes, and in particular poverty reduction, a line of analysis that had not previously been associated with the World Bank. But is empowerment instrumental in reducing income poverty? Stern, Dethier, and Rogers (2005) contend that empowerment is a driver of growth. Sen (1997) argues that capability expansion generates both economic pro- ductivity and social change, citing empirical studies that demonstrate a relationship between expanded female education, reduced gender inequal- ity within the household, and reduced fertility rates. However, before it can become powerful in development discourse, particularly at a policy level, the hypothesis that empowerment is a means toward progressive governance and poverty reduction has to be proved empirically. There is growing evidence that empowerment is instrumental in reduc- ing income and consumption poverty. But the empirical evidence associating empowerment with positive development outcomes has been strongest in the political sphere. A recent cross-country study on social accountability Table 2.1. Summary of the Analytic Framework Determinants and outcomes Macro level Intermediary level Local level Domain Subdomain A OS DOE A OS DOE A OS DOE State Justice Politics Public service delivery Market Labor Goods Private services Society Intra-household Intra-community Source: Authors. Note: A = Agency. OS = Opportunity structure. DOE = Degree of empowerment. Agency is measured through endowment of psychological, informational, organizational, material, financial, and human assets. Opportunity structure is measured through presence and operation of informal and formal rules. Degree of empowerment is measured through presence of choice, use of choice (direct or indirect), and effectiveness of choice. 23 24 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION and government effectiveness demonstrates a significant empirical rela- tionship between voice/accountability and effective public institutions that deliver efficient and equitable public services (Kaufmann, Kraay, and Mastruzzi 2004). At the national level, in India, higher voter turnout and lit- eracy rates serve to increase social spending and public food distribution (Besley and Burgess 2002).Arecent poverty assessment in Ethiopia revealed that better access to information through radio ownership had a signifi- cant positive marginal impact on poverty levels (box 2.4). The five country studies in part 2 of this volume recognize the intrinsic value of empowerment, but all try to go beyond that. Although each does so in a different context and in different ways, they all explore the instru- mental role of empowerment and, with differing degrees of success, attempt to link increases in empowerment of individuals or groups to positive changes in development outcomes. One of the clearest examples of efforts to prove the instrumental value of empowerment can be found in the country study from Nepal (chapter 11). That study has as one of its main objectives to understand the relationship between empowerment and other development outcomes. The working hypothesis is that higher levels of empowerment and social inclusion among women belonging to traditionally excluded castes are associated with (a) primary positive changes such as reduced domestic violence, less restric- tion and public intimidation of women, less violence toward low castes, and greater self-esteem and community influence among previously excluded groups, and (b) secondary positive changes such as improved health-seeking behavior and greater contraceptive use by women, increased female involvement in the market economy, and higher incomes for women. Though the study was still under way in 2005, it so far suggests that the communities where cooperation and inclusion of women from traditionally excluded castes are greatest--and where women's groups themselves have been most inclusive and effective--are also the communities where the World Bank­funded water and sanitation project has achieved the best results. In contrast, the most polarized communities, dominated by a single powerful group, are those where the project has been least effective in bring- ing sustainable drinking water to the community. The other country studies come to similar conclusions about the corre- lation between empowerment and development outcomes. For example, the Brazil study (chapter 7), which looks at the empowering effects of participa- tory budgeting on citizens and local government, also examines the hypoth- esis that empowerment can have direct tangible developmental benefits. The study finds that an increase in budget participation in many of the munic- ipalities translated into better overall governance--that is, participatory budgeting effectively empowered people. This is attributed both to the information returns associated with participation and to the increased meas- ure of accountability.16 Perhaps more exciting evidence of the robustness of EMPOWERMENT: AN ANALYTIC FRAMEWORK 25 Box 2.4. Poverty-Reducing Effects of Access to Information in Ethiopia The proportion of rural households owning a radio was estimated at only 12 percent in rural Ethiopia in 1999, which is the lowest in Sub-Saharan Africa and about three times less than the average for that region.a Interestingly, there appear to be striking poverty-reducing effects of providing access to information through increased access to a radio, as well as important externality effects of radio ownership (that is, people without a radio experience a positive effect when others in their commu- nity have a radio). Using multivariate regression techniques, we find that in rural areas of Ethiopia households with a radio are 17.5 percent richer than those without, and that a 10 percentage point increase in the propor- tion of households owning a radio in a community increases average household consumption in that community by 3.9 percent. As expected, given much wider dispersion of individual radio owner- ship, the externality effect is lower in urban communities. Given that only 12 percent of rural households in the country own a radio, there is clearly tremendous scope for reducing poverty by enhancing access to informa- tion, especially given the current isolation from the outside world. It is estimated that providing the poorest quintile of the population with a radio would increase average consumption by 5 percent, would reduce poverty incidence by 11 percent or 4 percentage points, and would reduce the poverty gap by about 40 percent. This impact is over and above peo- ple's wealth position as captured by their possession of livestock, farm equipment, and consumer durables, and their quality of housing. Moreover, this effect is in addition to the effect of other public infrastruc- ture (electricity, access to roads, health centers, markets, water and sanita- tion) as well as the overall wealth position of the communities. Source: World Bank 2004b. a. Results are from Demographic and Health Survey (DHS) data for 1995­2000 from selected Sub-Saharan African countries--that is, nationally representative household surveys that are comparable and standardized. Even in the urban population, radio and television ownership is among the lowest in Africa. the relationship is the outcome of econometric analysis that proves that participatory budgeting as a mechanism of empowerment not only results in empowerment but also has a strong and positive association with the reduction of extreme poverty. Understanding Empowerment: Summary This chapter has drawn on historic and recent discussions of the substance and nature of power to construct a framework that can be used to meas- 26 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION ure empowerment and to understand the operational needs of interven- tions designed to bring about empowerment. Two groups of factors, referred to as agency and opportunity structure, have been identified as associat- ing with empowerment, and the idea of different degrees of empowerment has been put forward. Because the empowerment of individuals and groups varies according to context, the framework also suggests that analysts and designers of interventions need to assess the level at which a person or group is empowered, from the local up to the macro. They must recognize that people may be empowered in one or more domains of life--as civic actors in the state domain, as economic actors in the market domain, or as social actors in the societal domain. Finally, the role that people play in each domain will affect all factors. In the state domain they can be state officials or ordinary citizens; in the market domain they can be suppliers or clients; in the social domain people with different attributes will have different assets and opportunity structures. While it is increasingly clear that empowerment can play an instrumental role in improving development outcomes, more research is needed on the specific entry points. In terms of asset building and institutional reform, this means identifying which interventions are likely to have the highest impact on increasing choice and translating increased choices into improved outcomes. What types of price-based or rule-based reforms, for instance, will encourage institutional changes that create greater equality of oppor- tunity? Similarly, what types of asset investment or redistribution are most likely to increase the bargaining or decision-making strength of individu- als and groups in different contexts?17 Theoretical and empirical insights from complementary research traditions, as discussed in the introduction to this chapter, will no doubt play an important role in this continuing research effort. The next chapter picks up on some of the themes raised above and exam- ines the methodological challenges associated with applying this frame- work to information gathering and analysis. Notes 1. Key among these authors are Foucault (1980), Giddens (1984, 1993), Gramsci (1971, 1978), Lukes (1974), Parsons (1937, 1963), and Weber (1904, 1905). 2. In addition to the authors mentioned above, the development of this framework owes muchtotherecentworkofBennett(2004),Bourdieu(1977),Clegg(1989),Kabeer(1999),Krishna (2003), Malhotra, Schuler, and Boender (2002), Sen (1985, 1992), and Petesch, Smulovitz, and Walton (2005). Readers are referred in particular to the Bennett, Giddens, Lukes, and Petesch publications for a discussion of the theoretical underpinnings of this framework. 3. As the figure indicates, agency and opportunity structure and degrees of empower- ment are assumed to be in a reciprocal relationship. Similarly, enhancements in a person's EMPOWERMENT: AN ANALYTIC FRAMEWORK 27 degree of empowerment are expected to lead to higher quality and quantities of assets and more supportive opportunity structures. 4. However, while anecdotal and case evidence--including that found in part 2 of this book--suggests an instrumental purpose in empowering people, statistical data demonstrat- ing a clear association between empowerment and development outcomes remain limited. The association between indirect indicators of empowerment--agency and opportunity struc- ture--and development outcomes is well documented. However, because of a paucity of representative data on direct indicators of empowerment and the statistical problems associ- ated with analysis of multiple dimensions, the authors recognize the need for further analysis to test the relationship between empowerment and development outcomes. In part, this is strategic--statistics are persuasive in policy discourse--and in part it is simply a matter of rigor. The validity of too large a proportion of the nonrepresentational analysis can be questioned. 5.Alkire (forthcoming) surveys a series of subjective measures of agency, that is, measures that capture people's self-evaluation of whether or not they are free to act as agents. There is also a rich literature available on measuring social capital (see www.worldbank.org/socialcapital). Social network analysis predates these more recent attempts to measure social capital and is arguably a better approach (Krebs 2005). Appadurai (2004) was the first to clearly articulate the idea of a "capacity to aspire." 6. Nussbaum (2000) argues that rational choices are "deformed" by underlying differ- ences in capabilities, which include the capacity to aspire and to imagine alternative options. 7. Toddy is an alcoholic drink brewed from palm tree sap. Toddy tappers cut incisions in the trunks of these trees and collect the sap for fermentation. In high Hindu culture, the consumption of alcohol is traditionally frowned upon, giving the police ample license to abuse the tappers. 8. Research in India found that the addition of one more source of information to a per- son's repertoire increased participation in local-level governance by more than 5 additional percentage points (Alsop, Krishna, and Sjoblom 2001). 9. Behavior reflects an individual's or group's attitudes. Attitudes, in turn, reflect a belief in "what is right and possible" and result in part from the interaction between assets and rules. Attitudes are an expression of psychological assets, and those psychological assets result in part from a combination of other assets--such as information, education, and social capital--and from the formal and informal rules of the game. 10. There is an important distinction between institutions defined as the "rules of the game" and organizations defined as groups of individuals bound by a common purpose, subject to a defined set of authority relations, and dedicated to achieving objectives within particular rules of the game (North 1990; Uphoff 1986). 11. However, without the force of the constitution, women would have experienced a much longer and harder fight to enter the political arena and operate effectively there. 12. In this example, one Indian woman may well experience different degrees of empower- ment from another. These differences can largely be explained by assets, such as education, information, and social capital, and by opportunity structure, such as social norms of behavior associated with caste and gender or formal rules allowing access to loans, markets, or services. 13. The concept of domains was originally developed and tested by Schuler and Hashemi (1994). In their work on women's empowerment and use of contraception in Bangladesh, they 28 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION identified seven domains of empowerment: income; employment; physical mobility; aware- ness of political life; involvement in political life; physical violence; and reproductive behavior. 14. There is considerable debate on the appropriate unit of measurement within the social domain (see, for example, Guyer and Peters 1987). In this volume it is recommended that analysis in this domain focus on intra-household and intra-community relations. This is essentially a pragmatic decision. The core debate in this area focuses on whether the unit of analysis should be the household or the family--but, as O'Loughline (1999) and others sug- gest, given that most major surveys work with the household as the unit of data collection, the household remains the unit most appropriate for policy-oriented data collection (Besley and Burgess 2002; A. Evans 1991; Folbre 1996; Kabeer 1994). This does not imply acceptance of a household production function. Rather, emphasizing the intra, it indicates that data collection and analysis needs to be undertaken on household members' characteristics and-- depending on the subject matter of a survey--on the relations between those members. 15. The same will apply at the international level. For example, coffee growers in West Africa will experience different levels of empowerment in international markets than in national or local markets. 16. More participation should in effect lead to scarce resources being targeted where they are most needed and to a reduction in the incidence of rent seeking. To the extent that partic- ipatory budgeting promotes such participation, therefore, it can positively impact develop- mental outcomes. 17. Research using a game theory approach has shown that women who have rights to property, for example to land, occupy a stronger bargaining position in their personal relation- ships and suffer less domestic violence (Bina Agarwal, personal communication). 3 Methodological Issues in Measuring Empowerment Poverty reduction activities traditionally focus on providing resources and services.Afocus on empowerment brings an additional emphasis on enhanc- ing people's choices and opportunities. The analytic framework introduced in this book is specifically concerned with the relationship between institu- tions as sets of rules and individuals or groups as civic, economic, and social actors or agents. Empowerment as an intervention involves strengthening people's asset-based agency on the one hand and changing the institutional rules that shape human behavior and interaction on the other. Increases in assets--such as information or credit--can improve the basis from which an agent bargains. Institutional change can empower by creating new sets of rights and obligations, altering incentives and sanctions, and lowering the economic and social costs of expressing choice--that is, it can create greater "equality of opportunity" (World Bank 2005c) for those who are tradition- ally disempowered. Measuring empowerment allows for changes in the availability and exer- cise of choice to be tracked over time and compared across populations. This enables, first, analysis of the relationship between empowerment and other developmental goals. Although empowerment is now seen as a legitimate developmental goal in its own right, there is a growing body of anecdotal and case study evidence, including some of the country studies in this volume, to suggest that empowerment also brings improved poverty reduction and other development outcomes. However, robust analysis on the positive associa- tion between empowerment and development outcomes is far from wide- spread. Continued efforts to measure empowerment therefore allow for the relationship between changes in empowerment and other development out- comes to be further tested. Second, beyond testing these associations, moni- toringandevaluationofempowermentimpactsgeneratesthedataandanalysis necessary for improved policy design and interventions. This chapter briefly reviews some of the challenges involved in mea- suring empowerment, drawing on existing literature and on the five coun- try case studies. It then shows how empowerment indicators that address these challenges can be generated. Finally, the chapter examines the use of 29 30 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION empowerment indicators as part of a mixed-methods approach to describ- ing and explaining trends and patterns in empowerment, and discusses the limitations of current analytic approaches. Challenges to Measurement To date, there have been relatively few attempts to provide the means of assessing whether policy interventions are having an empowering effect.1 Beyond aggregate measures of concepts such as rule of law, accountability, and corruption, data on and statistical analysis of empowerment are at an early stage of development (Stern, Dethier, and Rogers 2005). This is true even though some themes, such as women's empowerment, have been sub- jected to more frequent attempts to measure change (Kabeer 1998). This is not surprising given the challenges involved. Poverty is usually measured by using an approach based on per capita income or consump- tion that measures "money-metric" outcomes. But an approach to measur- ing and analyzing empowerment has to capture dynamic processes and relational changes that are less predictable, less tangible, more contextual, and more difficult to quantify in data collection and analysis (Graham and Pettinato 2005; Malhotra, Schuler, and Boender 2002; Uphoff 2005). Furthermore, while poverty measurement is applied to individuals, or to households as aggregate units, the process of empowerment often requires the collective expression of choice (Kabeer 1999). These challenges are trans- lated into methodological hurdles relating to issues of meaning, causality, andcomparabilitythatwediscussbelow(Bastagli, Coudouel, and Prennushi 2004; White 2005; Markus Goldstein, personal communication). The challenge of identifying meaningful measurements of any dimension of poverty requires, in the first instance, neatly capturing the essence of that aspect of poverty. Rates of malnutrition in children under age five or pri- mary school net enrollment rates are unambiguous indicators for the Millennium Development Goals of eradicating hunger and achieving uni- versal primary education, respectively. Beyond capturing what a dimension of poverty means, indicators are only useful if they allow us to observe a change in that dimension that is meaningful in both its direction and mag- nitude. The direction of change must be consistent, so that when measur- ing progress, more (or less) of the value of the poverty dimension is invariably better. Identifying the magnitude of this change then allows us to state "how much" better that dimension has become. So, if under-five malnutrition consistently declines, and does so by increments of 5 percent, then we can conclude that there has been meaningful change.2 In the same way, indicators of empowerment should allow measurement of empowerment characteristics if we are to conclude that a person or group is "more empowered" and specify how much more empowered that person or group is.Attempts to characterize empowerment have been hampered by METHODOLOGICAL ISSUES IN MEASURING EMPOWERMENT 31 a lack of conceptual rigor in conferring meaning on different types of choices. For example, Kabeer (1999) points out that indicators used to measure women's access to and control over resources have often failed to distin- guish between different resources according to their degree of strategic value to women in a given context. Part of the problem lies in the multidimen- sionality of empowerment. Different dimensions require different metrics, making the task of high-quality statistical analysis difficult. Some argue that for these reasons, sophisticated statistical analysis of empowerment is inap- propriate and analysis should therefore focus on simple descriptive statis- tics and narrative reporting. However, as this chapter will make clear, there are good strategic reasons for pushing statistical analysis as far as possible, and there are equally good reasons for using mixed-methods approaches as standalone exercises or to enhance large-sample survey work. The second major challenge is that of attributing causality, or making a linkbetweenadevelopmentinterventionandtheoutcomesandimpactshoped for. Attributing cause and effect can be relatively straightforward when the causal chain is fairly short and when other variables can be held constant, such as when considering the effect of an immunization campaign on the incidence of disease in a target population. Changes in power relations, how- ever, are not single-event outcomes, but are dynamic and process-based, tied up with bargaining, cooperation, conflict, co-option, rent seeking, and other forms of contracting. In short, empowerment is messy. The picture is further complicated by the cross-sectoral nature of empow- erment, which can happen (or not happen) in different ways in different domains and at different levels. Individuals or groups may have different experiences of empowerment in society, where they are social actors; in the market, where they are economic actors; or in the state, where they are civic actors. Their experiences also may differ depending upon whether they are on the supply or demand side of an activity. Changes in power among indi- viduals and groups may occur rapidly at the micro level, for example, and yet be blocked by institutional resistance or inertia at the intermediary and macro levels. Women who are increasingly able to exercise their sexual and reproductive rights within the household, and who are increasingly mobile and visible in their communities and more active in community decision making, for example, may still be denied political choice or access to justice at the macro level and may continue to face a very limited choice of jobs in the labor market. Civil servants in charge of project benefits may be in a strong position to solicit bribes from intended beneficiaries, and providers of marketing services--such as middlemen for agricultural products--are likely to pay producers well below market rates in locations where they have a monopoly. The potential problems of endogeneity are therefore con- siderable in any statistical analysis of empowerment. Additionally, choice can often be implicit and therefore difficult to observe and measure. Indeed, empowerment often allows people to choose not to 32 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION take action. Even when people do choose to take action, it will be difficult to determine, for example, that this is not a strategic or dependent form of action. In other words, how can a definitive statement be made as to whether people are really empowered? The third challenge is to infer comparability across populations. "Aggre- gation" describes whether the data generated can be aggregated across pop- ulations so that conclusions about impact and change can be inferred for larger population groups. Here the major problem is that empowerment often involves relative rather than absolute changes in states of being. Hence, an observable move toward a higher state of empowerment for one person or group cannot be assumed to apply to other individuals or groups. This holds both within countries and across countries. The economic empowerment of a low-caste, poor male villager in India, for example, is likely to be qualitatively and quantitatively very different from the eco- nomic empowerment of a high-caste urban male. Similarly, social empow- erment for women in Western Europe or North America is very different from social empowerment for a woman in an African village. The signifi- cance of context can vary across time as well as space. This creates problems for longitudinal tracking of empowerment, particularly when such change is rapid. Data from the early 1990s on contraceptive use in rural Bangladesh suggested that empowered women were more likely than others to use contraception. Now, however, contraceptive use is the norm and is there- fore unlikely to indicate a woman's empowerment through the exercise of strategic choice (Malhotra, Schuler, and Boender 2002). As the five case studies in part 2 of this document demonstrate, these chal- lenges are considerable. Some have been more or less overcome, whereas others remain issues in the country case analyses in part 2--thus raising questions about the results of the studies. This serves as a sharp reminder both of the complexity of this field of analysis and of the fact that this frame- work and the case studies applying it still constitute only a first step toward operationalizing empowerment and making it measurable.3 In that way, the methodologicalapproachintroducedbelowisbasedonthepremisethatempow- erment cannot be measured in a way that does justice to its inherent com- plexity and that satisfactorily meets these three criteria of meaning, causality, and comparability. What can be done, however, is to identify measurements that capture, albeit imperfectly, important dimensions of changes in power, and that can be complemented by more interpretive and explanatory forms of qualitative research. Toward a Methodology for Measuring Empowerment The methodological challenges discussed above crystallize around the choice ofspecificindicatorsofempowerment.Thissectionfirstsuggestsandillustrates METHODOLOGICAL ISSUES IN MEASURING EMPOWERMENT 33 a typology of indirect and direct indicators of empowerment that map onto the empowerment framework introduced in chapter 2. These indicators can be reviewed in conjunction with the Empowerment Base Survey Module that is included as appendix 6. The discussion of indicators is followed by a discussion of the added value of a mixed-methods approach to measur- ing empowerment that draws on the comparative advantages of both survey-based and interactive research methods. Identifying Empowerment Indicators Chapter 2 distinguished between indirect and direct measures of empow- erment. Empowerment can be measured indirectly by measuring either asset endowments or opportunity structure. With respect to asset endow- ment, existing survey instruments modeled on the World Bank's Living Standards Measurement Survey (LSMS) generate plenty of data, includ- ing measures of human capital, social capital, and access to productive assets. Table 3.1 provides some examples of indicators of asset endowment. With respect to opportunity structure, in selecting indicators it is impor- tant to recognize the gulf that often exists between the presence and the operation of institutions. In Ethiopia, for example, the government has sought to address the opportunity structure for women's empowerment by making a commitment to gender equality through the National Policy on Women (1993) and by enacting a new constitution (1995) that repeals discriminatory laws. Yet these institutions are poorly enforced, with no Table 3.1. Indirect Indicators of Empowerment: Asset Endowment Indicator Existing sources Asset endowment (sex-disaggregated) or instruments Psychological assets Capacity to envisage IQMSC change Informational assets Frequency of radio IQMSC listening/access to different media sources Organizational assets Membership in IQMSC organizations Material assets Ownership of productive LSMS assets such as land Financial assets Value of household savings Household Budget Survey in last year (cash and other forms) Human assets Literacy level LSMS education module Sorce: Authors. Note: IQMSC = Integrated Questionnaire for the Measurement of Social Capital. LSMS = Living Standards Measurement Study. 34 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION provision in the penal code for adjudicating them and a tendency among judges to adjudicate in ways that do not take account of women's rights. International agencies such as Freedom House have identified indicators and developed indexes that track progress in the operation of institutions, particularly in the state domain. These and other national indicators can meas- ure the functioning of institutions to ensure and protect economic, politi- cal, and social freedoms. Table 3.2 presents some examples of indicators for measuring the operation of (not the presence of) empowering institutions. In addition to measurement of assets and institutions, it is also possible to gather data, or information, on direct indicators of empowerment. This requires identifying indicators that can measure degrees of empowerment at the interface of agency and institutions, as discussed in chapter 2. Three degrees of empowerment can be measured directly: 1. Existence of choice: There is an opportunity to make a choice. 2. Use of choice: A person or group takes the opportunity to make a choice (either directly or through representation). 3. Achievement of choice: The choice is transformed into desired outcomes. Table 3.3 presents examples of concepts or themes relating to degrees of empowerment in each of the domains introduced in chapter 2, and then sug- gests indicators to measure these concepts. Two types of relevant data can be elicited using a household survey instrument. Information on observable phenomena is produced when respondents recall the frequency and types of specific, concrete actions they have taken; this generates indicators with "face validity" (Malhotra, Schuler, and Boender 2002). Perceptual information is produced when respondents apply a score to their own evaluation of the qualitative dimensions of those interactions; this generates indicators that recognize the intersubjective nature of the empowerment process (Kabeer 1998). So, for example, to generate indicators of accessible justice, respon- dents might be asked to recall the type and frequency of their use of dif- ferent aspects of a justice system and then to score the quality of those in terms of their effectiveness, inclusiveness, and efficiency. Perceptual data can also be captured using interactive or participative techniques of data col- lection. These techniques will usually raise the quality of information above that which can be obtained using a questionnaire, but they require both more time to engage respondents and higher-level interviewing skills. In the context of expost monitoring, use of recall is a necessary approach to determining impact. Reliance on recall is, however, a contentious issue because of the opportunities for confusion, inaccuracies, and respondent bias (White, personal communication). Because respondents tend to "telescope," that is, to think things happened more recently than they really did, it is important to make sure that the reference period is clearly established. If pos- sible, a well-known event from the reference period should be identified. Table 3.2. Indirect Indicators of Empowerment: Operation of Institutions Domain Indicator Existing sources or instruments State (justice) Index of civil liberties Freedom House State (political) Index of political rights Freedom House State (public service delivery) Percentage of budget allocation in line with PRSP PRSP policy matrix Percentage of population unable to access at least one basic World Bank Country Policy service in the previous year due to (a) cost, (b) physical distance, and Institutional Assessment (c) social distance Market (labor) Percentage of employers complying with state regulations on -- core labor standards Market (goods) Access to and use of productive asset ownership LSMS-type survey by income quintile Market (private services) Percentage of women or ethnic or religious minorities accessing -- specified financial services in previous year Number of formal transparency and accountability mechanisms -- for financial service providers Society (intra-household) Number of formal justice cases filed against violators of women's World Bank Country Policy rights legislation (domestic abuse) per year and Institutional Assessment Society (intra-community) Exclusion from community associational life based on social -- identity Source: Authors. Note: PRSP = Poverty Reduction Strategy Paper. LSMS = Living Standards Measurement Study. -- Not available. 35 36 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Table 3.3. Direct Indicators of Empowerment Domain Concept or theme Indicator State (justice) Accessible justice Recall data on types of justice Frequency of use of justice system used, frequency of use system Perceptual scoring of fairness of Fairness of justice system treatment and outcomes, Ability to complain about social difference in treatment, justice system accountability, ease of use Safety and security of citizens State (politics) Participatory democracy Recall data on frequency of Critical and independent elections at different levels, voting voting choice entitlements, voting behavior Use of accountability (including independence of mechanisms decision making) Perceptual scoring of interest in elections, knowledge of parties, involvement in political processes (including aspirations), fairness of electoral process, accountability of elected officials State (public Citizen voice and social Perceptual or recall data on service accountability serviceavailability, accessibility, delivery) Accessibility, quality, and making complaints relevance Recall data on services used Perceptual scoring of quality, accessibility, and effectiveness of complaints (distinguishing by social group) Market (labor) Freedom to enter labor Perceptual scoring of ease of entry market (skill-based) and movement within labor Freedom to withdraw or market withhold labor Perceptual data on constraints to Ability to command market above price for labor Recall data on union membership Perceptual data on labor rights Market Ability to access productive Perceptual data on access to (goods) inputs productive inputs Ability to access Recall data on consumer behavior consumption goods Market Access to and use of formal Recall data on credit access and (private or informal credit credit use services) Access to infrastructure Recall data on use of transport, Access to information telecommunications METHODOLOGICAL ISSUES IN MEASURING EMPOWERMENT 37 Table 3.3 (continued) Domain Concept or theme Indicator Recall data on use of different media and information sources Society (intra- Influence and control over Perceptual data on household household) strategic choices decision making Control over body and Perceptual scoring of control reproductive choices over body Society (intra- Freedom of association Perceptual data on constraints to community) Freedom of mobility association Participation in community Perceptual data on mobility decision making Recall data on decision making in community Source: Authors. Note: Indicators are disaggregated by social and economic group in analysis. Literature on the accuracy of recall data suggests that after a period of 24 hours, the accuracy of recall decreases.Accuracy can be improved, but only to a limited extent, by attaching the recall period to a key event in the respon- dent's life. The issue of bias stems from the tendency of respondents to think that things used to be better than they are now. There is no easy way to address this, although triangulation of responses between respondents can increase trustworthiness. It may sometimes be safer just to ask about the change rather than to attempt to compare two points in time. It is also important to note that indicators can be either objective or hypo- thetical. For example, respondents can be asked about their actual experi- ence with legal process or law enforcement, or they can be asked how they think they would fare hypothetically. The Empowerment Base Survey Module (appendix 6) contains both types of questions, because respondents with no experience in a particular area cannot be asked to evaluate that experience. However, subsequent analysis cannot rest only on those who have had such experiences, both because the sample size may be small and, impor- tantly, because of sample selection bias if one does so. Sequencing methods and data It is important to avoid reducing the measurement and analysis of empow- erment to a simple set of objectively verifiable indicators. The empower- ment framework focuses on an active and iterative relationship between agency and opportunity structure. It recognizes that there is an internal dynamic between assets, as there is between institutions. And it allows for the contextual nature of power relations, which militates against a standard- izing rationale of indicators. 38 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Through careful sequencing of interactive and extractive methods of data collection, which correspond roughly but not completely to the collection of qualitative and quantitative information, emerging trends and patterns in empowerment can be probed and explained. The added value of mixed- methods approaches to evaluating development interventions is now well established (Rao and Woolcock 2003; Baker 2000; Bamberger 2000; Carvalho and White 1997). Given the conceptual complexity of empowerment, the value of a mixed-methods approach to measuring and tracking empowerment becomes even more evident. Standard survey instruments at the national level are powerful tools for normalizing and tracking changes in empowerment at a broad macro level. But the contextual sensitivity of empowerment makes aggregation through standardization problematic, as discussed above.4 Furthermore, these broad- coverage methods when used alone are rarely adequate for in-depth under- standing of subjective and often contextual decision-making behavior. Interactive methods of research that generate both qualitative and quan- titative data can be applied to a subsample of sites that are linked to the survey data, thereby showing how typical or atypical those selected cases are.5 Inter- active methods enable a better understanding of the context in which data are generated, providing in-depth insights into the social relations that under- pin power and into the value systems that frame effective choice.6They allow researchers to probe, for example, the way people consider the trade-offs and costs involved in making choices, the implicit nature of empowerment when effective choice is an option but is not exercised, and the relationship between individual empowerment and collective empowerment. Among the country cases used for this study, the explanatory depth of interactive methods is demonstrated most powerfully in Woolcock and Gibson's analy- sis of conflict mediation in the context of community-driven development projects in Indonesia (chapter 10). However, as reflected in debates over the quantifiable nature of qualitative information, the terminology used to describe mixed methods is often mis- leading. For clarity's sake, practitioners using mixed methods can increase efficiency by differentiating between (a) the type of information being gath- ered, (b) the instruments or approaches used for collection of different types of information, (c) ways of recording information, and (d) approaches to analy- sis of information. Furthermore, additional work on presentation of non- quantified information could increase the effectiveness of communicating with users of the analysis. For example, see the presentation of narrative information and analysis in the Brazilian case (chapter 7). Information can be quantitative, that is, expressed in direct and discrete numeric units such as the number of days or a quantum of finance. Or it can be qualitative, that is, not directly numeric--such as descriptions of degrees of feeling or explanations of a phenomenon or experience. As documented in the huge literature on participatory techniques of data and information METHODOLOGICAL ISSUES IN MEASURING EMPOWERMENT 39 management, qualitative information in many instances can be effectively quantified and can be used for poverty analysis and interventions (Barahona and Levy 2002).7 The instruments and approaches used for gathering information range from standard questionnaires to semi-structured interviews and participa- tory appraisal tools (such as wealth ranking, mapping, diagramming, and so on) to anthropological approaches that include ethnographies and inter- active dialogue with respondents. Each approach to information gathering requires a corresponding method of recording information. Questionnaires are a simple tool in which infor- mation gathering and recording use the same document. Systematic record- ing from participatory appraisal tools has often proved more difficult, as scales tend to differ in diagramming and mapping and those involved in data col- lection can be diverted into discussion surrounding the exercise rather than focusing on quality recording. Semi-structured interviews using checklists also often suffer from poor recording; as with participatory appraisal tech- niques, it is worth spending time prior to fieldwork thinking through how to systematically record the information for ease of analysis. Moving away from semi-structured interviews and into more anthropological approaches to information gathering increases the difficulties of systematic recording and may require unraveling massive amounts of narrative information during the analytic stage. As a result, while this approach can be extremely helpful for in-depth understanding of empowerment cases, it is a costly undertak- ing that can only rarely be used in tracking and monitoring systems. Statistical techniques of data analysis tend to be referenced as "quantita- tive." However, increasingly sophisticated econometric models allow not only for straightforward continuous values to be analyzed, but also for "qualita- tive" ranked or noncontinuous information to be subjected to statistical processes. Further work and cross-disciplinary collaboration are required to overcomesomeoftheproblemsassociatedwiththemultidimensionalityinher- ent in analysis of empowerment; these include the personal and professional concerns of econometricians and noneconomic social scientists. The creation of composite indexes is one way in which some analysts have dealt with the problem of multidimensionality, but such constructs require extreme caution in terms of variables chosen and weights given by the analyst. While not fool- proof, techniques such as factor and correspondence analyses offer opportu- nities to improve the reliability of indexes that can be used for regression estimates. Literature from the field of psychometry--evolving out of work on theeducationsector'sclassicaltesttheory--suggeststhattheproblemofweight- ing variables can be overcome, or, given satisfactory tests of indicator relia- bility, is nonexistent (Crocker andAlgina 1986; Nunnally and Bernstein 1978). However, the prerequisites of rigorous sampling and robustness of variables required for self-weighting may not be realistic in the conditions under which much poverty and empowerment monitoring or research is undertaken. 40 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Some data, though, cannot practically be collected in sufficient quantity for findings to be statistically validated. Such information can be either analyzed using simple descriptive statistics or subjected to techniques such as word counts or narrative analysis. Econometric approaches to qualitatively dynamic "systems" such as power relations raise additional concerns related to the metrics identified and the associated risk of endogeneity.8 There is a potential trade-off here between wanting to understand more about the extent to which variables are predictive in terms of their impact on empowerment and other development outcomes, and risk of endogeneity through attribut- ing cause and effect based on the application of regression methods to the complex process. Many practitioners working with empowerment will face this dilemma and will have to make the choice of whether regression analy- sis is appropriate to certain kinds of data--and if it is, how to ensure that analytic specifications are sufficiently sophisticated to account for differ- ent metrics, types of data, and potential endogeneity implied by a highly interactive and multidimensional model. The methodologies adopted for the research projects presented in this book illustrate the added value of combining methods and data effectively to measure and understand empowerment. In their evaluation of the empowerment impacts of participatory budgeting (PB) in Brazil (chapter 7), Baiocchi and colleagues first applied matched-pair analysis to a sample of 10 municipalities, comparing empowerment trends in five PB municipali- ties with trends in a control group of matched municipalities that did not participate in PB. The second stage of the research used in-depth analysis, conducted by local research teams using the process tracing method, to unpack budgeting processes and impacts within the 10 paired municipalities. It then established a nationwide database of cross-sectional municipal-level data between 1991 and 2000. These data were merged with social, economic, governance, and civil society indicators from census and other surveys, along with data on participatory budgeting reforms covering the same period. Legovini's study of the empowerment impacts of the Women's Develop- ment Initiatives Project in Ethiopia (chapter 8) combined a secondary source review with interactive methods of data collection and a questionnaire-based survey. The literature review identified the formal and informal institu- tional context for women's empowerment in Ethiopia. Interactive fieldwork was then conducted in eight sampled kebeles through in-depth interviews with key informants, individual interviews, and focus group discussions with beneficiary and nonbeneficiary groups. The findings were used to focus and structure the questionnaire survey that was then implemented with a ran- domly selected sample from a project list of women selected or rejected as beneficiaries of the project. In the survey, data on indirect and direct indicators of empowerment were gathered from 1,000 households. While the questionnaire was generally quite standard in terms of questions with direct and limited-response answers, some METHODOLOGICAL ISSUES IN MEASURING EMPOWERMENT 41 types of information required more engagement with respondents than others. For example, when examining the degree of happiness women experienced, respondentswereaskedtochoosefromarangeofillustrationsthatmostclosely matched their psychological state.Also, when collecting ranked data, enumer- ators resorted to using interactive techniques such as drawing a line on the ground. While their use was limited in this survey, simple interactive tech- niques that generate easily and systematically recordable data can at times be extremely helpful when dealing with subjects a respondent is unfamil- iar with or has trouble quantifying. Information from the interactive survey was also used to assist in interpre- tation of results. In Ethiopia, preliminary results were not taken back to respon- dents, but in some other cases this has proved to be an ideal way to refine interpretation and analysis.9 The sample from the 1,000 households was used to measure empowerment and study its determinants through the develop- ment of regression models for key outcome variables to test hypotheses relat- ing to each domain.10 The impact evaluation part of the study compared average results in outcome variables and measured impact as the simple dif- ference between average outcomes in the treatment and the control group. As there was no baseline study, more reliable difference estimates were not possible, but the results were treated with caution and, where possible, econo- metric analysis controlled for possible differences in initial conditions. Di Gropello and Heinsohn's evaluation of the impact of the Proheco school- based management project on empowerment outcomes in Honduras (chap- ter 9) sequenced questionnaire surveys of school directors, school councils, and households with key informant interviews and interactive community interviews. Sixty treatment schools that were participating in the Proheco program and 20 control schools were selected from a random-stratified sam- pling frame. The data from the surveys were used to develop direct indica- tors, capturing empowerment of school councils and community members, as well as indirect indicators, assessing the agency and opportunity structure for these two groups. The direct indicators of empowerment were used to create a school council empowerment index, whereas the agency and oppor- tunity structure variables were entered first into a multiple and then a step- wise regression to see which of the variables were significant in terms of their correlation with school council empowerment. The qualitative or inter- active research methods were then used to probe the dynamics, processes, and relationships underpinning the project's empowerment impacts. Gibson and Woolcock's study of the impact of Kecamatan Development Project (KDP) activities on marginalized groups' engagement in conflict res- olution in Indonesia adopted an "iterative" mixed-methods approach to pro- ject evaluation (chapter 10). The research team began with qualitative work in the shape of over 70 "conflict pathway" ethnographic case studies.11These used a process-tracing method informed by over 800 focus group discus- sions, in-depth interviews, and participant observations. This ethnographic 42 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION research examined the mechanisms by which conflict is initiated, intensi- fied, and resolved (or not resolved) in different contexts. Quantitative methods were used to help select the appropriate sites for this qualitative module. Districts were first stratified according to their capac- ity to manage conflict, with research sites selected randomly within the districts sampled. Propensity score matching was then used to match these KDP treatment villages with counterfactual control villages within sam- pled districts. This sampling procedure allowed for a comparative (with and without) analysis of the influence of KDP projects on a village's capacity to manage conflicts and on the "collective capability" of marginalized groups to engage in conflict resolution. The results of the qualitative module then fed into the design of a new quantitative survey instrument, administered to a sample of 226 households in the two provinces where KDP had been operating for at least two years. Since the first survey round, additional nonproject villages have been added for comparative purposes. Analysis of the ethnographic information is pro- vided in chapter 8, but at the time of writing no statistical analysis was avail- able from the country team. Therefore, assessment of related indicators and of how statistical and narrative analysis were integrated cannot be discussed in this volume. Finally, Bennett and Gajurel's longitudinal mixed-methods research pro- ject seeks to understand the processes of empowerment and social inclusion in the context of the Nepal Rural Water Supply and Sanitation project (chap- ter11).Theiranalysisisbasedonsurveydatageneratedfrom1,000households in a sample of 60 villages. The methodology built on the idea of identifying the main domains where empowerment and inclusion operated and then choosing locally relevant indicators of individual status in those domains. Survey data generated separate indexes of empowerment and of social inclu- sion.12 The empowerment index was developed using a range of variables that sought to capture the respondents' sense of agency. The social inclusion index comprised a number of variables chosen to reflect the degree to which the respondent was able to exercise his or her agency effectively to actually achieve the desired outcomes. These indexes were analyzed separately and also combined into an empowerment and inclusion index for further analy- sis. In addition, gender-specific indicators were used to create a women's empowerment and inclusion index, and a multiple regression analysis was applied to examine the influence of caste and ethnicity on women's empow- erment and inclusion. Multidimensionality and Interactive Variables: Summarizing Methodological Issues The complexity of empowerment as a concept has discouraged many devel- opment theorists and practitioners from attempts to measure and monitor METHODOLOGICAL ISSUES IN MEASURING EMPOWERMENT 43 it. Certainly, researchers and practitioners will encounter serious method- ological challenges when they attempt to identify indicators of empower- ment that conform to the principles of meaning, comparability, and causality. But these are not insurmountable, particularly if data can be generated through a combination of indicators that count specific, concrete choices that people have made and that score people's perceptions of the quality of choices open to them. Furthermore, when the surveys that generate these data to gain representative breadth are sequenced effectively with in-depth enquiry using interactive techniques for data collection, analysis, and ver- ification to probe the processes and dynamics underpinning power rela- tions, then those undertaking research or monitoring are in a strong position to be able to both describe and explain patterns and trends in empower- ment in a range of policy and project contexts. In the same vein, it is appar- ent from both the country studies and the scattered but growing number of mixed-methods studies that analysis--and in particular, analysis of poverty outcomes and determinants--can be approached by economic and noneco- nomic social scientists in more constructive, creative, and insightful ways. Notes 1. See Narayan (2005) for a summary of multidisciplinary perspectives. 2. For this reason, changes that are slow, such as in life expectancy, are less useful than those that can be relatively quick, such as in under-five malnutrition. 3. Two problems for statisticians are immediately apparent in several of the case studies. The first is the issue of endogeneity in the variables used for regression estimations in the Honduras case, and to some extent in the Ethiopia and Nepal cases. The second is the use of composite indexes for regression estimations. While they are often useful tools for simple monitoring systems or for crudely tracking change, indexes are fraught with problems and can be ill-suited to this kind of statistical analysis. Factor analysis is one approach to screen- ing variables for relevance and co-linearity before creating an index, but questions about weighting remain. 4. For example, government health department surveys that address, for example, women's roles in decision making and attitudes to spouse abuse have content that varies by country because of the cultural relativity of gender-based power relations (White, personal communication). 5. Interactive research methods are a means of data collection in which the interaction between researchers and respondents forms part of the analysis (Fischer et al. 2004). This can be contrasted with positivist-empiricist research methods in which the influence of the researcher on the data gathering process is removed. 6. See, for example, Rao and Ibáñez (2005) on the use of qualitative data to contextualize a quantitative impact evaluation. 7. See Participatory Learning and Action (formerly PLANotes), the journal of the International Institute for Environment and Development, to begin tracking this literature (http://www. iied.org/sarl/pla_notes/about.html). One technique for strengthening the comparability of 44 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION quantified subjective responses is the "anchoring" of a subjective evaluation to a small number of vignettes (such as one or two lines describing a person and his or her influence over something), where the cases have a fairly clear subjective ordering. The respondent rates these vignettes and then says where they lie in relation to the different cases (King and Wand 2004). 8. A metric is a quantitative assessment of a process that is to be measured and defines what is to be measured. Endogeneity occurs when the predictor is correlated with the errors in the regression model that describes its relationship with the outcome. A correlation between a predictor and the residuals in a hypothesized regression model violates one of the fundamental assumptions of ordinary least squares (OLS) regression analysis. 9. See, for example, Alsop, Krishna, and Sjoblom (2001). Preliminary research results in India were taken back to the field for multi-stakeholder workshops during which respon- dents were asked to verify and explain findings. The workshops were also used to draw out operational implications. 10. As the sample was not a random sample of the population but a random sample of beneficiary women and women who had been excluded from the project, mean values for indicators should be interpreted accordingly and not compared. 11. Conflict pathways describe discrete stages in the evolution of conflict, including conflict triggers and the factors or mechanisms that sustain conflict, allowing it to either escalate, stagnate, or move toward resolution. 12.As will be further explained in chapters 4 and 11, the Nepal researchers conceptualized empowerment and social inclusion as two separate phenomena. Their definition of empower- ment is more or less equivalent to this framework's perception of agency, and their definition of social inclusion captures what this framework terms the opportunity structure. 4 Using the Empowerment Framework: Five Country Studies Summarized Part 2 of this volume summarizes the five country studies that were part of the effort to develop and test the empowerment framework. Reflecting the realities of operational and analytic work, the country studies present experi- ences of using a common framework to analyze and measure empowerment in different interventions and contexts. This chapter introduces the case stud- ies and reviews these similarities and differences, reflecting on how the frame- work can be applied in different operational contexts. The chapter has two sections. The first compares the structural aspects of the five country studies and their evolution, commenting on their analytic and methodological design. The second section summarizes the findings of the five studies and compares the indicators developed to measure empow- erment in each context. Design and Structure of the Studies The five case studies represent disparate fields of intervention and are linked to different kinds of interventions. The Brazil country study (chapter 7) looks beyond project-based interventions to examine the impact of participa- tory budgeting,1 a reform that was introduced in Porto Alegre in the south of Brazil in 1990 and has since spread to well over 200 Brazilian municipal- ities. The remaining four studies examine empowerment as a component of ongoing World Bank or government-funded projects in Ethiopia, Honduras, Indonesia, and Nepal. The Women's Development Initiatives Project in Ethiopia focuses on empowering women by mobilizing them in self-help groups and by increasing their social and economic welfare. The Honduras Community-Based Education Project (Proheco) is an education decentrali- zation program that devolves authority over school and education-related matters to community-based school councils. The Kecamatan Development Project in Indonesia focuses on alleviating poverty and empowering citizens and local government through block grants to villages for community-level projects. The Nepal Rural Water Supply and Sanitation project focuses on implementing a community-based, demand-driven approach to the deliv- ery of water and sanitation services. 45 46 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Three of the country studies form part of larger studies that focus on issues broader than those associated with measuring empowerment. The case study on Honduras formed part of an impact evaluation of the Proheco project that, in turn, contributed to a comparative study on educational decentralization in four Central American countries. The Indonesia case study is part of a larger study on local conflict and participatory development projects in Indonesia.TheNepalcasestudyispartofalargerstudyonMeasuringEmpow- erment and Social Inclusion (MESI) and has also contributed to the 2005 Country Gender and Social Exclusion Study. The Brazil and Ethiopia stud- ies were carried out specifically as a contribution to this work on develop- ing a framework, indicators, and instruments for measuring empowerment. As discussed in earlier chapters, all the country studies also shed some light on the question of the relationship between empowerment and development outcomes. However, because of methodological problems--including the critical issue of endogeneity that is endemic to statistical analysis of highly interactive variables--in most cases these findings must be treated as indica- tive of associations rather than as definitive statements of causality. This chap- ter therefore touches upon findings only briefly, concentrating instead on the indicators and instruments used for data collection. It makes no attempt to draw out policy recommendations from findings. Although the five country studies all use the empowerment framework as their conceptual point of departure, the studies have in some cases mod- ified the definition of empowerment presented in chapter 2 to fit their indi- vidual focus and analysis (table 4.1). Application of the Framework Each country study generally uses the notion of "power to" as its analytic reference point. This contrasts to the view of empowerment as a zero-sum game in which one person or group gains power at the expense of another person or group, and thus has "power over" them (see appendix 2, box A.2.1). The country studies also recognize an iterative relationship between individual or group agency and structure of opportunity, but differ in their analytic emphasis according to the project or initiative context (table 4.2). Those initiatives with delivery and accountability dimensions, particularly the ones in Brazil and Honduras, bring a stronger analytic focus on insti- tutional change and transformed development outcomes through citizen participation. The projects in Ethiopia and Nepal give more weight to agency building by increasing investment in or access to a range of assets-- including the development of psychological assets of individual actors by stimulating collective action. The Indonesia case study looks at the iteration of structure and agency by identifying "deliberative spaces" created by community-driven development projects for inclusion and voice in con- flict resolution. These features are discussed in greater detail below. USING THE EMPOWERMENT FRAMEWORK: FIVE COUNTRY STUDIES 47 Table 4.1. Definitions of Empowerment Used in the Country Studies Country Project or initiative Definition of empowerment used Brazil Participatory budgeting An empowered civil society has initiative implemented by a the autonomous capacity for growing number of self-expression (agency) and municipalities in the country the opportunity to effectively and meaningfully engage the state (Baiocchi et al. 2005). Ethiopia Women's Development Effective agency (Legovini 2005). Initiative Project Honduras Honduras Community-Based The ability of local actors to Education Project effectively participate in school related decision making (Di Gropello and Heinsohn 2004). Indonesia Kecamatan Development The process of enhancing Project individual or group capacity to make choices and transform those choices into desired actions and outcomes [with a focus on] the capacity to manage conflict (Gibson and Woolcock 2005). Nepal Rural Water Supply and Enhancing an individual's or Sanitation Project group's capacity to make purposive choices and transform those into desired actions and outcomes (Bennett 2004). Source: Authors. As indicated in chapters 2 and 3, the analytic framework is flexible in that it identifies what is important to understand, measure, or identify for action but does not prescribe the depth of substantive coverage of data collection, analysis, or intervention. It also does not hypothesize directions of causality. This is deliberate and is left to the discretion of those undertaking the meas- urement. Some analysts may be uncomfortable with this flexibility, but they should remember that hypothesizing associations and directions of causal- ity before developing survey instruments is common. One of the most wide- spread uses of participatory techniques in survey work involves identifying or verifying the key relations and variables that a questionnaire-based survey will then collect information on and analysis will test. In developing this approach to measuring empowerment, core elements of empowerment that could be measured and used within and across a range of countries and sit- uations were identified. However, while the framework focuses on generic domains, levels, and degrees of empowerment, the indicators, variables, and their values are likely to be country- and context-specific. Table 4.2. Case Study by Analytic Focus Country Emphasis in analysis of empowerment Level of measurement Actors Brazil What are the changes in financial assets (measure Groups at municipal level General citizen organizations, of agency) and formal institutions (measure of elected representatives, and opportunity structure) for civil society government staff participation and exercise of choice (measure of empowerment)? Ethiopia What is the position of Ethiopian women in terms Individuals as members of Women of formal and informal institutions (legal status, households and groups at political participation, status in the household)? village level Honduras What is the capacity of citizens to participate in School management groups at General citizens, with some management of school responsibilities, and has village level, individual focus on women, indigenous 48 there been an institutional change in the function household members people, and poorest of the school councils within the decentralization population process (interaction between changes in formal institutions and measures of empowerment)? Indonesia What is the capacity of groups of villagers to Interest groups at village level General manage given conflicts seen as small processes (measure of the empowerment effect of new formal institutions on informal institutions)? Nepal Has an intervention empowered previously Households disaggregated by Marginalized groups, excluded people in terms of delivery of gender and communities women, low castes development benefits (the empowerment effect disaggregated by caste and of new formal institutions on traditional, informal ethnicity at village level institutions)? Source: Authors. USING THE EMPOWERMENT FRAMEWORK: FIVE COUNTRY STUDIES 49 In the five countries where the framework was applied, each country team identified the domains and levels of analysis relevant to their inter- ests, as well as the indicators to use for measurement and the values placed on variables. The empowerment framework provides users with an "ana- lytic lens" that sets out areas of enquiry and then allows them to adapt aspects of the framework to a specific situation. In addition, in the common situation of scarcity of research or monitoring resources, the framework assists in prioritization of focal areas. The following sections outline how each of the country studies approached the task of measuring and tracking empowerment, and provides examples of specific indicators used in each.As will be seen, the indicators used differ from study to study, yet many have similarities despite the variation in focus and scope of the studies. There is also convergence in the domains in which the studies focus their analysis. Only one of the five studies, Ethiopia, works with all three domains (state, market, and society). The other four studies all focus on the state and the societal domains. Amatrix of indicators used by each study in the different domains, structured by the framework, is pre- sented in appendix 4. Brazil: The Participatory Budget Initiative TheBrazilcasestudyisnotconnectedtoanyproject,buttothewell-documented participatorybudgeting(PB)initiativeinBrazilianmunicipalities.Twohypothe- ses are examined: (a) that the direct and continuous form of participation created by the PB initiative empowers civil society and helps build capac- ities through problem solving, communication, and learning by doing; and (b) that empowerment can have direct, tangible developmental outcomes. A two-stage research approach was employed for analysis. In the first stage, local research teams conducted in-depth analysis of the budgeting processes in five matched pairs of cities. Each pair shared basic attributes such as political party configuration, region, and size, but included one city that had adopted PB and one that had not. This allowed for isolation of the effects of PB as well as analysis of the mechanisms through which PB had an impact. In the second stage, a data set was constructed for all 5,507 Brazilian munic- ipalities. The researchers then undertook statistical analysis of (a) the deter- minants of participatory budgeting, that is, the factors that explain the choice to adopt PB, and (b) the impact of participatory budgeting on a wide range of financial and developmental indicators. Empowerment was measured within the state domain and in the sub- domains of politics and service delivery. The study also included a small num- ber of indicators related to the societal domain and, more specifically, its intra-community subdomain. Because the locus of the analysis was munic- 50 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION ipal budgeting, analysis concentrated on the "intermediary level" (the dif- ferent levels of empowerment are described in chapter 2). Analysis of empowerment involved assessment of changes and the effects of changes in both opportunity structure and agency. The effect of changes in opportunity structures as a result of the introduction of PB institutions was assessed by analyzing trends in the participatory nature of municipal bud- geting. Opportunity structure was measured on a continuum, with clien- telistic politics and no participation at one end and associational politics with binding forms of participation at the other. Amunicipality's position on this continuum was determined by analyzing four sets of indicators: (a) the mode of participation--whetheritwasdirect,delegative,mixed,orentirelynonexistent; (b) formalization of the process--whether it was formal, informal, or nonexis- tent, and the existence of rules and procedures governing the participatory inputs; (c) decision-making power--whether citizens' deliberations are bind- ing, merely consultative, or nonexistent; and (d) the scope of discussion, cap- turing the range of governance functions the participatory processes sought to influence--none, making general demands, budgets, policies, or mixed. Changes in agency were assessed by comparing the evolution of civil soci- ety's capacity for autonomous action in municipalities with and without PB. Key variables included (a) civil society's access to information, seen as vital for active participation, and (b) social and human capital factors within civil society. Changes in agency centered around two axes: (a) mode of intermedi- ation, that is, the level of associationalism, the level of clientelism, and the level of exclusion of civil society organizations, and (b) the degree of self-orga- nization, that is, the extent to which the organizations were dependent on other organizations or individuals politically or economically or were autonomous in terms of management as well as ideology. This analysis of the vitalityofcivilsocietyorganizationsconfirmedthehypothesisthatthepresence of a vital and plural civil society was clearly important for empowerment. The study found that all five cities that had adopted PB experienced an increase in the presence and use of opportunity structures, but that only two of the five experienced an improvement in both opportunity and levels of agency. This led to the conclusion that improved opportunity structure does not by itself trigger increased agency. However, when combined with increased financial assets, the empowering mechanism of participatory budgeting has a positive effect on reducing extreme poverty. Ethiopia: The Women's Development Initiatives Project The study in Ethiopia had two objectives: (a) to examine the general empow- erment status of poor women in Ethiopia, and (b) to assess the impact of the Women'sDevelopmentInitiativesProject(WDIP)onwomen'sempowerment. Communities were selected from two different regions of the country.Amixed- methods approach combined interactive techniques, which included semi- USING THE EMPOWERMENT FRAMEWORK: FIVE COUNTRY STUDIES 51 structured individual key-informant interviews and focus group discus- sions, with a questionnaire-based household survey. Results from the qual- itative enquiry were used to develop testable hypotheses and indicators, which then framed the questionnaire-based survey. The Ethiopia study was exceptional among the five studies in the sense that it gathered data on women's assets, their opportunity structure, and their degrees of empowerment in all three domains--the state, the market, and society. Indicators on assets and degree of empowerment mainly cor- responded to the local and intermediary levels, and included human, social, material, and psychological assets. Indicators on opportunity structure also encompassed the macro level (see chapter 8 and appendix 4 for details of variables). These indicators considered both formal rules, such as the exis- tence of laws ensuring women's equal treatment within the justice and political systems, and informal rules of the game, such as those operating through widespread traditional courts or through cultural traditions influ- encing women's behavior. To illustrate, in the political subdomain of the state domain, indicators on the degree of empowerment included the ratio of women to men serving on village and district councils as well as the question of how women are affected by the country's judicial environment. The latter included treatment of women both by the national judicial system and by the laws upheld in the traditional courts still operating in the country. A number of assets were hypothesized to help women both increase their representation and influ- ence in the councils and get fair treatment in the judicial system; these included previous participation in associations (organizational assets), num- bers of people they could call on for help (social assets), their level of edu- cation (human asset), their self-confidence(psychologicalasset),andtheextent of their awareness of their rights (informational asset). Opportunity structure indicators included formal rules of the game such as laws granting women equal treatment within the judicial system and representation in commu- nity groups and councils. With regard to informal rules, indicators gauged the extent to which traditional practices made women less likely to obtain justice than men and less able to engage in political and public life. In the labor subdomain of the market domain, one of the indicators of degree of empowerment was the extent to which women are able to choose their type of employment. Asset indicators captured women's education and income levels, their possession of job-specific skills, and the extent to which they had access to different sources of information. Opportunity structure indicators captured the extent to which cultural restrictions deter- mined the nature of professions women were allowed to pursue, the amount of time women had to dedicate to household chores, and gendered rules governing access to productive assets and markets. The project's own impact evaluation also established a series of empower- ment indicators for the household subdomain of the society domain, such 52 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION as the percentage of women who had decision-making power equal to that of their husbands over the number and spacing of children, the use of con- traceptives, and conjugal relations. The interactive enquiry led the study team to hypothesize that the extent to which women had a say in these matters associated with such assets as women's education, income, and self-confidence levels; their awareness of reproductive health issues; and their participationinwomen's groups.Analysis also tested for associations among opportunity structure indicators, such as customs that influence whether women are allowed to disagree with their husbands and whether they are expected to play a subservient role regarding sexual conduct. The country team analysis suggests that both assets and opportunity struc- tures are restricting the empowerment of Ethiopian women in terms of polit- ical participation, equal and fair treatment in the formal and informal judicial system, participation in the labor market, and position in their households. The study also found that WDIP did improve economic and social empower- ment--but that interventions of this nature needed to also encompass, or work in tandem with, reforms at the macro-institutional level that expand economic opportunitiesforall,narrowthegendergapineducation,deepenlegalreforms, and strengthen enforcement of women's rights. Honduras: The Community-Based Education Project The measurement of empowerment in Honduras formed part of an impact evaluation of the Honduras Community-Based Education Project, known as Proheco. This evaluation measured empowerment in the context of two power relationships that were potentially affected by education reform: (a) power relations between school councils and education authorities, including school staff; and (b) power relations among different community members. It was hypothesized that the project empowered school councils to have greater decision-making authority and autonomy in relation to education authori- ties, and that it empowered different community members to participate and exercise effective agency in school council meetings, irrespective of their gender, socioeconomic status, or ethnicity. As with the other country studies, the Proheco evaluation used a mixed- methods approach, sequencing institutional and household surveys with key- informant and community interviews. The evaluation sampled treatment and control groups, allowing for a comparison between schools managed by a community school council and those that were part of the traditional state- run system. Identifying power relations and where they play out helped the research team to select the domains and levels where empowerment would be meas- ured. It also helped determine the units and locus of data collection and analy- sis. The Proheco evaluation addressed several aspects of empowerment as a USING THE EMPOWERMENT FRAMEWORK: FIVE COUNTRY STUDIES 53 result of decentralization. The study's indicators focused on the public ser- vice delivery subdomain of the state and sought to address whether the school councils were able to carry out the devolved tasks. Information on the social, ethnic, and economic attributes of community members were gathered to ascertain which parents were able to participate in the school councils. Using the concepts of assets and opportunity structure, researchers iden- tified a series of factors that influenced the councils' capacity to assume their new tasks. Examples of these collective assets that were found to help councils carry out their newly assigned functions included the amount of relevant information and training they received. The opportunity struc- ture for school council empowerment was found to encompass both formal and informal rules. Formal rules included, most notably, the regulations specifying the decentralization reform and the nature of the powers to be devolved to the school councils. Informal rules included a series of factors that shaped how the reform was implemented in practice, including the technical capacity of the ministry to provide communities with adequate information and training to manage the school, and the regularity and time- liness of ministry financial transfers that enable councils to buy school sup- plies and pay teachers. With regard to community empowerment, indicators on degrees of empow- erment related to a person's involvement in school council activities. Exam- ples of assets included parents' prior engagement in or experience with other community organizations, as well their awareness of their right to partici- pate in the council. The opportunity structure, in this case, referred to formal or informal rules of inclusion and exclusion, such as those that determined whether members of excluded groups could participate in public meetings and decision making. The study findings suggest that the decentralization process empow- ered school councils and to a lesser extent individual community mem- bers. In both cases, the researchers' analysis showed that opportunity structure variables demonstrated a higher impact than variables related to asset building. The degree of empowerment was linked to the power rela- tions that prevailed in the councils and communities, determining who could participate in local decision making (informal institutions). Indonesia: The Kecamatan Development Project The case study in Indonesia examined local-level conflict in the context of community-driven development initiatives under the Kecamatan Devel- opment Project (KDP).Although the KDPwas not designed as a conflict reso- lution program, the core question of the study was whether the KDP was building the conflict management capacity of villagers through unexpected spillovers. 54 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION To test this, the study generated over 70 ethnographies of how particular social tensions and incidents of conflict played out in their local context, look- ing at how the different actors reacted to the different conflicts and either nego- tiated or failed to engage.2 Each case study included a summary of the case, its prehistory, evolution, attempts at resolution, impacts, and aftermath. Col- lection of this qualitative data followed a quantitatively oriented sampling frame so as to construct a plausible counterfactual. Villages that received KDPfundingwerecomparedtovillagesthatdidnot,andqualitativeinterviews were used to confirm the accuracy of those matches. The Indonesia researchers focused their analysis mainly within the state and the societal domains and concentrated on the intermediary and local levels. In the state domain the study worked with the legal subdomain, looking at the ability of citizens to approach police and the ability of police and courts to apply the laws correctly and solve conflicts. It also worked with the political subdomain, considering the functioning and accountability of local authorities and citizen participation in the local decision-making pro- cesses. Within the societal domain, indicators were focused in the subdomain of the community, relating to the associational as well as the social interac- tion among people with different identities. The study used the terms "countervailing power" and "routines" to struc- ture its analysis of how relationships between different groups and individ- uals worked in practice. People's countervailing power in a given conflict relates to their agency, while the idea of routines determining the outcomes of conflicts can be compared to the concept of opportunity structures. The study suggests that local institutions of conflict management are often mar- ginalizing, but that everyday conflicts can themselves serve as flashpoints for forms of collective action and demonstrations of agency that are able to overcome those institutional barriers, allowing marginalized groups to man- age conflict through collaboration with other social groups. The study also demonstrates, however, that collective-action solutions to institutional chal- lenges cannot be engineered through technical interventions. The success- ful convergence of countervailing power through asset building is difficult to achieve, but where the KDP intervention has managed to cultivate agency, the result has often been the empowerment of groups through local institu- tions, catalyzing peaceful transition and improving development outcomes. Following the above, a central finding of the study was that empower- ment for marginalized groups in conflict situations involves making tran- sitions in the nature of forums (representing opportunity structures) used for deciding conflicts, as well as in the types of power used to acquire legit- imacy in such forums. The study suggested that, in terms of opportunity structure, the key task for development projects is to create the spaces, resources, and incentives that enable the poor to build effective and durable bases of countervailing power or assets in those domains where decisions USING THE EMPOWERMENT FRAMEWORK: FIVE COUNTRY STUDIES 55 most immediately related to their welfare are made. One of the primary challenges in terms of agency is enhancing the psychological assets, mean- ing the capacities to aspire and (perhaps more important) engage, that help marginalized groups conceive of and achieve alternative futures. Nepal: The Rural Water Supply and Sanitation Project The ongoing Measuring Empowerment and Social Inclusion (MESI) study in Nepal sought to measure empowerment as one of the components of the Nepal Rural Water Supply and Sanitation Project (RWSS I and II). The study aimed at developing gender-sensitive ways to measure levels of empow- erment and social inclusion. The second phase of the study, which was still under way in 2005, will also assess in more detail whether and how the RWSS project has empowered communities in relation to government and non- government service providers, and will ask whether marginalized groups have been empowered and have experienced greater social inclusion as a result of the RWSS intervention. The Nepal study was designed as two phases of research on empower- ment and inclusion, integrated into the RWSS project cycle. Each phase included a qualitative and a quantitative component. The first phase spanned nine months and the second phase is expected to span three years, ending in 2007. The first phase was a self-contained research design that estab- lished the baseline of the longitudinal study that is the keystone of this research. The quantitative portion of the design involved comparing com- munities that had already received an intervention with communities that had not. The Nepal country study covered the domains of the state and society and concentrated on the intermediary and local levels. Data were collected on a wide range of assets hypothesized to have a relationship with empower- ment outcomes.Asset indicators included standard measures such as literacy and land ownership, as well as group membership, participation in train- ing, and knowledge of rights. Indicators of opportunity structure focused on the rules (institutions) that govern social positioning, social interaction, physical mobility, violence, and economic security or vulnerability. Indicators on degrees of empowerment included the ease with which people can approach legal services, their voting behavior, and the degree of control over various aspects of domestic life that different household members enjoy. Indicators of intra-community engagement and the manner in which people behave or are able to behave in that setting were also gathered. The study found that people's caste and ethnicity had a very strong influ- ence on their level of empowerment and social inclusion, even though the caste system in Nepal has been legally abolished for decades. But it also found that gender was an important factor determining the degree of empowerment 56 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION and social inclusion, making it especially hard for low-caste women to exer- cise their rights, gain access to services, and live free of violence. Thus, the opportunity structures and especially the informal rules and norms still enforc- ing the caste system and gender roles played a crucial role in the degree to which men and women are empowered in the study villages in Nepal. More encouragingly, however, the study also found that this negative influ- ence of caste and ethnicity on empowerment can be reduced by development interventions.Thesecanbuildtheassetsofthedisempoweredthroughimproved access to education, through income earning, and through opportunities to accumulate other human, economic, and financial assets. Interventions can also create empowering organizational assets by supporting the mobiliza- tion of such people to engage in collective action and join together in groups. Conclusion: A Robust but Flexible Framework The five case studies summarized here, and presented in more detail in part 2 of this volume, provide an unprecedented opportunity to examine the application of an analytic framework for understanding empowerment to project and sector contexts. They demonstrate the flexibility with which the framework can be applied to a contextually appropriate understanding of changes in power. They also illustrate the value added in terms of gen- erating indicators that capture elements of empowerment and contribute understanding to the operational challenge of empowerment through devel- opment interventions. The next chapter considers, somewhat more sys- tematically, how the empowerment framework can be effectively applied to operational work. Notes 1. Known in Portuguese as orçamento participativo. 2. The study team terms these "conflict pathways case studies." 5 Applying the Empowerment Framework to Operations Chapters 3 and 4 focused on understanding concepts related to empower- ment and their application in terms of information collection, indicators, instruments, and analysis. This chapter considers how the framework can be applied to the design, implementation, and monitoring of development policies, programs, and projects. The scope of development interventions is broad, with varying impacts on structure and agency in different domains and at different levels.At the policy level, interventions can change institutions by altering incentives and impos- ing new forms of sanctions and creating new sets of entitlements that have an impact on levels of agency among different groups. Macroeconomic, fiscal, financial sector, and public finance reforms affect prices, access to credit and public services, and employment levels in the public sector. Trade, exchange rate, and agricultural reforms will influence the opportunity structure for both producers and consumers through changes in prices and levels of access to goods and services. Land reform can shift property rights, with significant implications for asset-based agency. Labor market policies will often have importantopportunitystructureimpactsonwageratesandemploymentlevels along with core labor standards and the rights of employees. Utility reform can affect the level and quality of access to utilities that are essential for asset investment and economic choices. Institutional policies that include civil ser- vice reform, justice system reform, and decentralization can significantly alter the opportunity structure of rights, obligations, incentives, and sanctions within government, between government and citizens, and within society. Finally, policy changes in state transfers through social safety nets and pen- sion schemes are likely to change the opportunity structure that determines entitlements and access for different social groups. At the program and project level, interventions can influence opportu- nity structure and agency through a range of sectoral interventions. Appendix 3 presents a review of World Bank projects with empowerment aims. Assets can be enhanced through extension of infrastructure and ser- vices, including road networks, water and sanitation, microfinance, and agricultural extension, to geographically remote or excluded communities. 57 58 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Projects and programs can challenge informal institutional inequality--that is, social rules governing access to and use of such assets--through the provi- sion and enforcement of new formal rules that enhance local rights and pro- mote social inclusion. Projects and programs can also directly build agency by investing in other types of assets, for example, through group mobiliza- tion and social accountability initiatives, or through direct investments in pro- ductive assets or health and education initiatives. They can also do much to stimulateamovefromthefirsttothesecondandthentoperhapsathirddegree of empowerment by providing much-needed assets such as injections of cash for discretionary spending at the local level. In the case of decentralization projects, this can act as an incentive for local citizens and government coun- terparts to exercise the use of formal institutions for engagement (Raich 2005). While operational application of the full framework discussed in this book is limited by its relative newness, the long sociological discourse from which it has evolved (see appendix 2) underpins many development efforts to enhance agency and change opportunity structures. This chapter reviews a selection of these interventions, along with some more recent actions to directly empower people. The examples serve to illustrate how the frame- work can be employed to identify and prioritize activities in projects, pro- grams, or other forms of operational work including policy analysis, formulation, and monitoring.1 The next section of the chapter revisits the framework, describing how it can be systematically laid across the design of projects or programs that specify empowerment as either a means or an end. This is illustrated with examples of past and current investments in assets and institutions. The fol- lowing section explores opportunities for using the framework to assist in policy analysis and design. Both sections incorporate a brief discussion of the systems required for monitoring empowerment at the project and policy levels. Monitoring is critical when tracking effects and evaluating the assump- tions of causal relations that are built into the empowerment framework; it generates the data and analysis for improved project and policy discourse, formulation, and operationalization. The final section summarizes the appli- cation of the framework to operations and takes up some of the outstand- ing issues in dealing with analysis of multidimensional information. Framing Program and Project Interventions To recap, the empowerment framework proposes that a person's or group's agency and opportunity structure interact, giving rise to different degrees of empowerment. In terms of measurement, assets are used as indicators of agency (the capacity to make purposive choice), and institutions are used as indicators of opportunity structure (the informal and formal institutional context). In terms of action to enhance empowerment, investments in both APPLYING THE EMPOWERMENT FRAMEWORK TO OPERATIONS 59 assets and opportunity structure are key to achieving the "three degrees" of empowerment, realized when the opportunity to choose (a) exists, (b) is used, and (c) achieves desired results. For the World Bank and many other development agencies, interven- tions historically have tended to be sector-specific in their primary focus-- for example, an education project, a sanitation project, or a watershed project. In such cases the project's development objective and sectoral focus determine any related empowerment activities. Within this sectoral struc- ture, the empowerment framework can be used to identify and prioritize activities, structures, processes, and monitoring indicators. In addition, the framework aids in determining what monitoring information needs to reside where, with whom, and for what purpose.2 For example, in the case of the Tamil Nadu Empowerment and Poverty Project in India, the empow- erment framework was explicitly used to identify project activities and to structure the project's monitoring and learning system (World Bank 2005b). The objective of the project is to empower the poor and improve their liveli- hood by (a) developing, strengthening, and synergizing pro-poor local institutions and groups (including village-level elected government); (b) enhancing skills and capacities of the poor (especially women and the vul- nerable); and (c) financing productive, demand-driven subproject invest- ments related to livelihoods for the target poor. Focusing primarily on the local domain, the project therefore seeks to enhance the assets of the poor and the opportunity structures within which they live and work. As increasing attention has been directed to programmatic and budget support, however, the nature of interventions has broadened. With these broader, "upstream" activities, policy strategy and intervention in support of a country's empowerment objectives can be identified, prioritized, and monitored with the help of the empowerment framework. For both pro- grammatic and sector work, the framework can be used to structure ex ante and ex post analysis of conditions and effects. Empowerment--Means or End? At the outset, a team designing a sector lending project, grant activity, or pro- grammatic strategy has to make a choice about the importance it will attach to enhancing people's empowerment. Empowerment can be an objective of an intervention or program (an end in itself), or it can be an instrument in the process of reaching a project or program objective (a means to an end). It can also be of such limited value to the core purpose of the proposed interven- tion that future action to enhance people's capacity to make effective choice may not be worth the effort. Simply put, the importance of empowerment to a future intervention will determine how the empowerment framework is used during preparation and, later, implementation. 60 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Chapter 2 stressed that, except in an in-depth research study, it is unlikely that all elements of the empowerment framework will be either operationally useful or financially cost-effective. Even when empowerment is an end in itself, hard choices have to be made about the most strategic needs and related activities. For such interventions, the framework can be used to · first, define the empowerment objective; · second, specify first-order actions required to meet this objective and priori- tize them according to the strength of their association with the objective; · third, identify critical links to second- or third-order actions in other domains and levels required to achieve the primary objective. Box 5.1 provides an example drawn from Guinea. As a means, empowerment becomes subordinate to the main develop- ment objective. In this case, steps taken to empower people are supposed to lead to the achievement of another overall development objective. Steps for identifying the importance of empowerment to the design and imple- mentation of interventions are discussed in the final part of this chapter. Box 5.2 presents a case in which empowerment actions were instrumen- tal in a watershed project in India. The table in the box identifies not only the domains in which first- and second-order actions were to be taken, but also the level at which issues were addressed. With regard to both activities and the allocation of financing, this project differs considerably from the Guinea case, in which economic and social empowerment was treated as an end in itself. Decisions on empowerment as a means or an end help determine deci- sions on what critical actions need to be undertaken and in which domains of people's lives. By providing a "theory of change" predicting how inter- ventions will have an impact at different levels and across different domains, the empowerment framework also identifies at which level activity needs to occur. The following section unpacks these areas of action further into those related to agency and those related to opportunity structure. Investing in Agency and Opportunity Structure Agency is largely determined by a person's or group's asset endowment.3 As both background searches and the five country studies found, very few efforts have been made to directly measure empowerment. Similarly, while nongovernmental organizations (NGOs) have amassed considerable experi- ence in changing power relations, until recently few of the larger bilateral or multilateral development agencies working with governments on a large scale invested in action specifically focused on empowering individuals or groups. However, examples of their investments in individual and group APPLYING THE EMPOWERMENT FRAMEWORK TO OPERATIONS 61 Box 5.1. Empowerment as an End in Guinea The Guinea Village Communities Support Program began in 1999 with the objective of promoting social and economic empowerment of rural population including women, youth, and other marginalized groups. The project appraisal document clearly articulated empowerment as an objec- tive and specified the areas of action in which the project was to work, and with whom. Using the language of the empowerment framework, this project sought to address the empowerment of people currently excluded in the social and market domains.a Project documentation further refined these general areas of work into first-order actions that would allow these people "to develop the capabili- ty to identify, prioritize, plan, and manage their own infrastructure and service needs." The first-order actions identified can be classified neatly as involving investments either in assets or in institutions. They include · establishing new demand-driven decision-making processes (presence and operation of organizational rules and institutions), · establishing new structures (organizational asset), · increasing information (information asset), · increasing services available (organizational asset), · enhancing local resource mobilization (financial asset), · establishing the capacity to sustain resource mobilization (presence and operation of institutions). Documentation also identified critical links to other issues to ensure that these first-order actions were successful. Underpinning social and market exclusion were social institutions that defined the statusb of women, youth, and other disadvantaged people and that resulted in a lack of capacity to make effective choices. Again, using the language of the empowerment framework, links were made with other domains of people's lives. The project needed to take action to influence the institu- tions present and operating within the social domain that also affected the people's capacity to make effective choices in the market domain. Source: World Bank 1999a. a. There is a lack of clarity within the project documentation, however. Much of the service delivery demand was to be met by government, making the state domain another area in which the project would need to operate--but there is limited discus- sion in documentation of how this would be achieved and monitored. b. Status is the power, authority, and prestige embedded in a role. 62 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Box 5.2. Empowerment in a Watershed Development Project in India The Integrated Watershed Development Project was implemented in five Indian states. Its primary objective was to "improve the productive poten- tial of the project area ... using evolving watershed treatment technolo- gies and community participatory approaches" (World Bank 1999b). The empowerment of the communities was regarded as a means of ensuring technical effectiveness and sustainability. Investments in assets are summarized in the table below. They included · at the local level: watershed treatment; fodder and livestock develop- ment; creation or rehabilitation of rural infrastructure; establishment of organizational capacity (Village Development Committees and links to local elected government bodies); building the financial assets of vil- lage accounts through a combination of project finance and community co-contributions; · at the intermediary level: building of technical and procedural capaci- ties of project staff; establishment of district coordination committees; · at the macro level: creation of project implementation units and coordina- tion committees across and in each of the five states; · increasing services available (organizational asset); · enhancing local resource mobilization (financial asset); · establishing the capacity to sustain resource mobilization (presence and operation of institutions). (Continued on the following page.) assets are numerous. Education and health interventions increase the value of human capital; infrastructure and agricultural projects improve mater- ial assets; decentralization and some poverty alleviation programs enhance the organizational assets available to both local governments and citizens; community-driven development activities increase social capital; and sav- ings and credit programs improve financial assets. Only psychological assets, which are key to converting assets into agency, have rarely received direct support or analysis in the context of development interventions. Opportunity structure is framed by the institutions that exist and oper- ate in different domains of individuals' and groups' lives. Again, while robust data and information on degrees of empowerment are hard to find at the national or project level, information on the institutional landscape, along with actions to support the presence and operation of equitable insti- APPLYING THE EMPOWERMENT FRAMEWORK TO OPERATIONS 63 Box 5.2. (continued) Investments in institutions included · village level: establishing rules for natural resources and financial governance, · intermediary level: establishing rules for operation of coordination committees and district-level project implementation units. Level Macro Intermediary Local (federal and state) (block and district) (village and hamlet) Domain Subdomain A OS A OS A OS State Justice Politics Public Secondary Primary Secondary Secondary Secondary Secondary service action action action action action action delivery area area area area area area Market Labor Goods Secondary action area Private Secondary services action area Society Intra- household Intra- Primary Primary community action action area area Note: A = Agency. OS = Opportunity structure. No cell notation indicates not a priority area for this project. tutions, is more readily available. Many interventions take action to ensure an appropriate policy environment at the macro level and put in place rules of engagement and operation at the intermediary and local levels. In some cases, interventions have also looked at the links between institutions--as in the case of Guinea, where it was recognized that social institutions would influence the way in which actors were able to take advantage of market institutions. The Madhya Pradesh Poverty Initiatives Project further demonstrates the use of the empowerment framework in working out from the core 64 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION empowerment objective to trace critical secondary actions required at dif- ferent levels and in different domains (table 5.1).4 The shaded cells show where project activities were concentrated. Starting with empowerment in the market domain (darkest cells), the framework allows for tracing of actions that need to be undertaken in other domains (lighter cells). The darkest cells illustrate the primary focus of the project, that is, the devel- opment of common interest groups. The arrows to lighter cells illustrate which activities were considered a priority if the primary project objectives were to be met. The example in box 5.3 further illustrates how the empowerment frame- work can be laid across interventions to identify activities required to build or activate assets and transform institutions. The use of the framework here is retrospective, but could also have been used beforehand to improve under- standing of what would be required to take effective action toward empower- ment. This example shows how action has taken into account different levels and different domains of women's lives and focused on the empowerment of individual women through collective action as an end in itself.5 Project-Level Monitoring Systems At the level of a project, use of empowerment indicators and analysis encour- ages project design that is better informed by an understanding of power relations and that is linked to ongoing measurement of empowerment impacts. At this level, the monitoring system and indicators chosen emphasize attri- bution of causality of outcomes and impacts to project inputs. If monitoring revealsthatprojectactivitiesarenotachievingdesiredresults,thisthenprompts projectrevisioninthenextcycleofactivity.Thewidelyusedlogframeapproach to project management often provides an instrument for managing infor- mation flows that include · projectinputsandoutputs,linkedtoadministrationandresourceaccountability; · project outcomes, enabling project participants, implementers, and manage- ment to understand the effects of the intervention, to link this to inputs and processes, and to compare effects across project locations; · project processes, allowing analysis of how and why things are happening and prompting corrective action on the basis of that new knowledge; · project impact, necessary to finally understand what the project invest- ment and action have achieved. Inlinearprojectsystems,monitoringandevaluation(M&E)areoftentreated as add-on activities that are somewhat disconnected from the project as a pro- cess of continuing evaluation. Typically, monitoring information is gathered at local levels by implementers and fed upward in the management systems Table 5.1. Actions to Develop Assets and Opportunity Structures in the Madhya Pradesh Poverty Initiatives Project Project development objective: To improve opportunities for the poor and vulnerable, especially women, to meet their own social and economic development objectives. State level District or block level Local or village level Domain (macro) (intermediary) (local) Opportunity Opportunity Opportunity Agency structure Agency structure Agency structure Created rules for finan- Created decentralized Established rules Established project State cial flows to local level project units or teams for integration with staff performance with line depart- appraisal systems ments and elected partly dependent on bodies client evaluations 65 Market Built assets (human, (a) Created CIGs Established rules orgnizational, and psy- (b) Provided financial for managing chological) of project support to CIGs group activities staff to (a) support (c) Developed business and finances development of CIGs skills and (b) develop CIGs' business skills Society Established incen- Established ground tive systems to rules to ensure ensure marginal- marginalized peo- ized people engaged ple included in in project project Source: Authors. Notes: Dark shaded cells indicate primary areas of action or intervention; light shaded cells indicate secondary areas of action or intervention; indicates relationship between primary area of action and requisite secondary action; indicates relationship between secondary area of action and requisite tertiary action; no shading indicates not a priority area for this project. CIG = common interest group. Box 5.3. Women in India: Enhancing Assets and Opportunity Structures The Rural Women's Development and Empowerment Project (RWDEP) sought to begin a "long-term process to improve women's economic and social status." In theory, the project design did not assume that the results of many centuries of institu- tionalized gender inequality could be completely changed within the (original) five years of the intervention.a The objectives of the intervention were focused and realistic. The project was designed to (a) establish self-help groups (SHGs) of women that would engage in savings and credit activities to satisfy short-term needs for small loans and establish group eligibility for accessing larger loans from banks; (b) build the capacities of organizations working on women's issues to support these SHGs; (c) develop women's entrepreneurial and business skills; and (d) enhance women's access to services and development funds. The project design prioritized improvements in women's status in two subdomains of women's economic lives, financial and public services. It also recognized several second-order areas of intervention that would be required to achieve success in these. The primary level of project engagement was the local level: SHGs would be established in villages or hamlets. While recognizing links 66 to other domains and levels, the design was clear in its primary locations and subjects of intervention and did not--at least by design--overextend implementers with a very wide range of activities. To address social rules governing women's behavior in the home and community (social domain), SHGs were used to provide (a) peer solidarity in challenging "unfair" gender norms, and (b) a forum for sharing information on different behaviors. Apart from this, no direct action was taken to change social rules at the vil- lage level. At the intermediary levels (block and district) the emphasis was on working with the banks to change the behavior of their staff toward village women as clients and on building the capacities of intermediary NGOs to work with the SHGs. At the macro (state and federal) levels, the project focused on the key organizations called Women's Development Corporations (WDCs), which were mandated to support women's interests in the long term. Directly in line with the project's objectives, activities at the village level focused on enhancing women's social, psychologi- cal, financial, organizational, and productive assets, in part through the formation of SHGs. It also established an institutional framework associated with the SHGs that would support the effective use of these enhanced assets. At the intermediary level, the project sought to change the informal institutions governing bank staff behavior that undermined formal regulations allow- ing women as groups to access bank loans. And at the macro and intermediary levels, the activities were geared toward build- ing the assets of the staff of NGOs and WDCs. (Continued on the following page.) Box 5.3. (continued) RWDEP Summarized Using the Empowerment Framework Level Macro (federal and state) Intermediary (block and district) Local (village and hamlet) Domain Subdomain Agency Opportunity structure Agency Opportunity structure Agency Opportunity structure State Justice Politics Encourage women's participation in village meetings associated with elected bodies (gram sabhas) Service (a) establish project Create project rules Create project Enhance women's delivery implementation enabling NGO structures enabling knowledge of structure; (b) develop intermediation NGO intermediation government- human assets of supported develop- 67 WDCs ment opportunities Market Labor Develop women's business skills Goods Private (a) Create group Change institutions (a) Establish SHGs Make effective exist- services savings accounts in governing bank as means of obtain- ing rules enabling banks; (b) develop staff behavior ing bank loans SHGs to access bank human assets of (b) Develop women's loans NGOs capacity for financial management and accessing mainstream financial credit Society Intra- Increase women's Create rules for peer household awareness of rights, solidarity and support Intra- gender injustice, and and collective action community options for change Notes: Dark shaded cells indicate primary areas of action or intervention; light shaded cells indicate secondary areas of action or intervention; no shading indicates not a priority area for this project a. The project has been extended beyond the original time frame and geographic coverage. In addition, the initial clarity of the project's objectives has been diffused during operations. 68 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION to be used for decision making by managers distant from the ground realities of implementation. This rather technocratic approach to project management has been increasingly challenged by approaches geared to the management of process, as opposed to "bean counting" of inputs and outputs. New man- agement thinking promotes a shift from hard to soft systems methodologies in which bureaucracies using information systems become learning orga- nizations that behave flexibly to improve project outcomes (Peters 1989; Senge 1990; Pasteur 2003). The characteristics of this learning approach when applied to empowerment monitoring, as with projects with a sig- nificant process element, include an emphasis on · strengthening the links between monitoring and evaluation information, · supporting information gathering and analysis by primary users, · strengthening the link from process analysis to project decision making, and · developing optimal information flows between these groups of users or decision makers. These elements have been applied to the development of a monitoring, information flow, and decision-making (MID) system for project manage- ment (Alsop and Farrington 1998). This "nests" discrete and independent flows of information into a composite system (figure 5.1). The independent systems generate data using a variety of methods, use information for dif- ferent purposes, and support decision making in different forums. The MID system brings these systems together in a way that enhances understand- ing of the process impacts of project interventions. The system empowers those performing different roles in implementation to collect, analyze, and react to monitoring information appropriate to their needs (thus producing the independent information systems). It also allows for the flow of mini- mum data sets between the different actors, again carrying information adequate to their decision-making needs but no more. This type of project monitoring system in turn has significant implica- tions for the institutional arrangements for project interventions. It ensures that those involved in implementation understand the purpose and bene- fits of data and information gathering. The nested information flows link the technical "nodes" of information synthesis and management with polit- ical "niches" of debate and decision making. Baker (2000) cites the methodology employed by a World Bank evaluation of the impact of social investment funds on community-level decision making, traditional power structures and relationships, and community capacity, trust, and well-being (Weiss 1998; World Bank 2002c). The poten- tial value of the empowerment framework to this type of evaluation is clear: the framework is based on clearly laid out assumptions about the impact APPLYING THE EMPOWERMENT FRAMEWORK TO OPERATIONS 69 Figure 5.1. Schematic Representation of Nested Monitoring and Learning System Niches (debate and decision making) Independent Independent Independent information information information system system system Node (information synthesis and management) Source: Alsop and Farrington 1998. on empowerment of asset building and achieving greater equality of oppor- tunity. With careful sequencing of methods and data, these assumptions can be tested by tracking project impacts on assets, institutions, and degree of empowerment. This kind of system supports the improved operation of projects--through a better mix or sequencing of activities--for better empowerment outcomes and impacts. Thistypeofongoing"evaluation"ormonitoringtoenablelearningbydoing is illustrated in table 5.2, using an example from the Madhya Pradesh District Poverty Initiatives Project in India. It should be noted that this project dif- fered from most others in terms of its management of the baseline survey. Instead of having the survey undertaken by an external agency as an "objec- tive" research exercise, the project management unit commissioned an exter- nal agency to apply the survey with the participation of project staff. The methodology involved a questionnaire as well as a range of participatory techniques for data collection and analysis. Project staff were involved in pretesting all instruments and underwent several rounds of training on use of the instruments. The results of the survey went beyond a statement of the initial situation. Project staff greatly increased their knowledge of Table 5.2. Summary of a Project's Monitoring and Learning System Type of information Subject Instrument Frequency Responsibility Administrative Inputs and Pro forma reporting Monthly Monitoring cells tracking outputs Effectiveness Outcomes Interactive assessment Quarterly Implementers at local levels Pro forma reporting Quarterly Implementers or monitoring cells 70 Causal factors Project processes Self-assessments Every six months Implementers Exchange visits Annual (minimum) Implementers or monitoring cells Pro forma reporting Annual (minimum) Implementers or monitoring cells Evaluation Project impacts Baseline survey By project launch Implementers Impact evaluation End of project Monitoring cells Exchange visits for Mid-term and at end of External organization peer review project to feed into impact evaluation Source: Authors. APPLYING THE EMPOWERMENT FRAMEWORK TO OPERATIONS 71 the areas in which they operated and felt fully engaged in what otherwise would have been seen as a boring and expensive data collection effort under- taken only to meet the needs of the government and the World Bank. Engage- ment in the survey also helped them better conceptualize the issues that the project was trying to address and provided an incentive for them to engage in monitoring progress over time. Policy Formulation and Implementation The introduction to this chapter summarized the scope of policy interven- tions that, to a greater or lesser degree, provide entry points for opera- tionalizing empowerment. In many cases the organizing framework for policy formulation is a poverty reduction strategy or equivalent policy framework. Indeed, a number of countries presently specify empowerment as integral to their poverty reduction strategies (see appendix 5). However, in many cases those implementing the policies struggle to operationalize the commitment to the concept of empowerment. Can the empowerment frame- work be useful in refining the policy formulation and in identifying areas for policy formulation and action? As discussed briefly above, the application of the empowerment frame- work needs to take into consideration that a range of macro-level policies will influence the opportunity structure within which individuals and groups attempt to exercise agency. If economic opportunities are limited, for example, improving agency by increasing different types of assets will have only a limited effect on outcomes. To give a simple example, farmers may be empowered by gaining access to a cell phone that allows them to obtain information on prices of agricultural products in different regions. This "asset," though, may have no effect on development outcomes in terms of increased income or reduced poverty if the lack of road infrastructure pre- vents the farmers from transporting their produce to any of those markets, or if the prices are so low (as a result of market distortions) that they cannot make a profit. The strategies that are necessary to create the economic oppor- tunity structure may be different from those necessary to promote agency, and sequencing issues can become important. While the framework has so far been used to embed empowerment issues into policy analysis and design in only a small number of cases, it has proved useful in several ways: · First, to identify areas for enquiry and information or data collection to inform policy formulation · Second, to establish a transparent set of assumptions regarding the pre- dicted empowerment impact of a policy intervention or sequence of interventions 72 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION · Third, to prioritize, sequence, and structure empowerment interven- tions that would have the greatest impact in terms of poverty reduction Application of the empowerment framework to Ethiopia provides an exam- ple of how this can work. The framework was used to inform analysis of a World Bank poverty assessment that in turn led to a focus on empowerment in the World Bank's forthcoming country assistance strategy for Ethiopia. The macro policy setting of the poverty assessment and country assistance strat- egy is provided by Ethiopia's Poverty Reduction Strategy Paper (PRSP).6 The Ethiopia PRSP, known as the Sustainable Development and Poverty Reduction Program (SDPRP), identifies empowerment as one of its eight major thrusts. It specifies as one of its objectives "improvements in governance to move for- ward in the transformation of society, improve empowerment of the poor, [and set a] framework/provide [an] enabling environment for private sector growth and development." In this context, the following discussion focuses on how the poverty assessment used the empowerment framework in its analysis of the nonmonetary dimensions of poverty.7 The Ethiopia poverty assessment describes itself as comprehensive in scope, but it deals with some issues in greater depth than others because of their prominence in current policy debates in Ethiopia (World Bank 2004b). Empowerment is one of these priority issues. The poverty assessment sought to complement and extend the government of Ethiopia's "Poverty Profile of Ethiopia" (2002) and used analysis from several World Bank eco- nomic sector works as well as numerous other documents.8 Drawing on current dialogue within the government and donor community related to revisions in the SDPRP, the poverty assessment provided an ideal opening for discussion of empowerment in relation to poverty outcomes. This assessment conceived of poverty not just as insufficient income, consumption, or spending, but also as a person's inability to choose to move toward a higher standard of living or state of well-being. Understanding poverty in this way added a useful dimension to the more common utilitar- ian analysis of poverty assessments.An empowerment approach treated pov- erty as a matter of deprivation in the exercise of choice, thus helping to explain why some people are more likely to be impoverished than others. It moved analysis from just the technical, involving the measurement of income, con- sumption, or expenditure, to the relational. This way of understanding pov- erty complemented the utilitarian approach and extended poverty analysis to include assessment of people's relative capacity to achieve a desired outcome given their asset base and institutional environment. Ascreening of existing documentation and data found that in Ethiopia, as in other countries, little representative empirical information exists at the country level on either empowerment or its causal factors (table 5.3). Narrative analyses of power relations were found, but these were scattered Table 5.3. Data Availability and Critical Actions for Poverty Reduction in Ethiopia Level Macro Intermediary Local (federal) (region or woreda) (kebele) Domain Subdomain A OS DOE A OS DOE A OS DOE State Justice Y X Y Y Y Y Y Y Y Politics X X X X X/Y Y X Y Y Public service X Y Y X/Y Y Y ­ ­ Y 73 delivery Market Labor X Y Y ­ ­ ­ Y Y Y Goods ­ Y ­ ­ ­ ­ Y Y Y Private services ­ Y Y ­ ­ ­ Y Y Y Society Intra-household Y Y ­ ­ Y ­ Y Y Y Intra-community Y Y ­ ­ Y ­ Y Y Y Source: Authors. Note: A = Agency. OS = Opportunity structure. DOE = Degree of empowerment. X = Representative empirical data or robust statistical analysis and documentation (for example, legal frameworks) available. Y = Partial data available: case study, small sample size studies, and narrative information available (this does not include monitoring activity). ­ = not available. Shaded cells indicate primary areas of action or intervention. No shading indicates not a priority area for empowerment as a means of poverty reduction. 74 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION and usually focused on assets and institutions without testing their associ- ation with the capacity to make effective choice. Analysis of empowerment for the poverty assessment therefore marshaled evidence from this frag- mented base, using the empowerment framework to examine relations between different groups of indicators. The framework was first used to structure the enquiry, that is, it allowed identification of the key areas for which information was needed in each domain and subdomain and at each level. Available information was then screened to assess which assets and institutions influenced the degree of empowerment found at the interface of domain and level. Finally, the frame- work was used to examine the relationships between different arenas of empowerment and to prioritize the activities most critical to enhancing the government's empowerment goals. The core generic findings of the review were as follows: · The vast majority of Ethiopians suffer from low asset bases and limited opportunities to act on their own choices in their interactions within soci- ety and the market and with the state. This is particularly true for women and pastoralists. · Citizentrustandinterestinformalmarketandstateinstitutionsremainslow, both overall and in comparison to traditional and nonstate institutions. · Informal institutions--including the norms, traditions, and cultural values of Ethiopia--inhibit citizens' overall empowerment by perpetuat- ing social divisions and discouraging participation in the market and gov- ernment decision-making processes. Box 5.4 summarizes the response to these finding in terms of identifying the key issues to be addressed if the government is to further its empow- erment priorities. In terms of statistical analysis, more could have been done with the avail- able data. Nonetheless, this poverty assessment is valuable in that it repre- sents probably the first attempt within the World Bank to seriously take into account the empowerment dimension of poverty reduction. Further efforts of this nature could include greater attention to a more integrated analysis of both the monetary and nonmonetary dimensions of poverty reduction. Policy-Level Monitoring Systems At the national level, the generation of data and analysis on empowerment needs to feed an evidence-based policy process that improves the design and implementation of policies to build assets and institutionalize equality of opportunity. At present there is often a gap between the supply of and the demand for data--as commonly seen in PRSP monitoring systems that are APPLYING THE EMPOWERMENT FRAMEWORK TO OPERATIONS 75 Box 5.4. Empowerment for Poverty Reduction in Ethiopia The actions outlined below focus on what is required to operationalize the policy-level objectives of empowerment. As of 2005, these actions were under discussion with the government and other donors in Ethiopia. First, increase trust in formal institutions: · At the local level of government that is closest and most visible to citi- zens, focus on improving kebele (village) governance to prove the value and relevance of government to citizens. Extend public sector reform programs to enhance the capacities of kebele councils and bureaus to engage meaningfully with citizens over kebele plans, information shar- ing, monitoring, and--where appropriate--implementation. Comple- mentary initiatives include making small quantities of untied budget available for capital spending at the kebele level, and improving the responsiveness of woreda (district-level) council and public service providers and their accountability to kebele representatives. · Promote mutual understanding among local and national government officials and the staff of NGOs of the respective value and limitations of state and nonstate organizations. · Continue commitment to improving the business environment. Key in this are (a) limited state interventions in the market through taxation, land use restrictions, competition, and use of state or party enterprises, and (b) stability in and enforcement of market laws and regulations. Second, address the norms, values, and beliefs that inhibit choice: · Disseminate information on the negative consequences of some tradi- tions and norms practiced across the country through print and broad- cast media, capacity building programs for government officials and citizens, and by financing nonstate organizations to work with tradi- tional leaders. · Retool teacher training, teaching methodologies, and school learning materials to emphasize creativity and inspire students to express them- selves in their own voices. · Improve enforcement of laws relating to violence against children and women, coupled with local government training to recognize and work with communities to address such practices. Third, pursue and support development of more robust analysis of empowerment in Ethiopia: · Investigate barriers to and potential of individuals as empowered eco- nomic actors, focusing on how formal and informal institutional struc- tures influence citizens' ability to make economic choices. · Include indicators for empowerment in national-level monitoring systems. Source: Kurey 2005. 76 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION unable to collect the right kind of information in a timely way (Booth and Lucas 2001; World Bank 2004a). In comparison with project-level monitoring, with its emphasis on coun- terfactual methodologies, policy-level monitoring cannot effectively pay the same degree of attention to causal attribution of empowerment impacts and outcomes. This is because upstream policy instruments that are guided by multidimensional empowerment goals require a complex mix of budget allocations alongside price-based and institutional policy decisions. Attribution of causality within this complex environment becomes a highly questionable and costly task. At this level, therefore, the focus of empower- ment tracking is on an information system that gives greater weight to impact and outcome monitoring, and then explains emerging trends and patterns using qualitative methods. Therefore there is, as discussed in chapter 3, a need for empowerment indi- cators that move consistently in the same direction and that move sufficiently quickly to be useful for understanding the relationship between policy and outcomes.9 These need to inform the analysis and design of policy interven- tions--including budget allocations and institutional reforms--that expand people's choices in the state, market, and political domains. To serve this pur- pose, surveys should be sufficiently light and agile to provide "just-in- time" information, complemented by qualitative research that probes specific policy issues as they emerge. The Empowerment Base Survey Module attached as appendix 6 represents a generic and nonexhaustive founda- tion module that demonstrates the potential scope and content of an empow- erment survey module. Clearly, in many instances this type of modular approach will not be appropriate or desirable. The emphasis in this chap- ter is on identifying options for generating "fit-for-purpose" data systems for national monitoring and evaluation in a context where there is often a lack of technical capability or political will for effective M&E. Such options include the possibility of using panel communities for just-in-time mixed- methods M&E data generation. Box 5.5 summarizes the key purpose and functions of national poverty monitoring systems. Such systems are increasingly being decentralized, particularly in the context of PRSP monitoring, to the regional and district levels. The information generated through local-level monitoring of decen- tralized M&E enhances the accountability of government at the local level, in particular the "short-route accountability" that exists between the providers and users of services (World Bank 2003). Depending on the mech- anism used to channel and deploy information, local-level monitoring can bring citizens and providers together through participatory mechanisms that include participation in health sector committees, community-based school management, and the delivery of social protection instruments. Beginning with early efforts on social audit and citizen report cards, a con- APPLYING THE EMPOWERMENT FRAMEWORK TO OPERATIONS 77 Box 5.5. Purpose and Functions of Poverty Monitoring Systems PRSPs and monitoring strategies usually identify the goals of a monitor- ing system as some or all of the following: · Measuring progress in PRSP implementation against national targets and international measures of development success, such as the Millennium Development Goals · Supporting government decision making on budget priorities, devel- opment policies, and the improvement and updating of the PRSP · Supporting the management and development of sectoral policies and strategies · Accounting for development expenditures, particularly Heavily Indebted Poor Country (debt-relief) funds · Accountability of government to the public for development policies and outcomes · Promoting fact-based dialogue with development partners and civil society · Sustaining and institutionalizing civil participation in the policy process, beyond PRSP formulation · Sustaining informed public debate on development goals and policies Source: Cox and Thornton 2005. siderable amount of work has been undertaken on mechanisms of social accountability, particularly in relation to service delivery.10 Despite this trend, effective national poverty monitoring systems remain an important priority for strengthening policy delivery.As shown in box 5.6, national systems incorporate a somewhat different range of institutional requirements than those at the local level (Cox and Thornton 2005). There is a need to create demand-driven poverty monitoring systems in which managers and policy makers seek monitoring information to improve their decision making. For this reason, institutional lessons from results-based management systems, in which systems of incentives and sanctions are brought to bear on managers, are increasingly being applied to public policy management (Booth and Lucas 2001). In addition to these internal accountability systems built into results- based management, national monitoring systems also encourage external accountability by generating information that fuels evidence-based dia- logue between government policy makers, development partners, and civil society.11 In this institutional environment, it is the transparent flow of infor- mation generated through national poverty monitoring systems that 78 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Box 5.6. Organizational Requirements of Poverty Monitoring Systems The function of national monitoring systems include: · Coordinating among different actors, including developing indicators and targets; agreeing on common information needs, standards, and technical platforms; and filling gaps and eliminating duplication in primary data · Building capacity · Organizing information flows · Compiling data from different sources · Organizing analysis and evaluation · Generating annual progress reports and other monitoring outputs · Disseminating outputs across government and to the public · Providing advice and support to policy makers · Organizing the participation of civil society, both as a means of improving monitoring, and as an end in itself Source: Cox and Thornton 2005. empowers citizens to hold government policy makers accountable for their choices and actions. For information to fuel change, however, it needs to be generated in an institutional context that encourages reflection and action. Experience with public policy monitoring at the national and local levels provides a grow- ing body of knowledge on ways to transform the institutional structures that govern the policy process. This knowledge can usefully inform donor inputs to governments on the design and delivery of PRSPs and other policy frameworks. As with project-level monitoring discussed above, this entails creating new "nested" channels of information flows, including those that involve citizens as primary gatherers of information. These information flows connect citizens with technical specialists and political actors in an empowering realignment of policy makers, bureaucrats, and citizens. At the national level, support to the process of public policy monitoring within the Ghana Poverty Reduction Strategy, for example, seeks to promote a system of outcome-based policy monitoring (Enterplan and CDS 2004). This process relies on a demand-driven and just-in-time approach to infor- mation flows while maintaining a separation of functions among the three sets of actors in the accountability triangle that comprises state, service providers, and citizens (World Bank 2003). Outcome information is gener- ated by using a mix of methods, including through partnerships with civil society actors. This enables policy makers to monitor the performance of the APPLYING THE EMPOWERMENT FRAMEWORK TO OPERATIONS 79 ministries, departments, and agencies delivering public policy while leaving input-output monitoring and the design and delivery of policy up to these bureaucratic entities. In turn, the ministries and civil society are in a stronger position to hold the policy directorate to account for allocating sufficient resources and delegating sufficient authority to deliver public policy. A similar process of outcome monitoring is being developed in Jamaica to monitor progress toward its social policy framework goals (Holland et al. 2005). Groups of community volunteers collect monitoring data that measure progress at the local level. Communities then compare their progress to that of other communities and feed back this evaluation into an ongoing dialogue with bureaucrats and ministers responsible for social policy delivery.12 Operationalizing the Empowerment Framework: A Summary with Questions This chapter has sought to illustrate the applied use of the empowerment framework, showing how concepts can be laid across project and policy interventions, as well as their monitoring systems. A number of key ques- tions can be used to structure application of the framework during design, implementation, and monitoring of an intervention. The first part of this chapter indicated key steps in the process of refin- ing discussion of empowerment in the development of a project or pro- gram intervention. The same basic questions can also be applied in the development of policy operations and monitoring. These questions need to be revisited regularly during any design process, but preliminary answers to the following will help to structure and refine discussions, decisions, and potential operational designs: · Consideration of the empowerment objective. What is the empowerment objective to be pursued or tracked? Is it a means or an end? What domains and levels are involved in the pursuit of this objective? · Clarification of assumptions about predicted impact of intervention. What is the "theory of change" by which the intervention will achieve its empower- ment objectives? How does this theory break down into a series of track- able processes? How do these processes affect and link domains and levels? · Prioritization of primary actions. What is the strength of association of arenas of empowerment (domains and levels) with the objectives of the project, program, or policy? What are the first-order actions or types of information required to meet this objective? · Identification of critical links to secondary arenas of action. Which arenas of empowerment (domains and levels) are secondarily linked to the empow- erment objective? What are the second-order actions or types of infor- mation needed to achieve or track progress toward the primary objective? 80 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Following this, a series of questions can further refine actions required by stakeholders in the process of designing and implementing a project or policy: · Who are the primary stakeholders on both the demand-client and supply- provider sides?13 · What are the stakeholders' roles in different stages of an operation? Does this group or person need to influence design, be involved directly in implemen- tation, or be involved in monitoring, analysis, and management responses? · What degree of empowerment is required? Should the stakeholders (a) just be given the opportunity to engage, so that the opportunity to choose exists; (b) be actively encouraged to engage, that is, use the opportunity; or (c) be able to articulate and demonstrate the effectiveness of their engagement? · Whataretheprerequisitesforthisdegreeofempowerment?Foroptions(a)through (c), what does the project or policy process need to do to ensure that stake- holdersonboththesupplyanddemandsideshaveboththerequisiteagency and opportunity structure? Which assets are needed for these different levels of engagement? Which institutions need to exist and function? This chapter has emphasized the critical role of M&E, or monitoring and learning (M&L) in improving project and policy cycles for empowerment. Effective monitoring relies on the timely flow of good information and on the alignment of organizational actors to respond effectively to this infor- mation (Cox and Thornton 2005). In many instances, policy and project cycles are poorly informed by data and analysis generated on the ground because of a lack of effective demand for data and a failure to generate the right kind of data at the right time (Booth and Lucas 2001). This is partic- ularly true when the subject of monitoring is as new or difficult to capture as empowerment. Therefore, while those seeking to improve project and policy outcomes through better M&E face a number of generic challenges, the need to develop effective systems carrying timely and appropriate infor- mation is all the more important in situations where empowerment is an objective of either a policy or development intervention. Three key areas must be addressed to ensure the generation of useful and institutionally embedded information: · Ensuring evidence-based policy analysis and project management. Are there effective institutional links between the providers and users of policy- and project-relevant information? · Getting the best out of the data. Are data sources and methods for analysis of empowerment being combined and sequenced effectively? · Information and power.Are monitoring instruments being implemented in a way that is sensitive to power differentials within and between social APPLYING THE EMPOWERMENT FRAMEWORK TO OPERATIONS 81 groups?14 Is policy and project M&E fueling a continuing process of institutional transformation and empowerment at different levels of gov- ernance? These questions can be integrated into design processes, and can be used to reflect on and adapt operational design during implementation. Finally, they can assist in identifying the information, institutional arrangements, and actors required for effective monitoring systems.15 Notes 1. Most of the cases and examples cited are drawn from the activities of the World Bank. However, the authors' experiences with other agencies and organizations suggest that the framework is applicable to the work of many other entities involved in development policy and action. 2. Monitoring systems carry information, which in itself is a critical asset for empower- ment (see this chapter's section on monitoring systems). 3. This endowment can either be owned (for example, land) or available for use (for exam- ple, social capital or access to credit). 4. One of the most interesting aspects of this case is its illustration of the value attached to ensuring that intermediaries' skills are appropriate and used to take forward the empower- ment objectives relating to the project's core clients. 5. Supervision and monitoring information for many projects, including those used in this publication for retrospective analysis, indicates that without the use of a framework, a project usually loses focus over time. Adaptations in action, analysis, and financing are all desirable and likely during an intervention. But in many cases where a strong conceptual framework is lacking, activities have broadened, and analysis--where it exists--has become diffuse. However, the purpose of this book is not to evaluate or critique interventions, but rather to use them to illustrate how elements of the empowerment framework can assist in design and implementation. 6. A Poverty Assessment (PA) is a policy document, agreed with the country's government, that is produced every five years by the World Bank.ACountryAssistance Strategy (CAS) frames the World Bank's investment, analytic, and advisory work in a country. Anew strategy is agreed with the country's government approximately every four years. A Poverty Reduction Strategy Paper (PRSP) describes a country's macroeconomic, structural, and social policies and programs to promote growth and reduce poverty, as well as associated external financing needs. PRSPs are prepared by governments through a participatory process that involves both civil society and development partners, including the World Bank and the International Monetary Fund. 7. The framework has also been used to structure analysis of empowerment in the recent participatorypovertyanalysisandtomorebroadlyinfluencetrackingandanalysisofempower- ment as part of the SDPRP monitoring system. 8. World Bank documents include the country economic memorandum and country status reports on education and health. 82 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 9. Poverty monitoring is a necessary but not sufficient component of national monitoring systems. It is necessary to keep poverty outcomes in constant view, but insufficient because of the relatively slow response of poverty variables to policy intervention and the difficulty of attribution (Cox and Thornton 2005). 10. See the work on social audit by the New Economics Foundation, London, at http:// www.neweconomics.org/gen/. See also the World Bank's Community of Practice on Social Accountability (COPSA) and the World Bank Institute's learning module on social accountabil- ity in the public sector (World Bank 2005a). The COPSAWeb site is at http://web.worldbank .org/WBSITE/EXTERNAL/TOPICS/EXTSOCIALDEVELOPMENT/EXTPCENG/0,,content MDK:20509313~pagePK:148956~piPK:216618~theSitePK:410306,00.html. 11. See GTZ (2004) for a discussion of the PRSP initiative as a paradigm shift from conditionality to (evidence-fueled) dialogue. 12. The Jamaica social policy evaluation process is part of a broader emerging paradigm of "democratizing" research, based on the proposition that broad ownership of the genera- tion and analysis of evidence will lead to a more effective and sustainable policy process. 13. Processes of stakeholder identification and analysis can be found in World Bank (2002b). 14. Many guides have been written on how to address gender and other social dimensions within research communities through careful composition of research themes and design of research methods. See, for example, Rennie and Singh (1999). 15. Guidance on monitoring indicators, instruments, and analysis is provided in chapters 3 and 4. Further details of indicators and a base monitoring module can be found in appendixes 4 and 6. 6 Operationalizing a Difficult Concept: Summarizing Empowerment Analysis and Action This book places empowerment firmly within the actionable remit of those involved in designing policy and development interventions. It has done so with good cause. Putting aside the valid humanistic arguments for invest- ing in empowerment, it is clear that there is a growing constituency that identifies empowerment as critical to development outcomes, including poverty alleviation and growth. The discussion in the previous chapters has outlined an approach to operational and analytic work on empower- ment that is based on a review and translation of historic and more recent discussions of the substance and nature of power. The resulting framework can be used to measure empowerment and to better understand the oper- ational needs of interventions to bring about empowerment. This summary chapter briefly revisits the use of structure and agency as determinants of empowerment and summarizes their application to the measurement and operationalization of empowerment. It then recaps how the analytic framework can be used to focus and prioritize both analytic and operational work. Finally, it explores options for use of the framework in monitoring and analysis, described as the "missing link" in the appli- cation of the empowerment concept to development. Structure and Agency: Prime Concepts The material in this book demonstrates that, while empowerment is increas- ingly appearing in the logic and documentation of development agencies, there is diversity in definitions used and in the inference and addressing of causality. Both definitions and determinants tend to be specific to the interven- tion they reference or to the subject matter of the analysis. Diversity is certainly crucial to a relational concept such as empowerment. However, diversity in interpretation is both more credible and more persuasive if it is rooted in solid principles and in a well-articulated theory of social change. The five coun- try studies used in the iterative development of the framework have shown that such principles do, in different degrees, underlie the use of this approach in each location. 83 84 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION The literature referenced in chapter 2 (see also appendix 2) shows that the interaction of agency and structure in power relations has long been posited as the fulcrum for analysis of power. Even when these concepts have not been explicitly recognized, much analysis of empowerment has circled around these ideas. The premise of this analytic approach is that power is a relational concept, as articulated by Giddens (1984) and others, com- prising constituent parts of structure and agency (see for example Rao and Walton 2004). For theoreticians--and some practitioners--this distinction between structure and agency might seem somewhat mechanistic and "binary," but as an applied analytic tool it has proved effective in framing thinking and interventions and in focusing on empowerment as contin- gent on the interaction of these two concepts. In this book, structure, or opportunity structure--sometimes termed the "relative space for achievement" of capabilities--is conceptualized as both formal and informal sets of rules of the game that guide choices, frame relation- ships and procedures, and influence the allocation of tasks and responsibil- ities. The opportunity structure is developed and transformed through the mix of rules, incentives, obligations, and sanctions that govern human interac- tion.Agency is defined as an individual's or group's ability to make purpose- ful choices--that is, to envisage and purposively choose options. Agency is built by strengthening the actor's endowment of assets. These encompass human and productive assets, social assets, informational assets, and psycho- logical assets--the latter including the individual attitudes and the "capacity to aspire." In shorthand, empowerment is defined as a group's or individ- ual's capacity to use their agency effectively through the existence of choice, the use of choice, and the achievement of choice. This conceptualization of effective choice is based on a constructive cri- tique of "utilitarian" or preference-based approaches to choice (Elster 1986). Large parts of economic theory assume that economic actors make rational decisions that maximize utility.Apower-based analysis of institutions holds that rules are rarely neutral in their construction and impact, but tend to emerge to support the dominant ideology or frame of thinking in any given context. Rules are devised by those with sufficient bargaining strength and often "in the interests of private well-being rather than social well-being" (North 1990, 48). By the same token, there is often no incentive for people to give up power in the interest of some common good or utility. In the absence of enforceable contracts and perfect information, those with control over resources, information, and decision making can use their power to compel other people to do what they would not choose to do. Sometimes this is done through coercion or violence, but often it is accomplished more subtly through the creation of consensus about "rules of the game" that are in fact skewed in favor of the powerful. Institutions resist change not only because of the conscious use of power by the powerful, but also because of the lack of con- OPERATIONALIZING A DIFFICULT CONCEPT 85 sciousness among the powerless. Bourdieu (1977) argues that in many contexts choice becomes habitualized and practical rather than reflective and strategic. Going beyond the question of individual choice, the country cases sum- marized in part 1 and presented in greater detail in part 2 of this book sup- port hypotheses regarding the significance of collective empowerment and of the relationship between collective action and individual empowerment. In the case of participatory budgeting in Brazil, analysis confirmed the hypothesis that the presence of a vital and plural civil society is important for empowerment. In the case of school-based management in Honduras, empowerment was built in response to the institutional exclusion of spe- cific social groups from the "community" and related collective action. The complexity of collective action and group-based power is articulated in the Indonesian case study through the notion of "countervailing power," exer- cised in this case by traditionally excluded groups in the space opened up by group-based conflict resolution. The cases of a women's development project in Ethiopia and a rural water and sanitation project in Nepal stress the interactive relationship between individual agency building and col- lective action for institutional change. The five country case studies have also shown that it is possible to iden- tify areas of people's lives in which empowerment can take place and which are common across the globe. While this suggests an opportunity for cross- country analysis, a great deal more work needs to be done to ensure that this deals with anything more than comparisons of trends in domains in dif- ferent locations. Specifically, given the strategic power of statistical analy- sis, the temptation to embark on cross-country regressions must be tempered with serious consideration of the comparability of metrics and values used in each country, at least until further work can be done on the possibilities and limitations of econometric analysis of multidimensional empower- ment data. Using the Empowerment Framework in Operations and Analysis The analytic framework introduced in this volume provides both a theory of change to explain the instrumental value of empowerment and an oper- ational tool with which to design empowering interventions and tools to measure the impact and outcomes of these interventions. The framework goes beyond income-based and utilitarian approaches to poverty "in which real incomes are presumed to translate unproblematically into well-being via utilitarian consumption choices" (P. Evans 2002, 57). Empowerment is based on tackling the differences in capabilities that deny actors the capac- ity to make transforming choices, and hence influence their ability to move out of poverty. 86 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Twogroupsoffactors,referredtoasagencyandopportunitystructure,have beenassociatedwithempowerment,andtheideaofdifferentdegreesofempow- erment has been presented. Because empowerment varies according to context, the framework also suggests that analysts and designers of inter- ventions need to assess the level--macro, intermediary, or local--at which empowerment is to occur. They should further consider the specific domains of life in which individuals and groups may be empowered: as civic (citi- zens or local government) actors in the state domain; as economic actors in the market domain; or as social actors in the societal domain. Operations also need to be based on a clearly understood theory and articulated set of assumptions about the connections that trigger changes in the relationship between agency and structure. Understanding the interaction among assets and the effect of this inter- action on agency presents both analytic and operational challenges and opportunities. Chapter 2 described how the endowment of a single asset, such as ownership of land or capacity to aspire, can affect a person's or group's ability to make meaningful choices. Command over one asset can also, however, affect the endowment of another asset. For example, educa- tion (a human asset) often gives an actor greater access to information (itself an asset) and at times improves his or her capacity to envision alternative options (a psychological asset). Similarly, for groups of people, collective sav- ings (a financial asset) could give access to land (a productive asset). In these cases, more than one asset contributes to the capacity to make meaningful choices. However, an imbalance of assets can have negative effects. For exam- ple, in the market, a person or group might be able to access credit yet have no labor opportunities or purchasing power. Also described in chapter 2 is the interaction that takes place between assets and institutions.Agroup or per- son may have a high level of asset endowment but a poor opportunity struc- ture framing engagement in any domain, thereby reducing or negating the capacity to make effective choice. Unfortunately, these high levels of interac- tion among variables can cause endogeneity problems in statistical analysis. However,giventhepolicy-levelimpactofsuchanalysis,oneshouldnotassume thatanythingotherthandescriptivestatisticsareworthless.Rather,moreatten- tion should be paid to integrating interactive and survey-based methodologies and working with econometricians capable of and interested in understand- ing the options, opportunities, and limitations of regression techniques. Analytic work and operational interventions also need to respond to the distinction between presence and operation of institutions, particularly as mediated by informal norms and value systems. The Ethiopia study, for exam- ple, focused both on the formal rules of the game, such as the laws ensur- ing women's equal treatment within the judicial and political systems, and on the informal rules, such as those enforced by the traditional courts still common in the country. OPERATIONALIZING A DIFFICULT CONCEPT 87 It is in understanding, tracking, and addressing the interplay of formal and informal institutions that change toward empowerment can be better effected. The framework highlights the importance of the enabling environment for empowerment, that is, an opportunity structure that allows people to trans- late their asset base into effective agency through raised consciousness, better information, more equitable rules, and expanded entitlements. The framework prompts policy and project designers to consider and identify the rules, incentives, obligations, and sanctions that will create that enabling opportunity structure for empowerment. Finally, the framework and country cases provide insight into the need for contextual sensitivity when intervening to strengthen collective action in contexts such as Nepal, where there is a proliferation of group-based activity but extremes of power and inclusion and exclusion. The Indonesia case suggests that, in terms of opportunity structure, the key task for devel- opment projects is to create the spaces, resources, and incentives that enable poor people to build effective and durable bases of countervailing power in those domains where decisions most immediately related to their wel- fare are made. The Missing Link: Need for a Coherent Approach to Empowerment Beyond providing a framework to support analysis and operations, this book also points to the importance of a coherent approach to integrating analysis and measurement of empowerment into the processes of project and policy design and implementation. There is a pressing need to under- stand what is missing in existing systems and identify what should be measured. Chapters 3 and 5 applied the framework to operational work and the development of empowerment indicators in different domains and at different levels. Chapter 5 also outlined an approach to developing sys- tems of information gathering that can facilitate "evidence-based" policy making and project design, based on the tracking and diagnosis of empow- erment trends and patterns. The subject of empowerment remains difficult to deal with in terms of quantitative analysis. This perhaps partly explains why the concept is, as yet, poorly researched. While empirical work directly measuring empow- erment is slowly emerging, it remains methodologically and spatially lim- ited, leading many practitioners--particularly those responsible for policy analysis, formulation, and implementation--to consider it less than robust. In addition, as of 2005, no direct indicators of empowerment were to be found in any country's poverty reduction monitoring system, making track- ing and evaluation of empowerment outcomes at the country level virtu- ally impossible. Even taking into account the point conceded in chapter 88 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 3--that empowerment measurements are often contextual and difficult to "universalize"--countries will find it hard to validate the instrumentality of empowerment articulated in their poverty reduction strategies without these data. Similar problems arise in project systems. Of the empowerment indicators that are captured, most are intermediary indicators, leaving ana- lysts with the problem of hypothesizing the impact of empowerment on poverty reduction or other project outcomes. The "theory of change" described above in connection with the design of development interventions also comes into play in the monitoring and evaluation (M&E) of empowerment. Theory-based evaluation explains the patterns and trends revealed by M&E systems in terms of the impact that interventions have had on the relationship between structure and agency and ultimately on empowerment. But theory-based evaluation needs to be supported by empirical data. This means recognizing the importance of designing more effective mixed methods and data for empowerment mon- itoring, to generate information that can effectively describe and explain the indirect and direct impacts of interventions on empowerment outcomes. Toward this end, one message of this volume is a plea for a greater degree of cooperation and mutual respect between noneconomic and economic social scientists. To make use of opportunities to improve analysis and enact empowering policies and practices, as suggested by the empowerment framework, there is need for a pragmatic loosening of the professional "standards" by which anthropologist, sociologists, and econometricians are judged. Most of those involved in applied development work are aware that life is messy and complex, and that this implies analytic problems. However, the disciplinary tendency to avoid work that may be critiqued as either too reductionist or lacking in analytic rigor does nothing to help move forward applied analysis and action. Understanding, measuring, and applying the concept of empowerment to development work implies accepting a degree of professional pragmatism. This does not mean ignor- ing good conceptual grounding. Indeed, the reverse is true: if application is to be effective, anyone brave enough to follow this path should be well able to articulate their rationale and assumptions for the chosen course of action. Systems of deliberation that make effective use of evidence generated on the ground have been recognized as critical to empowerment efforts. The use of "nested" systems of information flows that connect actors (cit- izens, technocrats, and policy makers) in new ways have also been identi- fied as a means of helping them respond effectively to the lessons being learned through the information and analysis of project and policy impacts. Finally, it has been argued that the more coherent use of data through better information systems and institutional connections provides the "missing link" for the effective application of the empowerment concept to devel- opment policy making and project design. PART TWO Measuring and Evaluating Empowerment: Five Country Studies 89 Preface to the Country Studies Part 2 of this volume summarizes five examples of efforts to apply the empow- erment framework. Five different teams undertook the studies in five coun- tries, each working in a different sector. The results are diverse in their scope and depth, demonstrating that whatever prime concepts are used, applica- tion of the empowerment framework must always be context-specific. How- ever, it is also clear from the country case studies that the prime concepts embodied in the framework stand up to application. Part 1 provided an overall guide to what is important to understand, work on, and measure in relation to empowerment, but it deliberately did not pre- scribe a design for data collection, analysis, or an intervention. Rather, a frame- work was offered that can be laid across efforts to measure empowerment and operational work, thereby systemizing approaches to and application of analysis and operations. In each of the five country studies, the research team enjoyed autonomy in terms of deciding the extent to which they would use the framework, the type and scope of data collection they would undertake, the choice of indicators to use, and the kind of analysis they would undertake. The advantages of this approach were that, in five very different applied contexts, the researchers developed a range of indicators, tested various instruments, and used a variety of analytic approaches. However, because of this experimental approach to testing the application of concepts, it was clear from the outset that the case studies would not provide "ideal" exam- ples of how to analyze and measure empowerment. Rather, each case afforded the opportunity to honestly refine the framework, develop indicators, and assess different ways of collecting, recording, and analyzing data. These five country cases constitute a first and valuable step toward applying the frame- work to concrete contexts and interventions. An in-depth evaluation of the strengths and weaknesses of each country study is beyond the scope of this volume. While lessons are drawn out from each case, the volume authors make no judgment on the quality of the researchers' work. Rather, readers are encouraged to approach each case study in a critically constructive manner, using the experience to help hone their own ideas about how the empowerment framework can be used. Each 91 92 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION study can help others working with empowerment by illustrating the clear advantages of having a framework to guide the task, as well as the possi- ble pitfalls when working with as complex a field as empowerment. Chapter 3 discussed the challenges of meaning, causality, and comparabil- ity that arise when operationalizing and analyzing empowerment at any level. The country case teams also encountered these challenges to varying degrees. All the studies faced the first hurdle of identifying meaningful measurements of the kinds of empowerment that each wanted to capture. This challenge was perhaps the easiest of the three to overcome, as all the studies managed to develop solid sets of indirect and direct indicators of empowerment in each field after identifying domains and subdomains relevant for the partic- ular contexts. Though the indicators needed to measure empowerment are obviously context-specific, the ones identified and used by the studies can serve as inspiration and can provide examples of how indicators could be shaped in a specific field of work. A matrix of the direct and indirect indi- cators used by the studies within each domain can be found in appendix 4. The second challenge of causality is one faced in most development con- texts and sociological studies, so it is not surprising that all the research teams struggled with it. In some cases it seems relatively safe to conclude that an increase in a particular asset--for example, a human asset such as level of education--has led to greater empowerment of a specific group. In many other cases, however, it becomes problematic to clearly attribute empower- ment to the increase of one specific asset, because it can be extremely dif- ficult to rule out the possibilities of reverse causality. Did an increase in financial assets of a specific community group, for instance, lead to a greater degree of empowerment, or did a greater degree of empowerment result- ing from changes in agency and opportunity structure cause the group to become economically better off? Four of the five research teams (Brazil, Ethiopia, Honduras, and Nepal) chose to use regression methods to determine the relationship between their different empowerment indicators. The country teams' choices reflect the current disparities found in poverty and development analysis. Some analysts will always be happier with statistical proof of cause-effect relations. Others, however, believe that the field of empowerment is too complex and multidimensional to be appropriately and rigorously analyzed using econo- metrictechniques.Inparticular,thehighdegreeofinteractionbetweenempow- erment variables makes endogeneity always a potential issue in regression estimations. Many practitioners working with empowerment will face this dilemma and will have to choose whether regression estimates are appro- priate for the analysis of certain kinds of data. If they are, researchers must take care to ensure that analytic specifications are sufficiently sophisticated to account for different metrics and types of data as well as the potential endogeneity implied by a highly interactive and multidimensional model. As yet, there is no simple answer. PREFACE TO THE COUNTRY STUDIES 93 Each case offers an example both of what can be achieved by using this type of analysis and of some of the associated problems. The Indonesian study stands alone as one that used ethnographic and descriptive techniques, rather than attempting statistical analysis. While, as a country case, the Indonesia study is useful in highlighting what can be achieved by nonquantitative analysis, it is part of an ongoing broader study that will attempt to carry out statistical analysis of highly interactive variables. The challenge of comparability was encountered in different ways in the studies. Two of the teams, Honduras and Nepal, chose to use indexes as a means to assess different degrees of empowerment of certain groups instead of working with individual variables. Other studies found that this was not a meaningful approach to take in their specific contexts. Opinions on the use of indexes vary; questions can be raised as to whether, in these cases, they add positively to the methodology of an empowerment analysis or lead to the inappropriate mixing of metrics or the use of poorly justified systems of weighting. Readers are therefore encouraged to view these examples of indexes as candid attempts to come to terms with the statistical analysis of multidimensional data and to treat these experiences as useful in making their own calls on when to use or not use such instruments. These methodological issues, discussed in more detail in chapter 3, rep- resent some of the challenges faced by the teams in each study. They are worth noting as significant themes underlying the five country case stud- ies in part 2. Although not all of these problems have been overcome, the case studies can still serve as useful examples for other practitioners who are faced with the task of operationalizing empowerment through project formulation, analysis, or measurement. 7 Evaluating Empowerment: Participatory Budgeting in Brazilian Municipalities Adapted from a study report by Gianpaolo Baiocchi, Patrick Heller, Shubham Chaudhuri, and Marcelo Kunrath Silva For citizens and local government officials alike, local government is a crit- ical domain for the exercise of democratic rights and for making effective choices about public policy. A range of factors, however, conspire against good governance, democracy, and equity at the local level throughout much of the developing world. The social and economic power of local elites often gives them disproportionate influence over the political process, while top- down, insulated, and nontransparent decision-making structures make it dif- ficult for ordinary citizens to have a voice. Democratic deepening begins with the democratization of local government--that is, the empowerment of cit- izens and local government. That is precisely what participatory budgeting (PB) initiatives have tried to achieve in a variety of municipalities in Brazil. Introduced in the city of PortoAlegre in 1990 and implemented since then in at least 200 other munic- ipalities throughout the country, PB introduces direct participation into the process of municipal budget formulation. This chapter reports the findings of a two-part analysis. First, using data from a 10-city, matched-pair study carried out in 2004, researchers examined the impact of PB on empowerment. The study asked whether these reforms promoted empowerment and, if they did, under what conditions this took place. Second, using an extensive data set covering 5,403 municipalities of Brazil along with a range of other sources, researchers generated a data set spanning 1991 to 2000. The effect of PB was then estimated for develop- ment outcomes including municipal finances; public service delivery; human development; and growth, poverty and inequality. The paired analysis showed that cities in which PB took place provided for much more effective forms of engagement than their non-PB counter- parts. The scope of citizen influence ranged from making general demands to specifically shaping patterns of investment and service delivery. Though there was great variation between PB cities, they were all marked by an expansion of the opportunity structure. The introduction of PB had a much more mixed impact on agency, as measured by the capacity of civil society to self-organize. The evidence from statistical analysis of the large data set 95 96 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION indicated a striking positive association between the introduction of PB and reduction in extreme poverty, especially in contexts where the initial incidence of extreme poverty was high. Impacts were also indicated in rela- tion to reducing overall poverty, poverty among children, and inequality, and increasing access to public services and human development. This case demonstrates an interesting use and integration of different approaches to data collection and analysis. The persuasiveness of findings in this case is due in part to the recognition of the different value and differ- ent limitations of each approach. In particular, the potential issue of endo- geneity--which is almost endemic when trying to attribute causality in an area such as empowerment--is largely avoided in the econometric analy- sis. Additionally, the careful sampling and matched-pair approach used in the descriptive analysis allows for a high degree of comfort with findings. The Participatory Budgeting Initiative The highly publicized successes of PortoAlegre in introducing and expand- ing PB have had a significant demonstration effect. Though there is wide variation in the design and implementation of PB, the basic process begins with neighborhood assemblies in which citizens deliberate and set bud- geting priorities. It concludes when delegates directly elected by the neigh- borhood assemblies formulate a citywide budget that incorporates the citizens' demands. In principle, by empowering citizens and their organi- zations to engage in budget decisions, PB marks a dramatic break with the patronage-driven politics that has long dominated municipal budgeting in Brazil. The case that has been made for PB follows the logic of decentralization more generally. The devolution of decision-making authority downward and into the hands of local actors increases transparency, taps into local sources of information, improves accountability of elected officials and public service deliverers, and encourages innovation. In the case of PB it has also been argued that expanding the spaces in which citizens can directly influence resource allocation creates incentives for citizen engagement and strengthens civil society. PB, in other words, is an institutional mechanism for building an empowered citizenry. In its design, PB specifically seeks to expand the opportunity structure for empowerment, both by reducing the transaction costs of participation for the poor and by increasing the trans- action costs for traditional elites.1 Existing research provides evidence that these initiatives have expanded the range of actors participating in the political arena. In Porto Alegre, an estimated 100,000 adults have participated at some point in the budgeting assemblies. Other cities that have adopted some form of PB have also expe- rienced very active participation, including municipalities with little in the EVALUATING EMPOWERMENT: PARTICIPATORY BUDGETING IN BRAZIL 97 way of civil society organization. What the existing research does not allow assessment of, however, are the empowerment outcomes of these reforms across contexts. Particularly confounding is the issue of selection bias. Because these are bottom-up reforms that have evolved organically, it is difficult, on the basis of case studies alone, to separate the effects of poten- tially unusual background conditions from the impact of the reform itself. The research design adopted in the study directly addresses this concern and in doing so offers succinct analysis of both empowerment and other devel- opment outcomes. Research Framework An empowered civil society must have an autonomous capacity for self- expression (agency) and an opportunity structure in which it can effectively and meaningfully engage the state.2 While participatory budgeting creates aformalopportunitystructureforstate-societyengagement(thefirstdegreeof empowerment),the effectiveness ofcollectivechoice(thethirddegreeofempow- erment) is not automatic. It is conditioned on the one hand by the capacity of civil society organizations to make purposive choice (agency) and on the other hand by the nature of both the formal and informal opportunity structure. PB was explicitly conceptualized as a means by which the state could expand the opportunity structure for citizen and local government engagement. Agency was conceptualized in two dimensions: self-organization and mode of engagement. Self-organization refers to the level of internally devel- oped and self-sustaining organizational resources and guiding principles. Civil society organizations may be said to be either dependent (those that do not have the capacity for self-organization and self-determination with- out external support) or autonomous (those that have the capacity for self- organization and self-determination). Mode of engagement refers to how civil society organizations engage with the state. Three modes are identi- fied: associationalism (rule-bound and transparent procedures of demand mak- ing), clientelism (discretionary demand making contingent on loyalty to broker or patron), and exclusion (no access to make demands). Opportunity structure was conceptualized as the presence and operation of rules (institutions) that determine the scope and quality of opportunity for civilsocietytointerfacewithlocalauthorities.Theopportunitystructureincludes both the institutional surface area (the extent and degree of inclusiveness of spaces and points of contact between state and public, that is, formal institu- tions) and institutional processes (how social demands are processed, that is, the operation of these formal institutions). Patronage (demands processed con- tingent on loyalty) and participation (demands processed contingent on open participation) are identified as two possible institutional processes that reflect informal institutions. Political parties are also considered as part of the insti- 98 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION tutional context, and two forms of behavior are used to characterize how infor- mal political rules function. "Oligarchical" parties are considered to be those dominated by powerful individuals, in which the party's identity is largely indistinguishable from that of a specific individual, family, or group of cohorts. More "modern" political parties have an identity that is mainly associated with an organizational program or platform. The matched-pair analysis was relational and context-sensitive. It asked, first, who are the key players involved in the budgeting process and how does PB transform the playing field? Understanding the way that infor- mal institutions and individual and group assets either block or facilitate participation was of particular interest. The role of civil society organi- zations, political parties, and the bureaucracy and the relations between these actors was specifically examined. The design of PB is meant to create new links to decision making for civil society organizations and indi- vidual citizens. This model of empowerment explicitly recognizes that an increase in the role of civil society requires a change in the balance of power. The very design of PB seeks to break traditional monopolies in decision making of clientelistic parties and insulated and unaccountable bureaucracies. The matched-pair analysis considered the balance of power among actors as PB was introduced and the way that this may have affected agency and opportunity structure. The frame and sequence of analysis is shown in figure 7.1. Following the empowerment framework, it was hypothesized that empow- erment can have direct and tangible developmental benefits, and that an increase in participation should translate into better overall governance. Empowerment Impact of PB: The Matched-Pair Analysis Discussion of the matched-pair analysis has four parts. After discussing the selection and matching of pairs, the chapter looks at changes in the ways citizens have been included in budgeting in each location. The sub- sequent section looks at changes in agency and opportunity structures. Selection of Pairs Local research teams began by working in five matched pairs of municípios (municipalities or cities). Each pair shared basic attributes of political party configuration, region, and size, but included one city that had adopted PB and one that had not. The paired analysis addressed two key methodological concerns that have not been fully dealt with in existing research. The first is the need to appropriately construct the counterfactual in implementing an evaluation so as to address concerns regarding the possible confounding effects of unobserved or hard-to-quantify features of the context (for example, the Figure 7.1. Brazil Study: Frame and Sequence of Analysis PB qualities Non-PB Outcomes Context Mode of Opportunity Agency participation Agency structure 99 Formalization of process 5 municipalities Development Opportunity outcomes 5 municipalities Decision-making Institutional Political Model of Self- structure process setting parties engagement organization Scope of discussion PB Source: Baiocchi et al. 2005. 100 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION local history of social movements). In the evaluation literature, this is termed selection bias. A second concern is that existing research does not adequately take into account possible heterogeneity in treatment effects. This raises the possibility that the effects of institutional innovations such as PB might vary with the institutional setting and the political, socioeco- nomic, and historical context. The researchers set out to match PB municipalities with non-PB munic- ipalities based on their degree of similarity in the vote shares of key polit- ical parties in mayoral elections held between 1997 and 2000. To allow for possible heterogeneity in treatment effects, the matching was carried out for different categories of municipalities, defined by size, region, and level of prosperity. Five pairs were selected: one in the south, two in the southeast, one in the northeast, and one in the north. This roughly follows the spatial pattern of adoption of PB in Brazil between 1997 and 2000. The researchers first identified all municipalities in Brazil where the Partido dos Trabalhadores (PT), or Workers' Party, had won or lost by less than 10 percentage points in the 1996 mayoral election.3 The 274 munici- palities that met this criterion were then divided by region, again by size, and finally by the electoral strength of other political parties, and lined up into columns of adopters and nonadopters of PB. Clusters were then iden- tified in which the PB adopter was a city where the PT had won election and where one or more matching nonadopters of similar size had a simi- lar difference in vote shares and a similar configuration of other significant political parties. This yielded a list of 23 PB adopters, each possibly match- ing between one and five nonadopters. Pairs were finally selected keeping the regional distribution in mind, and following the principle of greatest possible similarity between pairs. The Matched Pairs The 10 cities selected were grouped into the following five matched pairs. Mauá and Diadema in São Paulo state are midsize industrial towns, each considered a birthplace of the PT. Diadema's civil society is especially active and contentious, so much so that it has been wary of institutionalizing any engagement with government. In Mauá, the victory of the PT in 1996 ushered in PB, but the city's weak civil society quickly became dependent on the ruling party. This pair of cities produces the most unusual and coun- terintuitive outcome. Diadema emerges as a case of noninstitutionalized empowerment where citizens have significant voice, but largely through contentious politics. In Mauá, the adoption of PB resulted in a form of dependent participation that led to a weakening of the associational auton- omy of civil society. This is the only pair in which the PB city fared less well than its counterpart. EVALUATING EMPOWERMENT: PARTICIPATORY BUDGETING IN BRAZIL 101 Gravataí and Sapucaia do Sul in Rio Grande do Sul have similar socioe- conomic indicators. Both have a solid industrial base, but confront prob- lems of rapid growth and an impoverished low-income population with little access to urban infrastructure. Until 1996, both cities were microcosms of Brazil's political culture. Political power was vested in fragmented oli- garchical parties whose electoral support was built on the strength of clien- telist politics. At the same time, civil society actors who had emerged in the 1980s, in particular the public employee unions, actively built a base of support for the PT. Gravataí adopted PB and developed the most insti- tutionalized, robust, and PB processes of all the study cases. In Sapucaia, despite the rise of new organized political actors, little has changed. The budgetary process remains firmly in the hands of the executive, despite the existence of a formal "consultative system." João Monlevade and Timóteo are both, in effect, company towns in the industrial belt of Minas Gerais, an area known for labor union activism and PT sympathies. Both cities are marked by the strong presence of the steel industry, and they have similar histories and social structures. João Monlevade, literally built by the Belgo-Mineira Steel Company, is today described as a "leftist town" because of the strong presence of the PT, which first ran an administration in 1989­92 with backing from unions and com- munity movements. Countering these political forces are organized busi- ness interests as well as the influence of the steel factory itself. Timóteo is a city with two centers, one "downtown" and one nearACESITA, a steel fac- tory. Like João Monlevade, the town is politically defined by PT-friendly organized labor and organized commercial interests that orbit around the steel factory. Both towns had social movements that fought for proposals, such as more accessible housing. With the introduction of PB in João Monlevade, however, there was a significant opening of governmental access to citizens in the town. Despite the fact that the budgeting process there did not involve large numbers of people when compared to some of the other cases studied, it altered the form of mediation and engagement with civil society. Camaragibe, in Pernambuco, and Quixadá, in Bahia, are both in the northeast. Northeastern Brazil is infamous for its low levels of develop- ment and for the political dominance of traditional oligarchs. Well into the 1990s, political life in both Camaragibe and Quixadá was dominated by traditional families. Though there has been significant growth of civil soci- ety, especially in organizations linked to the Catholic Church, social move- ments have had far less impact than in the south and southeast. In Quixadá, very little has changed. Politics remain personality-based, clientelistic, and very hierarchical. Camaragibe has experienced some reform from above in a pattern that Tendler (1997) has documented for the northeast state of Cerea. A state reformer with links to civil society opened up some participatory 102 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION space in the form of health councils in the 1994­7 municipal government. The popularity of these councils helped build support for the PT, which then won the 1997 election. The introduction of PB in Camaragibe created a wide range of new opportunities for state-citizen engagement, but the overall impact was limited by the city's poor resource base and civil soci- ety's relative inexperience and lack of organizational clout. São Miguel do Guaporé and Mirante da Serra, in the state of Roraima, are frontier towns, where small-scale agriculture dominates. Both are rel- atively new, having been incorporated as part of the expansion into the Amazon in the 1980s, and both are largely free of the entrenched political practices that characterize the rest of Brazil. In the absence of a dominant local elite, politics has been a relatively open and egalitarian affair. The 1990s witnessed the increasing organization and assertiveness of civil society organizations, in particular small farmer associations. In São Miguel, this propelled the PT to power in 1997 and saw the implementation of a fairly robust form of PB. In Mirante da Serra, a continued interest in commercial issues dominated voting outcomes. Civil society remained largely excluded from decision making on the local government budget, but in the absence of clientelistic politics, was able to remain assertive. Citizen Engagement in Budgeting The regional teams examined participation in the budgeting process in the selected cities between 1997 and 2000. In each municipality key informants included administrators at various levels in the 1997­2000 administration, notably officers in charge of budgeting, planning, and popular participation. The mayor and heads of municipal departments were interviewed, along with legislators from the ruling and opposition parties, leaders of civil soci- ety organizations, and heads of local unions, business organizations, and political parties. For each municipality, respondents were asked a series of questions about how the budget is made and how, if at all, citizens are involved in the process. The questions were designed to tease out all forms of citizen engage- ment in budgeting, whether through informal mechanisms such as direct lobbying of the mayor or formally through PB, PB-like processes, or other formal structures such as constitutionally mandated health councils. The type of input was then assessed through four qualitative measures (table 7.1): · Mode of participation: none, direct, delegative, or mixed. "Direct" refers to participation by citizens in open decision-making forums, such as neigh- borhood assemblies. "Delegative" refers to instances in which citizens del- egate authority. It is important to underscore that "delegative" refers only to new forms of representation (in most instances delegate councils) and EVALUATING EMPOWERMENT: PARTICIPATORY BUDGETING IN BRAZIL 103 not to the elected city council structures (formal representative struc- tures). "Mixed" refers to both direct and delegative. · Formalization of the process: none, formal, or informal. This refers to the exis- tence of rules and procedures governing participatory inputs. · Decision-making power: none, consultative, or binding. To what extent are the deliberations of citizens' forums or delegates binding? Given that participatory processes have no legally binding authority, "binding" in this context is a matter of influence and was evaluated on the basis of the observed degree to which municipal authorities took citizen demands into account. · Scope of discussion: none, demands, budget, policies, or mixed. Over what range of governance functions (or domains) did participatory processes exert influence? "Demands" refers to general expressions of needs, "budget" refers to discussion of specific projects and costs, and "policies" refers to discussion of the modalities of coverage and delivery by government departments. "Mixed" refers to both budget and policy discussions. For the period studied (1997­2000), eight municipalities experienced an expansion of the opportunity structure for citizen engagement, specifically the introduction of new forms of participation. The outcomes are summa- rized in table 7.2. As might be expected, all of the PB cities saw the intro- duction of direct and delegative forms of participation (first degree of empowerment--the opportunity to make choice). Diadema was the only non-PB city in which direct participation took place. In all PB cases except Table 7.1. Citizen Participation in Budgeting in 10 Brazilian Cities, 1997­2000 Decision- Scope Mode of Formalization making of Municipality participation of process power discussion Camaragibe Mixed Formal Binding Mixed Quixadá None None None None São Miguel do Guaporé Mixed Informal Consultative Demands Mirante da Serra None None None None Gravataí Mixed Formal Binding Mixed Sapucaia do Sul Delegative Informal Consultative Demands Mauá Mixed Formal Consultative Demands Diadema Mixed Informal Consultative Demands João Monlevade Mixed Formal Binding Mixed Timóteo Delegative Formal Consultative Demands Source: Baiocchi et al. 2005. Note: PB cities are in bold. 104 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION São Miguel, PB was formalized to some degree. Because São Miguel is so small, respondents reported that formalization was not seen as necessary. It is quite clear, however, that the introduction of PB does not always translate into effective choice (the third degree of empowerment). Only in João Monlevade, Camaragibe, and Gravataí did the deliberative process qual- ify as binding (the upper right quadrant of the table) and cover a wide scope of developmental areas (budgeting and policies). In São Miguel, Mauá, and Diadema, participatory inputs were largely consultative in nature (the upper left quadrant of the table) and were limited in scope to the expression of demands (the second degree of empowerment). Finally, in the lower left quadrant are two cases where only delegative structures were introduced (first, but indirect, degree of empowerment) and where decision-making power was consultative (second degree of empowerment). These three quadrants can be further categorized respec- tively as full PB/binding participation, partial PB/consultative participa- tion, and state-controlled participation. A fourth category, contentious participation, may potentially also be found in all three quadrants. Full PB: Binding Participation Gravataí, João Monlevade, and Camaragibe are all cases of binding partici- pation. Acloser look at the process in Gravataí illustrates characteristics of this cluster and indeed of what could be construed as an "ideal type" of PB. With respect to mode of participation, Gravataí combines direct participation in microregional and regional plenaries (more than 80 for a city of 230,000) with instances of representation (forum of delegates and the PB council). The PB process in Gravataí is extremely formalized, with a detailed set of pro- cedures and rules that define the roles, responsibilities, and criteria for the distribution of resources and the manner in which delegates are chosen. Table 7.2. Synthesis of Participation in Budgeting, 1997­2000 Decision-making power Mode of participation Consultative Binding Direct and delegative (mixed) São Miguel (I) Camaragibe (F) Diadema (I) Gravataí (F) Mauá (F) João Monlevade (F) Delegative only Timóteo (F) Sapucaia (I) Source: Baiocchi et al. 2005. Notes: I = Informal. F = Formal. PB cities are shown in bold. Quixadá and Mirante do not appear in the table because no form of participation was introduced in those cities. EVALUATING EMPOWERMENT: PARTICIPATORY BUDGETING IN BRAZIL 105 During the plenaries, the discussion in Gravataí focuses on infrastructure demands and public services. In the PB council, those demands are trans- lated into financial values, which must be considered in the formatting of the budget proposal. Decision-making power is binding, as the process in Gravataí empowers the PB council to deliberate and decide on the public works and services demanded by the population. In each of the four years of PB, the council's budget was incorporated into the final budget. TherehasbeenamarkedexpansionoftheopportunitystructureinGravataí. Citizens are directly engaged in making choices in 85 microregions of 5,000 to 30,000 inhabitants each. Microregional demands and preference hierarchies are aggregated to the regional level where delegates from the microregions participate, and then once more to the level of the budget council, where councilors from the region participate. Because the plenaries are at the neighborhood level, they make participation fairly easy. Participants in Gravataí's PB generally had very little associative experience.4 Because neighborhood associations accustomed to clientelist forms of intermedia- tion opposed PB in the first years, many of the participants chosen as del- egates and councilors were first-time participants in associative life. Participation was high in the first year (6,900 participants), dropped in the second year (3,500), and climbed again in the third year to 13,000, reflect- ing impressive use of choice--the second degree of empowerment. In pro- portion to population, participation in Gravataí is four times as high as participation in Porto Alegre. The voicing of demands is also linked to the decision-making process by the election of delegates who participate in the PB council. The council not only actively discusses actual projects and services but also translates these into budgetary allocations.As a result of these changes, expenditures in health and sanitation as well as in social services increased significantly in real terms as well as proportionally over the four years in question. Social service expenditures went up to 10.76 percent of the 2000 budget, from 1.58 percent in 1996, and health expenditures rose to 11.15 percent from 2.15 percent of a budget that also increased in real terms. Despite being a poor municipal- ity with a per capita budget a fraction of neighboring PortoAlegre's, Gravataí registered some improvements in basic access to education, in adult liter- acy programs, and in the building of new health clinics. Empowerment, more- over, took place across a number of domains, including general governance (specifically in the determination of allocative priorities) and public service delivery. During the period in question, councils in health and in social ser- vice delivery became active. The study found that the opportunity structure also expanded in João Monlevade and Camaragibe. Binding and direct participatory processes for budgeting were established, and the opportunities for citizens to make choices increased.5 106 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Partial PB: Consultative Participation, Even When Contentious In all three cases that fall into the upper left quadrant of table 7.2, namely Diadema, Mauá, and São Miguel do Guaporé, delegative and direct forms of participation have been instituted, but decision-making power remains largely consultative. Here, the impact of engagement is less clear. Measured against the PB ideal, in which citizens are de facto (if not de jure) empow- ered to shape the budget, this form of participatory governance falls short. This does not imply that consultative participation is an inappropriate way to empower citizens. The PB ideal of binding participation is quite rare in even the most "developed" democracies. Citizens can choose or not choose to exercise choice directly. Insofar as civil society is judged to have an important role in Western democracies, its impact has more to do with the "politics of influence" than with binding authority (Cohen and Arato 1995). Most discussions of the politics of influence generally focus on fairly diffuse mechanisms such as opinion formation through the media and efforts to sway decision makers through the "strength of the better argu- ment." Findings on consultative participation suggest that analysis of the second degree of empowerment (use of choice) has to differentiate between direct and indirect use of choice. Consultative participation creates forums in which opinions can be discussed, formed, and publicized; these fre- quently result in effective choice (the third degree of empowerment). Even if outcomes are not binding (or effective), there are a number of ways in which the open and public expression of demands can increase the lever- age of civil society. First, it gives an opportunity for groups traditionally excluded from the decision-making process to form and express choices. Second, it provides new points of accountability for politicians and offi- cials. Third, the public articulation of direct demands can to some extent short-circuit traditional patronage politics. In Mauá the district-level PB plenary meetings did not seek to identify local demands and priorities. Because of the local administration's lack of ability to carry out new investments, PB took place mainly through regional meetings that were largely educational about the state of municipal finances. An additional function of these meetings was to elect councilors to a "par- ticipatory council," where councilors did not decide on the budget but rather were mandated to bring the priorities of their regions and neigh- borhoods to the administration. Councilors described this as an opportu- nity to exert political pressure on the administration and said it had resulted in some significant investments in areas such as health. They also felt that the creation and discussion of documents that listed regional demands and projects fostered accountability. In addition, they reported that limited PB processes led to creation of other participatory forums, such as in the health sector. EVALUATING EMPOWERMENT: PARTICIPATORY BUDGETING IN BRAZIL 107 The case of Diadema is something of an anomaly. Diadema falls into the consultative participation category even though it is not a PB city. Here the influence of citizens is found not in a formal structure but rather in the overall strength and contentiousness of civil society. In Diadema a PB exer- cise was attempted during the first two years of the mayoral administration in response to the demands of social movements. The process, however, was limited to town meetings. The legitimacy of the PB process soon came into question because of a lack of formalized rules, resulting in ultimate decisions being made by the administration, and an apparent failure to attract large numbers of unorganized citizens, resulting in claims that orga- nized groups were able to exert undue pressure. Because the raising of local demands was not linked to empowered citizen decision making (direct or indirect) on the overall budget, or to knowledge of budget constraints, there was a severe mismatch between demands raised and actual projects under- taken. This discredited the project and led to its eventual abandonment. Social activists were, however, able to pressure the administration into publishing an annual listing of projects for each district and neighborhood, along with information on the municipal budget, and organizing training courses on the budget for citizen activists. Social movements were also active in starting participatory councils on health, social services, and edu- cation that allowed citizens to monitor and influence service delivery. The promise of PB mobilized certain organized sectors that, dissatisfied with stillborn participatory attempts, demanded more access and decision making in governmental affairs. Movements were able to gain influence, not through the creation of a regular forum but through sporadic contention. This is very much an instance of the politics of influence, albeit predicated on the strength and militancy of a highly mobilized civil society. This point is fur- ther developed in the next section, but it is important to underscore here that citizens can experience an increase in agency even in the absence of insti- tutional change. State-Controlled Participation The cases of Timóteo and Sapucaia, where participation was limited to the indirect delegative form and was only consultative in nature, represent some- thing of a paradox. On the one hand, new forms of participation were for- mally introduced. On the other hand, the outcome was not empowerment. Insofar as participation in both cases was carefully controlled, even orches- trated by the local government, it actually had the effect of weakening civil society (see next section). There were significant continuities between these participatory schemes and earlier clientelistic forms of mobilization. In both municipalities, politicians selected the delegates to the citywide council. In Sapucaia the mayor appointed the president of the Union of Neighborhood 108 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Associations of Sapucaia do Sul as the director of community relations and exerted tight control over neighborhood associations that participated, even to the point of providing financial support for the creation of new associa- tions in areas where existing associations did not support the mayor. Civil Society Capacity: Agency The second part of the matched-pair analysis focused on the evolution of civil society, looking specifically at the impact of new governance struc- tures on the vitality of civil society organizations. The importance of a vital and plural civil society for empowerment is widely accepted by develop- ment practitioners. In evolving democracies, as a general rule, the greater the density, diversity, and capacity for self-organization of civil society organs, the greater the agency of historically disadvantaged groups in rela- tion to the three key domains of market, state (including political parties), and society. Associational life is in large part an artifact of institutional con- text (Baiocchi 2005; Abers 2000). The Brazil study measured change by comparing the state of civil soci- ety organizations before and after the 1997­2000 period, asking a set of questions to establish the state of civil society in 1996 and the same set of questions again after 2000. The results can be represented along two axes of intermediation and self-organization. As table 7.3 shows, four of 10 municipalities experienced no change in the state of civil society. All were non-PB cities. In contrast, all the PB cities experienced changes in civil society. It is also clear from the table that PB mat- Table 7.3. Civil Society Before and After 1997­2000 Degree of self-organization Mode of intermediation Dependent Autonomous Cities before Cities after Cities not Cities before Cities after Cities not PB adoption adopting PB adopting PB PB adoption adopting PB adopting PB possible possible (pre-1997) (pre-1997) Associationalism Camaragibe Diadema Diadema Gravataí João Monlevade Mauá São Miguel Clientelism Camaragibe Mauá Gravataí João Monlevade Sapucaia Sapucaia Quixadá Quixadá Timóteo Timóteo Exclusion São Miguel Mirante Mirante Source: Baiocchi et al. 2005. EVALUATING EMPOWERMENT: PARTICIPATORY BUDGETING IN BRAZIL 109 tered more for improving links between civil society and governments than for improving autonomy. In Gravataí and Camaragibe, the mode of intermediation shifted from clientelism to associationalism, and PB has promoted greater inclusion of traditionally marginalized social groups. In both cases PB created the formal institutional context for interaction between citizens' organizations and government, with clearly defined and publicly divulged rules (for the most part) that broke with the practice of discretionary demand making that has fueled clientelism. Also, in both cases participation increased every year, and as projects were completed, an ever-larger number of community orga- nizations were drawn into the process. Nonetheless, while there has been empowerment in the form of greater use of choice, this has not led to a strengthening of civil society's capacity for self-organization (an asset). Civil society was unorganized to begin with, and the new actors that PB invited in were mobilized mainly through circles of primary relations such as relatives, neighbors, and friends.6 In the absence of independent civil society organizations, civil society agency remains entirely dependent on the political process and specifically on the support of the PT. In São Miguel, an autonomous civil society took advantage of new avenues of engagement created by PB; this led to greater associational activity, includ- ing the formation of producer groups. Similarly, in João Monlevade, the see- saw of clientelism and contention was displaced by associationalism as the main mode of engagement. Civil society organizations could participate more actively in decision making through PB and through the conference on regional development, and the opening up of the administration's books has led to greater oversight and demands for accountability from civil society. Two municipalities actually experienced a contraction of civil society orga- nizations. Most intriguing was the case of Mauá, which went from having an autonomous civil society based on links of clientelism to having a less autonomous civil society linked through associationalism, or bonds of citizen- ship. In Timóteo, an autonomous civil society became much more dependent. Diadema did not experience a significant increase in formal participation in 1997­2000. Nonetheless, a well-organized civil society exerted significant pressure through more contentious activities such as demanding access to city hall books and demanding improved health delivery. Dissatisfied with the attempt at PB, social movements were able to achieve gains by protesting and mobilizing against a relatively sympathetic and left-leaning administra- tion. Because the contentious mode did secure significant influence, it became self-sustaining. In Quixadá and Sapucaia, where there have been no institutional changes, civil society organizations continue to engage the state through clientelis- tic structures. Mirante continued to be characterized by the exclusion of subordinate groups during this period, in large measure because a well-organized local 110 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION business faction was able to maintain control over the municipality. However, in the absence of clientelistic practices, autonomous and active organizations in Mirante responded to exclusion by mobilizing and protest- ing in the municipality during 1997­2000. These organizations, which had the strong support of progressive Catholic clergy and close ties to the PT, drew together the population of the rural region. The PT won the 2001 elec- tion in part as a result of this mobilization. Agency, Opportunity Structure, and Empowerment This section summarizes findings by combining the two sections above into an aggregated evaluation of empowerment, presented in figure 7.2. The hor- izontal axis represents changes in the opportunity structure, specifically the institutional setting and political parties. It ranges from the most restricted opportunity structure, marked by clientelistic politics, no institutional chan- nels of participation, and personalistic parties, to the opportunity structure most conducive to empowerment: associational politics combined with a direct,broad,andbindingformofparticipationinapoliticalcontextdominated by programmatic parties. The vertical axis represents changes in agency and ranges from the most limited case--a civil society that is not self-organized (dependent) and can access the state only through clientelism--to a civil society that is self-organized (autonomous) and engages the state actively without compromising its autonomy (associationalism).7 Each municipality is represented in 1997 and in 2000. PB cities are shaded, while non-PB cities are not. The general flow of the chart indicates just how much substantive change took place in this relatively short time frame. Only Quixadá and Mirante experienced no change. Given the overall impact that decentralization had in the late 1980s, a significant amount of institu- tional churning would be expected. The presence of the PT is obviously also a driving factor.8 As the figure shows, there are dramatic differences between the trajecto- ries of PB and non-PB cities. All five PB cities experienced an expansion of the opportunity structure, and four experienced an increase in agency. Of the non-PB cities, only Diadema and Sapucaia experienced an expansion of the opportunity structure, and only marginally so. No non-PB city saw an increase in the agency of civil society organizations during this period. Overall, there was more movement along the horizontal axis (opportunity structure) than along the vertical axis (agency). Given that PB primarily attempts to directly change the institutional setting, this is not surprising. However, it is clear that an expansion of the opportunity structure does not necessarily translate into increased agency. The paired analysis was also revealing. All of the case cities were paired by region, size, and PT vote share. In four of the five pairs, the PB city out- 111 Sour Figure Agency ce: associationalism associationalism Baiocchi Autonomous Autonomous Autonomous PB clientelism exclusion Dependent clientelism Dependent cities 7.2. Changes et al. Non-PB Sapucaia Quixadá Camaragibe Clientelistic politics Personalistic 2005. No participation Mirante São cities Miguel Exclusionary politics in No participation political Agency parties Sapucaia Clientelistic politics Delegated, consultative, narrow participation and Opportunity João Mauá Timóteo Gravataí Clientelistic politics No participation O pportunity Timóteo Clientelistic politics Delegated, consultative, narrow participation str Structure uctur Diadema Pr ogrammatic e Mauá Associational politics Delegated, consultative, narrow participation in political 10 Diadema Associational politics parties Brazilian Direct, consultative, narrow participation Camaragibe Clientelistic politics Direct, broad binding participation João Cities, São Gravataí Miguel Associational politics Direct, broad binding 1997­2000 participation 112 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION performed its non-PB mate. The exception was Diadema-Mauá. The paired analysis therefore concludes that · the introduction of PB makes a clear difference for use of choice, · the introduction of PB has a mixed impact on agency, and specifically the organizationalassetsofcivilsociety.InthecaseofMauá,PBhasactuallybeen used as an instrument of political control and has weakened civil society. These outcomes should be understood in terms of the delicate balance between party agency, institutional reform, and civil society. Thus, three important observations on this point emerged from the data: · The impact of PB depends in large part on the preexisting nature of civil society. In the cities where civil society was autonomous to begin with, an opening in the opportunity structure was far more likely to produce an increaseinagency.ThecaseofMauástandsasanimportantcautionarytale. Even a comparatively well-organized civil society can be weakened if it depends too much on access to the state. · The case of Diadema illustrates that a strong and autonomous civil soci- ety can lead to some degree of empowerment even in the absence of institutional reform. · As the cases of Gravataí and Camaragibe illustrate, even when civil soci- ety is weak, concerted reform from above can transform the nature of political intermediation. Poverty Outcomes: Estimating the Effects of Participatory Budgeting The second part of the study used quantitative data from all 5,507 municí- pios in Brazil to examine whether the introduction of PB translated into direct, tangible, developmental benefits. Quantitative evaluation methods were used to estimate the effects of participatory budgeting at the munic- ipality level along a number of dimensions, all related directly or indirectly to the well-being of ordinary citizens and in particular the poor. This sec- tion summarizes the findings from this analysis. Variables in each of the four analytic clusters are given in box 7.1. Methodology and Data Set In the second part of the study, the matching rule was implemented using regression techniques, as in other empirical studies based on the regression- discontinuity design. This provides an example of how regression tech- niques can usefully be employed to measure the effect of empowerment. EVALUATING EMPOWERMENT: PARTICIPATORY BUDGETING IN BRAZIL 113 Box 7.1. Assessing the Development Benefits of Participatory Budgeting: Dimensions and Indicators Four dimensions of PB were explored. In each, indicators included: Municipal finances: Share of capital expenditures Share of expenditures on education and culture Share of expenditures on health and sanitation Share of expenditures on housing and urban development Share of expenditures on social assistance and pensions Share of the budget deficit Public service delivery: Percentage of population with access to municipal trash collection ser- vices Percentage of population with access to municipal sanitation system Percentage of population with access to municipal water system Percentage of population with an electricity connection Human development: Human Development Index Infant mortality rate (per 1,000) Under-five mortality rate (per 1,000) Percentage of children ages 7­14 who are not in school Percentage of children ages 15­17 who are not in school Growth and inequality: Per capita annual income Percentage of population poor Percentage of population indigent Percentage of children ages 14 and below who are poor Percentage of children ages 14 and below who are indigent Income share (%) of poorest 20 percent of the population Income share (%) of richest 10 percent of the population Gini coefficient of income inequality Source: Baiocchi et al. 2005. To assess the performance of municipalities--both PB and non-PB--in terms of each of the indicators in box 7.1, the study constructed two mea- sures. The first measure, one that has been widely used in other contexts, was simply the percentage point change in the relevant indicator between a year preceding the introduction of PB and the year 2000. This measure was constructed for all the indicators. 114 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Certain indicators had a natural target level, that is, a maximum or min- imum attainable level that could not be exceeded. For instance, the poverty rate could never be lower than zero, and the coverage of the municipal water network could never exceed 100 percent. For these indicators the study considered a second measure, an index of progress between two dates, which was defined as the proportionate reduction in distance to the target. For example, a reduction in the headcount index of poverty (for which the natural target level is 0) from 25 percent to 15 percent is a drop of 10 per- centage points, or 40 percent of the original 25 percent level. This would translate into an index of progress of 40 percent (= 100 (10/25). Similarly, an increase in the school attendance rate from 50 percent to 60 percent would translate into an index of progress of 20 percent. For aggregate welfare indicators that had a natural upper bound (or lower bound), defining the index of progress in this way avoided some of the conceptual difficulties that arise in simply measuring progress in terms of percentage point changes. For instance, in assessing progress in poverty reduction, if the measure of performance is taken to be simply the per- centage point difference in poverty rates between two years, a city where the poverty rate fell from 30 percent to 20 percent (a 10-percentage-point decline) would be seen as having made greater progress than a city where poverty fell from 5 percent to zero (a 5-percentage-point decline) and fur- ther reductions were just not possible because the minimum attainable level had been reached. To construct these measures of performance, data on the indicators for all 5,507 municipalities in Brazil for at least two years were compiled from multiple sources. For each group of indicators, the first set of data came from a "pre-year" preceding the introduction of PB, and the second came from the year 2000. For the municipal finance indicators, which were obtained from the Brazilian Institute for Municipal Government (IBAM), the pre-year was 1996. For the remaining measures the pre-year was 1991. Data in the other three areas were either generated from the national censuses of 1991 and 2000 or obtained from the municipality database compiled by the Brazilian government's Institute of Applied Economic Research (IPEA) and the United Nations Development Programme (UNDP), which in turn draws on the census and on a national health database. Because the time frame for the PB initiatives whose impact the study examined was the 1997­2000 mayoral administration, the choice of 1991 as the pre-year, which was driven by data availability, was not ideal. However, unless there were significant differences in performance between PB and non-PB municipalities even prior to the introduction of PB, that is, in the period between 1991 and 1996, any impact of PB during the 1997­2000 period would still be discernible by comparing the performance of PB and non-PB municipalities over the entire decade between 1991 and 2000. EVALUATING EMPOWERMENT: PARTICIPATORY BUDGETING IN BRAZIL 115 Effects on Growth, Poverty, and Inequality Table 7.4 presents basic descriptive statistics, comparing the municipali- ties that adopted PB during the 1997­2000 mayoral administration with the 5,403 municipalities that did not. What becomes immediately clear from this table is the fact that the PB municipalities were, to begin with, quite different from the other municipalities, underscoring the importance of con- trolling for selection bias in evaluating the impact of PB. Municipalities that adopted PB in 1997 were, even in 1991, more prosperous; they had higher levels of access to municipal public services and considerably higher levels of human development and lower levels of poverty than non-PB municipalities. Table 7.5 displays the core results in summary form. For every indicator that had a natural target level, the estimates indicate the impact of PB on the index of progress. Thus, a positive estimate indicates that the introduction of PB had a positive impact in furthering progress toward the desired target.9 The most striking finding was the estimated impact of PB on the percentage of the population that is indigent or extremely poor (second row of the bot- tom panel). The four columns show the estimated effects of PB obtained under each of four different statistical models. Column 1 reports the result of a straight (naïve) comparison of PB and non-PB municipalities in terms of the index of progress between 1991 and 2000 in reducing the incidence of indigence. The reported estimate of ­16.48 indicates that, on average, the index of progress was lower in PB munici- palities than in other municipalities by 16.48 percentage points. This should (and does) exactly match the statistics reported in table 7.4, where it is seen that on average the index of progress in reducing extreme poverty was only 3.94 percent in PB municipalities but 20.42 percent in other municipalities, a difference of exactly 16.48 percentage points. Anaïve comparison would therefore suggest that the introduction of PB was an impediment to poverty reduction. There are, however, at least three problems with such a comparison. First, the naïve comparison does not control for differences in initial con- ditions--in this particular case, the fact that PB municipalities had a lower rate of extreme poverty to begin with. Once initial conditions are controlled for, the estimated "shortfall" in performance of PB municipalities is reduced to an 8.76 percentage point gap, as indicated in column 2. Thus, this estimate too suggests that the introduction of PB was an impediment to poverty reduction. But this estimate is problematic as well because the adoption of PB was at least in part due to unobserved attributes of the local context and the estimate does not take account of this. Column 3 therefore provides the mean "treatment" effect obtained by applying a regression discontinuity model that accounts for the unob- Table 7.4. PB Municipalities Compared with Other Municipalities Means for 103 municipalities with PB in 1997­2000 Means for remaining 5,300 municipalities Percentage Percentage point Index of point Index of Indicator 1991a 2000 change progress 1991a 2000 change progress Municipal finance indicators % municipal expenditures capital 18.05 14.10 3.95 n.a. 15.83 13.93 1.89 n.a. social assistance and pensions 3.84 7.19 3.35 n.a. 7.80 6.51 1.28 n.a. education and culture 30.02 27.51 2.51 n.a. 25.53 33.79 8.26 n.a. housing and urban development 16.49 10.38 6.10 n.a. 11.02 9.09 1.93 n.a. health and sanitation 7.49 22.29 14.79 n.a. 17.53 17.47 0.06 n.a. deficit 8.02 7.48 15.49 n.a. 6.97 6.30 13.27 n.a. 116 Public service delivery indicators % population with trash collection services 70.21 93.12 22.91 77.05 53.01 80.42 27.41 65.43 % of population with water from municipal network 70.62 75.72 5.10 31.37 46.11 58.69 12.57 30.36 connected to municipal sanitation network 35.32 37.17 1.86 0.09 14.74 15.49 0.75 0.45 with electricity connection 90.45 96.30 5.85 47.10 69.62 86.88 17.26 56.91 Human development indicators % children ages 7 to 14 not in school 14.74 3.34 11.40 77.16 26.61 6.04 20.57 77.23 % children ages 15 to 17 not in school 46.07 20.61 25.46 54.10 56.65 27.69 28.96 50.61 Infant mortality rate (per 1,000) 32.97 22.06 10.91 32.76 49.27 33.97 15.31 31.03 Under-5 mortality rate (per 1,000) 43.17 26.76 16.41 38.31 67.12 44.46 22.66 35.31 Human development index 0.00 lowest to 1.00 highest) 0.71 0.78 0.070 24.14 0.61 0.70 0.090 23.08 Growth and inequality indicators Per capita annual income (% change) 215.40 297.60 38.20 n.a. 121.20 170.30 40.50 n.a. % population indigent 15.48 10.84 4.64 3.94 32.70 24.67 8.04 20.42 % population poor 34.87 25.71 9.16 18.66 58.59 46.38 12.21 21.94 % children ages 14 and below indigent 20.57 16.12 4.45 4.93 41.08 33.25 7.83 12.09 % children ages 14 and below poor 42.89 35.68 7.22 1.45 67.09 57.83 9.26 14.10 Gini coefficient of income inequality (0.00 lowest to 117 1.00 highest) 0.54 0.55 0.020 0.040 0.53 0.56 0.030 0.070 Income share of richest 10% of households (% point change) 42.94 44.17 1.23 n.a. 42.15 43.92 1.78 n.a. Income share of poorest 20% of households (% point change) 3.54 2.99 0.55 n.a. 3.98 2.69 1.29 n.a. Source: Baiocchi et al. 2005. a. Pre-year for the municipal finance indicators was 1996. n.a. Not applicable. Table 7.5. Estimating the Effects of Participatory Budgeting on Multiple Dimensions of Development Performance Estimate of mean "effect" of participatory budgeting during 1997­2000 on the CHANGE in indicator between 1991 and 2000 1 2 3 4 Controlling Controlling for Controlling for initial Naïve for initial initial conditions conditions, selection, Indicator estimate conditions and selection and heterogeneity Municipal finance indicators (percentage point changes) % municipal expenditures: capital 1.65 0.34 0.27 0.14 social assistance and pensions 4.50 0.46 0.79 0.38 118 education and culture 10.56 5.49 0.29 0.27 housing and urban development 4.15 0.20 0.28 0.40 health and sanitation 14.94 4.15 6.23 6.32 deficit 3.19 3.05 0.16 0.46 Public service delivery indicators (index of progress) % population with trash collection services 11.63 9.14 11.07 5.29 with water from municipal network 1.01 0.96 33.12 9.86 connected to municipal sanitation network 0.55 13.07 18.90 15.29 with electricity connection 9.81 25.16 17.34 19.35 Human development indicators (index of progress) % children ages 7 to 14 not in school 0.07 1.27 5.57 1.14 % of children ages 15 to 17 not in school 3.49 5.84 4.09 1.22 Infant mortality rate (per 1,000) 1.72 3.13 7.89 6.60 Under-5 mortality rate (per 1,000) 3.00 1.53 2.09 6.43 Human development index (0.00 lowest to 1.00 highest) 1.13 0.31 0.84 3.76 Growth and inequality indicators (index of progress except where indicated) Per capita annual income (% change) 0.00 0.03 0.02 0.02 % population indigent 16.48 8.76 40.86 73.83 % population poor 3.27 5.60 11.78 14.72 % children ages 14 and below indigent 17.03 4.86 40.61 21.31 119 % children ages 14 and below poor 5.68 5.99 10.64 16.99 Gini coefficient of income inequality (0.00 lowest to 1.00 highest) 0.04 0.02 0.07 0.05 Income share of richest 10% of households (% point change) 0.54 0.37 4.65 5.78 Income share of poorest 20% of households (% point change) 0.74 0.39 0.27 0.21 Source: Baiocchi et al. 2005. Note: Estimates in bold are significant at the 10% level or lower; estimates shaded and in bold are significant at the 5% level or lower. 120 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION servable attributes of locations adopting PB. Controlling for possible selec- tion bias brings about a striking reversal of the results reported in the first two columns. The estimated effect now indicates that, once steps have been taken to control for initial conditions and selection bias, the introduction of PB increased the index of progress in reducing extreme poverty by over 40 percentage points. And the estimate is significant. Column 4 reports results from a model that also allows for variation in the "treatment" effect resulting from initial conditions. This in essence provides an estimate of the effect of PB for each initial starting point. This column reports the treatment effect estimated at the mean level for the indicator in the non-PB municipalities. The estimate indicates that had PB been intro- duced in a municipality with a level of extreme poverty in 1991 equal to the mean level of extreme poverty in non-PB municipalities, the index of progress in poverty reduction would have increased by 73 percentage points. While the impact of PB on the incidence of extreme poverty was the most striking, the impacts on several other dimensions were also noteworthy. Estimates of these appear in the other rows of table 7.5 and could be inter- preted just as the estimates of the poverty impacts were. Specifically, the estimates in columns 3 and 4 (the preferred estimates) indicate that the introduction of PB · increased the percentage of municipal expenditures allocated to health and sanitation by over 6 percentage points; · raised the index of progress toward universal access to the municipal water network by over 33 percentage points; · would have raised the indexes of progress in reducing overall poverty and poverty among children by over 14 percentage points and nearly 17 percentage points, respectively, at the mean levels of these indicators among non-PB municipalities; · would have raised the index of progress in reducing the Gini coefficient of income inequality by 0.05 percentage points and reduced the share of income earned by the richest 10 percent by over 5 percentage points, at the mean levels of these indicators among non-PB municipalities. Worth noting as well is the fact that while the estimated impacts on access to municipal sanitation networks and electricity were not signifi- cant, they were all positive and large in magnitude. Conclusion The analysis in this chapter suggests that power relations rooted in polit- ical associations can be challenged, and that changes in opportunity struc- ture can both build assets and empower citizens. EVALUATING EMPOWERMENT: PARTICIPATORY BUDGETING IN BRAZIL 121 Local government is a critical site of empowerment. Traditionally mar- ginalized groups face far lower costs of participation at the local level than at the intermediary or national level, and the possibilities for institutional reform are greater at the local level as well. Yet in countries such as Brazil, local politics has long been dominated by elite interests and local govern- ment has more often than not been an instrument of those interests. The transition to formal democracy has opened up new possibilities for insti- tutional reform. PB represents a comprehensive effort at empowerment. It not only targets a critical source of power--the allocation of local public resources--but does so explicitly by offering incentives for agency and link- ing agency to authoritative decision making. The mixed methodology used in this research for both data collection and analysis provided reliable results showing quite clearly that the adop- tion of PB is predicated on certain political and social conditions--as is true of all substantive institutional reforms. First, PB is inseparably linked to the rise of the PT. Second, PB is much more likely to be adopted in munic- ipalities where civil society is comparatively well developed. Because civil society is such a hard concept to measure, the general statistical models did not provide any direct evidence for this association, but the matched- pair narrative analysis did provide more than plausible evidence. PB is more likely to be adopted in larger municipalities where associational life tends to be more developed. Furthermore, the presence of NGOs did have a measurable effect in the northeast, where development of civil society is more uneven. The fact that most of the cities adopting PB in the northeast were much more populated than cities in other regions and were often provincial capitals reinforces this observation. Third, the negative associ- ation of PB with social exclusion that was discovered in the general model also points to the importance of civil society. What does Brazil's experience with PB say about empowerment? To the extent that the adoption of PB represents an open challenge to vested inter- ests and to traditional politics of clientelism, it represents a form of empow- erment in its own right. More accurately, the adoption of PB can be taken to mark the presence of political and civil society alignments that favor empowering reforms. But does the adoption of PB itself, independent of its political determinants, have an impact on empowerment? And does empowerment result in tangible developmental benefits? The study sought to answer these questions, the first through a detailed qualitative analysis of 10 selected paired cities and the second through a statistical analysis using data from all Brazilian municipalities. The paired analysis of PB and non-PB municipalities yielded a range of important findings. By pairing cities by important attributes (size, region, and strength of the PT), the study was able to develop a more rigorous understanding of the actual impact of PB. A careful examination of the 122 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION processes through which citizens engaged local government between 1997 and 2000 in these 10 cities revealed that PB cities provided for much more effective forms of engagement than their non-PB counterparts. The degree of effectiveness ranged from consultative participation, in which citizens were able to express their demands in an open and organized manner in dedi- cated forums and did informally influence decision making, to cases of binding participation, where citizens were directly involved in shaping the municipal budget. The scope of citizen influence ranged from making gen- eral demands to specifically shaping patterns of investment and service deliv- ery. Though there was great variation among PB cities, there is little doubt they were all marked by an expansion of the opportunity structure. Even the most restricted version of PB had the baseline effect of increasing the flow of information about municipal governance, creating spaces for citizens to voice their demands and subjecting what were once highly insulated and discretionary processes of decision making to public scrutiny and even iter- ated bargaining. The introduction of PB had a much more mixed impact on agency. In some cases, creating new avenues of participation did have the effect of increasing the self-organization of civil society. In other cases, including one in which PB was fully implemented, civil society organizations were not able to develop independent organizational capacity. This suggests that one should be cautious in assuming that an expansion of opportunity (and even incentives for participation) will necessarily strengthen agency. The study found that agency was very much predicated on the existing self- organization and independence of civil society. Moreover, one of the cases also showed that increased access to the state can in fact result in the weak- ening of civil society independence. With respect to material development outcomes, the evidence from the statistical analysis indicated a striking positive association between the introduction of PB and reduction in extreme poverty, especially in contexts where the initial incidence of extreme poverty was high. Significant posi- tive impacts of PB were also estimated in terms of reducing overall poverty, poverty among children, and inequality, and in terms of increased access to municipal water networks. On other measures of access to public ser- vices and human development, the estimated effects were also positive and large in magnitude, but were not significant. Worth noting as well is the fact that the statistical analysis clearly demonstrated the importance of con- trolling for selection bias and heterogeneity in treatment effects in evaluating the impact of PB. Naïve comparisons of PB and non-PB municipalities that do not control for these can obscure the true impact of PB and may possi- bly even yield misleading estimates. While questions may still be raised about the limitations of variables available from secondary sources, this study suggests several important EVALUATING EMPOWERMENT: PARTICIPATORY BUDGETING IN BRAZIL 123 lessons. First, it shows how narrative and statistical analysis can be combined in a complementary manner. Unlike many "mixed-methods" studies, this research creatively used the narrative analysis to do far more than just develop the parameters for a questionnaire-based survey. Second, it shows how careful research design can support robust econometric analysis. And finally, it demonstrates how good sampling can support convincing nar- rative analysis. Notes 1. It does this through four mechanisms. First, it gives citizens a direct role in city governance by creating a range of public forums (micro-regional councils, district councils, sectoral committees, plenaries, delegate councils) in which citizens and delegates can publicly articulate and debate their needs. Second, it links participatory inputs to the actual budgeting process through rule-bound procedures. Third, it improves transparency in the budgeting process by increasing the range of actors involved and publicizing the process. Fourth, it stimulates agency by providing tangible returns on grassroots participation. 2. Participation in PB is arguably far more empowering than voting, the traditional form of citizen participation. PB entails direct participation and active intervention by citizens in shaping outcomes, as opposed to the delegation of decision making that is the hallmark of representative democracy. Moreover, participation in PB typically continues over a period of time, while voting represents an intermittent exercise of choice in the political domain (and is only an indirect way of exercising choice in other domains). Such direct participation can build capabilities through problem solving, communication, strategizing, and what is in effect learning by doing. Because it provides direct forums for engaged citizenship, PB is a potential "school of empowerment" in which people develop skills that are readily trans- ferred to other activities. Such multiplier effects are all the more notable because collective action is a potential resource that the disempowered may have in relative abundance. 3. The choice of this matching rule was motivated by the idea that vote shares for politi- cal parties are likely to reflect (and hence capture) important aspects of the sociohistorical and political economic context. The assumption is that two municipalities in which a party garnered similar vote shares are unlikely to differ much in terms of, for example, traditions of political activism or the degree to which clientelistic relations are ingrained in local politi- cal culture. However, even small differences in vote shares can lead to large (discontinuous) differences in political outcomes--for example, which party ends up controlling the munici- pal administration--which in turn lead in many cases to large (discontinuous) changes in policy, such as the introduction of PB. A matched comparison of municipalities with similar vote shares but large differences in political outcomes that coincide with large differences in policy therefore provides the opportunity to cleanly identify the impact of the policy differ- ence, which in this case is the introduction of participatory budgeting. Under the maintained assumption that vote shares capture the relevant aspects of the local context, this research design is therefore a variant of the regression-discontinuity design. 124 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 4. The median was someone from a developing or underdeveloped urban area with specific demands and problems, but no experience in civil society. 5. More information on these two municipalities can be found in the full country report. 6. This research, with its focus on civil society organizations, did not seek to assess whether PB resulted in empowerment for individuals. 7. From a methodological point of view, this representation of narrative findings demonstrates the effectiveness of diagrams. Narrative information is often difficult for a reader to synthesize and in an effort to put points across clearly, authors are often forced into a reductionism that undermines their message. This case, though, nicely illustrates how a complex set of nonnumeric information can be presented. 8. As a political party that built its presence by working with social movements and by focusing on democratizing municipal government, the PT is by definition an agent of change. But clearly, the presence of a determined and programmatic political agent is only part of the story. The fact that there is significant variation across the cases, despite the constant of the PT presence (the party's vote share in the 1996 election ranged from 36 percent to 49 percent in the study municipalities), suggests that local configurations and institutional designs do matter. 9.Apositive value for the index of progress is always to be desired as it indicates progress toward the desired target level, whether that is a poverty rate of zero or a municipal water network coverage rate of 100 percent. 8 Measuring Women's Empowerment in Ethiopia: The Women's Development Initiatives Project Adapted from a study report by Arianna Legovini The Women's Development Initiatives Project (WDIP), a World Bank­aided project, is currently under implementation in Ethiopia. The project's recent impact evaluation assessed the empowerment status of Ethiopian women in both rural and urban areas and evaluated the effect of participation in WDIPon women's empowerment. The study used indicators in the state, eco- nomic, and social domains, on the assumption that WDIP, a project designed to expand economic opportunities, strengthen self-reliance, and build aware- ness, would affect both empowerment and broader development outcomes.1 As with the other case studies, the main interest for readers of this volume lies in the methodology and indicators used. Results are also reported, which serves two purposes. First, while questions can be raised about sam- pling and parts of the analysis, the results are interesting and largely plau- sible, and they highlight areas requiring further investigation. Second, the case presents some of the methodological problems that any analyst will confront when trying to deal with the difficult question of measuring empowerment. These concern in particular the relation to potential endo- geneity (and hence attribution of causality), the robustness of certain asso- ciations, and the limitations of the data set. Gender Inequality and the WDIP Project Gender inequalities in Ethiopia are pronounced and well-known, and have been widely studied.2 Areview of secondary sources reveals a grim picture of women's human and economic assets, implying severe constraints on their ability to exercise agency. Ethiopia ranks well below the average for Sub-Saharan Africa in terms of the representation of women in govern- ment, and it has a wider-than-average gender gap in primary school enroll- ment and adult literacy (World Bank 2002a). Women consistently have lower levels of education than men, and over 75 percent of women in the country have no formal schooling at all (compared to 50 percent of men). 125 126 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION While in some health indicators females perform slightly better than males, particularly in life expectancy and child malnutrition, women over- all are at higher health risk than men.3 Women are more likely to be infected with HIV/AIDS, and less likely to have ever heard of the disease or to know about and make use of effective prevention mechanisms. A signifi- cant factor contributing to women's lack of knowledge in this regard is their lack of access to outside information through media outlets. Less than 14 percent of women in Ethiopia have access to the media, and women are much less likely than men to have heard of HIV/AIDS through media sources (Central Statistical Authority 2001). Women have a significantly lower employment rate than men. Nearly 43 percent of Ethiopian women are unemployed, and over 36 percent are chron- ically unemployed. Women also have little representation in decision-making positions. For example, although 40 percent of government employees are women, 71 percent of these employees are concentrated at lower levels (World Bank 2004b). In addition to income-generating activities, women work long hours cooking, providing child care, and carrying out domestic chores such as fetching fuelwood and water. They have greater responsibility than men for a wide range of other domestic time- and labor-intensive tasks. The government of Ethiopia recognizes the disadvantaged position of women and has implemented a number of policies, laws, and initiatives to promote women's empowerment, including removal of discriminatory laws fromtheconstitution.WiththeannouncementoftheNationalPolicyonWomen in 1993 and promulgation of the new constitution in 1995, Ethiopia high- lighted its commitment to the equal developmentofwomen.Article25ofthe constitutionclearlyguaranteesequality and makes any discrimination on the grounds of race, color, and sex illegal. However, although violations such as female genital mutilation, domestic violence, and sexual harassment are out- lawed in the constitution, the penal code contains no provisions for adjudi- cating them, and existing laws are often applied by judges in a manner that does not take account of women's rights (World Bank 2004b). The strength of traditional systems is damaging Ethiopian women's abil- ity to live freely as equal citizens (World Bank 2004b). A major component of this appears to be limits to the capacity of the government to enforce the laws that are in place to protect women. Informal rules and norms are strong, and they present severe obstacles to efforts to change views on women's positioning. In the words of one Ketema elder, "A woman should be like the ground: whether you spit, pee or pound on it, it does not answer back." The WDIP is a community-driven development project that seeks to enhance women's empowerment and participation in development inter- ventions by mobilizing them at the grassroots level and capitalizing on their potential to support development processes. It does so by facilitating the for- mation of self-help groups, strengthening existing grassroots groups, and enhancing women's capacity to act collectively, on the assumption that this MEASURING WOMEN'S EMPOWERMENT IN ETHIOPIA 127 will increase the social and economic welfare of their households. The pro- ject also provides training to enhance women's business skills and public awareness of women's issues. The project components include (a) a demand-driven fund that finances activities of women's groups, such as grain and spice processing, handicraft production, and the rearing of animals and poultry; (b) capacity building and training for women's groups and other project stakeholders on orga- nization, facilitation, project design, appraisal, and monitoring and evalu- ation; and (c) information, education, and communication activities that enhance gender awareness. The following section outlines the conceptual framework and method- ology of the study.Analysis of women's empowerment and the determinants of empowerment in the three domains of their lives is then presented, fol- lowed by an overview of the analytic approach and research team's assess- ment of the impact of WDIP. Conceptual Framework of the Study The impact study treated opportunity structure and agency as key determi- nants of empowerment. Because of a closer conceptual alignment with frame- works for measuring participation, rather than empowerment, the original WDIP evaluation's treatment of these elements differed slightly from that suggested in chapters 2 and 3.4 Opportunity structure was treated as com- prising both institutions and assets, and thus was originally measured in the WDIP evaluation as a combination of the presence and operation of "rules of the game" and the "asset endowment" of respondents (figure 8.1). This was hypothesized to give rise to agency. However, because of definitions used, there are no fundamental conceptual differences between the two models. As the variables shown under opportunity structure in figure 8.1 indicate, the WDIP study framework was measuring effective agency, or as it is termed in the rest of this volume, empowerment. Thus, the differences between the conceptual framework used in this study and the one used to structure analy- sis in the rest of this volume are no more than semantic.5 Research Design and Methodology This section outlines the instruments and sampling frame used in this study, and provides a review of the types of variables collected. Instruments and Sampling The study used a mixed methodology, combining a secondary source review with interactive methods of data collection and a questionnaire-based household survey. 128 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Figure 8.1. Conceptual Framework for WDIP Evaluation Agency Opportunity structure Institutions: Assets: market human state physical Influence society financial religion social psychological Outcomes Source: Legovini 2005. The secondary source review screened both legislation and other research work on women's empowerment. This provided information on the exis- tence of the formal institutional framework under which women live. Screening research documentation furthered understanding of the reali- ties of women's lives and helped refine the focus of the study. The interactive fieldwork was undertaken in four kebeles, or villages, in the Amhara region and four kebeles in and aroundAddisAbaba. It involved 80 in-depth interviews with key informants (kebele officials, business women and men, community leaders, elders, WDIP representatives, religious lead- ers, and civil servants). In addition, 24 individual interviews were conducted with community men and women (both WDIP beneficiaries and nonbene- ficiaries), and 16 focus group discussions were held with beneficiary women, nonbeneficiary women, men, and mixed gender groups. This enabled the study team to characterize the situation of women and helped them under- stand how poor local women operated with their environment and how they perceived empowerment. It also enabled researchers to identify outcomes that could be used to measure the impact of WDIP. Results from this interactive fieldwork informed the preparation of the questionnaire survey instrument. Communities for the interactive fieldwork were selected from among the sites where WDIP was operating and where random selection of women by the project had taken place. The selection was purposive. Both rural and urban communities were included as sites for potentially different devel- opment outcomes. MEASURING WOMEN'S EMPOWERMENT IN ETHIOPIA 129 The questionnaire survey covered 1,000 households located in five kebe- les in the Amhara region and six kebeles in Addis Ababa. Households of WDIP beneficiaries (the treatment group) made up approximately half the sample. The other half consisted of households of women who had wanted to participate in WDIP but were eliminated during random selection of project beneficiaries (the control group).6 This sample was used to measure empowerment and study the deter- minants of empowerment. Regression models for key outcome variables were developed to test hypotheses relating to each domain. As the sample was not a random sample of the population, but a random sample of ben- eficiary women and women excluded from the project, mean values for indicators should be interpreted accordingly and should not be compared, for example, with means from population samples (as in Demographic and Health Surveys). This flaw in sampling design illustrates one potential pit- fall that can undermine confidence in findings. The impact evaluation portion of the study compared average results in outcome variables and measured impact as the simple difference in aver- age outcomes between the treatment and the control groups. Because of the absence of a baseline survey, more reliable difference-in-difference esti- mates of impact were not available. Results were therefore interpreted with care, and, where possible, econometric analysis controlled for possible dif- ferences in initial conditions. Responses from the self-assessment portion of the survey were also used to gauge the direction of change in outcomes. However, as self-assessments are notoriously generous on program results, analysis also takes this into consideration. Indicators Both interactive and questionnaire survey instruments gathered data on indicators in the market, state, and social domains.7 In this study, outcome indicators combined those related to empowerment (presence, use, and effectiveness of choice) and broader development outcomes (changes in health, education, and other assets). These mainly corresponded to the local and intermediary levels. The full list of variables used was long, and many intermediary variables (assets and institutions) cut across domains. Box 8.1 provides an illustrative list, grouping indicators appropriate to analy- sis of the three main domains of women's lives. Empowerment Outcomes and Determinants As secondary sources clearly demonstrate, Ethiopian women have low levels of empowerment across the economic, political, and social domains. Even though reforms at the macro-institutional level have taken place, the analysis of primary data undertaken as part of this study supported the 130 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Box 8.1. Illustrative Indicators Used in Three Domains of Women's Lives The following indicators were used in the analysis of women's empowerment: State domain Asset variables include literacy, information, and awareness of women's rights and female representation in government; personal experience; the number of people women can rely on for support; household attributes Institutional variables include legal framework; social rules governing women's participation in local councils; presence and operation of judicial rules and processes Outcome variables include participation in kebele and woreda councils; speaking in public; breaking traditional rules of conduct Market domain Asset variables include human assets, including health, education, market- able skills, and market experience; financial assets, including household income; material assets, including household landholdings; psychological assets, including aspiration and self-perceived capacity to negotiate; social capital, including number of credit relationships Institutional variables include enactment of institutions governing gen- dered access to markets; capacity to negotiate in markets; access to finance Outcome variables include participation in labor force; demand for and receipt of loans; change in income; change in assets Social domain Asset variables include family background; personal authority; knowledge of traditional harmful practices (THP); capacity to aspire Institutional variables include gendered roles and traditional rules of con- duct; domestic violence; norms governing THP Outcome variables include freedom to make domestic and reproductive choices; sending girls to school; getting involved in THP training and stopping THP at home; going out alone; riding a cart; willingness to make independent decisions; happiness Source: Legovini 2005. review findings that discriminatory tradition still rules over most aspects of life for the majority of Ethiopian women. Results from the interactive part of the study suggested that women suffer from low levels of asset endowment in areas such as education, skills, MEASURING WOMEN'S EMPOWERMENT IN ETHIOPIA 131 and experience. Poor services for households combined with women's tra- ditional roles as providers of household labor put high demands on women's time and constrain their participation in paid employment. There are few job opportunities for women, partly because of poor growth in the Ethiopian economy and partly because women have little access to input, output, and credit markets and information. Women are also not particularly happy. Half the women in this survey said they felt sad in the previous week, 31 percent cried a lot, 34 percent did not feel like eating, 29 percent did not feel like doing their work, and 42 percent suffered from restless sleep. Eighty-two percent of Ethiopian women claimed to suffer from stress over economic resources and oppor- tunities, 50 percent over uncertainty about the future, 39 percent over their children's future, 35 percent over control to make decisions in their lives, and 28 percent over loneliness and isolation. Fifty-eight percent of the women said that this stress interfered with their normal activities. However, women's capacity to aspire was apparently undiminished. Eighty-seven percent of the respondents believed their children would have a better life, with more freedom and opportunities. Investigating the determinants of happiness was a challenging task. Women were asked to place themselves on a five-step ladder by matching their level of happiness to one of five expressions on a "smiley face" (very happy, happy, neither happy nor sad, sad, very sad). The resulting vari- able was regressed against age and education, physical and mental health, employment status, location, family background, current income, social capital, and the (infrequent) experience of having been raped. The model developed was able to account for a surprisingly large amount of the variation in women's response to their level of happiness (64 per- cent). Significant variables include parental background: the more edu- cated the father, the less likely is the woman to feel happy. This is intuitive, given the fact that the regression controls for current income levels; given the same income, women of higher socioeconomic background are less happy than those of lower socioeconomic background. Social capital, mea- sured as the number of people one can rely upon for support, significantly increases happiness. Having had days of sickness within the last six months also significantly increases the level of happiness. Although this is coun- terintuitive, it may result from a feeling of relief after having recovered from sickness. Poor mental health, on the other hand, as measured by stress over economic resources, has a large and significant negative effect on hap- piness--as would be expected. Participation in WDIP was significant and positive. This was in addi- tion to a large and significant effect from being employed and to the absence of any significant effect from income, which implied that the mood bene- fits of WDIP participation come from factors other than employment, 132 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION income, or expanded social capital. This might indicate that participants were simply feeling better about themselves, about their place in society, and about their level of control over their lives and those of their children. The counterintuitive findings shown in table 8.1, particularly those relat- ing to rape, highlight a key issue. This is that impact evaluations of this nature can rarely be definitive, as pragmatic considerations often shape the scope of data collection and analysis. Researchers undertaking studies of this kind often face a dilemma when trying to balance quality of analysis and attribution of causality with limited time and other resources. There are two ways to deal with this. First, the analyst needs to be clear before beginning a study about the limitations of the data set and the time available to undertake reliable and sophisticated analysis. At times it may be better to resort to the use of simple descriptive statistics such as cross- tabulations and correlation. This issue is central to the debate on to what extent questionnaire-based data collection and regression analysis are rel- evant and useful in studying complex issues such as empowerment. Second, any reporting has to make these limitations clear. In particular, care has to Table 8.1. Explaining Women's Happiness Dependent variable: Psychological well-being Unstandardized Standardized coefficients coefficients Std. Variable Beta Error Beta T test Sig. (Constant) 1.415 1.216 n.a. 1.163 0.267 Age 0.005 0.016 0.045 0.288 0.778 Schooling 0.050 0.061 0.128 0.810 0.434 Urban dummy 0.082 0.376 0.037 0.218 0.831 Father's education 0.797 0.377 0.417 2.116 0.056 Days of sickness 0.037 0.015 0.431 2.469 0.030 in last six months Stress over economic 0.879 0.327 0.443 2.688 0.020 resources Employed 1.105 0.586 0.403 1.886 0.084 Social capital (number 0.277 0.050 1.077 5.500 0.000 of people can rely on for small amount of money) Cash income 0.001 0.002 0.120 0.547 0.594 (monthly) You were raped 0.591 0.796 0.130 0.742 0.472 WDIP participant 0.702 0.338 0.354 2.077 0.060 Source: Legovini 2005. Note: n.a. Not applicable. MEASURING WOMEN'S EMPOWERMENT IN ETHIOPIA 133 be taken in the attribution of causality. Often it is better to use analysis of a limited data set to examine associations and--if resources allow--to take preliminary findings back to key informants for interpretations of causal- ity. Unfortunately, in this particular case the study team did not have the luxury of returning to respondents for verification and interpretation that would have permitted follow-up analysis of these potentially questionable findings. State Domain The review of secondary sources reveals that Ethiopian women's empower- ment is low in the state domain. Their political representation is among the lowest inAfrica, and so is their ability to affect political decisions. Represen- tation is only 8 percent in the national civil service, less than 2 percent in the parliament, and 5 percent in elected regional councils.8 Women's access to services is lower than men's, and the justice system continues to reference unfair informal norms rather than the more egalitarian formal rules of the state (Kurey 2005). The interactive stage of the study pointed to the problems associated with a dual justice system of regular and traditional courts that perpetuate the use of discriminatory rules against women by delaying the application of modern family law. During focus groups and interviews, women repeat- edly spoke of prejudice in the local courts. Traditional courts were found to encourage women to submit to their husband's will, obey his orders, and stay home. In some divorce cases, women were asked to leave the house- hold and their property to the husband; in the case of the husband's death, courts typically estimated the value of the woman's property at 100 Ethiopian birr ($12) and assigned no share of land or other household property to her. The study again used regression models to investigate some of the fac- tors that determine women's level of awareness regarding their right to, and practice of, engagement in the state domain. Empowerment variables in this domain included awareness or knowledge of female representation in kebele and woreda councils (political subdomain) and perceptions of the fairness of courts (justice subdomain). Explanatory variables included (a) key asset variables such as literacy, access to information (through family and friends), distance from information sources, participation in social groups, and location; and (b) institutional variables such as religious affil- iation, presence and operation of government rules, and social norms. A WDIP participation dummy was added to monitor the project's effect on these outcomes. The model had low explanatory power (16 percent at best), but it nonetheless pointed to some of the factors that were significant in explaining awareness. Using data from the questionnaire-based survey, the study found that, in terms of women's engagement in the state domain, a significant portion 134 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION of women were not aware of just how low the level of female representa- tion really was. Thirty-two percent of female respondents thought that women were as well or better represented in kebele councils as men.Another 10 percent simply did not know. More than 40 percent of women surveyed did not know that women were less well represented than men in kebele and woreda councils, in the courts, and in the civil service. Contrary to intuition, perhaps, rural women were much more aware of this than urban women. Only 12 percent of rural women, for example, thought women were equally or better represented than men in kebele councils. This might be associated with high female participation in women's associations in the rural areas (60 percent versus 39 percent in urban areas). The main rea- sons identified by respondents for low representation in kebele and woreda councils were that women have very limited education and experience, are home-bound, and do not present themselves for election. But the majority of women thought weak representation in the courts and in the civil service was overwhelmingly the result of poor female education (62 percent and 73 percent, respectively). Regression estimates for political engagement found that literacy increases the level of awareness significantly, and so does information (table 8.2). Par- ticipation in social groups, however, is ambiguous: participation in religious groups (tsewas) increases awareness, while participation in funeral clubs (iddirs) seems to be negatively associated with it. Women in Amhara were Table 8.2. Explaining Women's Political Engagement Dependent variable: Awareness of representation Unstandardized Standardized coefficients coefficients Std. Variable Beta Error Beta T test Sig. (Constant) 1.814 0.232 n.a. 7.806 0.000 Full literacy 0.193 0.091 0.067 2.126 0.034 Information from 0.250 0.123 0.063 2.033 0.042 family and friends Distance from 0.001 0.001 0.084 2.598 0.010 telephone (minutes) Social capital: iddir 0.099 0.046 0.075 2.154 0.032 Social capital: tsewa 0.220 0.106 0.068 2.068 0.039 Region 0.696 0.124 0.245 5.616 0.000 Muslim 0.350 0.119 0.123 2.932 0.003 WDIP participant 0.123 0.095 0.041 1.292 0.197 Source: Legovini 2005. n.a. Not applicable. MEASURING WOMEN'S EMPOWERMENT IN ETHIOPIA 135 more aware than their counterparts in Addis, and being a Muslim added significantly to a woman's level of awareness. Participation in WDIP did not have any significant effect. These results were quite consistent across aware- ness variables. Regarding women and the justice system, the majority of women said they are treated equally under the law (91 percent), and that they are protected against violence at home (77 percent) and against sexual violence (75 percent). This appears to contradict evidence provided by these same women. It is also surprising in light of the fact that the law has no provisions against domestic violence or marital rape. Until recently, even nonmarital rape was exonerated if the woman could be coerced into marrying her aggressor.9 When their family is attacked or robbed or when they are raped, women go to the police (80 percent and 74 percent, respectively), but in case of domes- tic disputes or violence they instead seek the help of neighbors, family, and friends (53 percent and 47 percent). As table 8.3 shows, after family and friends, the police receive the best fairness ratings, followed by the traditional courts, then the formal courts. These overall averages do not apply in cases of domestic violence. In such cases women do not use traditional courts, and they give them the lowest rating on fairness. Formal courts are also rated poorly in terms of fairness on domestic violence. The lowest rating for Table 8.3 Women and Justice: Proportion of Institutions Considered "Fair" in Different Circumstances Traditional Formal Friends and Circumstance Police courts courts neighbors N Mean N Mean N Mean N Mean I or my family are robbed or attacked 682 0.94 14 0.71 75 0.80 52 0.87 I or my children are beaten by my husband 554 0.79 13 0.69 67 0.70 43 0.84 In case of dispute with husband 551 0.77 13 0.92 67 0.70 44 0.84 In case of separation or divorce 594 0.84 14 0.79 68 0.72 45 0.93 In case of dispute with neighbors 591 0.87 13 0.92 65 0.75 45 0.89 In case of dispute in business 548 0.89 13 0.92 65 0.77 43 0.95 In case of rape 581 0.85 8 0.75 63 0.76 43 0.91 Weighted average 0.85 0.82 0.74 0.89 Source: Legovini 2005. 136 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION the police also relates to domestic disputes and violence. This suggests that protection of women may be weakest when the issue directly involves male versus female confrontations. It also suggests that wonmen's perceptions of legal protection are influenced by other factors. Logistical models of awareness of women's rights were estimated to investigate the impact of personal experience on the perceived gap between de jure and de facto rights of women. The dependent variable was defined as the difference between the respondent's knowledge of women's rights and her perceptions of women's ability to uphold those rights. Respondents who believed that women should have a certain right but cannot uphold it in their society were assigned a level one of awareness; the others a level zero. Awareness was then modeled against relevant personal experience and WDIP participation. Results reported in table 8.4 confirmed that personal experience is cen- tral to developing awareness around women's empowerment in the legal domain. For example, knowing at least one woman who was raped increases awareness of the gap between the right to be protected and actual protec- tion against sexual violence by 20 percent. Having chosen one's husband decreases awareness of the gap between the right and the ability to choose one's husband by 6 percent, while being abducted increases that aware- ness by 9 percent (table 8.5). On both political representation and protection under the law, therefore, Ethiopian women fare poorly on empowerment. Results from the WDIP evaluation suggest that there are ways to increase women's empowerment in the political domain, and the above analysis points to education as the most critical variable. Second, provision of information and steps to increase women's awareness of issues affecting the quality of their own lives is Table 8.4. Experience and Awareness of Legal Protection against Sexual Violence Dependent variable: Gap between right and ability to be protected against sexual violence Unstandardized Standardized coefficients coefficients Std. Variable Beta Error Beta T test Sig. (Constant) 0.224 0.025 n.a. 8.945 0.000 At least one woman 0.206 0.031 0.207 6.727 0.000 you know was raped WDIP participant 0.145 0.028 0.157 5.107 0.000 Source: Legovini 2005. n.a. Not applicable. MEASURING WOMEN'S EMPOWERMENT IN ETHIOPIA 137 Table 8.5. Experience and Awareness of the Right to Choose One's Spouse Dependent variable: Gap between right and ability to choose Unstandardized Standardized coefficients coefficients Std. Variable Beta Error Beta T test Sig. (Constant) 0.184 0.025 n.a. 7.336 0.000 You chose your 0.064 0.024 0.088 2.677 0.008 husband You were abducted 0.098 0.061 0.052 1.597 0.111 Muslim 0.043 0.025 0.059 1.738 0.083 WDIP participant 0.043 0.025 0.057 1.710 0.088 Source: Legovini 2005. n.a. = Not applicable. important. Particularly crucial is knowledge of the legal and political rights provided by the state and the impact of traditional beliefs relating to gen- ital mutilation, early marriage, and women's subordinate status. The impact of organizations on awareness is not conclusive in relation to the two orga- nizational forms used in the statistical analysis. But the interactive enquiry coupled with the apparent effects of participation in the WDIP project sug- gests that women's association, their capacity to aspire, and their belief in both their ability to speak and their right to be heard are also key to enhanc- ing their ability to make effective choice. Market Domain Around 90 percent of women reported good or excellent housework and cooking skills, but their marketable skills were far more limited: 32.8 percent of women reported good or excellent skills in animal husbandry; 21.4 percent in farming; 14.0 percent in weaving; and 5.5 percent in pottery making. On less measurable skills, women ranked themselves quite high, showing a cer- tain level of self-confidence. Between 60 and 80 percent of all women inter- viewed ranked their money management, leadership, "people skills," and creativity above average. The cycle of low investment in girls perpetuates differences in human capital and market outcomes. Girls continue being employed in nonremu- nerative and nonmarketable activities, while their brothers are allowed time to work on their schooling. Girls are responsible for every type of household chore to a much greater extent than their brothers, and are called to sub- stitute for their mother's labor two to three times more often than boys. 138 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION When a child is kept from school to help with household chores, it is nearly always a girl (92­100 percent of cases). On any weekday, 42 percent of boys and 36 percent of girls spend time on schoolwork, while 11 percent of boys and 63 percent of girls spend time on housework. Poor services continue to inflict costs on women. Only 3 percent of women have piped water inside their house and only 33 percent have it within their compound. The rest must walk to public water points, wells, and rivers to get water. On average, the closest water point is at a 14-minute walk. Lack of gas and electricity force women to use biomass as their main source of cooking fuel (81 percent). Seventy-one percent of rural women collect their own firewood and 11 percent purchase it. Even in urban areas, 66 percent of women use firewood. Together, water fetching and firewood collection together cost women on average 50 minutes a day, and in the rural areas, up to one hour and 20 minutes. Low human capital and poor services combine with inadequate access to markets, financing, and information. Four out of 10 women think it improper to ride a cart or an animal. So, women go to the trade market alone (81 percent) and on foot (64 percent), walking an average of 5 kilo- meters on unpaved road (64 percent). Lack of financing ranks number one among the economic constraints faced by women. This was confirmed by a measure of credit rationing: 56 percent of women claim that they would take a loan today at available rates if they could. For the most part, credit rationing is due to a generic lack of financial development in the country, but it also has a gender component. When women were asked whether lenders trust men or women more to repay debts, half of women thought men are more trusted. But when asked whether men or women are better at repaying their loans, three-quarters of respondents thought women are the better credit risk. In terms of access to information--and potentially also credit--illiteracy is a powerful impediment. Only 13 to 15 percent of women read newspa- pers or posted bulletins. Overall, 60 percent of the women sampled say they are illiterate, but in rural areas this figure increases to 75 percent. Education levels of a woman's father are significant: 66 percent of respondents report- ing a father with no education are illiterate, but this decreases to 32 percent if a father has primary education and reaches a low of 16 percent if a father has some secondary education. The majority of women rely on word of mouth for information, mainly family and friends (85 percent), community meetings (65 percent), and the radio (50 percent). About half of urbanized women watch television. In terms of access to information from any source, the situation is significantly better for urban than for rural dwellers. Cultural constraints notwithstanding, women have little choice but to seek income-generating activity. In the words of one, "We are too poor not to work or for our husband to have an issue with it." Labor force participation among MEASURING WOMEN'S EMPOWERMENT IN ETHIOPIA 139 female respondents is 96 percent, with an employment rate of 75 percent. This compares to their husband's employment rate of 92 percent out of a 95 per- cent participation rate. The much larger active unemployment of women than men (21 percent versus 3 percent) points to gender differences in employment opportunities and agency. Furthermore, employed women earn only 114 birr (US$13) per month on average, while employed men earn almost twice that (220 birr or US$26). When assets and institutional data from the questionnaire survey were used to explain the variation in monthly cash income of female respon- dents, the model successfully explained 89 percent of the variation in the dependent variable (table 8.6). Assets are strong determinants of economic outcomes. Among human capital variables, years of schooling increase income, albeit at a decreasing rate. Returns are highest for the first few years of schooling but fall sharply after that, reflecting the lack of opportunities for more educated women. Years of experience in work also have positive returns, each year of experi- ence adding 9 birr per month to a woman's income. Awoman's ranking of her negotiation skills relative to those of men is also a very significant signal of her ability to earn. For example, a woman who ranks herself "much better than a man" in her negotiation skills gains on average 62 birr more than the woman who ranks herself "somewhat better than a man." Sickness within the last six months significantly decreases monthly income, each day of sick- ness lowering income by the prevailing daily wage (4 birr). Social capital, as measured by the number of people a woman can rely on for a small amount of money, does not appear to be a significant determinant of income. Access to markets is important. Women live, on average, 2 kilometers from a produce market. Each additional kilometer decreases monthly income by 88 birr. Distance from the main trading market, on the other hand, affects income in an unexpected direction. There are fewer trading markets, and the average distance from a woman's home is about 5 kilometers, but every additional kilometer from the market adds to a woman's income an esti- mated 11 birr a month. Results therefore indicate that it is more important to be close to the produce market than to the trade market. This makes sense if women are mostly involved in food processing. In terms of information assets, women who watch television earn 130 birr more than the average. Since television is only available in the urban areas and even there less than half of the women have access, the variable must be a proxy for women living in a better-than-average urban milieu. On aver- age, the closest phone is at a distance of three-quarters of an hour on foot. The estimated parameter is positive but not significantly different from zero. Institutional variables affect income significantly. Parental background, proxied by the father's level of education, is negative and significant, reflect- ing the finding that women from "better" social backgrounds are less likely 140 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Table 8.6. Explaining Women's Economic Empowerment Dependent variable: Cash income (monthly) Unstandardized Standardized coefficients coefficients Std. Variable Beta Error Beta T test Sig. (Constant) 389.699 105.614 n.a. 3.690 0.014 Schooling 108.841 18.792 2.480 5.792 0.002 Schooling square 7.926 1.426 2.275 5.557 0.003 Years of working 8.723 1.934 0.523 4.509 0.006 experience Negotiation skills 61.882 15.676 0.458 3.948 0.011 Days of sickness in 3.924 1.032 0.437 3.803 0.013 last six months Distance to produce 87.986 26.657 0.424 3.301 0.021 market Distance to main 11.332 1.431 0.779 7.919 0.001 market for trading Information: watch 130.291 42.305 0.597 3.080 0.027 television Y/N Distance from 0.551 0.810 0.120 0.680 0.527 telephone (minutes by foot) Social capital (number 3.590 5.757 0.109 0.624 0.560 of people can rely on for small amount of money) Father's education 129.361 35.221 0.546 3.673 0.014 Economic constraint 13.124 7.974 0.234 1.646 0.161 (cultural restrictions) Region 79.497 38.216 0.349 2.080 0.092 WDIP participant 100.993 33.047 0.403 3.056 0.028 Source: Legovini 2005. n.a. Not applicable. to work outside the home. When they do engage with the market, they work fewer hours. Cultural restrictions, as a self-identified economic constraint, do not appear significant, while location--hypothesized to reflect different social milieus--definitely is. Living inAmhara rather thanAddis decreases earnings by 79 birr per month. This might also reflect purchasing power differences, and/or the level of economic opportunities for women in the capital city. Finally, the research team's analysis pointed to a strong positive effect of WDIP on economic outcomes. Participants earn between 35 and 167 birr more a month than nonparticipants (98 percent confidence interval). MEASURING WOMEN'S EMPOWERMENT IN ETHIOPIA 141 Social Domain In the social domain, within their home and their community, women sub- mit to traditional codes of conduct. Most women believe it is improper for them to engage in a whole series of activities and behaviors that men are free to do. They believe, for instance, that it is improper for them to go to a bar (84 percent), walk at night alone (67 percent), disobey their husbands (74 per- cent), ride a bike or a cart (42 percent and 39 percent), or wear trousers (35 per- cent). Their freedom is curtailed not only by force but by adherence to their own beliefs. In addition, women submit to and perpetuate a host of traditional prac- tices. Supporting trends identified by the National Committee on Traditional Practices in Ethiopia (Terefe 2002),10 this study found that respondents expe- rienced genital mutilation (82 percent), extraction of milk teeth (32 percent), and forced and early marriage (11 percent before age 12, 60 percent before age 18). In this sample, only 55 percent of women were able to choose their own husband and 4 percent were abducted; 31 percent of respondents know at least one woman who was abducted. While few admit to having been raped (2 percent), many say they know at least one woman who has been raped (26 percent). These practices are even more common in the rural areas, where 95 percent of women have been genitally mutilated, 40 percent have had their teeth extracted, 15 percent were married before age 12, and 75 percent were married before age 18. When asked whether their daughters will be subjected to the same practices, most of the women deny it. Regression estimation indicated that living in Amhara rather than in the capital city tends to decrease the age at which women marry by almost two years. Schooling will significantly delay marriage, each year of schooling adding almost half a year to the woman's age of marriage. A girl will also marry later if her social background is lower than that of the prospective hus- band. Finally, the model confirms that Muslim women marry later than Christian women by as much as one to two years on average. A host of traditional practices and beliefs affect marital relations and intra-household bargaining. Married on average at age 17 to a man of 25, the young wife typically moves to her in-laws' house and is immediately at a bargaining disadvantage. Her belief that she must obey her husband (74 percent) and that she cannot openly voice disagreement with him (69 percent) or with her in-laws (44 percent) limits her control over her own life and her ability to make household decisions. In the great majority of households, the husband's supremacy is enforced through physical beatings. "The woman is like a donkey," a traditional saying goes, "she must be beaten to ensure her docility." Wifely transgressions that can trigger a beating include cheating (82 percent), disobeying the hus- band (62 percent), saying or doing something the husband disapproves of (58 percent), or disagreeing openly with the husband (52 percent). Husbands 142 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION also batter because they are drunk (66 percent), jealous (68 percent), or simply want to maintain their authority (35 percent). Almost half of the female respondents identify at least one acceptable reason for which a man may beat his wife. This includes one-fifth of women who think their husband has a right to beat them when they disobey him, and 5 percent who uphold his right to beat them regularly. These results again reflect the findings of a nation- ally representative study (Central Statistical Authority 2001). Regression results indicated that stress in family life is significantly and positively correlated with all domestic beatings, whether they are done reg- ularly to maintain the husband's authority or occasionally for specific rea- sons. The only exception is beating resulting from a wife's infidelity. Living in urban areas significantly increases the frequency of beatings. While this may have various causes, the interactive enquiry suggested that the main explanatory factors are the availability of alcohol in cities, urban stress, and men's attempts to maintain tradition within an urban context where women have more opportunity to break with traditional rules and roles. Indeed, women's awareness of their rights (and of the lack of protection of those rights) tends to be associated with more spousal beatings, even though the direction of causality remains unclear. But so does being a more traditional wife, one who believes it is improper to go to a bar alone. Traditional women may suffer more from beatings because they are more willing to accept them. But possibly the most interesting result of all was that women who chose their own husband, which represented 45 percent of the sample, are beaten less than women whose husbands were chosen for them. The latter group includes women who married as a result of abduction and rape. Women's degree of autonomy within the household in terms of deci- sions on expenditures is limited, but they are not excluded from these deci- sions. Sixty-seven percent of women make small consumption decisions on their own but are consulted by their husbands about decisions on larger consumption items and investments. Sixteen percent of women would make their own decision on big food items, and about one in 10 would make a decision on adult clothing, children's clothing, medicines and med- ical care, and school supplies. It was in the realm of reproductive and sexual decisions where the study found women to be the least empowered. Thirteen percent of men will make the decision to have another child without consulting the wife, but no woman would make such a decision alone. Similarly 7 percent of men versus only 3 percent of women will make decisions about contraception without con- sulting the spouse. And while almost no wife will initiate sexual encoun- ters (0.4 percent) or even discuss the possibility with her husband (1 percent), 27 percent of men initiate sex with their wives without consulting them. What then are some of the factors that help to explain women's control over domestic decisions? One hypothesis of the study was that lower levels MEASURING WOMEN'S EMPOWERMENT IN ETHIOPIA 143 of economic dependency associate inversely with women's intra-house- hold power. This would result from both the authority that comes from having a job or running a business and from contributing a greater share to household resources. Another hypothesis was that relative social standing of women vis-à-vis their husband might affect their bargaining power. These hypotheses were tested through a series of logistic regressions look- ing at the effect of income, WDIP participation, and family background on the ability of women to make decisions on their own. The first hypothesis was confirmed through econometric results while the second hypothesis was not. Regarding decisions on large food items, adult clothing, children's clothing, medicines and medical care, school sup- plies, and children's schooling, woman's income significantly increases independent decision making while participation in WDIP does not. Living in urban areas also strengthens independent decision making, but family background, as measured by father's education, affects it negatively, when significant.Awoman'sincomealsoaffectsherabilitytomakedecisionsonpur- chases of animals and farm inputs, but not on purchases of business tools or inputs,householddurables,kitchenutensils,farmtools,oryardanimals.Finally, theeffectofparticipationinWDIPissignificantandpositiveinaffectingsheep and cattle purchases. This makes sense because many WDIP women are involved in sheep-fattening activities, and their business decision would be made within their group and not within their household. In terms of engagement within the community in which they live, the women interviewed all belong to at least one group. As table 8.7 illustrates, there were some differences between rural and urban respondents. The main reasons for participating in a group were to get emotional sup- port in the case of iddir (72 percent); to pray in case of a tsewa (73 percent); and to obtain financing in the case of iqqub (79 percent). The outcome of partic- ipation is consistent with the initial motivation. Eighty-three percent received emotional support from an iddir, 34 percent felt they had nurtured their spir- itualityinatsewa,and53percentearnedmoreincome(eventhoughonly13per- cent obtained financing) from an iqqub. Twenty-six percent of women also reported increased confidence in public speaking as a result of their partici- pation in iqqub. Usingdatafromthequestionnairesurvey,thestudyfoundthatmostwomen consider their attendance in political and social community meetings to be the same or less than that of men (about 75 percent). However, 63 percent of respondents feel that women speak less than men at these meetings, and almost half believe that women's views have less impact than the views of men (48 percent). Results from interactive data collection suggest that "it is one in a thousand the woman who can speak in public and make herself heard." But that one woman, men say, will speak for all women and make a difference. She will speak of water, sanitation, and harmful traditional 144 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Table 8.7. Women's Associational Life: Participation in Groups among Rural and Urban Women Percentage of women participating Group Urban Rural All Iddir (funeral group) 90 92 91 Iqqub (collective savings group) 17 9 14 Tsewa (religious group) 24 3 16 Farmers' association 1 21 9 Women's association 38 60 46 Credit and savings association 6 4 6 Teachers' association 0 1 1 Youth association 4 3 4 WDIP group 61 87 71 Other group 0 2 1 Source: Legovini 2005. practices--issues that would not otherwise be raised. Conversations with focus groups also reveal that continuous participation in meetings teaches women to be more forthcoming in expressing their opinions in public. Aregression model of women's influence on community decisions indi- cated that women considered attending meetings to be relatively unimpor- tant. What is important for influencing decisions is to make oneself heard and to have that voice be effective. Regression estimations showed that age and level of education has no bearing on whether women think they can make a difference, but personal authority does (table 8.8). Overallempowermentinthesocialdomainisweak,bothwithinthehouse- hold and within the community. Economic achievement is one factor help- ing women strengthen their control over social outcomes, but this is clearly not enough. Strengthened awareness and legal protection of women's rights and protection against violence and abuse has to be part and parcel of an approach to solidify women's social empowerment. And cultural change will be affected by increased education, information, and training. WDIP and Women's Empowerment The WDIP targets poor and disenfranchised women and combines economic opportunity and financing with social capital building, skills development, and increased awareness of women's issues. WDIP participants over- whelmingly claim that they feel happier, less lonely, and less isolated (80 per- cent), and have become someone others look up to (63 percent). In addition, as discussed above, when the country case study team controlled for edu- cation, family background, and social capital, among other variables, WDIP participation became a positive and significant determinant of happiness, MEASURING WOMEN'S EMPOWERMENT IN ETHIOPIA 145 Table 8.8. Women's Influence on Community Decisions Dependent variable: Women's capacity to affect decisions Unstandardized Standardized coefficients coefficients Std. Variable Beta Error Beta T test Sig. (Constant) 0.188 0.162 n.a. 1.159 0.247 Women attend 0.009 0.024 0.010 0.393 0.694 political community meetings Women speak up at 0.153 0.032 0.135 4.755 0.000 meetings Women's views are 0.810 0.027 0.791 29.781 0.000 taken into consideration Schooling 0.005 0.008 0.013 0.643 0.520 Age 0.003 0.003 0.017 0.849 0.396 Source: Legovini 2005. n.a. Not applicable. accounting for an almost one-step change in the level of happiness on a five-step ladder. The impact analysis summarized in this section focused on the economic and social domains of women's lives.11 It relied on two sources: participants' self-assessment, and simple difference in average outcomes between the treat- ment and control groups at the time of observation. The latter assumed that those differences were zero at the beginning of project implementation due to randomization of self-selected petitioners, but this assumption may not fully hold. Judging by average endowments, which are not likely to be affected by the project, the study team concluded that women in the control group are somewhat better off than women in the treatment group. This indicates that randomized selection of beneficiaries was not followed. For example, 45 per- cent of nonbeneficiaries are literate versus only 38 percent of beneficiaries. Nonbeneficiaries married at age 18 on average and beneficiaries at age 17 (which indicates lower social background or more rural setting). Because of this, measured differences in outcomes (measured impact) can be considered a lower bound on actual differences (true impact). Measured differences will underestimate true differences; therefore econometric analysis in this and the previous section was used to control for differences in initial conditions by using variables not affected during the life of the project. Self-assessment, on theotherhand,tendedtooverestimateimpact,andcanbeconsideredanupper bound on true impact. 146 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Impact in the Market Domain Beneficiaryself-assessmentindicatedthateconomicoutcomesimprovedgreatly under WDIP. Women reported on average a 59 birr change in monthly income (accounting for 48 percent of current beneficiary income) and a 420 birr change in accumulated assets (accounting for 20 percent of total household assets). Women also reported a 28-hour change in the hours of work per week. This implies a daily wage of almost 4 birr (the prevailing female daily wage in Amhara).Inaddition,65percentofwomensaidtheycannegotiatepricesbetter than before and 74 percent claimed they have gained business skills. Measured average differences between treatment and control, however, were negative. In the control group, women had 13 birr more of income per month and 211 birr more in household assets. This confirms that asset endow- ments need to be controlled for when comparing treatment and control groups because of differences in initial conditions. When controlling for asset endow- ment and access variables, the impact of WDIP participation was a 101-birr midpoint estimate of a 35 to 167 birr confidence interval (98 percent). From the econometric and self-assessment results, it could be concluded that WDIP is an effective mechanism for improving economic outcomes. Beneficiary self-assessment also indicated that the largest impact is obtained by older groups--the ones that received project funding more than two years ago (93 birr change in income and 587 birr change in assets). This implies that measured impact can be expected to improve as businesses age. Indeed, most women were well aware that they had not yet achieved satisfactory levels of income, but they were also confident that profits would increase in the future. Another finding was that the most lucrative busi- ness by far for WDIP groups is sheep and cattle husbandry (112 birr change in income and 613 birr change in assets). Market indicators suggested that beneficiary women are engaged in more tradable activities than nonbeneficiary women. Between 12 and 15 percent more beneficiaries than nonbeneficiaries go to the produce and trade mar- kets to buy inputs for their activity and sell outputs. While most women walk to the market, beneficiary women are more likely to be accompanied by some- one else. This might be because of the increased load they carry and might indicate that their activity is now recognized as economically viable and deserving of support. In addition, fewer beneficiary than nonbeneficiary women think access to market is a constraint to their economic activity. Only 21 percent of beneficiary women rank access to market among the top three problems affecting economic activity, while more than 28 percent of nonben- eficiary women do. Financial markets are still elusive. Beneficiary women are less likely to ask for a loan or to receive one. In addition, the average size of the loans obtained is less than half that of nonbeneficiary women. As a result, beneficiary MEASURING WOMEN'S EMPOWERMENT IN ETHIOPIA 147 women feel more credit-rationed than nonbeneficiaries, and are more likely to rank financial constraints among the top three barriers to economic activ- ity (87 versus 84 percent). Impact in the Social Domain The majority of participants said that since becoming engaged in WDIP they have more freedom (71 percent), have more power to make decisions in their household (59 percent), and have broken some of the unwritten rules such as the prohibitions against wearing trousers (13 percent), going out alone to public places (55 percent), and riding bicycles, animals, or carts (13 percent). Eighty-one percent said that they have learned more about tradi- tional harmful practices, have stopped them in their own households (64 percent), or are involved in teaching others about them (55 percent). Thirty- five percent of women said they are now sending their children to school. As shown in table 8.9, WDIP participants are less likely than nonpartici- pants to uphold informal social rules. They are 6 percent less likely to believe farming is improper for a woman (although 7 percent more likely to believe construction is); 7 and 6 percent less likely to say that a woman should not ride animals or carts; and 10 percent less likely to say she should not walk alone at night. But a greater percentage of WDIP participants believe that they should obey their husbands, and more of them uphold their husbands' right to beat them. Even though so many more WDIP participants received training on tra- ditional harmful practices than nonparticipants, they are still more likely than nonparticipants to genitally mutilate their daughters, extract their milk teeth, and marry them off at an early age. There were no consistent differences in beneficiary and nonbeneficiary participation in groups. Nonbeneficiaries tend to have a stronger showing in funeral clubs, religious groups, and credit and savings associations; ben- eficiaries show greater participation in women's associations and informal credit clubs. There are, though, significant differences between beneficiary and nonbeneficiary women in the number of people they can rely on for a small amount of money (five versus three). Seventy-eight percent of par- ticipants say they now have more friends and people they can rely on in case of need. In addition, 73 percent say they have more access to kebele officials, and 59 percent say that they have learned to speak up in public. This is reported to be a direct impact of group activity. The study's regression analysis (table 8.10) confirmed the descriptive statistic results. WDIP participation was positive and significant in deter- mining this social capital variable. Other significant variables included lit- eracy, creativity, and religious affiliation. As would be expected, urban location tends to decrease social cohesion. 148 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Table 8.9. Social Outcomes: WDIP Beneficiaries and Nonbeneficiaries WDIP WDIP nonbeneficiaries beneficiaries All Outcome (%) (%) (%) Women who believe it is improper for a woman to Farm (plow or harvest) 47 41 43 Construct or maintain a fence or house 19 26 24 Buy grain 3 3 3 Ride bike 40 43 42 Ride animal 26 19 22 Ride cart 43 37 39 Go to market alone 9 7 7 Walk at night alone 74 64 67 Go to a bar or club alone 88 83 84 Wear trousers 35 35 35 Trade cattle 26 23 24 Disobey her husband 72 75 74 Women who believe husband has a right to beat Regularly to maintain authority 3 5 5 When his wife disagrees openly with him 6 10 9 When his wife disobeys him 17 19 19 When wife does or says something he disapproves of 9 14 12 When he is jealous 6 12 10 When his wife cheats on him with another man 51 44 47 When his wife gives their money to someone else 37 27 30 When he is drunk 2 5 4 In other circumstances 20 29 25 Never 49 51 50 Women who received training on Female genital mutilation 51 77 68 Milk teeth extraction 49 76 67 Early marriage 51 77 68 Women's rights 53 78 70 HIV/AIDS 56 82 73 Tattooing 46 76 66 Daughters will Be genitally mutilated 6 11 9 Have their teeth extracted 3 10 8 Marry before age 12 1 10 7 Marry before age 18 6 13 11 Source: Legovini 2005. MEASURING WOMEN'S EMPOWERMENT IN ETHIOPIA 149 Overall, WDIP is found to have a positive impact on economic outcomes. However, results are ambiguous in terms of social empowerment. WDIP might have had some impact on social mores, but it does not appear to have had a major impact in this domain, at least not in the two-to-three-year period under observation. Conclusion The study team reviewed the level of empowerment in three main domains of women's lives--state, market, and social--and found that further improvements are needed in the conditions of women in Ethiopian soci- ety. In the state domain, the team's analysis suggests traditional beliefs and low education serve to severely limit women's participation. Efforts to orga- nize women help teach them to speak up and influence decisions. But the analysis indicates more effort must be made to educate, inform, and increase their levels of awareness, thereby increasing agency. More must also be done to close the gap between formal and informal practices and thus improve women's protection under the law. In the market domain, women start from a disadvantaged position in terms of both their asset endowments and the rules that affect them. While agency is high by necessity, the study team find that women face fewer opportunities than men. Gaps in economic outcomes persist. Expanding asset endowments and improving services and access to markets are nec- Table 8.10. Number of People Beneficiaries Can Rely on for Loans Dependent variable: Social capital (number of people respondent can rely on for small amount of money) Unstandardized Standardized coefficients coefficients Std. Variable Beta Error Beta T test Sig. (Constant) 2.135 1.281 1.667 0.096 Age 0.019 0.022 0.033 0.871 0.384 Full literacy 1.139 0.500 0.089 2.278 0.023 Urban dummy 1.371 0.596 0.108 2.301 0.022 Skills and creativity in 0.459 0.263 0.060 1.745 0.081 finding solutions Muslim 1.557 0.593 0.124 2.626 0.009 WDIP participant 0.847 0.458 0.065 1.849 0.065 Source: Legovini 2005. 150 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION essary to strengthen economic outcomes for women and close the gender gap in remuneration. In the social domain, women are severely disadvantaged, with a large majority still suffering from genital mutilation, forced marriage, and domes- tic abuse. Cultural change through training and awareness campaigns along with legal reform and enforcement of existing laws can help to reduce tra- ditional practices that humiliate women, lower their self-esteem, and reduce their bargaining position in relation to men. The study also found that interventions like WDIP can help in improving empowerment outcomes, but the study team suggest that these interven- tions cannot be small and isolated. Rather, they need to encompass reforms at the macro-institutional level to more significantly expand economic opportunities for all, aggressively closing the gender gap in education, and deepening legal reforms and strengthening enforcement to protect women's rights. WDIP has a clear impact on economic outcomes by expanding women's access to opportunities, financing, information, and training. These effects appear to spill over into the social sphere by expanding social capital outcomes by providing an expanded support network and greater access to government. WDIP also appears to improve psychological out- comes and women's capacity to aspire. In terms of indicators, this country study identifies a useful range of those suitable for analysis in the political, economic, and social domains of women's lives. The case also serves to illustrate both the pitfalls of cer- tain sampling designs and how more integration between approches to data collection and analysis enhances interpretation of results. For exam- ple, where results are identified as counter-intuitive or where limits to explanations are acknowledged, a different sequencing of questionnnaire- based and interactive information gathering could have been beneficial. In addition, the case raises questions over the cost effectiveness of advanced statistical analysis when the data set is limited in scope. Project impact eval- luations, such as that undertaken by the research team in this case, may often be subject to resource constraints. Restrictions on the time, skills, and financial resources available always affect the scale and scope of data col- lection and hence analysis and reporting. Reflection on these constraints early in the process of defining the boundaries for data collection and spec- ification of analytic techniques is clearly critical to robust and consistent conclusions. Notes 1.Arianna Legovini task-managed this study and is responsible for the statistical analysis of primary data and much of the related text. Background and secondary analysis draws heav- ily on preparatory work and papers for the World Bank's 2005 poverty assessment for Ethiopia. MEASURING WOMEN'S EMPOWERMENT IN ETHIOPIA 151 2. This background section draws on Alsop and Kurey (2004). 3. See Christiaensen and Alderman (2004) for evidence on child malnutrition. 4.Analytically, participation cannot be equated with empowerment. Participation may be treated as one indicator of empowerment, but this requires complementary analysis of indi- viduals' and groups' capacity to make effective choice about that participation and of how that choice is affected by agency and opportunity structures. 5. The terminology differences do not undermine the analysis, but they do require careful interpretation of final narrative reporting. Where the term "agency" is used in the full WDIP report, this chapter substitutes "empowerment." Where the term "opportunity structure" is used in the WDIP report, this book unpacks variables and subsequent analysis into institu- tions and assets. 6. It should be noted that the sample frame was drawn from women wishing to partici- pate in the project, thus introducing a self-selection bias and affecting the representativeness of the final study sample. 7. The original report by Legovini also considers the psychological as a domain of people's lives. However, because empowerment is a relational concept dealing with the relative power of different actors, this chapter treats psychological variables as assets, that is, as a stock from which benefits can flow. 8. Figures are from the Ethiopian News Agency Web site, http://www.ena.gov.et. 9. Both abduction and rape are criminal offenses under Ethiopian law, but articles 558 and 599 of the 1957 Ethiopian Penal Code provided that if marriage is subsequently agreed, then the husband is exempt from criminal responsibility for these crimes. The code was amended in 2005. 10. Data were collected for 44,000 women in all 10 regions. 11. At the time of writing, analysis of the effect of WDIP on women's empowerment in the state domain was not available. 9 Impact Evaluation of the Honduras Community-Based Education Project Adapted from a study report by Emanuela di Gropello and Nina Heinsohn In Central America, education reforms are granting communities greater decision-making powers over school-related matters. An impact evalua- tion examined the effects of the Honduras Community-Based Education Project, known by its Spanish abbreviation Proheco.1 This evaluation pro- vided the context for developing and testing empowerment indicators and approaches to information collection. It also clearly illustrates some of the problems associated with analysis of data of the kind required in assess- ments of empowerment. In its analysis, this study focused mainly on the empowerment impacts of education reform. Proheco is an education decentralization project, initiated in 1999, that devolves authority over school- and education-related matters to community- based school councils known as AECOs (Asociación Educativa Comunitaria).2 Members of the school council are elected from the community. Proheco schools are established in remote, rural areas of the country where com- munities have little or no access to a school. The project's objectives are to enhance access to and quality of education and to empower communities by enabling them to influence school management. TheAECOs are responsible for a wide range of educational and manage- ment functions, including monitoring school budgets. In traditional state- run schools, similarly composed parent associations (asociación de padres de familia) exist, but have no comparable management functions. In traditional schools these functions are the responsibility of the Ministry of Education and its departmental and district offices. Technically each AECO consists of two separate bodies: the general assembly (asamblea general) and the board (junta directiva).All community members are automatically members of the general assembly; the assembly in turn formally selects members of the board. The board, in consultation with the general assembly, is respon- sible for carrying out the association's devolved tasks. While in theory all community members are part of the AECO, in practice school-related deci- sions are made by the board. This chapter first outlines the conceptual framework before discussing the methodology used in the Honduras evaluation. The methodology sec- 152 THE HONDURAS COMMUNITY-BASED EDUCATION PROJECT 153 tion is particularly interesting because of some of the issues it highlights in relation to the analysis of data. Following the methodological discus- sion, key findings on school council empowerment and community member empowerment are presented. Readers are encouraged to view these find- ings and those in the following section in light of the discussion on method- ological problems.As such, findings on outcomes are interesting, but should be treated as indicative only. Conceptual Framework of the Study The design of Proheco's impact evaluation adapted the analytical and con- ceptual framework for measuring empowerment presented in chapter 2 to fit the specificities of the Proheco project. In practice, and despite the clear distinction between participation and empowerment, the evaluation used "participation in school management tasks" as a proxy indicator of empow- erment of school management committees and individuals. The impact evaluation sought to assess the extent to which Proheco had met its educational and empowerment objectives. In terms of educational outcomes, the evaluation measured and analyzed the nature and quality of teaching and learning environments in Proheco schools, the extent to which the project contributed to increased national enrollment rates, and Proheco's record on repetition, dropout rates, and test score results. In terms of empowerment, the impact evaluation addressed two sepa- rate questions. First, it sought to assess whether and to what extent the AECOs were able to carry out the school management functions devolved to them (school council empowerment). Second, it sought to determine which community members were engaged in the school council and the activities they undertook (community member empowerment). Questions aimed at determining the levels of community member empowerment asked whether women, indigenous people, and other traditionally excluded groups were involved in the school council. The empowerment framework proposes that empowerment levels asso- ciate with both actors' agency and their opportunity structure. In the con- text of Proheco's impact evaluation, this premise was used to develop the following hypotheses: (a) the extent of school council empowerment is shaped by a community's agency as measured by its average education and income levels; and (b) the extent of school council empowerment is influenced by the efficiency and organizational procedures with which the education author- ities support the operation of the Proheco program (opportunity structure). The first hypothesis suggests that the school councils a priori have the resources to become empowered, while the second proposes that school council empow- erment depends more on the administrative and financial support the coun- cil receives from the Ministry of Education and other education actors. 154 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION The equation the study developed for analysis of school council and community member empowerment combined both hypotheses: Ei = a + iXi + Empowerment (E) was treated as the dependent variable, and agency ( ) and opportunity structure (X) as independent variables. This assumes a linear relationship in which empowerment is the product of agency and opportunity structure. This is a reasonable postulate for analysis, but in reality, as chapter 3 discusses, empowerment cannot always be treated as dependent. Given more time, this study would have benefited from con- sidering (a) the effect of E on agency and opportunity structure variables-- that is, to begin further analysis by treating and X as dependent in turn; as well as examining (b) the interaction between different variables within and X and their exogenous impacts. Research Design and Methodology This section first outlines the instruments and sampling frame used in the study. It then describes the indicators of empowerment, agency, and oppor- tunity structures on which information was gathered. Proheco's impact evaluation was conducted using a mixed-methods approach. The questionnaire-based surveys (1­4 below) were designed to capture data on both empowerment and education outcomes, while the interactive tool (5 below) gathered information only on empowerment. The impact evaluation also made use of secondary information, such as data on financial transfers and statistical information on student enrollment pro- vided by Proheco's central coordinating unit, and manuals outlining how the AECOs operate in practice. The following instruments were used for primary data collection: 1. Aschooldirectorsurvey,whichaimedtocaptureschooldirectors'viewsabout the functioning of the school as well as its relations with the community. 2. A school council survey, which targeted the board members of the school council and mainly addressed questions about the functioning of the Proheco program. 3. Ahouseholdsurvey,which,inadditiontocapturingsocioeconomicdataabout the informants, was used to gauge the degree of community members' sat- isfaction with Proheco schools as well as the extent of their involvement in the school council and other school activities. 4. Key informant interviews, which were directed at a series of education authorities--including members of Proheco's central coordination unit, Proheco's departmental coordinators, promoters, and others--to collect information related to the implementation of the Proheco program. THE HONDURAS COMMUNITY-BASED EDUCATION PROJECT 155 5. Interactive community interviews targeting women, indigenous people, and individual community members who were not actively involved in the AECO. Question guidelines were designed to capture information on empowerment-related processes and outcomes. ThesampleofschoolsexaminedbythisstudyincludedbothProhecoschools ("treatment" group) and non-Proheco schools ("control" group). The con- trol group included communities that had a traditional state-run school not participating in the Proheco program. Communities with traditional schools tended to have parent associations. While these associations pro- vided parents with a platform to organize themselves and present their interests to school staff, they were not legally recognized by the govern- ment as having a school management mandate. A total of 80 schools were included in the study, comprising 60 treat- ment schools and 20 control schools. The study used a random-stratified sampling methodology, with 15 treatment schools and five control schools randomly selected in the regions in question. Within sampled school com- munities, 10 households were randomly sampled for the household survey. While having children in the local school was not a prerequisite for enter- ing in the household survey sample, most households either had or were related to school-age children. Finally, a subsample of 20 school commu- nities was selected for the interactive community interviews, four of these communities belonging to the control group. Three types of indicators were used in this study: (a) direct indicators of empowerment, (b) indirect indicators of empowerment (agency), and (c) indirect indicators of empowerment (opportunity structure). Direct Indicators of Empowerment Direct indicators of (a) school council empowerment and (b) community member empowerment tested the ability of councils and communities to engage in school management functions. Indicators of school council empowerment gauged the extent to which the AECOs were able to make school- and education-related decisions that in the pasthadbeenmadebytheMinistryofEducationanditsdepartmentalanddis- trict offices.3 Nine indicators measured the extent to which school councils were empowered to carry out their school management responsibilities. These indicators captured whether the councils functioned as established in the laws, regulations, and manuals that outline specific features of the decentralization program. That is, they assessed how well the formal opportunity structure created by the project functioned in practice. The nine indicators were · whether a school council board exists that was elected by the general assembly (EB), 156 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION · percentage of board members that are women (BMW), · whether the board knows when to submit an explanation of their expenses to the Proheco administration (EXE), · whether an explanation of expenses was submitted on time to avoid delays in government transfers (EXT), · whether the AECO is believed to have hired the teacher independently (HTI), · whether the AECO grants teachers leave of absence permissions inde- pendently (LAI), · whether the AECO controls teacher attendance (CTA), · whether the AECO or community has contributed financially to school construction or school activities (FSCH), · percentage of households that believe teachers should be hired by the AECO (THC). The variables captured a range of the functions for which the AECO or the board were responsible (EB, EXE, EXT, CTA, FSCH). They also gauged whether the AECO managed to carry out its responsibilities independently, that is, without influence from other local or municipal actors (HTI, LAI); while newAECOs initially needed support in carrying out their tasks, coun- cils were only considered empowered if they developed to the point of independent decision making. One variable (THC) also captured whether the communities accepted or internalized their new responsibilities. These nine indicators were combined into a school council empower- ment index. This was intended to capture all three degrees of empower- ment. That is, the presence of choice, the use of choice, and the achievement of choice. As discussed in chapter 3, there are numerous well-documented issues inherent in the use of indexes. These are particularly apparent when using indicators for more than very basic descriptive analysis. However, researchers undertaking the analysis in Honduras felt the advantages of using an index rather than working with individual variables prevailed over the methodological concerns surrounding the weighting of variables and interpretation of results. The rationale for their use in this case was not explained. This highlights another issue for those who are engaged in analy- sis of empowerment: the reasons for using particular methods of data col- lection, recording, or analysis need to be clearly presented if results are to be treated as reliable. Without satisfactory explanations the use of a com- posite index comprising a number of different metrics, and apparently sub- jective weightings, undermines findings. Questions of endogeneity can also be raised about using such a composite index as a dependent variable. That said, the variables used remain pertinent to and useful for the measurement of empowerment. Indicators of community member empower- ment were used to assess whether specific social groups in the community THE HONDURAS COMMUNITY-BASED EDUCATION PROJECT 157 had opportunities to participate in council activities. Attributes, including gender, occupational group, and marital status were tested for participa- tion in school-related decision making to establish whether certain social attributes were associated with lower levels of engagement in school coun- cils' activities. The evaluation developed an empowerment index for a community's most excluded members (CME). This index measured the extent to which different excluded community members were able to engage in AECO activities and decision making. Unfortunately, no comparator group infor- mation was gathered by the study team. AECO activities referenced the council's general assembly meetings as well as other related activities orga- nized by the AECO that were not part of its school management responsi- bilities, such as a fundraising event. AECO decision making referenced the school management tasks entrusted to the council's board. Again, questions can be raised about the cost-effectiveness of constructing indexes and these questions should be carefully considered before such an approach is taken in analysis of empowerment. Indexes have multiple uses, including some practical applications in monitoring systems that can give quick and reasonably reliable guidance on outcomes and related variables, such as might be required for local-level project tracking. However, the construction of indexes beyond such uses may require the use of statistical processes that can be time-consuming and hence costly. The research team in this case might have produced more persuasive evidence if regressions had been undertaken against a dependent variable less subject to endo- geneity issues or open to critiquing on the basis of weighting and variable metrics. This highlights the importance of having a fairly good idea of which associations will be tested--though those undertaking a study should not fall into the common trap of overprescribing future analysis at the begin- ning of a research effort. With such foresight, it is possible to build into the data collection and research process (a) exogenous variables, and (b) the opportunity for verification and interpretative discussions of preliminary analysis with key informants. Perhaps more importantly though is the problem of no comparator group. Without these, data results for excluded groups has to be treated with extreme caution. Indirect Indicators of Empowerment: Agency The study built on the idea, fundamental to the empowerment framework, that the agency of a group or individual contributes to the degree of empow- erment experienced. Agency indicators for school councils included4 · a community's poverty line (PL); · a community's unsatisfied basic needs (BN); 158 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION · the average education levels in the community, based on that of male heads of households (MEDL); · percentage of school councils that are informed of the requirement to acquire legal status to open up a bank account; · percentage of councils that possess an operational manual that lists their responsibilities; · percentage of councils that have opened a bank account to receive trans- fers from the government; · percentage of councils that file (a) bank statements, (b) receipts of pur- chases and payments, (c) receipts of transfers; · percentage of councils that have received training on budgetary matters; · duration of the training; · levels of perceived usefulness of the training. This list is limited. Arange of additional potentially useful agency indi- cators were not taken into account in this study due to lack of time and resources. Other agency indicators could have included (a) levels of finan- cial and administrative support that the communities received from NGOs or the government, (b) communities' access to urban or rural centers, and (c) the skill levels that communities acquired as a result of participation in other local organizations. Agency indicators for community members, capturing their assets, included · percentage of women and of poorest community members who are lit- erate, · percentage of women and of the poorest who are aware of when coun- cil meetings take place, · percentage of women and of the poorest who understand what types of decisions are made during the meetings, · percentage of women and of the poorest who have received training regarding the council's functions, · levels of perceived effectiveness of training, · percentage of women and of the poorest who are involved in other com- munity organizations, · types of positions held in other organizations, if any. The use of these indicators allowed the research team to analyze whether respondents' educational and financial resources influenced the extent to which they engaged in school council matters. Indirect Indicators of Empowerment: Opportunity Structure Again, two sets of indicators were used to assess the opportunity structure for AECOs and community members. THE HONDURAS COMMUNITY-BASED EDUCATION PROJECT 159 Theopportunitystructureindicatorsweredevelopedtomeasuretheformal and informal institutions that affected the functioning of school councils. These indicators focused on the efficiency of organizational procedures and the effi- ciency with which the Ministry of Education and its departmental and dis- trict offices followed rules laid down to support implementation of the Proheco project.5 In theory, the education authorities were to guarantee the efficient operation of theAECOs (and thereby their empowerment) by providing dif- ferent sources of administrative and financial support. However, in practice the project might not operate as efficiently as envisioned. For example, the Ministry of Finance was mandated to transfer funds to the school councils on a monthly basis so that theAECOs could pay teacher salaries and buy school materials. If, however, the ministry failed to transfer funds to the school coun- cils on a regular basis, the council could not undertake these financial trans- actions. Similarly, the communities were to receive regular administrative support and training by project "promoters." These government employees were supposed to visit Proheco communities regularly to support them in carrying out their school management tasks. If the Ministry of Education failed to ensure that the promoters visited the communities on a regular basis, the councils would not receive sufficient training and would experience dif- ficulties in carrying out school management tasks. The opportunity structure indicators focused on the organizational processes surrounding the implementation of the project, that is, the enact- ment of formal rules. These indicators included the following: · duration of the training (DUTR) · whether the AECO consults the manual outlining the functions of the council (CMAN) · amount of time that passes until AECO receives legal status (TIMELS) · amount of time that passes between AECO obtaining legal status and receiving the first government transfer (TPTRANS) · amount of time AECO needs to reach the bank where it has opened up its bank account (TIMEB) · whether the promoter helps the AECO with administrative procedures (PROAD) · whether changes in the position of the promoter has affected govern- ment transfers (CHPRO) · whether delays in paying the promoters has negatively affected admin- istrative procedures (DEPAY) · number of transfers AECOs received in 2003 (NOTRANS) · extent of delay with which AECO receive the government transfers (DETRANS) The opportunity structure for community members was defined as the oper- ation of social rules influencing the participation of the community mem- 160 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION bers in the councils and the local decision making related to the educational decentralization process. More specifically, these social rules were hypoth- esized to determine which members were able to participate in local decision making, including theAECO's school management tasks. However, because thesamplewasrelativelysmall,nospecificopportunitystructureindicatorswere developed. Instead, and problematically, the analysis of community member opportunity structure used data on respondents' attributes, specifically: pov- erty level (poor/not poor), ethnic group (indigenous/not indigenous), gender, and profession. In other words, a mix of asset and (arguably exogenous) indi- vidual attribute variables were used as proxy indicators of the opportunity structureforanalysisofabilitytoparticipateinschoolcouncildecisionmaking. This highlights the need to plan analysis at the time of developing data col- lection instruments to ensure that appropriate variables are captured. School Council Empowerment Outcomes This section begins by presenting the study team's findings on school council empowermentandasummaryoftheirexplanationoftheseoutcomesinterms of actors' agency and opportunity structure. In nearly all schools surveyed, the opportunity to exert choice (the first degree of empowerment) existed. In 98 percent of schools a board had been elected in a general assembly--an important finding in its own right. This led the study team to ask whether this opportunity was used. Did boards choose to perform functions mandated to them? In other words, was the second degree of empowerment achieved? Table 9.1. Empowerment of School Councils Variable % school councils performing task Election of board in general assembly (EB) 98.3 Submission of explanation of expenses (EXE) 98.3 Timely submission of explanation of expenses (EXT) 89.5 Independent selection of teachers (HTI) 71.7 Leave of absence permissions (LAI) 88.3 Control of teacher attendance (CTA) 93.3 Financial support to school (FSCH) 91.7 Belief that teachers should be hired by AECO (THC) (% of households) 91.7 Women represented on the board (BMW) (% of board members who are women) 38.3 Source: di Gropello and Heinsohn 2004. THE HONDURAS COMMUNITY-BASED EDUCATION PROJECT 161 The data in table 9.1 show that the majority of school councils carried out their mandated responsibilities. In terms of action independent of the Ministry of Education, the lowest value was for selection of teachers. The research team also included the percentage of women on the board as a variable in this index, because women were never involved in school man- agement decision making under the previous system controlled by the min- istry. It is the variable with the lowest value, but even so its appropriateness as a measure of school council empowerment is questionable. The nine variables were used to create the following school council empowerment (SCE) index: SCE = EB + BMW + EXE + EXT + HTI + LAI + CTA + SCM + THC All the variables were given equal weight. As discussed above, the ratio- nale for this has not been explained, highlighting once again some of the prob- lems in attempting to develop indexes for anything other than very basic tracking purposes without a clear statistical rationale. For the purpose of the evaluation, the level of school council empowerment was defined to be high for schools scoring more than 70 percent on average, medium for schools scor- ing between 40 and 70 percent, and low for those scoring less than 40 percent. As with the explanations of weightings, the rationale for these break points is unclear, illustrating yet another issue that researchers will commonly have to deal with, that of categorizing data. In this case it may have been more appropriate to use standard quartile classes. The data show that the SCE index is high for a majority of schools (table 9.2). The index is higher for the northeast and southwest of the country than for the northwest and the southeast. Despite issues with the school empowerment index, these are interesting indicative findings in themselves. The research team then tested the effect of agency and opportunity structure variables on school empowerment, using firstamultipleandthenastepwiseregression.Fiveofthe13agencyandoppor- tunity structure variables were entered into the multiple regression (table 9.3). The five were selected on the basis of the fact that they had the highest level of correlation with the empowerment index. Table 9.2. Level of School Council Empowerment, by Region Region Medium (%) High (%) Northeast 16 83 Northwest 40 60 Southeast 40 60 Southwest 20 80 Source: Authors. 162 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Table 9.3. Multiple Regression against School Council Empowerment Index Independent variable Coefficient (std. error) Average (head of household) community education level (MEDL) .0272 (.01496)* Delays in paying the promoters (DEPAY) .0698 (.0407)* Consultation of the manual (CMAN) .0497 (.0384) Promoter helping with administrative procedures (PROAD) .0473 (.0592) Time passed until AECO receives legal status (TIMELS) .0204 (.0347) Number of observations 45 R squared .09 F 1.8827 Source: di Gropello and Heinsohn 2004. Note: The standard error is given in parentheses. * = Significant at the 10% level or lower. The research team's multiple regression estimates indicate little associ- ation between the independent variables and school council empowerment outcomes, apart from a positive association between the average education levels of heads of households and a negative association between delays in promoters receiving their salaries and levels of school council empowerment. While the strength of this finding is undermined by the type of variable used and the use of a composite index, these remain interesting indicative findings and suggest that more rigorous analysis could help in under- standing which associations are stronger than others. The stepwise regres- sion (table 9.4) provided results on the association between agency and opportunity structure variables and empowerment outcomes: · The higher the average education level of heads of households in a com- munity (MEDL), the higher the school council's empowerment. · The more theAECO consults the manual outlining its school management responsibilities (CMAN), the higher its empowerment. · The farther away the bank where the AECO has its bank account and to which the Ministry of Finance makes the financial transfers (TBANK), the lower the council's empowerment. · The larger the delays with which the promoters receive their salaries (DEPAY), the lower school council empowerment. · The larger the delays with which the AECO receives government trans- fers (DETRANS), the lower the council's empowerment. THE HONDURAS COMMUNITY-BASED EDUCATION PROJECT 163 Among the five variables that prove to have a significant correlation with school council empowerment, one falls in the agency cluster (average educa- tionlevels)andtherestbelongtotheopportunitystructurecluster.Theaverage education level of heads of households mattered for school council empow- Table 9.4. Stepwise Regression against School Council Empowerment Index, by Number of Variables Entered Number of variables entered into the regression (std. error) Independent variable 7 8 9 10 11 12 13 Duration of training (DUTR) .0048 .0031 .0032 .0060 .0071 .0145 .0258 (.0076) (.0080) (.0082) (.0054) (.0061) (.0083) (.0683) Consultation of the manual .0538 .0881 .0987 .1373 .1135 .0077 .0528 (CMAN) (.0436) (.0488) (.0556) (.0606) (.0748) (.1109) (.3918) * * * Amount of time passed until .0059 .0088 .0195 .1019 .0874 .0563 .0363 AECO receives legal status (.0382) (.0436) (.0479) (.0598) (.0683) (.0682) (.1522) (TIMELS) Amount of time passed .0046 .0178 .0205 .0003 .0190 .0176 .0217 between receipt of legal (.0474) (.0514) (.0568) (.0664) (.0775) (.0778) (.1112) status and first transfer (TPTRANS) Amount of time AECO needs .0771 .0761 .0837 .2228 .0223 .3448 .9196 to reach the bank (TIMEB) (.0748) (.0783) (.0808) (.0648) (.3135) (.4178) (3.4709) ** Promoter helps in .0440 .0370 .0428 .0623 .0001 .1890 .3653 administrative procedures (.0590) (.0661) (.0687) (.0773) (.1266) (.1940) (1.0838) (PROAD) Changes in promoters .0309 .0475 .0332 .0257 .1664 .3606 .6695 affect transfers (CHPRO) (.0417) (.0491) (.0520) (.0550) (.2225) (.2603) (1.8740) Salary delays affect .0966 .1091 .1459 .1787 .1745 .2083 administrative procedures (.0493) (.0520) (.0468) (.0712) (.0660) (.2211) (DEPAY) * ** ** * Number of transfers in 2003 .0296 .0497 .0730 .0856 .1514 (NOTRANS) (.0248) (.0204) (.0418) (.0401) (.3956) * Extent of transfer delay .2359 .2239 .1395 .0668 (DETRANS) (.0490) (.0560) (.0864) (.4490) *** ** Index of community's poverty .0039 .0065 .0110 line (PL) (.0060) (.0059) (.0284) Index of community's .0033 .0063 unsatisfied basic needs (.0027) (.0185) (BN) Average education level .0635 (MEDL) (.3779) R squared .0555 .0311 .0207 .6953 .6447 .6950 .4068 F .6696 1.1485 1.0823 4.1950 3.3097 3.6586 1.7384 Source: di Gropello and Heinsohn 2004. Note: The standard error is given in parentheses. * = Significant at the 10% level or lower. ** = Significant at the 5% level or lower. *** = Significant at the 1% level or lower. 164 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION erment because literacy and numeracy were key to the AECO's ability to draft documents and fill out forms necessary for managing schools. In terms of hiring teachers, for instance, the AECO needed to be capable of putting together a job advertisement for the teacher position and reading applica- tions from different candidates. Similarly, theAECO needed to know how to write checks to pay teacher salaries, calculate prices for school materials, and keep a record of its expenses to carry out its budget management respon- sibilities. In terms of opportunity structure explaining empowerment of school committees, it appears that levels of knowledge of rules governing processes, the physical ease of accessing finance, and good functioning of administra- tive processes are important. The majority of AECOs (80 percent) indicated that they possessed a manual outlining the AECO responsibilities, and 70 percent of these stated that they used it. The rest of the councils did not use the manual, or mem- bers did not know whether the council used it, or did not know what the manual was. Second, the distance to the bank (TIMEB) where the council has its account proved to be a challenge for some of the AECOs. Proheco specifically tar- geted communities in remote rural areas, where infrastructure is particu- larly deficient. AECOs need a half to a full day to travel to the nearest bank, reducing time available for other tasks such as budget management. Third, delays in government transfers (DETRANS) negatively affected school council empowerment. TheAECOs needed the funds to pay teachers and buy school materials, two of the core responsibilities devolved to them. However, transfers were not made until the AECO received legal status. In practice this meant that the longer it took for Proheco's central coordina- tion office to process the legal status of an AECO, the longer it took for the school council to receive its first government funds. The AECO councils waited from one to five months before receiving legal status and the trans- fer of funds, and on average, they received three transfers per year instead of the envisioned four.6 Finally, the delays with which the promoters received their salaries (DEPAY) provedverysignificantasanexplanationofpoorschoolcouncilempowerment. The promoters were government employees and key figures in the imple- mentation of the Proheco program, providing the link between the commu- nities and Proheco's central, departmental, and district offices. They were responsible for providing ongoing advice and training to the communities on school management functions. However, 86 percent of the communities vis- ited as part of this evaluation stated that promoters were replaced more than once. According to the promoters, these high levels of rotation reflect attri- tion due to salaries not being paid on time. On average, promoters receive their salaries with a three-month delay. As a consequence, many promoters THE HONDURAS COMMUNITY-BASED EDUCATION PROJECT 165 use credit to survive and to be able to pay for their visits to the AECO com- munities. This has led to high levels of frustration and to promoters either carrying out their work reluctantly or deciding to quit their jobs. Many of these findings have intuitive appeal. However, they cannot be con- sidered robust because of the issues related to the use of indexes and the selec- tion of variables, discussed above. In addition, the effort of this research team raisesaquestion--againonethatotherstudyteamswilloftenface--ofwhether regression analysis is appropriate to certain kinds of data, and if it is, how to ensure that specification of the model is sufficiently sophisticated to account for different metrics and types of data, and potential endogeneity. Empowerment of Excluded Community Members This section discusses the findings on community member empowerment, analyzing whether traditionally excluded social groups such as women, a community's poorest members, and indigenous people became involved in AECO activities and decision making. Proheco schools specifically targeted rural and marginalized communities.Approximately 75 percent of the com- munities that have a Proheco school can be classified as poor or extremely poor; differences in wealth are relatively small. Once again, the research team constructed an index for empowerment of excluded community members through the exercise of choice to partic- ipate in school councils. This comprised three variables: · the extent of women's participation in AECO decision making (WPD) · the proportion of council board members that are agricultural laborers (BMAL) · the participation levels of single mothers and the community's poorest members in AECO activities (PPOOR) In the index, WPD and BMAL variables were each given a 30 percent weight and the PPOOR variable 40 percent weight. The CME index was con- sidered to be high when communities scored an average of more than 70 per- cent, medium if they scored between 40 percent and 70 percent, and low if the average was below 40 percent. While the same issues can be raised about the weighting, the variables used in this index are less problematic than those used in the community empowerment index. They use the same metric and are of a similar type--they all focus on the use of opportunity. The team found that, on average, empowerment levels for excluded com- munity members were high in 45 percent of the communities, medium in 42 percent, and low in 13 percent (table 9.5). Analysis by region showed that empowerment levels for excluded com- munity members were highest in the southwest of the country (73 percent 166 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Table 9.5. Empowerment of Excluded Community Members, by Region Empowerment level Region Low Medium High Northeast 20 47 33 Northwest 13 40 47 Southeast 13 60 27 Southwest 7 20 73 Average 13 42 45 Note: Percentage of communities falling into each empowerment level. scored "high" on the CME index). Coincidently, this was the region with the highest poverty levels and the largest presence of indigenous people. In terms of poverty, the communities in the southwest region were also characterized by their homogeneity: differences in poverty levels were hard to detect. This meant that in the poorest areas, the participation of the community as a whole was higher than in less poor areas. In the other regions of the country, empowerment levels of excluded community members, as measured by the CME index, ranged between medium and high. Data analysis revealed that it was easier for some community members thanforotherstoengageinschoolcommittees.Whiletheempowermentofthe most excluded was considerable, as far as the limited data set can demon- strate, it was still lower than that of less excluded groups. While equal oppor- tunity existed for men and women (the first degree of empowerment), men used the opportunity more frequently (the second degree of empowerment). Indigenous people were present in 25 percent of all the communities that were part of this evaluation; in each of these communities they were active in the AECO, but they were less active in communities where they constituted a minority population. Finally, those members of a community that were con- sidered among the poorest participated the least in the council and its activi- ties--again, the opportunity to choose existed, but its use (the second degree of empowerment) was less frequent for poorer people. Examination of these patterns through the community interviews in both treatment and control communities suggested that these patterns were caused in large part by a lack of agency, particularly with respect to the low literacy levels of excluded groups. Income was less significant in explaining non- participation, as income levels in most communities were uniformly low. In addition, discussions of these patterns with community members provided helpful insights into the key role that traditional power relations (governed by informal institutions) played in constraining or enabling levels of empow- erment. For excluded people, both agency and opportunity structure proved relevant to explain levels of empowerment achieved through the project. THE HONDURAS COMMUNITY-BASED EDUCATION PROJECT 167 Thisisbrieflydiscussedbelowinrelationtoempowermentofspecificgroups regarded as excluded within Honduran society: women, indigenous people, and poor people. A concluding section summarizes the findings. Women Women'sparticipationinAECOdecisionmaking(WPD)wasselectedasavari- able of the index of empowerment of excluded people, because women are typically among the poorest in rural Honduras.7 Comparative analysis with control group communities raised the ques- tion of how to ascribe meaning (as discussed in chapter 3) to the chosen indicator of women's empowerment. While the participation of women was slightly lower than that of men in Proheco communities (47.1 percent of participants were female and 52.9 percent male), the participation of women in the parent associations in non-Proheco communities was three times higher than that of men. One interpretation of this finding is that the introduction of the school council system has challenged traditional gender- based role assignments through increased male participation in schooling matters. This would suggest that women in Proheco communities were now in a position to choose not to participate, whereas previously this implicit choice did not exist. A more plausible interpretation of data is that male participation in Proheco AECOs is higher than male participation in non-Proheco parent associations because the AECOs, in contrast to the traditional parent asso- ciations, brought with them access to resources and political influence. This would mean that the AECOs have partly reproduced traditional gender patterns of male control over political decision making and financial resources, something that could be a hindrance to further empowerment of women in the Proheco communities. If this were the case then the fact that 47.1 percent of AECO participants were women can be interpreted as a sig- nificant level of empowerment of women through the exercise of choice. The qualitative module of the evaluation revealed that the picture was strongly influenced by context. In some villages women's participation in AECO was constrained by the multiple demands on their time or by men preventing them from participating. In another village, women partici- pated more in the AECO because the men from that community had migrated to the capital or abroad. In still other communities women were more likely to participate because men worked outside the home all day and came home tired in the evening. Indigenous People There are about 10 different indigenous groups in Honduras. They are a small minority within the country: of the sample population, only 6.6 per- 168 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION cent were classified as indigenous. Indigenous people lived in about 25 per- cent of the study's communities. Within these communities, the percentage of community members that belong to the ethnic minority varied. Indigenous people participated in the AECO in all communities where ethnic minorities lived. In 93 percent of these communities, indigenous people were actively involved in AECO decision making. Nevertheless, empowerment of indigenous people through participation on school coun- cils was comparatively lower than that for nonindigenous people. The study shows that indigenous participation was especially low in those communities where they constituted a very small minority, suggesting that in these con- texts social exclusion of indigenous people is more likely. Empowerment was relatively high only in those communities where indigenous people con- stituted a majority. Semi-structured discussions with indigenous people indicated that this related to the fact that the ladinos, those of Spanish or mixed descent, tra- ditionally assumed leadership roles within communities. This meant that power relations between indigenous and nonindigenous people--consti- tutive of the opportunity structure--mattered in terms of who could play a dominant role in the school council. Interestingly, indigenous people living in non-Proheco communities were less likely to participate in tradi- tional parents' associations than indigenous people in Proheco school areas. One interpretation of this comparison is that the AECOs provide a plat- form for participation to those indigenous community members who pre- viously were excluded through interaction of a constraining opportunity structure and a lack of agency. Further research is needed to explore the significance of gender for the ability of indigenous people to choose to par- ticipate in school councils. The Poorest Among the poorest people living in communities with a Proheco school were single mothers, agricultural laborers, and indigenous people (in the western part of the country). The majority of people living in Proheco com- munities were agricultural laborers (31.8 percent), farmers or stockbreed- ers (43.5 percent), or, to a lesser extent, housewives, craftsmen, and small merchants. Farmers and stockbreeders were much more active in theAECO than agricultural laborers: 63.7 percent of the former group but only 14.6 per- cent of the latter group engaged with the council as board members, as part of the general assembly, or through activities related to the council. Given the correlation between excluded people's lower education and income levels (agency), it might be assumed that excluded community members voluntarily delegated school management responsibilities to com- munity members with greater agency. However, the interactive community THE HONDURAS COMMUNITY-BASED EDUCATION PROJECT 169 interviews indicated that local power structures (caciquismo) played a more important role in rural Honduras than education and income. Families or individuals who traditionally held leadership positions were also those who played dominant roles in community associations such as the school council, actively preventing less prominent community members from having a say in local decision making. The rules governing local power relations were therefore an important part of the opportunity structure that influ- enced whether and to what extent community members were in practice empowered to engage in school management committees and activities. Conclusion In terms of impact, the conclusions that can be drawn from the statisti- cal analysis must remain tentative. The study has identified useful variables and demonstrated that certain associations may exist, but issues with com- bining different variables and the use of indexes as well as inherent issues of endogeneity mean these findings should only be treated as indicative. Theempowermentframeworkwasusefully,ifpartially,employedtodevelop a series of empowerment hypotheses and to develop a range of indicators. Thefirsthypothesiswasthatthedecentralizationhadthepotentialtoempower communities.Thesecondhypothesiswasthattheprogramempowersexcluded community members to participate in school councils and their activities. Seventy percent of the school councils achieved high levels of empower- ment,meaningthatthemajorityofAECOswereusingtheopportunitytomake choices about school management. However, data relating to the empow- erment of excluded groups indicates that these groups used the opportu- nity to engage in school management less than other community members. For both school council and community member empowerment, the empowerment framework proved useful to the research team in identify- ing and framing--no matter how problematically--analysis of the factors influencing whether and to what extent groups or individuals became empowered. For example, key "agency" variables were the educational levels of communities as a whole (school council empowerment) and of individual community members (community member empowerment). The higher the education levels, the higher the levels of empowerment. Again, despite methodological concerns surrounding the analysis, the study suggests that the effect of opportunity structure on community member's empowerment was mediated by the informal institutions gov- erning local power relations. Rates of engagement were low for tradition- ally excluded people in Proheco school councils. While only indicative, these findings suggest that it will be a difficult and long-term commitment for the government to address power imbalances (opportunity structure) within communities. These power imbalances are structural in nature and 170 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION often of historic origin, as is the case with traditional elite structures and the dominance of the ladinos over the indigenous population. Improving the opportunity structure for school council empowerment should be easier for the government to achieve. While findings have to remain tentative, they do suggest that this would involve addressing a series of operational deficiencies such as the delays in payment of the promoters and the con- sequent high degree of rotation among them, as well as the delays affect- ing the councils' receipt of government funds. Inmanywaysthestrugglesofthisteamtodevelopanduseadatasetappro- priate for analysis of "empowerment" serve to illustrate issues that other research teams may well encounter. Lessons emerging from this experi- ence include the need to · better integrate and sequence the use of questionnaire-based surveys and more interactive methods of information gathering; · question when regression or simpler forms of analysis are required, par- ticularly when a high degree of endogeneity is present; · be extremely cautious in the formulation and use of indexes; · ensurethatthetypeofdataandthemetricsusedforindexesarecompatible. Thisstudywasundertakenasanexperimentinapplyingtheframework.As such,ithasprovedextremelyusefulinhighlightingproblemsthatcanoccurin formulatingandundertakinganalysisofsuchmessysubjectsasempowerment. Aschapter3suggested,themeasurementandanalysisofempowermentimply theneedtopayverycarefulattentiontomeaning,comparability,andcausality. Notes 1. The impact evaluation "Evaluación Participativa del Programa Hondureño de Educación Comunitaria (PROHECO)" was designed and conducted by ESA Consultores, Honduras. It formedpartofalargercomparativeanalysisofeducationdecentralizationreformsinElSalvador, Guatemala, Honduras, and Nicaragua. 2. Asociación Educativa Comunitaria, or Community Education Association. 3. The Proheco program is administered and implemented by a central coordination unit, located within the Ministry of Education, and the ministry's departmental and district offices. The promoters are in closest contact with theAECOs, legally recognized by the government to carry out school management tasks. 4. The agency indicators for school council empowerment measured the agency of the com- munity as a whole, that is, not only of the council's board, because in theory all community members were part of the AECO through automatic membership in the general assembly. 5. The discussion of Proheco's opportunity structure did not address any political factors such as the influence of teacher unions, the extent of government support, or the nature of state-society links, because the scope of the evaluation was too limited. The discussion was THE HONDURAS COMMUNITY-BASED EDUCATION PROJECT 171 limited to an analysis of whether the education authorities supported the AECOs as determined in the reform design. 6. There are several reasons for these delays. In some cases AECOs have submitted incom- plete explanations of expenses to Proheco's coordinating office, and these incomplete docu- ments have been returned to the AECOs. The underlying problem here is that neither the school councils nor the promoters are sufficiently informed about and trained in filling out the explanation of expense forms. While delays in receiving legal status negatively affect school council empowerment, it should be noted that it is much more difficult for organizations or associations to receive legal recognition from the government outside the Proheco context. Other organizations on average need more than a year to obtain legal status. 7. The evaluation survey showed that while differences in poverty levels within Proheco communities were small, women living in these communities were slightly more likely to be poor than men; that is, they had fewer financial assets. 10 Empowerment and Local-Level Conflict Mediation in Indonesia Adapted from a study report by Chris Gibson and Michael Woolcock As discussed in chapter 3, relatively few efforts have been made to sys- tematically measure and track empowerment at the project level. In par- ticular, little reliable empirical work exists to show whether and how community-driven development (CDD) and related projects increase the influence of poor and marginalized people in shaping decision making and improving development processes and outcomes (Mansuri and Rao 2004). This chapter describes one attempt to make headway on these fronts using a mixed-methods approach to analyzing local-level conflict management spillovers from a CDD project in Indonesia. The country case team chose to analyze empowerment through conflicts, as they represent one critical con- text in which power relations are played out. The case presented in this chapter only relates to findings from part of the study--that undertaken using ethnographic methods of data collection. Unfortunately, at the time of writing the broader analysis and commentary on mixed methods was unavailable. Nonetheless, the case stands as a unique example among the five country studies of how nonquantitative data can be collected and man- aged to produce interesting and valuable results. The effort to measure empowerment in Indonesia formed part of a larger piece of research, the KDPand Community Conflict Negotiation Study, which included an assessment of the impact of the Kecamatan Development Project (KDP) on communities' ability to manage local conflict. While KDP was not designed as a conflict resolution program, the core question of the conflict negotiation study was whether KDP builds the conflict management capac- ity of villagers through unexpected "spillovers" from the deliberative processes it initiates. Begunin1998,KDPisamassivecommunitydevelopmentproject,thelargest in Southeast Asia, operating in over 28,000 villages across Indonesia. The objectives of the project are to (a) support decentralization and participa- tory approaches to improve local-level transparency, (b) provide a sustainable mechanism for locally managed development investments, and (c) strengthen the capacity of local governments to plan and manage suitable development expenditures. KDP distributes block grants in amounts ranging from $60,000 172 EMPOWERMENT AND LOCAL-LEVEL CONFLICT MEDIATION IN INDONESIA 173 to $110,000 directly to kecamatan (subdistricts) and ultimately to villages to fund locally prioritized subprojects, typically small-scale infrastructure, social, and economic activities. There are four project components: the development grants, technical assistance for implementation, monitoring support, and policy studies. Proposals for block grants are generated at the village, subvillage, and neighborhood levels and can entail either public infrastructure projects or economic activities. Each village council can ratify up to two proposals, which are then submitted to the kecamatan associa- tion of village heads for verification, review, and funding. This chapter first discusses the application of a conceptual framework, grounded in the tension between opportunity structure and agency, and introduces the study's central concepts of countervailing power and collabo- rative routines. It then moves on to describe the study research questions, explain the research methodology--designed to elicit data on the dynamic processes that are the focus of the study--and explain the use of the "con- flict pathways case studies" method employed. The following section sum- marizes case study examples that point to the importance of context in shaping the impact of KDP interventions on power dynamics and forms of collaboration. The chapter then draws out some analytical implications of the study and presents a typology to assess the empowerment impact of the KDP before presenting brief conclusions. Conceptual Framework and Research Hypothesis The Indonesia study defined empowerment as the process of enhancing individual or group capacity to make choices and to transform those choices into desired actions and outcomes. This is in line with the definition used by the analytic framework presented in chapter 2 of this volume. Akey assumption of the study was that project and country contexts ulti- mately shape localized manifestations of individual and group power. In the case of KDP, the project context was one of a very conscious movement away from the notion of projects as the deliverers of particular products and toward an understanding of projects as a way to trigger and support pro- cesses in which villagers exercise discretion in solving self-identified devel- opment problems. Far from merely tweaking incentives for individual involvement in civil society, KDP has systematically opened up the exercise of state-level power to collective decision making and influence by groups on the local level.Asimilarly decisive factor in the Indonesian country context has been the primacy of collective identities in shaping group-based deci- sion making. Precisely because of these project and country contexts, KDP's approach to empowerment has been distinctly collective. For this reason, the primary form of power that the study examined--conflict management capacity--was an inherently relational one. 174 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Akey process through which KDP aims to alleviate poverty and improve local governance is by convening a series of facilitated forums or meetings at the hamlet, village, and kecamatan levels to encourage and institution- alize broader community participation in decision making and priority set- ting. The Indonesia study proposed that in addition to KDP's complaints mechanisms, this community decision-making process might also incre- mentally shift power relations on local levels. As an example of what Peter Evans has termed collective capabilities,1 cases of marginalized groups influencing everyday conflicts amount to a kind of "meta-indicator" of empowerment. The study did not assume that KDP causes this form of empowerment. Rather, it examined a two-way association between certain KDP practices and the conflict management capacity of marginalized groups. The study was based on the hypothesis that as instances of everyday con- flict resolution come to mirror the "adaptive problem solving"(Heifetz 1994) processes of well-functioning KDP forums, then the distribution of con- flict management capacity often expands to include women, the poor, and other marginalized groups. That is, where marginalized groups use con- flicts to generate dense, diverse, and organized collective action, they may also have become more capable of exercising meaningful influence in con- flict management forums. The study explored conditions under which this may (or may not) have been the case, as indicated by heightened associa- tional contact between groups in formalized settings. The quality of the associations was also expected to change, with a shift toward more dis- cretionary and transaction-intensive decision making between alliances of less rigidly bounded and prescribed identity groups. Based on a substantial review of academic and operationally relevant literature on capability, power, and conflict in development, the study argues that reliable approaches for measuring empowerment must start from a firm understanding of which actions represent different kinds of power relations within particular contexts. Previous research from Indonesia and elsewhere has suggested that conflict management capacity involves groups using norms, processes, and formal and informal institutions to influence everyday clashes with other groups. As various authors point out, under certain conditions, local-level power may be manifested through a collective capacity that groups express in ongoing processes rather than in one-off actions by individuals.2 For this reason, everyday conflicts are particularly appropriate subjects for the study of power because of their dynamic and inherently relational nature. A synthesis of relevant theories of powerlessness suggests that marginal- ized groups are often shut out of everyday inter-group conflict management by a range of social, political, and cultural norms. The informal opportunity structure therefore becomes a significant focus of research to understand how these norms influence conflicts and their management. However, every- EMPOWERMENT AND LOCAL-LEVEL CONFLICT MEDIATION IN INDONESIA 175 day conflicts may also serve as "flashpoints" in which marginalized groups begin to develop the collective organization to overcome such structural barriers while building the agency required to "adaptively" and collabo- ratively manage conflict. Two conceptual distinctions were introduced in the study to help theo- rize the routines governing both local-level conflict management and inter- group power relationships: fairness-based versus purely interest-based decision rules, and adversarial versus collaborative types of forums. The former distinction serves to highlight whether the terms of debate enable the poor to participate more equitably, by invoking real or rhetorical appeals to alternative (countervailing) sources of power; the latter serves to frame the discussion in terms of whether parties to a given dispute see it in zero- sum (win/lose) or positive-sum (potentially win/win) terms. Distinct sources, forms, functions, and effects correspond to distinct combinations of each, and determining the qualities of various combinations was the task of empirical analysis. Although the growing literature on participatory col- laboration suggests that the most durable forms of empowerment require both countervailing power and collaborative forums, the evidence gath- ered in this study suggested that such combinations are rare indeed. More often, conflict management routines feature one, the other, or neither. The notions of countervailing power and collaborative routines of con- flict management may initially sound too abstract to be operationally rel- evant or measurable. Nevertheless, they are a pivotal feature of what project creators allude to when they hypothesize that CDD forums trigger and support processes in which villagers exercise discretion in solving self- identified development problems (Guggenheim forthcoming). As one of the more observable manifestations of the institutional basis of power rela- tionships between conflicting groups, a broad range of routines drive var- ious types and stages of conflicts--both those occurring within and those outside of KDP forums. Drawing comparisons across the two simultaneously helps researchers trace what contributes to effective conflict management and describe the institutional features that empower marginalized groups in the course of conflicts. The chapter next considers how the study's methodology enabled this form of comparative analysis. Research Design and Methodology The study generated evidence to test the core hypothesis about the project processes: that KDP's socialization, planning, decision-making, and imple- mentation processes improve the deliberative skills of community groups and build trust between these groups by forging associational ties (Barron, Diprose, and Woolcock 2005). To test this and other hypotheses, the research generated over 70 "conflict pathways case studies"--that is, ethnographies 176 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION of how particular social tensions and incidents of conflict played out in their local contexts. Researchers wrote these case studies based on over 800 focus group discussions, in-depth interviews, and participant observations that they conducted. This approach permitted the research team to estab- lish ongoing conflicts as the primary unit of analysis to be studied. Thus, the case studies examine how different actors--villagers, facilitators, local leaders, and so on--together managed (or failed to manage) different types of conflict in different settings. Using a version of case-based process tracing (George and Bennett 2005; Varshney 2002), the conflict pathways described discrete stages in the evo- lution of conflict, including conflict triggers and the factors or mechanisms that sustained conflict, allowing it either to escalate, stagnate, or move toward resolution. By following cases of everyday conflict, the studies revealed the factors that transform underlying tensions into different out- comes--violence or peace. In essence, each case worked backward from an outcome by asking what led to what. Importantly, they were selected in a way that controlled for some of the traditional weaknesses of qualitative approaches3 while capturing each stage of a conflict, attempts at its resolu- tion, and events that linked different stages. Each case study includes a sum- mary of the case, its "prehistory," evolution, attempts at resolution, impacts, and aftermath. Collection of the qualitative data used to construct these cases occurred in three stages and used a quantitatively oriented sampling frame to con- struct a plausible counterfactual. Propensity score matching was used to "match" villages that received KDP with those that would have been likely to receive it but did not, further verified by qualitative interviews to confirm the accuracy of those statistical matches.4 To control for endogenous fac- tors contributing to conflict management, matching of villages was con- ducted within districts with high and low capacities for conflict management. Within both high- and low-capacity districts, the study constructed a coun- terfactual that permitted meaningful comparisons between (a) cases of con- flict from "treatment" villages in which KDP operated for at least three years and (it was hypothesized) influenced conflict processes and outcomes; and (b) similar cases from "control" villages that would have been statisti- cally likely to receive KDP but did not.5 After conducting pre-fieldwork and devising this sampling strategy in this first stage of the research, the second and third stages of research applied the strategy through fieldwork. Stage two of the research used these treatment and control groups to assess conflict management "spillovers" from KDP processes into every- day conflicts unrelated to KDP. This approach permitted exploration of correlations between (rather than strong causal statements about) KDP's influence and empowerment in the form of the presence or absence of capacity to manage such conflicts. Also, construction of these comparison EMPOWERMENT AND LOCAL-LEVEL CONFLICT MEDIATION IN INDONESIA 177 groups permitted observation of "baseline" opportunity structures and conflict management routines that framed conflict pathways in treatment and control villages. In practice, stage-two cases proved especially useful for studying power relationships in the absence of KDP interventions. They permitted detailed analysis of the conditions under which marginalized groups of villagers did and did not gain access and exert influence in the course of everyday local conflicts. Stage three examined more directly the conflict pathways within KDP processes themselves and permitted comparison with conflicts that arose in other development projects in similar villages. These cases offered a slightly more powerful lens for examining claims about the "empower- ment" value of KDP, specifically whether and how the "treatment" of KDP permitted marginalized groups to express voice, influence decision making, or manage more capably collective conflicts that affected their lives. "Treatment" refers to the spaces, incentives, and resources that--when functioning properly--KDP introduces at the neighborhood, village, and subdistrict levels. By comparing these cases to similar conflict pathways that emerged out of other village-level development projects in similar vil- lages at roughly the same time, the study began to isolate whether and how marginalized groups who participated in KDP's inevitable conflicts also built what could be called the capacity to engage. Throughout, however, the study highlighted the process, not the outcome features of program- matic treatments.6 The Indonesian study is the only one of the five studies to use ethno- graphic and descriptive techniques alone, rather than attempting statisti- cal analysis. While this study is useful in highlighting what can be achieved by nonquantitative analysis, there is much potential added value in demon- strating how statistical analysis could be combined with the qualitative methods used. Although this added value will be demonstrated in a later phase of the study, no statistical analysis was available from the country team at the time of writing. An assessment of related indicators and inte- gration of statistical and narrative analysis could therefore not be demon- strated in this chapter. Assessing Empowerment: An Indonesian Case Study The study defined and tracked five basic types of conflict case studies: polit- ical office seeking, vigilantism, domestic violence, contested public resources, and publicly administered projects (KDP and non-KDP).7 To draw meaning- ful comparisons across these types and the many levels of heterogeneity within them, the study proposed that local conflict management posed a fun- damentally "adaptive" rather than "technical" challenge to villagers and marginalized groups (as well as to agencies that aim to structure effective 178 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION development interventions at the local level). That is, constructive solu- tions to conflict require contextually specific collaboration of a type that can rarely be determined ex ante by outside technocrats. Successful inter- ventions turn on the creation of spaces, incentives, and resources that make it possible for disputants to develop resolutions that all sides can own, uphold, and enforce. The presence and use of such conflict management practices may make the difference between everyday disagreements (such as disputes over agricultural theft, or political tensions) erupting into vio- lence instead of ending with more peaceful resolutions (Barron, Smith, and Woolcock 2004). As the case study examples below indicate, KDP, although not explicitly designed as a conflict management tool or intervention, aims to build the sorts of deliberative skills and collaborative decision-making spaces that characterize adaptive problem solving. State-Owned Resources in Kecamatan Badegan, Kabupatan Ponorogo, East Java In the KDP village of Biting in Ponorogo, East Java, an extended conflict over the repair of a leaky dam served as a flashpoint for the successful orga- nization of farmers and other villagers dependent on the now-empty reser- voir for irrigation. In this case, the farmers' group actually channeled an escalating conflict into a solidaristic routine of speaking and acting that generated new, more effective routines for promoting their interests from below. By appealing to a broad group of protestors, the farmers generated significant negotiating power. The farmers' use of the richly symbolic protest action of blocking a road with a crowd of villagers, and placing two empty chairs facing the dam, was a public performance that transformed their new association into a powerful force to be reckoned with. In the early days of the conflict, the group mostly used bureaucratic channels to request repair of the Sumorobangun Dam.After writing a series of letters to the district legislative assembly (DPRD) head and the district head starting in 1996, the farmers' group felt their demands for action had fallen upon deaf ears and began to feel rejected and angry. As farmers suf- fered more and more from the water scarcity, frequent arguments and small- scale violence broke out. As unrest peaked in 2001, the farmers' group changed its tactics, organiz- ing a public demonstration that mobilized a broad web of social networks including teachers, police, civil servants, rice paddy owners and farmers, and paddy workers from four subdistricts. This mobilization caught the attention of a candidate from a locally weak political party who was run- ning for a DPRD seat and who took the opportunity to apply pressure on the incumbent. Together, hundreds of villagers blockaded a key road con- necting two districts and in the middle of the road set up two chairs facing the dilapidated dam. By demanding that the two officials view the condi- EMPOWERMENT AND LOCAL-LEVEL CONFLICT MEDIATION IN INDONESIA 179 tion of the dam and witness the hundreds of villagers demanding its repair, the farmers' group finally solicited a response. The DPRD deputy head arrived on the scene and committed to fixing the dam, a promise the district gov- ernment ultimately fulfilled one year later.Additionally, a subsequent flurry of peaceful and fruitful activism ensued surrounding government com- pensation for lands previously inundated by the dam. The Sumorobangun Dam case illustrates how a coalition of marginal- ized villagers revised the dominant practical and discursive routines for managing an ongoing conflict and thereby managed to alter the existing opportunity structures, creating political space for increased empower- ment. The study concludes that transformative and solidaristic routines-- ranging from patterns of protest action and symbolic performance to patterns of argument within KDP-style forums--often constitute key mechanisms by which marginalized groups expand their influence in this and other ways. Relating to the empowerment framework, this shows a case where all three degrees of empowerment could be detected. The existence of choice was there, as the farmers had the choice of coming together to demonstrate and put forward their demands. As they did so, they also used the opportunity to choose, representing the second degree of empowerment.And as the author- ities committed to fixing the dam and to providing compensation for land, the choice made by the farmers also resulted in the desired outcomes, which indicates that the third degree of empowerment was attained in this case. KDP Management in Kecamatan Proppo, Kabupatan Pamekasan, East Java In a case from the village of Tattangoh in the East Java district of Pamekasan, researchers tracked the evolution of a conflict between elites and commu- nity members regarding the management of KDP itself.8 The improvement in the management of KDP in Tattangoh was quite noticeable over its three- year life there. An important aspect of this improvement related to the rou- tines plausibly inculcated by participation in KDP itself. While the institutions KDP created in Kecamatan Proppo involved broad groups of nonelite villagers in decision-making processes, these villagers nevertheless failed to gain much sustained purchase within these processes. Seeing that the same groups seemed to receive funding year after year, these nonelite groups reported diminishing interest in choosing to participate in the brain- storming and proposal submission process. So even though there was a change in opportunity structures, creating a formal space for participation that did not previously exist, this did not seem to result in any significant empowerment of the marginalized groups. In the first year, nearly all of the most important participatory components of the program failed, reflected in a lack of village representation on the kecamatan councils, no kecamatan forums being held, and no proposals 180 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION being made. Additionally, the kecamatan head appointed a family member as facilitator instead of opening the selection to a transparent election. Not surprisingly, interviews and focus groups targeted at a broad cross-section of villagers revealed that few even knew the program existed, much less participated in the series of deliberative and planning meetings required by the program. In addition to these miscarried processes, the program's out- comes also revealed a familiar pattern of elite capture in the village: the 750 million Indonesian rupiah (about $85,000) allocated to the program went directly to construct drains in a market owned by the village head. Amidst this overall failure, the use of an anonymous complaints mecha- nism by an unknown informant led to the facilitator's dismissal by the firm that was monitoring KDP implementation and performance. This last twist in year one epitomized the turn taken by the program in several KDP villages across Kecamatan Proppo over the following two years. While the routines by which villagers managed the program did not become dramatically more collaborative, incremental improvements in countervailing power began to emerge. As mandated by KDP, villagers held elections for the program's various teams and posts. Villagers also began participating in the program's consultative meetings.At a minimum, one could safely say that the associational interactions generated within these new institutional structures have the potential to become more for- midable bases of countervailing power. Nevertheless, a familiar picture emerged: despite the appearance of skeletal structures for representation, participants lacked the level of countervailing power required to funda- mentally revise the overall context of elite dominance in project decision making. Elites tended to dominate decision-making processes and the han- dling of project finances, while the different people elected were also involved in small-scale corruption. Land Conflicts in Mangarrai, Nusa Tengarra Timor In another case from an area with a tradition of violent land conflicts in the subdistrict of Mangarrai in Nusa Tengarra Timor, the role of mediators emerged as important in realigning institutions that underpin cyclical con- flict. Credible and knowledgeable mediators often play a pivotal role in orchestrating agreements because they may be the only parties able to even recognize the multiple cultural and social dimensions of a conflict, much less craft the correspondingly nuanced approaches required to resolve it. In this case the mediator managed to change the opportunity structures influ- encing the violent interaction between the farmers by setting up new rules and agreements for interaction, enabling the farmers to make new and effective choices to stop the violence. Tracing back to 1950, this inter-village conflict over a shared land bound- ary escalated significantly between 1990 and 1999, leading to property EMPOWERMENT AND LOCAL-LEVEL CONFLICT MEDIATION IN INDONESIA 181 destruction and a series of injuries and deaths. Through a successful medi- ation process conducted over the course of more than a year, a local parish priest managed to gain the trust of the conflicting parties and build a common appreciation of the background to the conflict. Apeace ceremony combined traditional (adat) and church rites that highlighted the inter- marriage links between the villages and their previously unrecognized sim- ilarities of ethnic background. The mediation process and ceremony yielded more than just a cessation of violence. After feuding for over 50 years, the groups began emphasiz- ing their shared ethnic roots, both in speech and in patterns of associational interaction with one another. By modeling a repeatable routine of interac- tion and by promoting an overarching "meta rule," or institutional change, regarding how to proceed, the mediator helped transform a vicious cycle of violence into a virtuous cycle of association through a revealed empha- sis on common identity.9 KDP in Kecamatan Pasean, Kabupatan Pamekasan, East Java In this case a change in opportunity structures--allowing women to par- ticipate in KDP decision-making processes--created the existence of choice. The case study also shows, however, that formal institutions, the official channels of participation, are often easier to change than the informal insti- tutions, in this case the norms governing women's (and men's) behavior in the villages. As a result, the change in opportunity structure did not nec- essarily lead to a rapid increase in empowerment. But it may have sparked a process, which with time could increase women's agency and empow- erment and eventually lead to a change in the power balances within vil- lage institutions. Kecamatan Pasean is located on the largely Muslim island of Madura. The district has a low capacity to manage conflict, with forms of violence such as carok embedded in traditional Madurese social and cultural institutions (box 10.1).Across Madura, gender roles and relations are also influenced by traditional Islamic norms that forbid women to speak with men unless at least one of the woman's male relatives or her husband is present. Within this localized context of inflamed, male-dominated routines of conflict management, the deliberative processes that KDP introduces quickly become controversial. KDP's requirement that women represent project proposals in subdistrict forums that typically draw contentious debate exposes women more directly to collective conflicts about scarce resources. Ethnographies of such conflicts in Pasaen note that women often gained greater access to decision making but did not necessarily become active contributors. Significantly, the way that KDP functionaries issue the invita- tions to women10--or indeed any members of socially marginalized groups-- to attend forums ultimately seems to help determine whether they will 182 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION attend and influence decisions. Women's increased access to decision making within KDP forums has, however, coincided with the rise of broader deci- sion-making routines in which women function as subjects rather than just as the passive objects of collective conflicts. Other Cases Worth noting is the study's finding that many cases featured two or more similarly marginalized actors battling against one another rather than against more organized or established interests. Whereas the dam case fea- tured more and less powerful groups in conflict with one another, some of the most tragic cases of conflict in the data (including carok cases) are those in which two similarly marginalized groups find themselves trapped in cycles of violence and retribution. Many of the cases from Sikka and Mangarrai--subdistricts in which the per capita gross domestic product approaches $200--fall into this category. Often both parties in such con- flicts are similarly isolated from economic, political, social, and even cultural power, yet find themselves clashing over the only available resources for sur- vival: land, water, and life itself. In such instances, intermediaries sometimes take more forceful form: for example, police interventions to stop village-level violence in Indonesia's Maluku province or military interventions in Aceh to stop senseless blood- shed. Thus, routines for immediately stopping violence need not neces- sarily be collaborative, nor must types of power be countervailing to be salient for development. Such cases of larger-scale violence, however, were beyond the scope of the study. Box 10.1. The Madurese Tradition of Carok On the island of Madura, carok is a tradition of dueling in which male combatants fight with sickles, often to the death. The duel typically fol- lows an otherwise nonviolent verbal altercation in which at least one combatant perceives a blemish on his social reputation. Researchers iden- tified several recent occasions in which carok expanded beyond two com- batants to include their family members and culminated in more than just two deaths--in some cases as many as 10. Carok typically arises following verbal conflicts whose chief character- istic is a highly paternalistic, male-dominated routine of adversarial deci- sion making in which women become objects but not subjects of conflict. In these conflicts, women lack not only influence but also access to the more or less public forums in which the adversaries clash. Source: Gibson and Woolcock 2005. EMPOWERMENT AND LOCAL-LEVEL CONFLICT MEDIATION IN INDONESIA 183 Analysis and Implications The case study data suggest that adaptive conflict management is a form of collaboration that relies at least partly on the presence of countervailing power. The sources of countervailing power seem to be quite capricious, arising from the diffuse public sphere to crystallize in the more familiar forms of inter- est groups, social movements, and lower barriers to collective action (Fung 2002). Nevertheless, there was some evidence in well-functioning KDP vil- lages that countervailing power may be on the rise. Countervailing Power and Adaptive Conflict Management One crude but revealing measure of this apparent trend in Ponorogo included the number and identity of people who participated in KDP forums. In the study's cases, for example, two and sometimes three main elite coalitions tended to fight about public decision making in KDP areas, whereas in non- KDP areas elites seemed to be more monolithic. Far from suggesting the formation of robust alliances of everyday villagers, this trend indicates that elites may be splintering into more and smaller groups because of the increased involvement of nonelite villagers. In a survey given to key infor- mants in Ponorogo, a majority reported that village decision making and conflict resolution had become at least somewhat more democratic since KDP's implementation.11 Perhaps more telling was one woman's comment from an in-depth interview: "Women have only just started to want to come together [in meetings] since KDP . . . Women were unwilling [to attend meet- ings for another development project] because . . . there was a cynical view of women who attended meetings in the village hall . . . But since KDP, women's participation has improved" (Rasyid 2004) . The study's ethnographies of local-level conflict showed that power is as much about generating shared meanings in a particular context of social and cultural norms and institutions as it is about anything else.12 Lacking influ- ence involves a vicious cycle: those unable to participate in generating shared meanings were precisely those who lacked the opportunities for the kinds of conjecture, refutation, inquiry, and criticism that might result in greater influence. Marginalized actors struggle to "get recognized" in these contexts and tend to oscillate between intense "loyalty" to core cultural values and "exit" from them, the latter often manifested in violent protest or total apathy (Hirschman 1970; Appadurai 2004). Finding the middle ground typically involves negotiating and adapting the rules and procedures of interest for- mation and defense, which at the same time makes the prospect of finding resolutions to conflict more palpable (Barron, Smith, and Woolcock 2004). But even adaptive conflict management routines--within or outside KDP processes--proved to be insufficient to reach the poorest of the poor. The most excluded need the most help to develop the countervailing power 184 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION necessary to challenge dominant norms of communication and revise oppor- tunity structures. Here, small experiments suggested that such efforts need to be couched in deeply symbolic cultural and social languages to become actions of resistance. Hence broad, far-reaching collaboration may require marginalized groups to find new ways of speaking and acting outside of intergroup collaborations such as those promoted by KDP. This should all confirm that countervailing power and context revision do not refer to the unrealistic notion that institutions rise and fall monolithically and sud- denly. Rather, the study concluded that revision is incremental and grad- ual,13 becoming most embedded when marginalized groups can develop means of countervailing power in their own personal lives or immediate social groups. Procedures for Countervailing Power The cases suggest that the procedures associated with different approaches to conflict management matter to the development of countervailing power and to the expansion of effective choice among different groups. Procedures accompanying formal and informal conflict management institutions frame the behavior and interaction of actors in group problem solving and deci- sion making, specifically by influencing which party or mediator has author- ity (in the case of two-party negotiations) or jurisdiction (in the case of mediation). Within KDP, context-revising routines flourished where rules were upheld for the election of key project team members, where open con- sultation and deliberation were embraced in forums, and where collabo- rative competition was pursued in proposal selection. Clearly, procedures can reinforce or challenge existing hierarchical struc- tures and roles. But not all routines are created equal. One type determines authorities and jurisdictions by borrowing directly from the hierarchies and stereotyped roles of society and culture. When conflicts invoked the traditional adat forums in Sikka for mediation of domestic violence disputes, for example, the authoritative mediators selected were generally groups of socially and culturally elite males. Not surprisingly, the routines fol- lowed a rigidly top-down and adversarial meting out of justice that only per- petuated the exclusion and suffering of victims. In contrast, conflicts featuring forums with more diverse intergroup associational ties, such as in the Sumorobangun Dam case discussed above, are likely to promote less hierarchically entrenched arbiters and encourage associations that are empowered to pursue collective interests constructively. By broadening its coalition to include an array of poor and relatively rich parties (especially landowners), the group maximized its countervailing power, thereby deep- ening the empowerment of all groups by making it more likely for the actors to see their choice result in a positive outcome. EMPOWERMENT AND LOCAL-LEVEL CONFLICT MEDIATION IN INDONESIA 185 Conflict management procedures are context-specific, determined in part by the underlying motivations for problem solving. The case studies revealed that the motivations that opposing groups had for starting a conflict and the motivations they had for ending it might differ greatly. For example, in some conflicts it was unrealistic to expect that groups with a history of vio- lence would develop resolutions to the contentious issue. Rather, they might simply seek a cessation of violence. This was especially common in cases where two very poor and marginalized groups clashed with one another over scarce common resources, such as land or water. Where less power- ful groups clashed with more powerful groups over a particular issue--as in the case of the Sumorobangun Dam--greater resources were often at stake, resources that could potentially be transferred from one party to the other. In such cases, marginalized parties in particular could have an incen- tive to develop resolutions. On the other hand, more powerful groups might have no incentive to end conflict because they were benefiting from the status quo and might not want to risk losing control over resources. Disaggregating conflict management In the cases discussed above, actors from marginalized groups typically succeeded either in developing sources of countervailing power or in devel- oping collaborative decision-making routines. But rarely did they achieve both. The cases might then be placed into three stylized categories. The first category concerned cases in which marginalized groups developed countervailing power, but lacked sustained collaborative routines. In the Sumorobangun case, for example, the network of associational ties that mobilized around a protest act represented a short-lived but formidable check on the otherwise dominant power of district and regional officials in charge of managing that public resource. Yet, without a formalized system of collaborative routines for embedding their cooperation, the network of protesters failed to transform the opportunity structure governing their relationship with public officials. When faced with the reality of navigating within adversarial, top-down public institutions, their coalition and its tac- tics ultimately proved unsustainable and lost its momentary influence. The second category accounted for a much larger share of the cases: those in which marginalized groups had moderate success in promoting routines of collaboration but could develop little in the way of counter- vailing power required to stave off elite domination of decision-making processes. Such was the fate of KDP in Kecamatan Proppo. There, formal procedures for brainstorming, creating proposals, and facilitating inclu- sion improved over the three years of the project, yet actors from margin- alized groups gained little sustainable influence over decision-making processes because of the embedded informal opportunity structure. 186 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION The third, most unfavorable category involves cases where marginal- ized groups neither won support for collaborative routines nor acquired a substantial or sustainable basis of countervailing power.14 Such cases are gen- erally characterized by extreme situations of elite capture. Afourth logical possibility, rare in practice, is a situation where highly collaborative routines of conflict management are matched by well-organized countervailing power. Relating to the measuring empowerment terminology, such a situ- ation presents no significant change in either agency or opportunity struc- ture, and therefore entails no significant degree of empowerment of the actors involved. KDP's Impact on Empowerment: A Typology A central finding of the study was that empowerment for marginalized groups in conflict situations involves making transitions both in the nature of forums--or opportunity structures--used for deciding conflicts, and in the types of power (or agency) used to acquire legitimacy in such forums. The findings suggest that, in terms of opportunity structure, the key task for development projects is to create the spaces, resources, and incentives that enable the poor to build effective and durable bases of countervailing power in those realms where decisions most important to their welfare are made. In terms of agency, the primary challenge is enhancing the capacities to aspire and (perhaps more importantly) engage. This means helping marginalized groups conceive of and enact alternative futures using modalities of exchange, refutation, and dialogue ("discursive routines") in which they have a com- parative advantage as a basis for more confidently and effectively negotiat- ing their entry into broader political spaces.Akey characteristic of KDP is the provision of forums for managing the conflict that the program itself nec- essarily introduces through competition over finite resources. But even relatively well-functioning KDP forums, which enable multiple face-to-face encounters between groups who may never otherwise formally interact with one another, may still create power vacuums. This all-too- common situation, well-known to theorists of collective action failure, is represented in box I of table 10.1. Box I represents forums that systemati- cally exclude already marginalized groups and that feature adversarial sets of relations and domination of decision making by one or a few groups that exercise autarchic decision-making power. Perhaps the situation of least power for marginalized groups, box I features both the lack of collabora- tive forums or opportunity structures as well as the lack of countervailing power or agency. Box II is marginally more empowering in that groups exert a small measure of informal, often disruptive power in political and social institutions. Partly because of the lack of adequately prepared organizations and similar problems of coordination between the poor, data from such exam- ples show few tangible gains for marginalized groups. EMPOWERMENT AND LOCAL-LEVEL CONFLICT MEDIATION IN INDONESIA 187 Table 10.1. Routines for Managing Local Conflict Style of Type of power associational interaction Autarchic Countervailing Top-down, I. Elite capture II. Interest group politics adversarial Corruption, collusion, Elected or appointed interest nepotism groups forced (often by ­ KDP Year 1, Kecamatan election cycles) into giving Proppo patronage to marginalized ­ Many other examples groups that may acquire Lack of agency, lack of interest group status opportunity structure ­ Ponorogo dam case Perhaps increase in agency, but no lasting change in opportunity structure Participatory, III. Access, little or no IV. Adaptive problem solving collaborative influence Well-functioning KDP forums: Poorly functioning KDP marginalized groups forums: marginalized groups maximize their seat at the have seat at table, but generate table by combining (a) public no inducement for elites to displays of their strength collaborate; elites dominate outside of collaborative forums via agenda setting and forums (via adversarial limiting, participant restriction, organizations, political parties) reduction of inputs to ignorable with (b) context-revising, advice discursive, and practical ­ KDP Year 2, Kecamatan routines within such forums Proppo resulting in "adaptive" Change in formal opportunity problem solving structure but informal ­ Few examples hierarchical structures still Actors empowered to transform embedded opportunity structure Source: Gibson and Woolcock 2005. The entrance of KDP onto the scene often helps marginalized groups make the transition from box I to box III. This movement implies a small measure of empowerment because groups that were previously excluded from the formal seats of decision making for development projects and microfinance activities ostensibly acquire an entrée. But groups rarely make the leap from box I to box IV. The lack of countervailing power here can arise for many reasons, including three directly related to KDP. First, intermediaries--in the form of facilitators--may do an inadequate job of including groups in discussions or may not ensure that invitations are broadly disseminated. Similarly, standard KDP procedures--for instance, posting prices of material for construction projects in public places and 188 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION announcing them at KDP meetings--may simply not be followed. Second, the meta-rules KDP introduces may be patently unfair where they place a premium on relatively sophisticated discursive and practical routines that women and the poorest may never have developed due to sickness, lack of time, or lack of education. In other words, gaining fully from the poten- tial of KDP might require assets, whether human, social, psychological, or informational, that the marginalized often lack. Third, the ways in which groups (both leaders and members) interact with each other and define, defend, and reconstruct their identity--or, more accurately, particular forms of identity in response to particular circumstances--ultimately shape the power of groups.15 For example, norms of decision making in KDP forums may militate against women, the poor, and individuals from certain cul- tural traditions. Where these individuals are not encouraged to participate, the (il)logic of their membership in a particular identity group may pre- clude quick involvement in a novel associational setting. The movement from box III to box IV represents the great empower- ment challenge and opportunity for community-driven development pro- jects, including KDP. The adaptive problem-solving style referred to in box IV requires a rare convergence of both a well-functioning participatory and collaborative process and the development of countervailing power among marginalized groups. This combination adds up to more than the sum of its parts. It is one matter for marginalized groups to gain access to devel- opment forums traditionally dominated by elites who invariably play a role in any development process. It is another matter altogether for mar- ginalized groups to have already built coalitions with other marginalized groups before CDD interventions arrive in a community, not least because such groups typically face a dearth of opportunities for formal associa- tional interaction. When they do build coalitions, however, such groups may enter collaborative processes from a point of strength and informal power. The combination of these two qualities has great potential to trans- late into sustained and formalized power by previously excluded groups. When CDD project creators refer to this synergistic reaction, it is often assumed that powerless groups already possess systems of countervailing power and a range of different assets when they typically do not. Thus, another key question is whether--in addition to providing inclu- sive decision-making forums and thereby changing the opportunity struc- ture--such participatory collaborative processes can help generate such points of strength for coalitions of marginalized groups to develop coun- tervailing power, and if so, how. Does KDP, in fact, empower villagers in this way? The study argues that KDP can do so over time, but that the initial stage of this brand of empowerment may be neither "flashy" nor defini- tive in its observable implications. The study's cases suggested that typically KDP forums catalyzed formal associational interactions and sustained EMPOWERMENT AND LOCAL-LEVEL CONFLICT MEDIATION IN INDONESIA 189 deliberation between diverse groups of women and the poor who might otherwise have found no reason to interact with one another. Key infor- mant surveys from well-functioning KDP districts revealed that villagers found its deliberative forums more democratic than most other forums in their villages (Barron, Clark, and Marwadi 2004). In addition to these findings, a good indicator of empowerment is whether KDP provides forums for the management of the conflict it necessarily introduces to villages and neighborhoods. By this measure, the project did exceptionally well in the study. In none of the specific cases did the study find evidence that conflicts caused by KDP forums (for example, competition between neighborhood groups vying for the ability to submit funding pro- posals to the subdistrict forum) went unresolved or became violent.16 The fact that the program provides and encourages a series of routines for allocat- ing the time, resources, and spaces required to resolve peacefully the con- flicts it inevitably introduces is an achievement in and of itself. Conclusion KDP in Indonesia represents a very self-conscious movement away from the notion of development agencies as deliverers of "projects" and toward a vision of development projects as ways to trigger and support processes in which villagers exercise discretion in solving self-identified development problems. Thus, a natural and indeed intended product of this approach is conflict. But beyond simply inducing conflict in this way, the study argued, KDP in the cases examined also cultivated a set of collaborative routines of conflict management that villagers could use to work with more organized and influential actors and thereby change the opportunity structures influ- encing their lives. Within sets of facilitated forums extending from neighbor- hood to village to subdistrict level, the simple act of participating in KDP planning and decision-making forums has often become the first occasion in which villagers from different identity groups ally themselves around purposeful collective action and decision making. The willingness and abil- ity of external agents to introduce new resources into communities, along with deliberative and administrative tools for managing the inevitable con- flict which arose over them, in many respects represented a radical depar- ture in development practice (if not theory), at least for large multilateral and bilateral agencies. Clearly, KDP is not without its flaws. Especially in relation to the most disenfranchised villagers, building this brand of conflict management capac- ity--and ultimately empowerment--depended on more than just collabo- rative routines and deliberation. The presence of countervailing power among the poor was both a product of, and breathed life into, deliberative spaces and attempts by the poor to influence decision making. People's 190 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION agency became as relevant as the opportunity structures surrounding them in evaluating the degree of empowerment brought directly or indirectly by the presence of KDP. For this reason, conflicts often arose--both within and outside KDP processes--in which marginalized groups lacked and could not develop ways of speaking and acting that would prevent elites from dominating decision making using adversarial tactics. In other words, the marginalized often lacked the assets, including human, informational, psychological, and other types of assets, to experience a significant degree of empowerment. Yet even in otherwise dark scenarios of elite capture, the data suggested that anonymous complaint mechanisms and other accountability measures within the program allowed those cut out of decision making to express their voices "defensively." In more than one case, those using this kind of recourse initiated a slow broadening of involvement and lessening of elite capture (for example, the reinstatement of elections where previously they had been illegally skipped). This is no small achievement in a country whose village-level institutions are still emerging from underneath the decades-long shadow of national and district-level political hegemony over neighborhood and village-level decision making. Nevertheless, the chal- lenge of providing resources, spaces, and incentives for more positive intra- group countervailing power will always remain for any development program. In cases where KDP cultivates conflicts featuring both collaboration and tangible points of political power for marginalized groups, the result can be not only a well-functioning school or medical clinic but a process of defin- ing and defending collective group interests and solidarity. The study argues that the beginning stage of such a transformation toward popular notions of local-level empowerment--in which unequal groups build the capabil- ity to peacefully engage one another in conflict--is a humble but nontriv- ial outcome for a development project, especially where violence has been a normal occurrence. Furthermore, deliberation and shared intergroup decision making have been followed in the data by incremental steps toward more equitable spaces for engagement between more and less organized and influential actors. The ethnographic approach taken to gathering information in this coun- try case generated a depth of data and quality of findings that would have been impossible to capture using questionnaire-based techniques. It also demonstrated the use of a recognized sampling methodology--propensity score matching--and thereby avoided some of the criticisms often leveled against ethnographic studies that finds are not representative. Ethnographic enquiry is an approach well-suited to in-depth, academic research and one that, with creativity, could be reduced to forms of application and integrated as recurrent (perhaps annual) special studies within a broader project or EMPOWERMENT AND LOCAL-LEVEL CONFLICT MEDIATION IN INDONESIA 191 policy monitoring system--particularly if capacity is built for self-recording of events in which power relations are critical. However, because of the need for a high level of interviewing skills, the startup costs of taking an ethnographic approach are generally high. In addition, because of the time needed to engage sufficiently to ensure reliability of complex information, replicability across a sufficiently large population to yield generalizable results is expensive. The resourcefulness of the country team responsible for this case took care of time and funding issues that may well act as a brake on other teams trying to use the same approach. Notes 1. Evans (2002) and Rao and Walton (2004) similarly develop the concept of "equality of agency" to consider how individual opportunity may be influenced by relational social and cultural systems that ultimately affect poverty and inequality. 2. This literature can at times be limited in its focus on collective action. Under certain conditions, such as in a mature democracy, individuals can experience a high degree of empowerment. Where democratic systems and processes function there is often little need for collective action. For example, where systems of transparency and accountability in local governance and justice function effectively, an individual is as likely as a group to use that opportunity structure and to be equipped with the assets that let him or her do so. 3. See Barron et al. (2004) for a more detailed description of the sampling methodology. 4. The propensity score is a statistical measure designed to calculate the probability of a given household or village being selected for inclusion in a program (for an introduction, see Baker 2000). The qualitative interviews helped to control for unobserved variables, which propensity score techniques alone cannot do. 5. See Barron et al. (2004) for details of the propensity score matching technique used to construct treatment and control groups. 6. After the initial round of qualitative fieldwork, a key informant survey was designed to assess the generalizability of some of these early conclusions. The preliminary results of this work are reported by Barron, Clark, and Marwadi (2004), but final conclusions await the findings from a companion survey conducted in non-KDP areas. These questions, it should be stressed, are largely subjective and attitudinal by nature and design, whereas the case study material reported here covers actual behavior, and does so in greater depth. For this reason, and given the rigorous sampling procedures that underpinned their selection, they should perhaps be given greater credence, at least for now. 7. These conflict categories come from Diprose (2004). 8. Discussion of this case draws heavily on Diprose (2004). 9. See Barron, Smith, and Woolcock (2004) for a more detailed discussion of meta-rules. 10. It was found, for example, that sending KDP invitations home with schoolchildren increased the likelihood of women attending forums because children were more likely to give the invitations to their mothers, the principal caregivers, than to their fathers. 11. These results are not statistically significant; see Rasyid (2004). 192 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 12. This has been suggested by a long line of anthropological research associated with Geertz (1977) and Horowitz (1985). 13. These incremental changes mirror what the psychological literature calls "small wins" (Weick 1984) and the legal institutional literature calls "small steps" (Sunstein and Ullmann- Margalit 1998). 14. This was most prevalent when KDP was poorly implemented or when local governance structures were thoroughly captured by local elites, setting up cycles of disappointment, disillusionment, and frustration whose primary beneficiaries were (once again) the rich and powerful (see Barron, Diprose, and Woolcock 2005). 15. Diprose (2004) and Barron, Smith, and Woolcock (2004) develop this notion as the "dynamics of difference." 16. This empirical finding is confirmed in the larger evaluation study (see Barron, Diprose, and Woolcock 2005). 11 Negotiating Social Change: Gender, Caste, and Ethnic Dimensions of Empowerment and Social Inclusion in Rural Nepal Adapted from a study report by Lynn Bennett and Kishor Gajurel This chapter summarizes the first phase of an ongoing study on social change in rural Nepal with focus on the gender, caste, and ethnic dimen- sions of empowerment and social inclusion. Apart from being one of the cases of this initiative, the study in Nepal also represents phase 1 of the Measuring Empowerment and Social Inclusion (MESI) study, a longitudi- nal mixed-methods research project. The MESI study seeks to understand the processes of empowerment and social inclusion to develop more mean- ingful ways to measure and track changes in these two outcomes--and also to learn something about how empowerment and inclusion affect other important development outcomes.1 The context for the Nepal study is an impact evaluation of the World Bank­funded Nepal Rural Water Supply and Sanitation (RWSS) project, which started in 1996. RWSS seeks to promote decentralization and to increase the involvement of beneficiaries and the private sector in rural water supply and sanitation service delivery. It does so by inculcating a demand-driven approach in the sector; enabling communities to take lead roles in the iden- tification, design, implementation, operation, and maintenance of their water supply and sanitation schemes; and developing adequate capacities in the gov- ernment and nongovernment sectors to support community initiatives. The project has entered its second cycle: RWSS I was completed and RWSS II was launched in 2004. By far the largest component of the RWSS project has been the construc- tion of water supply and sanitation schemes in 900 rural communities. 193 194 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Service agencies and support organizations, including nongovernmental organizations (NGOs), community-based organizations, and private-sector firms, help local communities define their needs and design suitable water and sanitation projects to be submitted for approval. Normally, a subpro- ject takes 36 months and has three main phases: (a) predevelopment, involv- ing feasibility studies and selection of support organizations; (b) development, in which the scheme is designed and the rural community is prepared to take ownership; and (c) implementation, involving construc- tion, establishment of a community action plan, and transfer of ownership to the rural community (Alsop 2004b). Because of the participatory and inclusive nature of the project and because of RWSS I's clear project objective of empowering women, the MESI team selected it as the case study for research on empowerment and social inclu- sion. Each "batch" of RWSS communities follows a 36-month process during which water user groups are organized to build and manage their own water tap schemes. Subsequent water schemes incorporate the lessons learned from previous ones. Over its eight years of operation, the RWSS Fund Board has shown itself open to ways to improve both the efficiency and the equity of its operations. This chapter first outlines the conceptual framework for the study and describes how this framework informed the identification and construction of a set of empowerment indexes. The chapter then examines patterns of empowerment and inclusion within the research communities, with a focus on gender and ethnic or caste distinctions, and discusses the determinants of these patterns, before briefly concluding. Conceptual Framework of the Study The MESI research began in late 2001, linking with the current study on mea- suring empowerment in 2003. As explained briefly below, the conceptual framework of the MESI study was slightly different from that of the empow- erment framework presented in chapter 2 of this volume in that MESI ana- lyzed empowerment and social inclusion as separate but interdependent and complementary dimensions of the overall process of social transforma- tion. Simply put, the MESI study conceptualized empowerment as centered around agency and the actors' assets, and social inclusion as capturing the opportunity structures. This is in contrast to the empowerment framework, which sees both agency and opportunity structures as factors influencing the degree of empowerment. The conceptual distinction between empowerment and social inclusion set the methodology of this case study slightly apart from that of the others, but did not change the approach significantly. At the outcome level, in terms of income and human development levels and the political influence of the poor on the institutions that affect them, the Nepal study conceived of NEGOTIATING SOCIAL CHANGE IN RURAL NEPAL 195 empowerment and social inclusion as virtually indistinguishable. However, at the level of development interventions to achieve these outcomes, the dis- tinction was seen as useful in broadening the range of strategic options available to the Bank and its clients. At the broadest level, the goal of the Nepal study has been to help the World Bank live up to its commitment to results-based management. If empowerment and social inclusion are now seen as important intrinsic as well as instrumental values and outcomes that Bank investments seek to help client countries achieve, then a means of measuring these outcomes is needed. The specific objectives of the study are 1. to develop gender-sensitive ways to measure levels of empowerment and social inclusion of different groups; 2. to determine influencing factors or covariates of empowerment and social inclusion broadly; 3. to assess changes in levels of empowerment and social inclusion achieved by different groups in the context of a specific development intervention (in this case, the Nepal Rural Water Supply and Sanitation project (RWSS I and II); 4. to understand the extent of the contribution of empowerment and social inclusion to broader development objectives such as higher incomes, improved health and education outcomes, and reduced fertility levels, and vice versa. This initial phase of the study addressed the first two of these objectives and made a preliminary assessment of the empowerment and social inclusion impact of the RWSS I, based on a with/without research design. However, a more rigorous impact assessment of RWSS must await the completion of phase 2 of the study when longitudinal panel data will be available. Only at that time will it be possible to fully address objectives 3 and 4. This chap- ter will therefore focus on how the study addressed the first two objectives. Defining Empowerment and Social Inclusion For the purposes of the MESI study, the social transformation required to achieve social justice and equality was conceptualized as comprising two interrelated but analytically distinct processes: empowerment and social inclusion. The study emphasized the measurement and analysis of social inclusion to map the relations between the two fields. The definitions of these processes were taken originally from the World Bank's Social Analysis Sourcebook (2002b): · Empowerment was treated, in the draft study, as the enhancement of assets and capabilities of diverse individuals and groups to function and to 196 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION engage, influence, and hold accountable the institutions that affect them.2 Following peer review commentary, this definition was modified to use the definition of empowerment presented in part 1 of this volume. · Social inclusion is the removal of institutional barriers and the enhance- ment of incentives to increase the access of diverse individuals and groups to development opportunities. Empowerment in this sense was conceived in the study as a process that occurs at the individual and group levels, through changes in internal self- perceptions and changes in the way external structures and institutions are under- stood by the excluded individual or group. Empowerment was perceived as occurring from below or from within, as marginalized individuals and groups become able to organize themselves and their resources to contest unequal structures and define their own place in the wider context of their lives. The idea of agency or self-actualization was therefore central. However, this study broke empowerment down into two parts. The first related to the improvement of capacities and assets and was referenced as livelihood empowerment. This was more precisely defined as an increase in the human and material resources and assets an individual or group can access. Closely related to this was the second aspect of empowerment, described as mobi- lization empowerment (Bennett 2003). According to the study, this involved increased self-awareness on the part of individuals and groups and the col- lectivization of action and purpose. Changes in empowerment, therefore, were seen as apparent and measurable mostly at the level of the individual, though these were aggregated so that the empowerment of social groups and their enhanced capability for collective action could be understood. Social inclusion was treated as the process whereby institutions and poli- cies are transformed so that all people are able to enjoy equal opportunities and participation. The study found that, in contrast to empowerment, social inclusion occurs at the systems level--that is, in terms of changes in the opportunity structure or the institutional environment in which people live their lives. This resulted from either changes in formal structures such as policies and laws, or changes in informal structures such as norms and belief systems. Social inclusion was seen as observable at four levels in the study: the household, the local community, the local government, and the general society and nation.Analysis of change at the different levels revealed the degree of inclusion as well as the links that were least developed. Complementary Dimensions of Social Change The interaction of empowerment and social inclusion was seen to lead to the kind of transformational social change that creates equitable and just soci- eties. The study team regarded the relationship between the two aspects NEGOTIATING SOCIAL CHANGE IN RURAL NEPAL 197 of the social change process as not straightforward, that is, it did not follow a predetermined or consistent path from one to the other. Rather, the two were deeply intertwined and interdependent. Without systems-level change, the team believed it unlikely that achievements in the empowerment of disadvantaged groups would have a sustainable impact. Nor would real and meaningful social inclusion be likely to occur without the empowerment that enhanced and recognized the agency of those groups meant to benefit from institutional changes. Conceptualized in these terms, empowerment and social inclusion repre- sent particular approaches to be used in development initiatives. They are also outcomes that are valuable (a) in their own right for the positive social transformation outcomes, and (b) for the assumed positive instrumental impact that less rigidly hierarchical social organization can have on other development outcomes. In this model of social transformation, an explicit expectation is that the creation of more equitable relations between social groups and between social groups and the institutions meant to serve them is positively correlated to the achievement of other development outcomes. Many studies have begun to illustrate and articulate the relationship between higher levels of social development and equality and higher levels of eco- nomic development and more stable political environments.3 Thus, the core hypothesis of the study was that investments in empow- erment and social inclusion would result in "positive" empowerment and social inclusion outcomes, and that these outcomes in turn would posi- tively influence human and economic development indicators such as poverty reduction, human security, and good governance. More specifi- cally, the study hypothesized that the empowerment and social inclusion of women belonging to traditionally excluded castes would result in primary positive changes such as reduced domestic violence against women; less restriction, public intimidation, and violence toward low castes; and greater self-esteem and community influence for previously excluded groups. Secondary positive developments would include improved health-seek- ing behavior, greater contraceptive use, increased involvement in the market economy, and higher incomes for women. Research Design and Methodology The data presented in the study were collected in a survey conducted in 1,000 households from a sample of 60 villages in the eastern, central, and western regions of Nepal.4 Great care was taken in developing and testing the survey instrument to ensure that it used the village vernacular (Deshi) rather than more literary and Sanskritized Nepali. Also, both male and female enumerators were used so that in most cases women respondents could be interviewed by women. In addition, despite an ongoing Maoist 198 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION insurgency, the anthropologists and sociologists on the research team were able to carry out a six-week interactive study in four of the sample villages. The study grouped the 59 different caste and ethnic groups covered by the survey into four main groups, based roughly on their rank in the tra- ditional caste system as codified in the Mulki Ain or National Code of 1854. Brahmans, Chhetris, and Newars of Kathmandu valley were clustered together as the BCN group, generally viewed as the most privileged.5 The study also clustered the middle-ranking Hindu castes together in a second group and the Janajati or indigenous peoples as a third group. The fourth group was the Dalits, the former "untouchable" castes at the bottom of the old social hierarchy. The study methodology built on the idea of identifying the main, more or less universal "domains" where empowerment and inclusion operated and then choosing locally relevant indicators of individual status in those domains. Midway into its methodology development phase in 2003, MESI joined forces with this empowerment study, and the shape of the final methodology and the instrument design reflected that collaboration. The Nepal study's conceptu- alized empowerment and social inclusion as two separate phenomena, and this was reflected in the methodological design. Thus, the study sought to measure social inclusion separately from empowerment in two different indexes, but it also combined the two together in an overall index. As discussed elsewhere in this volume, the use of indexes as a way to assess differentdegreesofempowermentshouldbeconsideredcarefullybeforebeing attempted.6 Indexes have proved useful for monitoring trends and patterns and have revealed interesting patterns to date in this ongoing evaluation, as shown in this chapter. But it is also important to recognize the limitations associated with indexes in terms of their usefulness for predicting or explain- ing complex processes. In addition to questions that can be raised about combi- nations of metrics within an index, these issues become particularly apparent when two or more indexes are themselves aggregated. In this case, the team found an index valuable, and the variables were designed especially for and were appropriate to this use--although a deeper explanation of, and perhaps some statistical testing of, the choice of variables and analytical technique would have protected the team from critics. The same could be said for the study's use of regression analysis, where a stronger focus on the arguments for using this method would have been valuable and would have made the results more robust. The Empowerment Index An Empowerment Index (EMI) was developed to measure empowerment, using a range of variables that sought to capture the respondent's sense of agency. The survey sought evidence and indicators of psychological, infor- mational, and social asset endowment, as well as evidence and indicators NEGOTIATING SOCIAL CHANGE IN RURAL NEPAL 199 that the individual had actively demanded access to services or tried to influence local community decisions. The EMI included some data from the "inner" psychological sphere, as well as data on social, economic, and political relations within the com- munity and between the community and various levels of the state. A set of indicators was developed to measure the extent to which an individual had actually engaged with the institutional environment (or opportunity structure) by seeking services from it or trying to change or contest it. The indicators comprised five dimensions: (1) knowledge and awareness of rights and procedures, (2) participation in local development services, (3) confidence and comfort level in accessing services and exercising rights, (4) social networks (economic and political), and (5) efforts to influence local government (box 11.1). The Social Inclusion Index The assumption behind the Social Inclusion Index (SII) was that it was pos- sible for an individual or group to become "empowered" by becoming con- scious of the injustice of the surrounding institutions and deciding to act to protest or challenge institutional norms. The resulting improved outcomes depended not only on actions or agency of the empowered person or group, but also on the impact of these actions on institutions. Following the empow- erment framework's logic, the country team indicated that it was quite pos- sible--in the short and medium term--for exclusionary institutions to continue to thwart individual and even collective efforts to exercise agency. A wide range of indicators were used in the SII, chosen to reflect the degree to which the respondent was able to exercise his or her agency effec- tively to achieve desired outcomes. These outcomes included being able to obtain justice, access police protection, remain free from intimidation in public spaces, influence the performance of a child's school, or access other public services and economic opportunities. The indicators attempted to measure the results of the individual's interaction with formal and informal institutions or "rules of the game" constituting their opportunity structure. The indicators for the SII included four dimensions: (1) self-perceived status of own caste or ethnic group, (2) restricted access and public intimidation, (3) effectiveness of local political influence, and (4) effectiveness in obtain- ing services and opportunities (box 11.2). The Composite Empowerment and Inclusion Index The Empowerment Index and the Social Inclusion Index were brought together into a Composite Empowerment and Inclusion Index (CEI) that was developed to test aggregate associations between empowerment and inclusion, project interventions, and development outcomes (figure 11.1). 200 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Box 11.1. Indicators Used for the Empowerment Index (EMI) The following variables were used in constructing the index: 1. Knowledge and awareness of rights and procedures Understanding of police procedures Understanding of court procedures Knowledge of National Code and rights of Dalits Knowledge of local services 2. Participation in local development services Seeking local services Participation in programs of child's school 3. Confidence and comfort level in accessing services and exercising rights Approaching the police Approaching the courts Approaching children's school 4. Social networks (economic and political) Connections for getting a job for oneself Ability to help others get a job Connections at ward level Connections to local service agencies Connections at Village Development Council (VDC) level Connections at District Development Council (DDC) level 5. Efforts to influence local government Suggestions or complaints at ward level Suggestions or complaints at VDC level Suggestions or complaints at DDC level Advice to school officials Source: Bennett and Gajural 2005. The Women's Empowerment and Inclusion Index Neither the EMI nor the SII included indicators of power differentials between men and women that were played out within the household. In general, societal norms favor men; even the Dalit man who may have to follow hierarchical caste norms and suffer humiliation in the wider com- munity is nevertheless "king in his own household" vis-à-vis the females in his family--especially his wife or other women who joined the family NEGOTIATING SOCIAL CHANGE IN RURAL NEPAL 201 Box 11.2. Indicators Used for the Social Inclusion Index (SII) The following variables were used to construct the SII: 1. Self-perceived status of own caste or ethnic group Relative economic status and success of own group Relative contentment and comfort with social status of own group Respectful treatment Relative access to opportunity Cooperation from other groups Respect in the community 2. Restricted access and public intimidation Whether the respondent is restricted from entry into certain public areas (such as temples or people's homes) or prevented from using public facilities (such as water taps) Whether the respondent faces verbal or physical intimidation, humilia- tion, or violence in public spaces such as the village or the nearest bazaar 3. Effectiveness of local political influence Result of complaints or suggestions they have made at ward, village, or DDC level 4. Effectiveness in obtaining services and opportunities Invited by agencies to participate Promptness of service Consulted for opinion Access to training opportunities Source: Bennett and Gajural 2005. through marriage. Therefore, in addition to whatever opportunities or con- straints women may experience in their wider social interactions on the basis of their caste or ethnic identity or their wealth or educational levels, women also face an additional set of institutional barriers to empowerment that are enacted--particularly in their role as wives--in the household and family sphere. The Women's Empowerment and Inclusion Index (WEI) was designed to also take into account all barriers that are part of the "opportunity struc- ture" (figure 11.2). The indicators for the WEI cover five dimensions: (1) domestic violence and intra-household behavior; (2) mobility and ability to travel to various destinations alone, and the need for permission; (3) con- trol over fertility; (4) control over self-earned income; and (5) household decision making (box 11.3). 202 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Figure 11.1. Composite Empowerment and Inclusion Index (CEI) Levels, by Gender and Caste and Ethnicity BCN Tarai Middle Dalits Janajatis All 0.00 0.10 0.20 0.30 0.40 0.50 0.60 CEI Index Male Female Source: MESI study. Analysis: Who Was Empowered and Who Was Included? The following analysis discusses the data generated to date for the different indexes and emerging variations in terms of gender, caste, and ethnicity. The first part of the MESI study focused mainly on measuring empower- ment and social inclusion for the people in the sample villages and assess- ing how the level of empowerment varied according to gender and caste. The MESI study is still under way, and the analysis has not yet gone much further in terms of assessing the extent of change in the levels of empowerment and social inclusion resulting from the RWSS project. This will be part of the second half of the study. This analysis, therefore, cannot discuss how the level of agency has changed among the villagers since the intervention, or to what extent opportunity structures have been altered as a result of the project. However, a summary is presented at the end of this chapter of how the study tried to assess the effect of the project in rela- tion to the idea of group organization and group membership, as the orga- nization of people into groups was one of the main characteristics of the RWSS project. NEGOTIATING SOCIAL CHANGE IN RURAL NEPAL 203 Figure 11.2. Women's Empowerment and Inclusion Index (WEI) Levels, by Caste and Ethnicity BCN Tarai Middle Dalits Janajatis All 0.45 0.50 0.55 0.60 0.65 WEI index Source: MESI study. Hierarchies Based on Caste and Ethnicity The study found that in a broad sense, the ranking of caste and ethnic groups that was set out in the Mulki Ain in 1854 was also still reflected in the way different groups ranked on the CEI in rural Nepal. As shown in figure 11.3, the dominant Brahman-Chhetri-Newar group scored higher than the Janajatis, who in turn scored higher than the Dalits. BCN levels on the CEI were 21 percent higher than those for Dalits and 10 percent higher than those for Janajatis. The Janajatis' CEI levels were 11 percent higher than those for Dalits. This pattern held true not only for the CEI index but also for the sepa- rate empowerment and social inclusion indexes. Even when helpfully dis- aggregated into their individual indicators, the EMI and the SII revealed that the BCN group consistently came out on top. Hence, the leading posi- tion of the BCN group was repeated in all the separate elements of the Empowerment Index, where this group "outscored" the Dalits as follows: · Knowledge of rights and procedures: BCN score was 1.3 times higher than Dalit score (BCN = .59, Dalit = .45). 204 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Box 11.3. Indicators Used for the Women's Empowerment and Inclusion Index (WEI) The following variables were used in the construction of the WEI: 1. Domestic violence and intra-household behavior Experience and frequency of verbal or mental abuse Experience and frequency of physical abuse Treatment by husband initially and now 2. Mobility and ability to travel Ability to travel to various destinations alone, and the need for permission 3. Control over fertility Discuss family size with husband Discuss contraception with husband Use contraceptive method 4. Control over self-earned income Earns cash income Keeps money Decides how to spend 5. Household decision making Difference between male and female household member's scores Source: Bennett and Gajural 2005. · Participation in local development services: BCNs scored twice as high as Dalits (BCN = .46, Dalit = .23). · Confidence and comfort level in accessing services and exercising rights: BCN score was 2.26 times higher than Dalit score (BCN = .43, Dalit = .19). · Social networks: BCN score was 1.5 times higher than Dalit score (BCN = .41, Dalit = .27). · Efforts to influence: BCN score was 3.7 times higher than Dalit score (BCN = .26, Dalit = .07). The indicators of ethnic group agency focused mainly on informational and organizational assets as knowledge of rights and procedures, requires access to information and knowledge of how to get that access. The impor- tance of organizational assets was reflected in the indicators on participa- tion in local services and social networks.Additionally, psychological assets were assessed by measuring the level of confidence and comfort of the respondents in accessing services and exercising rights, quantifying the extent to which the respondents perceive themselves as citizens with equal NEGOTIATING SOCIAL CHANGE IN RURAL NEPAL 205 Figure 11.3. Composite Empowerment and Inclusion Index (CEI) Levels, by Caste and Ethnicity BCN Tarai Middle Janajatis Dalits 0.00 0.10 0.20 0.30 0.40 0.50 CEI index Source: MESI study. rights and thereby gaining insight into their capacity to perceive alternative choices. Often BCNs have been associated with higher levels of material assets as well as higher levels of education (a human asset). But it is note- worthy that the BCN group also came out with the highest score on all three of the above-mentioned assets, including psychological assets. This gives an idea of the interaction among different assets and supports the notion presented in the framework in chapter 2 that the command of one asset can affect the endowment or use of other assets and thereby enhance the agency of an individual or group. The dominant position of the BNC group was similarly reflected in the SII: · Self-perceived status of own caste and ethnic group: BCNs saw the recog- nition and opportunity they got from their caste identity as nearly two times what the Dalits felt they received from their caste identity (BCN = .81, Dalit = .43). · Restricted space and public intimidation: BCNs were 1.8 times less likely to experience restriction than Dalits (BCN = .97, Dalit = .52). Ninety-nine percent of the Dalits had experienced restrictions while only 12 percent 206 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION of the BCNs had (and that was mostly related to women's temporary menstrual seclusion or to temporary pollution upon the death of a rel- ative). · Effectiveness of local political influence: The BCN influence score was 2.3 times higher than the Dalit score (BCN =. 06, Dalit = .026). · Effectiveness in obtaining services and opportunities: The BCN score was 1.7 times higher than the Dalit score (BCN = .13, Dalit = .076). This dominance of the BCN group across indicators showed that the local opportunity structure was also heavily determined by ethnic and caste divisions. As the caste system was officially abolished decades ago and the law formally prohibits any kind of caste discrimination, this gives an idea of the complex interaction between formal and informal institutions and the challenges to altering these structures to increase empowerment and social inclusion for all of the different caste and ethnic groups. Cross-Cutting Hierarchies of Gender and Caste-Ethnicity When simple male-female comparisons were constructed, the results were for the most part as expected. Men scored significantly higher on the CEI and on the individual indicators (figure 11.4). Men's CEI levels were 7 per- centage points higher than women's. Interestingly, the gender gap was higher (14 percent) for the empowerment measure and lower (3 percent) for the social inclusion measure. This was probably because the SII (which, as men- tioned above, was based on data that could be meaningfully collected from both men and women) was more effective at picking up community-level social stratification than intrahousehold-level stratification (such as age, but primarily gender relations within the family). Therefore, in assessing women's opportunity structure or degree of social inclusion, it is impor- tant also to keep one eye on the WEI measure that sought to pick up intra- household power differentials. Disaggregation of gender differences by caste and ethnicity revealed that the lowest level of mean CEI was found among female Dalits and the highest among male BCNs. The mean CEI of male BCNs was about two and a half times higher than the mean CEI of female Dalits. The intra­caste and ethnicity comparison of the mean male/female CEI index indicated that the highest gender differential was observed in middle-caste Hindu groups (13 percent) followed by BCNs (10 percent). The lowest gender differences were observed among Dalits (6 percent). The cross-cutting impact of gender and caste exclusions was again apparent in the high differential between the empowerment levels of BCN men and Dalit women, who fell 30 per- centage points lower on the EMI measure. NEGOTIATING SOCIAL CHANGE IN RURAL NEPAL 207 Figure 11.4. Composite Empowerment and Inclusion Index (CEI) Levels, by Gender Male Female 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 CEI index Source: MESI study. Examining the gender differentials in the community-level empower- ment index (EMI) was instructive. Men scored significantly higher on the EMI than women among all ethnic groups, the gender gap ranging from 4 per- cent for Dalits to 9 percent for BCN/Middle (figure 11.5). Similarly, on all disaggregated indictors of EMI, women fell below men, indicating that women's agency was significantly lower than men's: · Participation in local development services: Men participated in or took advantage of local development services 1.6 times more than women. · Confidence and comfort level in accessing services and exercising rights: Men were 1.5 times more confident than women in accessing services and exercising their rights. · Social and political networks: Men's levels of social and political net- works were one and a half times higher than women's. · Efforts to influence: Men tried to influence the institutions that were supposed to deliver services to them 2.7 times more than women did. When examining individual indicators for social inclusion that related most evidently to caste and ethnicity, it was revealed that there was no signif- icant difference between male and female outcomes. In other words, the exclu- sionary impacts of caste and ethnicity were not significantly further qualified by gender. But for the indicators that had to do with influence in the wider 208 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Figure 11.5. Community Empowerment Index (EMI) Gender Gap by Caste and Ethnicity: Result of Multivariate Analysis 10 9 8 7 6 centage 5 Per 4 3 2 1 0 BCN/ Janajatis Dalits All Tarai Middle Source: MESI study. community and relative effectiveness in getting access to services, the male- female gap reappeared: · Self-perceived status of own caste and ethnic group: No significant gender difference. · Restricted space and public intimidation: No significant gender difference. · Effectiveness of local political influence: Men were 4.8 times more able to actually influence their institutional environment than women. · Effectiveness in obtaining services and opportunities: Men were one and a half times more successful in actually getting services than women. Shifting Patterns in Women's Empowerment by Caste and Ethnicity When looking at women's CEI levels by caste and ethnicity, the study encountered some major surprises. One of these was a reversal of the find- ings of the Status of Women study (Acharya and Bennett 1981), which 25 years ago found that Brahman-Chhetri women were actually less empow- ered than Janajati or Dalit women in terms of their mobility, their input into household decision making, and a number of other measures. The data of the Nepal study showed, however, that women from the dominant BCN NEGOTIATING SOCIAL CHANGE IN RURAL NEPAL 209 castes had significantly higher CEI scores than Janajati or Dalit women (figure 11.6). A similar pattern appeared in the comparison of WEI levels across caste and ethnic groups. Here both BCN and middle-caste groups had significantly higher WEI scores than Janajati or Dalit women. To understand better this apparent shift, the study first examined the separate indicators in the CEI index that captured the interaction of caste and gender at the level of intra-community gender behavior. Examination of the separate empowerment indicators in the EMI showed that BCN women had significantly higher scores than women from other caste and ethnic groups on all the indicators. This showed that the pattern of advanced levels of agency for the BCN group in general was repeated when looking specifically at women of each caste. BCN women, like the BCN group in general, had more informational assets than women from other groups in terms of access to information and knowledge of rights, more organiza- tional assets in terms of high levels of participation, and more psycholog- ical assets in terms of confidence in demanding rights and envisioning a change in traditional gender restrictions. Figure 11.6. Composite Empowerment and Inclusion Index (CEI) Levels for Women, by Caste and Ethnicity BCN Tarai Middle Janajatis Dalits All 0.00 0.10 0.20 0.30 0.40 0.50 CEI index Source: MESI study. 210 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION These patterns were repeated in the measurement of opportunity struc- tures. The SII showed that BCN women scored highest on all of the separate social inclusion indicators except those reflecting the restricting impact on women's mobility of high-caste concern with ritual purity and menstrual pol- lution. Among BCN women the perception that their own caste and ethnic group provided them more security, status, and opportunities was signifi- cantly more widespread than among women in the other groups, and BCN women were significantly more effective than women from other groups in getting access to services and opportunities. The opportunity structures in this way seemed less restricting for BCN women than had been the case 25 years before, and less restricting than was the case for women in the other caste and ethic groups. Cross-Cutting Caste, Ethnic, and Gender Dimensions of WEI The study then looked separately at the indicators in the WEI, which primar- ily captured intra-household gender behavior. Much in line with the find- ings of the CEI index, regression results showed that BCN and middle-caste groups had significantly higher WEI levels than Janajatis or Dalits. However, the results were not as straightforward for the individual WEI indicators as they were for the individual CEI indicators. For the house- hold decision-making data the study found, surprisingly, that there were no significant differences between male and female decision-making levels. Furthermore, BCN women had the lowest decision-making levels relative to men in their households--though the difference was not statistically sig- nificant. In other related indicators of the WEI such as control of self-earned income, BCN women also scored the lowest, though again the difference was not statistically significant. This may be in part because of the traditional upper-caste view that high-caste women should not work for wages, but should instead work only within the household subsistence economy. In the study, only women who actually earned cash income from the market could get a positive score for the control over cash income indicator, so the low score of BCN women could reflect the continuing low status accorded to women who do wage work. The qualitative findings, however, suggested that BCN women, while they may not want to work for wages, were becoming increas- ingly interested in ways to earn cash income through micro businesses; this was one of the reasons they were so active in savings and credit groups. These findings give rise to interesting considerations regarding the inter- action among different assets. In this case it seems that higher levels of human and organizational assets attributed to the BCN women did not as such lead directly to their acquiring greater material assets or having greater control of the family's financial assets. The informal norms forming part of the opportunity structure constrained this development, while formal changes in these structures through, for example, increasing access to sav- NEGOTIATING SOCIAL CHANGE IN RURAL NEPAL 211 ings and credit groups did create opportunities for an expansion of asset endowments. For the remaining WEI indicators, BCN women scored higher than others. On the indicator of domestic violence BCN women were the least vulnera- ble, though the difference was not significant.Also, in terms of mobility, BCN women reported the highest levels of freedom to visit destinations alone, and their overall mobility was significantly higher than that of women in other groups. More than 70 percent of women from the dominant BCN group reported being able to visit the local market, the local tea shop, and a temple alone. Regression results showed that BCN women had greater freedom of movement, even when controlling for respondent's level of education, hus- band'slevelofeducation,region,householdwealthrank,andsourceofincome. This represented a major change from the earlier study. It suggests that the traditional concern with strict maintenance of the sexual purity of high- caste women through restrictions on their mobility may be weakening, and the opportunity structures slowly changing in this area. Still, women from all ethnic and caste groups overwhelmingly felt they needed permission from their family members to travel. More than 80 percent of all women needed permission to go to their maternal homes; almost 90 percent of all women needed permission to go to district headquarters; and more than 90 percent needed permission to go to a cinema hall. BCN women had significantly higher levels of knowledge of women's legal rights and more contact with the district Women's Development Office (WDO) than women from other groups. Finally, with respect to the last WEI indicator, control over fertility, the BCN women were once again among the most empowered, although there was no significant difference between groups for this measure. Encouragingly, the data showed quite high levels of con- trol over fertility for women in all groups. This suggests an opening in the opportunity structure, as norms of family planning seem to be changing gradually. This change is combined with an increase in informational, human, and psychological assets, making individual women more capa- ble of making their own choices in this area. While the study found it to be a good sign that BCN women were becom- ing empowered, it noted also that women from other groups in relative terms may be falling behind, strengthening existing caste and ethnic dis- parities while lessening gender disparities for some. However, the regres- sion results reported in table 11.1 showed that caste and ethnicity had no significant influence on the level of women's empowerment and inclusion. Significant variables included · age (the higher the age the higher the level of women's empowerment and inclusion); · level of education (the higher the level of education the higher the level of women's empowerment and inclusion); 212 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Table 11.1. Multiple Regression Analysis Examining Influence of Caste and Ethnicity on Women's Empowerment and Inclusion Std. Variable Beta Error Sig. BCN .914 1.368 .504 Middle 4.013 2.393 .094 Janajati .459 1.368 .738 Age 225 .062 .000 Completed grades .411 .184 .026 Wealth scale .137 .306 .655 Group membership 3.195 1.159 .006 Exposure to media and places 2.962 1.498 .048 Fluency in Nepali language 1.266 1.25 1.312 More than 5 km from district headquarters .441 .994 .657 Knowledge of women's legal rights 3.351 2.413 .165 Contact with WDO 3.800 1.648 .021 Constant 45.859 3.036 .000 Source: Bennett and Gajural 2005. Note: Shaded rows indicate significant variables. · group membership (members have higher levels of women's empow- erment and inclusion than nonmembers); · exposure to media and places (the higher the exposure, the higher the level of women's empowerment and inclusion); · contact with WDO (the higher the level of contact with WDO, the higher the level of women's empowerment and inclusion). Caste and ethnicity had no significant influence once these five variables were controlled for. Of the five, all but age are factors that can be influenced by government policy and by civil society and donor actions. The signifi- cance of group membership and contact with the WDO were particularly encouraging and suggested fruitful approaches for future efforts to empower women--and to reduce caste and ethnic disparities at the same time. What Determines Empowerment and Social Inclusion? Caste and ethnicity have a strong influence on empowerment and social inclusion, as shown by the range of CEI scores presented in figure 11.7. Furthermore, regression analyses that included both caste and ethnicity and gender variables showed that together they explained 33 percent of the variations in the CEI. In two separate bivariate analyses, one for caste NEGOTIATING SOCIAL CHANGE IN RURAL NEPAL 213 and ethnicity and the other for gender, the study found that caste and eth- nicity alone explained 26 percent of the variation and that gender alone explained 7 percent of the variation. This supported the hypothesis that caste and ethnicity are a stronger variable than gender in determining indi- vidual CEI levels.7 In each of the four communities covered by the in-depth studies, caste continued to be the most important factor determining economic status, community involvement, and level of education.Although the caste system has been legally abolished for decades, the study shows that caste remains a reality that pervades the everyday lives of Nepali people and enters into every aspect of their being. To overcome the kind of structural inequalities that the caste system has created is a task that demands systemic change over a long period. Even in Jamune and Dibya Nagar, the two communities where the most progress has been made in the empowerment and inclu- sion of Dalits and Janajatis, there is still a noticeable difference between the socioeconomic levels of different caste and ethnic groups. The study found that the influence of caste and ethnicity can be reduced, however, through development interventions that deliver livelihood empower- ment. Such interventions would improve access to education, enhance income earning, and increase asset accumulation opportunities of excluded groups, as well as support mobilization empowerment through participation in groups. When asset variables8 were held constant, the CEI gap between BCNs and Janajatis was reduced from 10 percent to 2 percent, while the gap Figure 11.7. Influence of Caste and Ethnicity on Composite Empowerment and Inclusion Index (CEI) Levels, by Type of Analysis 0.50 BCN 0.40 Janajatis 0.30 Dalits index CEI0.20 BCN Janajatis 0.10 Dalits 0.00 Bivariate Multivariate Source: MESI study. 214 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION between BCNs and Dalits was reduced from 21 percent to 10 percent. Simi- larly, higher educational levels, higher wealth status, and group member- ship significantly increase CEI levels when other things remain constant. This underlines the significance of agency and the importance of increas- ing people's asset base across both caste and gender. The analysis showed that being a member of one or more groups increased CEI levels by 5 per- cent net. Ten years of education increased CEI levels by nearly 19 percent, while an increase in wealth status by 3 points (on a scale of 1 to 9, where 1 indicates the lowest and 9 the highest level of wealth status) increased CEI levels by 9 percent. This means a Dalit who has three years of schooling (which raises the CEI level by 5.7 percent) and is a member of a group (up 5.9 percent) would have almost the same level of CEI as an uneducated BCN who is not a member of any group. In almost all sectors, development agencies rely heavily on group-based approaches, and the RWSS project, though building on many aspects, is one example of this. There are health groups, water user groups, irrigation groups, forest user groups, children's groups, mother's groups, and thousands of savings and credit groups, to name just a few.Arecent study estimated that there are some 400,000 development-oriented groups or local organizations currently in operation in Nepal (Biggs, Gurung, and Messerschmidt 2004). One-quarter of the MESI study respondents were members of at least one such group. However, the findings suggested that despite the overall empowering and inclusive impact of group membership documented in the quantita- tive analysis, the picture on the ground was a mixed one. BCN women were more likely to be group members than others. Thirty- five percent of them belonged to groups, compared to 24 percent of the middle-caste women, 18 percent of the Janajati women, and 21 percent of the Dalit women. On one hand, this was positive, and the study suggested that the high rate of BCN women's participation in groups may be one of the rea- sons for the apparently marked increase in their empowerment levels in the 25 years since the Status of Women study. Regression results showed that women's membership in groups had a significantly positive influence on their overall empowerment and inclusion index, other variables remain- ing constant. Nevertheless, one must caution that women's groups may go further in empowering women in relation to men and the government bureaucracy than they do in facilitating widespread social inclusion across ethnic, caste, and class barriers. This does not necessarily bring about any significant change in the opportunity structures governing life for Nepalis in general, and instead may only favor particular groups at particular times. The pattern of BCN domination in groups held not only for women but for men as well. The high proportions of BCN and middle-caste respon- dents who were members of groups, along with fieldwork findings on the Dalits' experiences of exclusion--not universally, but consistently--from these groups, revealed what appears to be a general pattern of high-caste NEGOTIATING SOCIAL CHANGE IN RURAL NEPAL 215 "capture" of development interventions. Members of the BCN communi- ties had significantly higher levels of participation in and influence over local development initiatives and services. The study found that members of the BCN group were twice as likely as Dalits to get services and oppor- tunities from agencies and projects in their community. It was not only Dalits who were left out. Janajatis, along with their low levels of human development, also had the lowest levels for two empow- erment indicators: awareness of rights and procedures and participation in development services. Their overall EMI was much closer to that of the Dalits than to that of the middle and upper-caste groups. Janajati women came out at the bottom of the WEI. While Janajati men and women fared well on the general SII, Janajati women had the lowest levels in terms of three of the women's inclusion indictors (contact with the WDO, knowl- edge of women's rights, and control over fertility). This was a significant finding insofar as Janajatis have generally been considered to fall evenly between the relative empowerment and prosperity of BCN communities and the relative disempowerment and poverty of Dalit communities. Still, it would be too simple to dismiss the potential of group-based efforts by saying that all groups are dominated by the higher castes and run solely for their benefit. As was noted earlier, the picture on the ground is mixed. For example, when comparing the women's groups associated with the RWSS project in the four communities where qualitative work was carried out (Dibya Nagar, Jamune, Naukilo, and Bishaltar), it was seen that in two of the communities Dalits and Janajatis were genuinely included in the women's groups. In the two other villages they were not. Furthermore, there is also an indication of a positive relationship between group inclusiveness and project effectiveness. Hence, the two communi- ties with the greatest degree of cooperation and inclusion of Dalits and Janajatis, and where the women's groups themselves have been the most inclusive and effective, are also the communities that have had the great- est degree of success in achieving the desired results of the RWSS project. In contrast, the most polarized communities, where a single powerful group dominates, are those where the RWSS schemes have been the least effec- tive in bringing sustainable drinking water to the community. The study suggests that projects and programs relying on group-based approaches need to pay more attention to organizational and governance issues in the formation and capacity building of user groups and other local organiza- tions if they are to fulfill their potential as mechanisms for social transfor- mation and more inclusive development. Conclusion The ongoing evaluation of the RWSS project described in this chapter shows that empowerment and social inclusion can be effectively measured and 216 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION tracked. Through the use of both indexes and disaggregated analysis of individual indicators, the study has demonstrated the utility, in terms of both project effectiveness and broader distributional equity of development outcomes, of interventions that focus on asset building for empowerment and institutional reform for social inclusion. Methodologically, although ques- tions can be raised about the use of mixed variable types and values in the three indexes constructed by the study team, the indexes are likely to prove their value as a monitoring tool. Perhaps the most valuable from a statisti- cal point of view has been the analysis of individual variables. Such subject matter is unusual in a study of this scale, and the clustering of key variables into like groups is very helpful to analysts. The regression analysis using an index as the dependent variable is, however, highly likely to be subject to criticisms of endogeneity. Again, this is an issue against which other researchers will struggle and one that suggests further collaborative efforts are necessary to establish the limits to, and opportunities provided by, data sets comprising highly interactive variables. Though the study has not yet concluded, it has so far found that the two communities with the greatest degree of cooperation and inclusion of women from traditionally excluded castes, and where women's groups themselves have been the most inclusive and effective, are also the com- munities that have had the greatest success achieving the results of the RWSS project. In contrast, the most polarized communities, where a single powerful group dominated, were those where the project has been least effective in bringing sustainable drinking water to the community. The study has indicatively found that the empowerment status of high- caste woman has progressed over a 25-year period. Nonetheless, it cautions that there is risk of a lag in the progress on asset building and social inclu- sion among middle- and lower-caste women. The overwhelming influence of caste and ethnicity among the most excluded groups has been demon- strated by showing that the marginal influence of gender on these groups' inclusion is very small. Future evaluative work on this project using the indexes and sequenced with qualitative explanatory research will provide further insights into the longitudinal impact of this intervention on geographical communities in Nepal and on the gender- and caste-based social groups within those com- munities. Notes 1. Phase 2 of the MESI study will permit a subsequent round of quantitative measure- ment after the RWSS II (Rural Water Supply and Sanitation Project) intervention is complete. 2. This definition of empowerment is very similar to that used in Empowerment and Poverty Reduction: A Sourcebook (Narayan 2002) NEGOTIATING SOCIAL CHANGE IN RURAL NEPAL 217 3. Some examples include Dollar and Kraay (2000); Kaufmann, Kray, and Zoido-Lobatón (1999); Kaufmann (2000); Murthi, Guio, and Drèze (1995); and Sen (1999a, 1999b). 4. Because one of the objectives of the MESI study was to assess changes in levels of empowerment and social inclusion achieved by different groups in the context of RWSS I and II, the sample villages were chosen on the basis of the operating area of the project. 5. Brahmans and Chhetris are both from the cultural-linguistic group called the Parbatiya, and are considered high-caste Hindus. Both speak Nepali (an Indo-Aryan language rooted in Sanskrit) and follow similar customs. Newars, on the other hand, speak Newari, a Tibeto- Burman language. They are of Mongoloid stock and were the original rulers of the Kathmandu valley before being conquered in 1768 by the invading Parbatiyas. The Brahmans/Chhetris are thus culturally, racially, and linguistically distinct from the Newars, but in Nepal today all three groups are regarded as relatively influential and can be considered part of the ruling class. Because there were very few Newars in the sample and their empowerment and social inclusion profile was quite close to that of the Brahmans and Chhetris, they were clustered with those two groups. The full report includes a list of caste and ethnic groups covered by the sample survey. 6. Use of indexes is discussed in chapter 3, in the preface to part 2, and in other country case studies in this volume. See in particular the discussion in the Honduras country study. 7. This may be partly because many of the indicators of female disempowerment and exclusion are contained in the WEI and would not show up in the CEI. The two indexes can- not be combined mathematically because they relate to different populations (WEI relates just to women and CEI to both women and men). That said, these additional barriers faced by women should be kept in mind as interventions are prioritized. 8. Income is often used as a proxy for financial assets. Care needs to be taken, however, because theoretically income is a flow, whereas assets are a stock. Appendix 1 Selected Examples of Approaches to Measuring Empowerment 220 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Definition Measurement Study, location, and scope of empowerment concepts and methods Albertyn (2001) Effective empowerment Concept refined to must occur at each of 61 statements (33 on the South Africa 3 levels: micro (attitude, micro, 15 on the interface, feelings, and skills), inter- and 13 on the macro Programmatic face (participation and level) after pre-testing, action immediately implementation, and around the individual), validity testing and macro (beliefs, action and effects) Bartle (2003) Having the capacity to do Discussion of the degree things that community to which the community Generic members want to do and has changed with respect going beyond political or to 16 elements: altruism, Community level legal permission to par- common values, commu- ticipate in the national nal services, communi- political system cations within and outside the community, confidence, political and administrative context, information, intervention, leadership, networking, organization, political power, skills, trust, unity, and wealth Brown (2005) Providing empowerment Q-sample (purposive opportunities as neces- selection of statements sary prerequisites to on what it means to be Generic altering a person's empowered), Q-sorting potential reality and (having participants Group level giving people the means rank-order them for 4 to better themselves or disagree to 4 or agree), and Q-factor analysis to show how opinions are patterned into overall perspectives Goetz and Gupta (1996) Control over resources Assessing women's vs. men's managerial control Bangladesh of loans, including pro- ductive process, market National level and inputs APPENDIX 1 221 Data sources Focus Outputs and conclusions Questionnaire Developing a standard- Instrument can be used implememented in an ized instrument to mea- for needs analysis (base- experimental design in sure the outcomes of line assessment), long 8 Free to Grow Groups empowerment and trends and effects, proof (company presenting increase accountability of impact, comparison self-development pro- in adult education of methods, and action grams in organizations), programs research. combined with qualita- tive data collection Local and personal interpretations of the definitions can be vali- dated with an objective measurement Interactive: Facilitator Capacity development Workshop presenting a calling for the observa- participatory method- tions of all community ology to measure com- members in an annual munity empowerment. evaluation meeting to reach consensus on the Participatory methods relative strength of and eliciting and using local changes in each item knowledge to measure progress toward people's own objectives Participatory method Poverty reduction Q-methodology provides a measure of how strat- egies that emphasize the material world outside the individual relate to poor people's realities. Interviews with 253 Testing gendered Micro credit programs women and 22 men assumptions of micro are not necessarily loanees in 5 regions credit extension empowering women as men often control loans given to women. 222 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Definition Measurement Study, location and scope of empowerment concepts and methods Goetz and Gupta (cont.) Grootaert (2005) Expanding assets and Average empowerment capabilities of poor peo- score based on 12 priority Albania, Armenia, ple to participate in, indicators along five ele- Azerbaijan, Georgia, negotiate with, influence, ments: state reform (gov- Kyrgyz Republic, control, and hold ernment effectiveness, Macedonia, Moldova, accountable institutions corruption perceptions Tajikistan, and Uzbekistan that affect their lives index, incidence of illicit payments); reform of National level legal system (rule of law, regulatory quality, pro- poor decentralization); democracy (civil liberties and political freedoms, voice and accountability, civil society strength); removal of social barriers (share of women in political offices, income inequality); and building social capital Kabeer (1998) Expansion of choice Perceived changes in where previously none women's self worth, Bangladesh existed agency, contribution to the household, and National level confidence in community interactions Kroeker (1995) Reversing the process of Seven months of partici- alienation and disbelief pant observation and Nicaragua in change and increasing four follow-up visits access to resources and Cooperative level control over the condi- tions and decisions that affect one's personal life and environment APPENDIX 1 223 Data sources Focus Outputs and conclusions Qualitative case-based study builds indicative findings International databases Non-income dimensions Many indicators are (governance database, of poverty available from inter- Transparency Inter- national databases national, WBES/BEEPS, although there is a lack Freedom House, UNDP, of indicators of decen- and World Development) tralization and of social capital. Overview of accessible indicators available at different points in time. Survey of 696 women Testing assumptions Micro credit has loanees and in-depth about gendered impact decreased the trade-offs interviews with 50 of micro credit extension that women have to women and 20 men on women involved in make between dimen- market activities funded sions of their well- by micro credit being. Mixed method research methodology using interprepretive frame- work generates working hypothesis. An agricultural coopera- Factors enhancing and Interrelated levels of tive in Nicaragua in 1989, impeding individual, empowerment are neces- literature review, inter- organizational, and sary: personal, organiza- views of key informants, societal empowerment tional, and societal. and visits to 15 coopera- tives around the country 224 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Definition Measurement Study, location, and scope of empowerment concepts and methods Kvinnoforum No set definition of Project tested a number (2001a; 2001b) empowment but focus on of qualitative and quan- two basic principles: titative methods: Circle Namibia, Botswana, (1) disempowerment as of Important Areas, Zimbabwe, and starting point for empow- House of Life, ratings Mozambique erment processes, and questionnaire, individual (2) empowerment cannot interviews, focus group Project and individual be given to someone by discussion, participatory levels somebody else; it has to observations, case studies start from within and be owned by the person herself Lokshin and Ravallion Taking actions that Respondents place them- (2005) selectively empower selves on Cantil ladder those with little power to (nine steps) for power; redress power inequality the interpretation of the Russia meaning of the lower and higher steps is left to Individual and participants household level McMillan et al. (1995) Gaining influence over Psychological empower- events and outcomes of ment measures using self USA importance. Empower- assessment scales for five ment seen mostly as: components (perceived Individual and outcome (influence, knowledge and skill organizational level achievements); participa- development, perceived tion; perception of com- participatory competence, munity; and inclusion. expectancies for future individual contributions, perceived group or organi- zation accomplishments, and expectancies for future group or organiza- tional accomplishments) Organizational empow- erment derived from two items (key informant telephone survey) rated on a four-point scale con- cerning the impact of the task force on their organi- zation policies or proce- dures and resources APPENDIX 1 225 Data sources Focus Outputs and conclusions Participatory methods Women's empowerment Methods that empower that ask women to reflect are important. on their own situation Concepts mean different things to different women. Indicators differ at indi- vidual, group, and societal levels, even for the same concept. Data on Russian adults Comparability of There is a significant but from the Russian expressed perceptions of seemingly weak associa- Longitudinal Monitoring power and economic tion between power and Survey (1998 and 2000) welfare welfare. The scope of empower- ment is not limited to the poor. Data from members of Predictors of empower- Measures predict: 35 community coalitions ment of individuals and (a) individual psycho- organized for the organizations in the logical empowerment prevention of alcohol context of community (participation levels, and other drug problems; coalitions sense of community, key informant telephone perceptions of a positive survey organizational climate), collective empowering of members (net benefits of participation, com- mitment and positive organization climate); and (b) organizational empowerment (psycho- logical empowerment and positive organiza- tion climate) There is an ongoing interactive process, link- ing the individual to the collective, that is both multilevel and context specific. 226 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Definition Measurement Study, location, and scope of empowerment concepts and methods Malena and Heinrich Enabling or giving power Aggregated empower- (2005) to (whom) to do (what). ment score based on the Empowerment in relation average of five indicators Pilot implementation in to the impact of activities scored from 0 to 3: 13 countries (South Africa, of civil society. Indicators activity and effectiveness Mexico, Uruguay, Estonia, related to aspects of of civil society in inform- Indonesia, Pakistan, inclusion and participation ing and educating, build- Belarus, Croatia, Romania, (structure), opportunity ing capacity for collective Ukraine, Canada, structure (environment action, empowering New Zealand, and Wales) and values), and influence poor people, empower- (impact). ing women, and building National level social capital Malhotra, Schuler, Enhancing assets and List of indicators cover- and Boender (2002) capabilities of diverse ing: different dimensions individuals and groups (economic, sociocultural, Review of 45 studies to engage, influence, and familial, legal, political, hold accountable the and psychological); and Different levels institutions that affect different levels of aggre- them gation (household, community, broader arenas), in which empow- erment was considered as an outcome or an intermediary process. Mason and Smith (2003) Extent to which some Six-item scale of women's categories of people are say in household eco- 56 communities in able to control their own nomic decisions; Pakistan, India, Malaysia, destinies even when three-point scale of Thailand, and the their interests are women's participation Philippines opposed by others with in family planning whom they interact decisions; Individual and group five-item scale of level women's freedom of movement; two yes/no items about women's exposure to coercive controls by the husband; and APPENDIX 1 227 Data sources Focus Outputs and conclusions Media review, focus State of civil society in Description of the groups, community case countries around CIVICUS Civil Society surveys, secondary data, the world Index with a scoring and fact finding matrix and 69 indicators encompassing four dimensions (structure, external environment, values, and impact of activities) A comprehensive approach to measure- ment (social forces, map of civil society; power relations in society, and relations within organizations) Review of 45 studies Women's empowerment Empowerment can occur using quantitative and and gender relations in one or more areas of qualitative data in international life, at various levels, development and be individual or collective. Most studies focus on the household level and are usually weak with respect to intervening processes; only few studies measure data over time. Section of a larger house- Women's empowerment Community character- hold questionnaire in the domestic sphere istics provide a far and fertility stronger predictor of women's empowerment than individual traits. Empowerment is multi- dimensional, and direct measures are preferable to proxies. Empowerment as a group-based process is influenced by culture at the group level. 228 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Definition Measurement Study, location, and scope of empowerment concepts and methods Mason and Smith (cont.) community-level gender attitude measures (means calculated and attached to the records for indi- vidual women) Moser (2005) Expanding assets and Composite indicators at capabilities of poor the individual level (self- Colombia people to participate in, esteem, importance of negotiate with, influence, gender identity, attitude Community level control, and hold toward peace, participa- accountable able institu- tion in meetings, time for tions that affect their conflict and peace related lives activities); organizational level (internal cohesion); and inter-institutional level (contact, coordination) Spreitzer (1995) Intrapersonal empower- Self-assessment using a ment as the component seven-point "Likert USA of psychological empow- response format" for four erment that deals with dimensions (sense of Individual level cognitive elements. Other meaning--beliefs and components are inter- attitudes, competence, actional (thinking about self-determination, and and relating to the envi- impact or efficacy) ronment) and behavioral averaged into a single (taking action and measure of intrapersonal engaging issues). empowerment UNDP (1995) Unspecified Gender Empowerment Human Development Measure (GEM) index Report based on women's parlia- mentary representation, Generic economic participation and decision making National level (combining administra- tive and managerial posi- tions and professional and technical positions), and the gender disparity in earned incomes, reflecting economic independence APPENDIX 1 229 Data sources Focus Outputs and conclusions Quantitative descriptive Empowerment of local Empirical issues concern- information on inputs communities through ing community percep- and outputs from written participation in ongoing tions of empowerment documentation and inter- peace processes include difficulties with view sources; qualitative conceptualizing indica- participatory evaluation tors; their context speci- workshop from two ficity; and delayed projects impacts. Survey data from a Interpersonal empower- Intrapersonal empower- sample of 324 middle ment in the workplace ment mediates the rela- managers from different tionship between the units of a Fortune 50 social context (support, organization information, resources, and unit culture) and behavioral outcomes. Data about percentage of Comparing abilities of Provides a measure of women in parliament, women and men to par- whether women can take among legislators, senior ticipate actively in eco- an active part in eco- officials, and managers, nomic and political life nomic and political and among professional domains. and technical workers, as well as male and female proportional income share Appendix 2 Understanding the Concept of Power Adapted from Csaszar (2004), Moncrieffe (2004), and Holtom (2004) The concept of power is complex and heavily theorized, but provides impor- tant conceptual entry points for the themes pursued in this book as work on empowerment is intrinsically linked to power and power relations. This appendix explores the key points of the debate on defining power. The first part of this appendix examines theories of power with a focus on the extent to which the theories see power as exercised through "consensual" or "con- flictual" mechanisms. This appendix takes as its departure the consensual work on power by Parsons (1987) and contrasts this with the conflictual perspectives of Dahl (1961), Bachrach and Baratz (1962), and Lukes (1974). The discussion then turns to the works of "middle ground" theorists--includ- ing Foucault (1980), Giddens (1984), and Clegg (1989)--who conceptualize power as having elements of both consensus and conflict. Table A.2.1 at the end of this appendix provides an overview of the theories summarized, while box A.2.1 presents a typology of power, widely referenced in the empower- ment literature that can be usefully applied to approaches to empowerment. Conceptualizing Power The consensual and conflictual theories of power are examined below, as are the theories of those who take the middle ground. Consensual Theories of Power Consensual power-theorists believe that power is not necessarily linked with conflict, and that it does not have to be a zero-sum game. Instead, they argue that power is the capacity to achieve outcomes, whether these are achieved by force or for the interests of certain sectors of society. Talcott Parson is an exam- ple of a consensual theorist, who believes that power is mostly consensual. Parsons: Power is Created and Legitimated by Society Parsons' (1937; 1963) contribution to the conceptualization of power is an important one because it draws attention to the generative aspects of social 230 APPENDIX 2 231 power. Parsons argues that power does not just exist--it has to be created. Power is produced or created by society, and can be expanded, so a gain in power by some is not necessarily at the expense of others. Parsons explains the production of power in terms of an analogy between the economy and the polity. Money only has value because we believe that it does. Similarly, the power of those in authority is based on self-perpetu- ating beliefs in legitimacy. Parsons does not deny that some power rela- tions are based on coercion or the threat of violence, but he perceives coercion as a poor substitute for consensual legitimate power. Parsons argues that complex political systems rely on the legitimation of power, as power is separated from its base of coercion. Legitimate power increases through effectiveness, and contracts through the misuse of authority. For example, political regimes that coerce their populations generally do so because they lack the type of authoritative, legitimate power as conceptualized by Parsons. Conflictual Theories of Power Parsons' contribution is significant because of his insight that power is not inherently contradictory to peoples' interests. Power according to Parsons is a constitutive force that is generated by consent. This perspective legitimizes power and therefore fails to ask important questions about the winners and losers in any given social order. Consensus is simply assumed. In contrast, power has been portrayed in far less legitimate and more conflictual terms as the means by which one group protects its interests against another group. Power in this model is maintained either violently through coercion or more subtly through an engineered consensus or "false consciousness." In contrast to Parsons' generative conceptualization, this model also assumes that power is a zero-sum game. Within the school of conflictual power, three main theorists' works dom- inate the field. Each of these theorists' ideas build on each other, creating what is called the "three-dimensional power debate," in which the theo- rists Dahl (1961), Bachrach and Baratz (1962), and Lukes (1974) explore dif- ferent "dimensions" of power.1 This three-dimensional power debate, outlined below, provides the foundation for the main theories of power used in the political and social sciences. Dahl: Power as Decision Making Within Dahl's (1961) framework, Ahas power over B to the extent to which Acan get B to do something that B would not have done otherwise. Within this paradigm, power refers to the act of prevailing in decision making and is not to be equated with power resources, which are only potential power. Dahl notes that resources may or may not be mobilized in decision making. For example, one wealthy person may choose to collect paintings, while another 232 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Box A.2.1. A Typology of Power Power over: This type of power can be characterized as controlling power, which may be responded to with compliance, resistance, or manipulation. This is the type of power most often debated in the political science and sociological literature. The three dimensions of power all seek to charac- terize this type of controlling power to find the root cause of the methods of control and the relationship between knowledge and power. "Power over" has many negative associations for people, including repression, force, coercion, discrimination, corruption, and abuse. In this view, power is seen as a zero-sum game: having power involves taking it from someone else and then using it to dominate others and prevent them from gaining it. In politics, those who control resources and decision mak- ing have "power over" those without such control. In terms of this type of "power over," empowerment is concerned with bringing people who are outside of the decision making process into it. Power with and power to: "Power with" is collaborative power, or having a sense of the whole being greater than the sum of the individuals, especially when a group tackles problems together. This type of power involves finding common ground among different interests and building collective strength. This type of power, along with the next two types, falls under the "positive power" category established by Foucault (1980), whereby power is seen as a positive, productive force. Because it is based on mutual support, solidarity, and collaboration, "power with" multiplies individual talents and knowledge. This type of power can help bridge dif- ferent interests, transform or reduce social conflicts, and promote more equitable relations. "Power to" refers to the unique potential of every person to shape his or her life and world. "Power to" is generative or productive power that creates new possibilities and actions without domination. It is expressed by people's ability to recognize their interests, and to realize that they have the power to shape their circumstances to achieve a situation that is more favorable to their interests. Both "power to" and "power with" involve the definition of actors' "true interests," in the conception that Lukes (1974) provides of power defined in the terms of interests.a In the realms of both "power to" and "power with," empowerment is concerned with the processes by which people become aware of their own interests. This takes place in the "relational sphere," as people devel- op the ability to negotiate and influence the nature of a relationship and the decisions made within it. Also, this type of empowerment can take place in the "collective sphere," with the development of cooperation between individuals and collective action, so that they can achieve more impact than each could have alone. Power from within: This type of power has to do with a person's sense of self-worth and self-knowledge. It is the capacity to imagine and to have hope, to believe that one is strong enough and has the right to APPENDIX 2 233 Box A.2.1. (continued) change one's circumstances. This is the least-mentioned type of power in power literature, perhaps because it is quite difficult to identify and mea- sure. Nevertheless, this type of personal power or self-actualization is con- sidered an important component of the exercise of power. "Power from within" can be characterized as the spiritual strength and uniqueness that resides in each of us and makes us truly human. Its basis is self-accep- tance and self-respect, which extend to respect for and acceptance of oth- ers as equals. In terms of this typology of power, empowerment requires building agency through psychological assets of individuals and groups, providing people with the "capacity to aspire" (Appadurai 2004) and to imagine alternative choices. Many grassroots organizations focus on this "power from within" to help people affirm their personal worth and to recognize their "power to" and "power with." Source: Adapted from Rowlands (1997) and Hughes et al. (2003). a. However, gains in power here are elastic (that is, this is not a zero-sum model) and can be achieved either through the c onsent of others or through independent action (Nelson and Wright 1995). collects politicians--both may have equal resources but only the latter is powerful, because of the political realm in which his resources are used. Thus, while resources are important in determining power, they determine potential power, not the degree of power itself. In his community power study of New Haven, Dahl analyzed who ini- tiated and who vetoed decisions in key issue areas (political nominations, public education, and urban renewal). He found that while there was an unequal distribution of resources in the community, there was no single elite that exercised power in all three of the main issue areas. From this he concluded that there was not one elite in New Haven but a plurality of elites, because there was a variety of competing power structures, none of which was dominant. Thus, Dahl argued that modern democracies deliver democratic outcomes through competition between elites, a form of gov- ernment he termed "polyarchy." Dahl's model of power was criticized by many, including Bachrach and Baratz, and Lukes. Both parties argued that his concept of power was too narrow, and the following sections examine the theories of power that grew out of their elaborations on Dahl's definition of power. 234 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Bachrach and Baratz: Power as Agenda-Setting Bachrach and Baratz (1962) add to Dahl's theory on power by emphasizing that not only does A exercise power over B in overt decision making, but A may also exercise power over B by limiting the scope of the political process to issues that A wants. Thus, Bachrach and Baratz add to Dahl's dimension of power as decision making by introducing a second "dimension" of power, a dimension that takes into account the manner in which decisions are made and can be influenced. The best example of this is the process of agenda-setting, where an issue of importance to B is deliberately left off the agenda by A. Thus, power is exercised not only by decision making, but also by preventing issues from reaching the agenda. This form of power is called non­deci- sion making, that is, the decision not to make a decision. This second dimen- sion of power can be exemplified by action such as excluding items from the agenda, creating selective precedents, defining matters as a private affair, excluding others by endless red tape, creating committees that never reach decisions, or "losing" files. Lukes: Power in Terms of Interests Lukes (1974) introduces a third dimension of power by examining power in terms of interests: A exercises power over B when A affects B in a manner contrary to B's interests. Lukes criticizes the behavioral focus of both Dahl and Bachrach and Baratz's power theories. Instead of emphasizing exclu- sively individual agency in decision making, Lukes takes a more radical view of power as being formed by the dialectical relationship between structure and agency.According to Lukes, biases are not necessarily reducible to indi- viduals' actions or deliberate non-actions, but rather they are inherited from the past in the form of structured or culturally patterned behavior of groups. The second aspect of Lukes' conceptualization of power is his focus on the relationship between power and knowledge. Here he emphasizes "false con- sciousness" or ideology, where the less powerful are not aware of their "real interests." Clearly influenced by Gramsci's (1971) notion of "hegemony," the main underlying premise of the third dimension of power is that power "dis- torts" knowledge or "the truth" in a direction that is beneficial to the specific interests of the dominant group. It also suggests that consensus and social order are bound up in a conspiracy of the powerful in which the "truth" is hidden from the powerless. While this concept of false consciousness is prob- lematic, as we discuss further below, Lukes significantly highlights the impor- tance of the relationship between the construction of knowledge and power. "Middle-Ground" Theories of Power A conflictual conceptualization of power is more promising analytically than Parsons' consensus-based approach but also has its limitations by APPENDIX 2 235 restricting power to a relational concept based on domination ("power over"): the intentional actions of one group of actors that achieve power over and gain consent from another group of actors in a zero-sum rela- tionship. In the "middle ground" between the conflictual and the consen- sual schools are theorists such as Foucault (1980), Giddens (1984; 1993), and Clegg (1989), who argue that power is constituted by both conflict and consensus. These theorists' frameworks are the most nuanced and the most complex of the main power theories. Foucault: Power is Constituted by Free Subjects The perspective of conflictual theory is that consensus and social order are bound up in a conspiracy of the powerful in which the "truth" is hidden from the powerless. An alternative, Foucauldian view of this consensus, rejects the notion that the "truth" is being distorted by hidden powers of the hegemonic state or dominant classes, through a single overriding "dis- course." Instead, power as an "authorless" but positive force enables people to construct their own subjective, and competing, discourses, and learn the habits of freedom. Foucault concludes that "there can be no freedom with- out power." Foucault's (1980) conceptualization of power evolves around the idea that we must examine the relationship between power and consciousness. He argues that power has no essence, that is, power is not situated in any particular place. Thus, Foucault asserts that power is not reducible to insti- tutions such as government bodies. Instead, power is always relational (between people, between different departments within a ministry, and so on) and exists only when it is exercised. Foucault also believes that power is "constituted" or created in a network of relationships among subjects who are free to act. The modern perception of the relationship between power and knowledge is a negative one where power distorts the truth--the third dimension of power. Foucault calls this phenomenon "negative power" and distinguishes it from "positive power," which is the power to say yes and to produce new realities. Whereas Lukes assumed that there exists a knowledge that is free from power and thus is objective, Foucault believes that all knowledge is created by society and thus it cannot be objective. For Foucault, power is not coercion or violence: violence takes place when the limits of power are reached. Foucault argues that conflict and war pre- suppose an absence of a shared truth and defines power as the "setting up of shared truths in order to avoid war." He characterizes power as a form of pacification that works by codifying and taming war through the impo- sition of socially constructed knowledge as truth. While Foucault does not articulate a comprehensive, coherent theory on power, his views of the social construction of power are an important addi- tion to the power debate. Foucault's writings highlight the idea that the 236 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION relationship between power and knowledge is not oppositional--it is mutu- ally constitutive. Giddens: Agency and Power The concept of "negative power" as critiqued by Foucault critically ignores the "agency" of the individual; that is, the capacity of social agents them- selves to make independent choices that generate power.Afocus on agency provides a more "optimistic" outlook on the capacity of humans indepen- dently to bring about social change (as discussed in box A.2.1 with respect to "power to"). But social theorists are also able to retain the important rela- tional aspects of power ("power over") as being bound up in structures; that is, the sets of rules that interact with agency to determine human behav- ior and interaction. Giddens (1993) has perhaps produced the most inter- esting and elegant thinking on the iterative relationship between structure and agency. Structures,2 he argues, allow people to live and make reflec- tive and knowledgeable choices but also tend to limit choice by habitual- izing actors to reproduce particular ways of viewing the world. Importantly, Giddens' theoretical framework allows for the possibility of new forms of social action to be created and this gives his thesis particular value in the context of this volume. Giddens' analysis of power is based on his theory of "structuration" (Giddens 1984). In brief; structuration was developed in response to the dualism that exists between subject-centered and object-centered social theories. (Subject-centered theories place emphasis on individuals as cre- ators of society, while objectivist theories focus on society itself and view agents as the effects of social order). Structuration attempts to bridge this divide by proposing that social structures exist in the moment that they are reproduced by agents, and agents define themselves as agents by repro- ducing social structures. The simultaneous moment of the reproduction of agency and structure is what is known as "structuration." Giddens argues that social structures give people a capacity for action as agents. Without the existence of social structures created by society, agents would not have the ability to act. According to Giddens, agency is only possible because of resources that exist as a result of the meaning they are given by society. For example, the resources of a wealthy person or of a political leader exist only because of the meaning of money and authority. Giddens' theory of power argues that power is intrinsic to human agency. People are never completely governed by social forces--they have agency--but neither are they autonomous beings. Even when they display outward compliance, people make ratio- nal assessments of the situation and the viable alternatives; thus, compli- ance is a choice and does not automatically entail agreement. Giddens argues that the analysis of power entails uncovering the subtle mix of what APPENDIX 2 237 actors do (and refrain from doing), what they achieve (and fail to achieve), and what they might have done (but did not do). Giddens believes that power is enabling as well as constraining, and that power is exercised as a process. Power cannot be attributed to resources; rather, it is constituted through processes of negotiation between individuals in society. Clegg: Frameworks or "Circuits" of Power Clegg (1989) frames his analysis of power in a theoretical perspective that is a synthesis of several contemporary debates on power. He characterizes power in terms of the paradigm of Lukes and Giddens, and combines this with the "creation of meaning" analysis of Foucault. Clegg's framework divides power into different "circuits." Dahl's model of Aexercising power over B is the first circuit of power--episodic power, that is, the power at the agency level where agents are autonomous. This circuit is a reflection of a deeper, second circuit--dispositional power, which is where meanings are created, recreated, and contested. Because meanings are tied to rules, dispositional power is reflected in the "rules of the game" that constitute real- ity. While actors may resist meanings as single agents, the meanings them- selves are a reflection of a deeper systemic form. This deep, facilitative power is the third circuit of power, constituting the general systemic set of relations. This systemic circuit of power is comprised of systems of rewards and punishments, and it defines power and powerlessness at the macro level. According to Clegg, power is essentially paradoxical. The power of an agent is increased by the agent delegating authority; the delegation of authority can only proceed by rules; rules entail discretion; and discretion potentially empowers delegates. Thus, there is a form of hidden power within the rules, which can alter the very opportunity structure that they constitute. This give-and-take in power relations produces the basis of "organizationally negotiated order." Conclusion While there is no coherent, universally accepted definition of power, this appendix has outlined the three main theoretical camps of the power debate: the consensual, the conflictual, and the "middle ground," while empha- sizing the connections between different theorists. One theme running through this theoretical review is the relationship between structure and agency and the recognition that this processual relationship provides the basis for social change and empowerment. It is the link between power and empowerment that is significant for this volume, which bridges the gap between the theoretical notions of power and the practical ways in which power relations can be altered. 238 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Table A.2.1. Summary of the Social Theoretical Models of Power Relationship between Theorist Central thesis Structure and agency A and B Parsons Power in society de- Agency partly constitu- Socialization generates (1937; 1963) pends upon the ability ted and constrained by B's recognition of A's of society to generate structure. authority, and hence and sustain belief in the power. legitimacy of authority. Dahl (1961) One dimension of Agency, with structure A has power over B to power: power refers to implicit in the political the extent that A can the act of prevailing in realm of competition. get B to do something decision making. which B would not have done otherwise. Bachrach and Two dimensions of Agency, with structure A exercises power over Baratz (1962) power: power as deci- implicit in the political B in overt decision sion making combined realm of competition making and in limiting with agenda setting scope of debate. ("non-decision making"). Lukes (1974) Three dimensions of Reaction against struc- A's ability to shape B's power: open conflict; tural determinism; subjective "interests" covert conflict attempts a dialectical secures B's consent to (non-decision making); relationship between their domination by A. and suppressed con- structure and agency. flict (false conscious- ness). "Real interests" help reveal false consciousness. Gramsci Dominant class(es) are Reaction against A's ability to prevent B (1971) able to present their (a) Marxist `econom- developing "good "interests" as if they ism'; and (b) Croceian sense" secures B's con- were universal, securing "idealism"; dominant sent to their domina- the consent of subordi- groups are able to tion by A. nate class(es): "Hege- manipulate "structure" mony" derived from to serve their own "false consciousness." "interests." Foucault Power is constitutive of Agency is both enabled Discourse's privilege (1980) subjectivity. If agents by, unintentionally certain constructions must "learn" the habits reproduces, and is con- of the world, giving of freedom, there can stituted by structures. `authority' to some, be no freedom without establishing relation- power. ships between A and B. Giddens Social action depends Agency is at the heart, Access to and ability (1984, 1993) upon actors who use but agency is both to use structure both both "rules" and enabled by and unin- enable and constrain "resources." By using tentionally reproduces A's and B's agency. rules and resources structure ("dualism"). (structures) agents unin- tentionally reproduce them (structuration) APPENDIX 2 239 Interests Normative position Critique Socialization produces Socialization of social norms, Socialization appears to "moral agents." sustains society. negate agency. Partly innate, partly derived Modern democracies oper- Narrow concept of power. from knowledge. ate through elite competition. Partly innate, partly derived Modern democracies oper- Narrow concept of power. from knowledge. ate through elite competition. "False interests" shaped in Emancipation of human's Prohibitive or coercive; if society, "real interests" are minds through uncoerced actors cannot know their innate. communication (cf. own minds, on what basis Habermas' ideal Speech can observers? Situation). "Worldviews" shaped in Emancipation of humans' Prohibitive or coercive; if civil society. "Real" interests minds through the "philoso- actors cannot know their are innate. phy of praxis" (Marxism). own minds, on what basis can observers? Actors wholly "subjectiv- Actors should have the free- Tends to marginalize ized," therefore no place for dom to change "regimes of knowledgeable intentional "real interests." truth." agency. Partly innate, partly derived Actors should be free to use "Instantisation" means from knowledge. structures. agency allowed to predom- inate over structure. 240 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Table A.2.1. Summary of the Social Theoretical Models of Power (continued) Relationship between Theorist Central thesis Structure and agency A and B Clegg (1989) "Circuits of power" Autonomous agency A has power over B. determine the creation occurs at the episodic B can resist in episodic and contestation of level, with contestation moments or through meaning. of meaning necessary the operation of deeper at deeper circuits. structural rules. Source: Adapted from Holtom (2004). APPENDIX 2 241 Interests Normative position Critique Partly innate, partly derived Actors should be free to Tendency to reduce agency from knowledge. contest meaning and change to episodic power circuit. rules. Notes 1. Much of the analysis of power theorists in this paper is based on Mark Haugaard's work (Haugaard 2002). 2. Structures according to Giddens are composed of both rules and resources. Rules comprise stocks of shared knowledge that enable the use of language as a medium of communication (1993, 103­106), while resources--both material (allocative) and social (authoritative)--are used according to the rules established (1993, 185). Appendix 3 Measuring and Monitoring Empowerment in World Bank Projects At the time of writing, about 1,800 World Bank lending operations include the term empowerment. Eighty-four of these specify empowerment as a project development objective. The following selection of projects repre- sents a small sample of these. 243 244 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Project Project objectives Empowerment objectives Tamil Nadu To improve livelihoods Improving the livelihoods Empowerment and and empowerment of the of the marginalized and Poverty Reduction poor, ultra poor, and facilitating their partici- (India) vulnerable. pation and inclusion in the planned economic PAD (#31806) development and growth (2005/06/02) strategies of the State. Liberia Trust Fund: To assist war-affected An enabling environment Community communities to restore for social cohesion and Empowerment Project infrastructure and ser- the socioeconomic vices and build their revival ... for improved PAD (#31337) capacity for collective economic and social (2005/02/24) action. governance. Bolsa Familia Project To reduce poverty and 1) To foster bridges (Brazil) inequality today, through between Bolsa Familia the provision of direct beneficiaries and com- PAD (#28544) monetary aid to poor plementary services so (2004/05/25) families; helping them as to enable an escape invest and grow out of from poverty. poverty in the future. 2) To promote the devel- opment of tailored empowerment strategies. APPENDIX 3 245 Empowerment concepts and indicators used Instruments and analytics Comment 1) Rate of increase in Aggregated and disaggre- Indicators 1 and 2 incremental income gated data on geographic measure financial assets. against base year. coverage, coverage of Indicator 3 measures 2) Proportion of target vulnerable groups, pro- organizational assets. households that have ject components, human Indicator 4 measures the increased their incomes. resources, capacity build- first degree of empow- 3) Proportion of the ing, services provided, erment (existence of identified vulnerable and number of groups choice). population organized and committees formed. into self-help groups and There is a meaningful accessing special assis- link between empower- tance funds. ment objectives and 4) Proportion of poor and indicators used. ultra poor women occu- pying decision making positions. 1) Measure of level of Beneficiary assessments. Indicators 1 and 2 social capital and organi- Technical and manage- measure opportunity zational development. ment audits. structure. Indicator 3 2) Measure of level of Community mobilization measures the third access to and use of social and environmental degree of empowerment and economic infrastruc- studies. (the achievement of ture and services. choice). 3) Proportion of sub- projects that reflect com- There is a meaningful munity priorities. link between empower- ment objectives and indicators although indi- cator 1 appears difficult to measure. 1) Proportion of Bolsa Baseline survey devel- Indicators 1 and 2 Familia recipients receiv- oped and piloted in at measure financial assets. ing income transfers. least 10 municipalities. Indicators 3 and 4 2) Proportion of total System for verifying measure the second transfers going to fami- eligibility for cash trans- degree of empowerment lies in the bottom quintile. fers developed and (use of choice). 3) Proportion of primary- piloted in at least 5 age children in extremely municipalities. There is a meaningful poor beneficiary families link between empower- attending school. ment objectives and 4) Proportion of benefici- indicators used. ary children with health cards. 246 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Project Project objectives Empowerment objectives Second Indigenous and To promote the empow- Strengthening the Afro-Ecuadorian Peoples erment and improved necessary aptitudes, Development access to natural and skills, and abilities that financial resources for enable full participation PAD (#28968) indigenous and Afro- and contribution in the (2004/05/21) Ecuadorian communities improvement of their and organizations within own communities, and the framework of their on a regional and own development and national level under interculturalism. conditions of equality. Community Develop- To improve livelihoods 1) Improving access of ment and Livelihood and quality of life for the village communities to Improvement "Gemi rural poor. basic economic and social Diriya" Project infrastructure and (Sri Lanka) opportunities. 2) Empowering commu- PAD (#27642) nities with respect to the (2004/03/05) selection, preparation, and direct use of finan- cial resources and implementation. APPENDIX 3 247 Empowerment concepts and indicators used Instruments/ and analytics Comment 1) The Institutional Project's progress tracked Indicator 1 measures Capacity Index of the through Management organizational assets. facilitating entities (FEs). Information System. Indicators 2 through 5 2) Proportion of commu- Baseline study and measure oppor- nity's share of co- annual beneficiary tunity structure. financing for the sub- assessments. Indicator 6 measures the projects. Impact evaluation second degree of 3) Proportion of FEs studies. empowerment (use of receiving additional Standard auditing and choice). funding from other in- supervision missions. stitutions and total in- There is a meaningful vestment received by link between empower- the FEs. ment objectives and 4) Level of investment in indicators used. local governments and non-governmental organizations (NGOs). 5) Government adopts National Development Policy for Nationalities and Peoples. 6) Proportion subprojects financed and executed by women's groups and proportion with gender equality focus. 1) Rate of incremental Key informants and focus Indicators 1 and 2 income increase. group discussions con- measure financial assets. 2) Proportion of house- ducted by subproject Indicators 3 and 4 holds with enhanced appraisal committee. measure the first degree income. Direct observation. of empowerment 3) Proportion of house- Documentation review. (existence of choice). holds accessing commu- Annual assessment. Indicator 5 measures nity infrastructure. opportunity structure in 4) Proportion of women the project context. and youth in decision- making positions Indicators 3 through 5 (for example, Chair- provide the most mean- person or Treasurer of ingful measurement of various subcommittees) empowerment objectives at village level. although these objec- 5) Proportion of Village tives are very broad. Development Plans with with at least one sub- project activity completed that benefits at least 80% poorest community members. 248 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Project Project objectives Empowerment objectives Third East Timor To reduce poverty, Inclusive patterns of Community strengthen social capital, growth and development; Empowerment and and improve local media projects produced by Local Governance and communications. communities for Project communities. PAD(#24119) (2002/10/24) Ecuador Poverty To strengthen local Supporting human capi- Reduction and Local empowerment, improve tal development and Rural Development quality of local services, broadening opportunities (PROLOCAL) Project and increase access to for excluded groups. productive assets to PAD(#22286) to improve the (2001/06/08) well-being of poor households in selected micro-regions. Madhya Pradesh To improve opportuni- Building capacity of the District Poverty ties for the poor and poor to voice their Initiatives (India) vulnerable, especially demands and to pressure women, to meet their local governments PAD (#21020) own social and economic and antipoverty pro- (2000/10/05) development objectives. grams to be more responsive to them. APPENDIX 3 249 Empowerment concepts and indicators used Instruments and analytics Comment 1) Measure of level of Periodic studies and Indicators 1 and 4 mea- access to markets, ser- beneficiary assessments. sure the first degree of vices, credit, and water Community Develop- empowerment (exis- supply. ment Working Group tence of choice). 2) Measure of council Reports. Indicators 2 and 3 mea- participation in broader Project Management sure the second degree range for development Unit monitoring reports of empowerment (use of activities. Cultural heritage man- choice). 3) Level of participation agement teams reports. of male and female Community radio sup- Broad empowerment members at meetings in port center reports. objectives make it diffi- 40 subdistricts. cult to identify mean- 4) Number of community ingful indicators. radio stations operating. 1) Proportion of female Project reports. Indicator 1 measures the participants in commu- Management information second degree of nity development plan- system. empowerment (use of ning process. Beneficiary assessments. choice). 2) Proportion of benefici- Impact studies. Indicators 2 through 5 aries that are poor. measure financial and 3) Measure of level of productive assets. productive asset Indicator 6 measures the ownership. first degree of empower- 4) Beneficiary income ment (existence of levels. choice). 5) Beneficiary under- employment rates. There is a meaningful 6) Measure of level of link between empower- beneficiary access to ment objectives and quality rural develop- indicators used. ment services. 1) Measures of quality, Household surveys by Indicators 1 and 2 diversity, and quantity an independent agency measure assets. of income, expenditure, (pre and post project). Indicator 3 measures the and consumption. Development audit. first degree of empower- 2) Measures of house- ment (existence of hold asset positioning choice). (human, material, social). 3) Measure of level of The indicators do not access to government clearly measure objec- programs and services. tives of enhanced voice and responsiveness. 250 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Project Project objectives Empowerment objectives Ethiopia Women's To improve the welfare Improving the status of Development Initiatives of poor households in target group through Project poor districts through social and economic increased productivity of means. PAD (#20428) women and enhanced (2000/07/27) awareness of important welfare issues. Andhra Pradesh District To improve opportunities Building the capability Poverty Initiatives for the rural poor and of community to demand Project (India) vulnerable to meet their a critical minimum base own social and economic of governance and PAD (#20089) development objectives. accountability for the (2000/03/20) delivery of basic services amongst government (Panchayat Raj) and non-government organizations. APPENDIX 3 251 Empowerment concepts and indicators used Instruments and analytics Comment 1) Number of women PIU and management These indicators measure with new or improved information system human, financial, infor- skills. reports. mational, psychological, 2) Number of trained Supervision mission and organizational women who enter micro- reports. assets. credit projects. Qualitative participatory 3) Number of groups evaluations. There is a meaningful created or strengthened. Micro-project reports by link between empower- 4) Measure of level of intermediary unit. ment objectives and satisfaction with group- indicators used. and community-level outcomes. 5) Proportion of micro- enterprises managed by women. 6) Proportion of funds repaid by individual beneficiaries to their group. 7) Proportion of profits generated compared to grants for raw materials and inputs. 8) Measure of increase in assets two years after project intervention (excluding direct assistance). 1) Measure of level and Independent survey Indicators 1 and 2 and impact (in terms of reports. 6 through 7 measure inclusiveness of local Reports provided by the three degrees government) of organi- District administration of empowerment (exis- zations of the poor. and State/District Project tence of choice, use of 2) Measure of level of Management Unit choice, and achievement perception of the quality reports. of choice). and quantity of govern- Reports and feedback ment services. from NGOs working in Indicators 3 through 5 3) Measure of level of the district. measure financial and investments in produc- material assets. These tive activities. indicators measure 4) Measure of diversity organizational, informa- in sources of income. tional, financial, materi- 5) Measure of level of al, and human assets. productive assets. 252 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Project Project objectives Empowerment objectives Republic of Guinea: To promote social and Local communities and Village Communities economic empowerment their representative local Support Program of the rural population, governments develop including women, youth, the capability to identify, PAD (#17934) and other marginalized prioritize, plan, and man- (1999/01/25) groups. age their own infrastruc- ture and service needs. Republic of Madagascar: To promote, on a pilot Enhanced community A Third Social Fund basis, empowerment of organizational capacity Project poor rural communities to undertake community- and communes to prioritized projects. PAD (#18967) achieve decentralized (1999/02/16) social and economic development. APPENDIX 3 253 Empowerment concepts and indicators used Instruments and analytics Comment 6) Measure of level of Indicators 8 through 10 access to government measure the second programs and services. and third degree of 7) Measure of level of (girls') empowerment satisfaction with facili- (use of choice and tating NGOs in project achievement of choice). villages. 8­10) Measures of girls' There is a meaningful enrollment and comple- link between empower- tion rates in bridge ment objectives and schools and their transi- indicators used. tion from bridge to regu- lar schools. 1) Number of new Beneficiary assessments These indicators measure demand-driven decision and citizen satisfaction the first degree of making processes and surveys. empowerment (existence structures. Periodic reports by of choice), second degree 2) Measure of level of prefectures, CRDs of empowerment (use of transparency and (Communautés Rurales choice), and third degree accountability at all de Développement) and of empowerment levels. service providers. (achievement of choice). 3) Measure of level of Mid-term and final pro- access to higher quality, ject evaluations. There is a meaningful lower cost, and sustain- link between empower- able basic services. ment objectives and 4) Measure of local capac- indicators used. ity to mobilize resources for community infra- structure management. 5) Measure of level of representation (women, youth, migrants, poor) to give voice in local affairs. 1) Level of health center Project impact monitor- Indicators measure the attendance. ing system. third degree of empower- 2) Level of availability of Quarterly and annual ment (achievement of water at the village level. reports prepared by FID. choice). 3) Level of agriculture Financial and technical production at the village reports submitted by There is a meaningful level. financial auditors and link between empower- 4) Level of community independent consultants. ment objectives and income through transpor- Supervision mission indicators used although tation of agricultural pro- reports by World Bank questions are raised over duction to rural markets. staff. attributing measures of 254 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Project Project objectives Empowerment objectives Benin Social Fund To improve access to The ability to organize, Project social services, increase analyze needs, identify income-generation, and and evaluate feasible PAD (#16403) empower communities. solutions, practice such (1998/04/22) solutions, and manage their implementation. Rainfed Natural Empowerment of local The provision of skills to Resource Management rural communities. organize, to manage com- Project (Mauritania) mon resources, and to obtain access to SAR (#16384) financing. (1997/05/22) Rural Women's To strengthen processes Inculcating confidence Development and that promote economic among women, generat- Empowerment Project development of women ing awareness of their (India) and create an environ- rights and privileges, ment for social change. training them for eco- SAR (#16031) nomic activity and (1997/03/04) employment, and bring- ing them into the main- stream of national development. APPENDIX 3 255 Empowerment concepts and indicators used Instruments and analytics Comment Beneficiary assessments empowerment to project by independent inputs. consultants. 1) Number of people (by Mid-term evaluation, Indicators 1 and 2 gender) with access to end of project evaluation. measure the first degree improved services: Beneficiary assessments. of empowerment (exis- schools, clinics, water tence of choice). points, storage facilities. Indicator 3 measures the 2) Number of people (by second degree of gender) with increased or empowerment (use of more stable income due choice). to credit or improved infrastructure. There is a meaningful 3) Number of communi- link between empower- ties and NGOs using new ment objectives and capacities for self-help. indicators used. Measure of level of Mid-term and final pro- Indicator measures the women's participation ject evaluations. second degree of in decision making. empowerment (use of choice). There is a meaningful link between empower- ment objectives and indicators used although measurement needs to capture the quality of women's participation. 1) Measure of income Baseline surveys for Indicators measure the control within the assessing project impacts. first and second degrees household. Follow-up surveys at of empowerment (exis- 2) Measure of decision- midterm and completion tence of choice and use making power regulat- of project. of choice). ing expenditures within the household. There is a meaningful 3) Measure of member link between empower- participation in political ment objectives and processes in the indicators used. community. Appendix 4 Indicators Used in the Five Country Studies The following matrices show some of the indirect and direct indicators of empowerment used in the five country studies in each of the three domains (state, market, and society). Indicators do not always reflect their actual catego- rization in the study. In this appendix, we have adapted the classification (not content) of indicators to illustrate the analytical framework. 257 258 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION A.4.1. State Domain Empowerment Case country indicator Macro level Brazil Agencya Opportunity Structureb · Measure of availability of information about budget (budget accessible through internet, all posts on budget public, and so forth) Degree of Empowermentc · Number of budget allocations with civil society participation · Number of projects moni- tored with civil society participation · Number of government activities with participation from marginalized groups · Number of decisions regard- ing service delivery with civil society participation Ethiopia Agencya · Measure of educational attainment · Measure of level of self- confidence · % women participating in a community group · Time passed since joining the group, if any · % women members of a political party Opportunity Structureb · Existence of laws that treat men and women differently · Existence of informal rules that treat men and women differently APPENDIX 4 259 Intermediary level Local level · Number of NGOs per city and per capita · Per capita income of civil society · Literacy rates of civil society · Measure of degree of self-organization of civil society organizations (registra- tion, members, and so forth) · Measure of educational attainment · Measure of level of self-confidence · % women participating in a commu- nity group · Time passed since joining the group, if any · % women members of a political party · Existence of discriminatory laws · Existence of discriminatory laws · Existence of discriminatory practices · Existence of discriminatory practices · Measure of disconnect between for- · Measure of disconnect between for- mal laws and informal practices mal laws and informal practices · Measure of extent to which men · Measure of extent to which men regard women as equal to them regard women as equal to them 260 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION A.4.1. State Domain (continued) Empowerment Case country indicator Macro level Ethiopia (cont.) Degree of Empowermentc Honduras Agencya Opportunity Structureb · Legislation grants school councils the right to oversee their budgets Degree of Empowermentc APPENDIX 4 261 Intermediary level Local level · Measure of likelihood of a woman · Measure of likelihood of a woman obtaining justice in disputes between obtaining justice in disputes between a man and a woman a man and a woman · Extent to which women are equally · Extent to which women are equally represented in district councils represented in village councils · % school councils informed of the requirement to acquire legal status to open a bank account · % councils that possess an opera- tional manual that lists their respon- sibilities · % councils that have opened a bank account to receive transfers from the government · % councils that file (1) bank state- ments, (2) receipts of purchases and payments, (3) receipts of transfers · % councils receiving training on budgetary matters · Duration of the training · Measure of level of perceived useful- ness of the training · Legislation grants school councils · Legislation grants school councils the right to oversee their budgets the right to oversee their budgets · Extent to which other actors (school · Provision is made to inform principal, mayor, council member, councils about their budget Proheco district or departmental responsibilities officer) intervene in budget oversight · Provision is made to train councils about budget responsibilities · Measure of extent to which other actors (school principal, mayor, council member, Proheco district or departmental officer) intervene in budget oversight · Measure of extent to which school councils are able to comply with the tasks devolved to them (number of tasks managed, timeliness of compli- ance, and so forth) · % women and the poorest members of the community participating in school council meetings 262 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION A.4.1. State Domain (continued) Empowerment Case country indicator Macro level Indonesia Agencya Opportunity Structureb Degree of Empowermentc Nepal Agencya Opportunity Structureb APPENDIX 4 263 Intermediary level Local level · Ability of citizens to approach police · Ability of citizens to approach police · Villagers' ability to access inform- tion on KDP funds and functions · Efficiency with which police solve · Measure of efficiency of police to conflicts (% conflicts reported, solve conflicts % conflicts solved, and so forth) · Measure of ability of the police to · Ability of the police to apply the law apply the law correctly correctly · Measure of ability of courts to · Ability of courts to solve conflicts solve conflicts · Measure of efficiency of village gov- ernment in addressing problems · Number of villages where councils had been elected according to offi- cial KDP procedures · Number of women participating in the KDP councils · Number of conflicts resolved in KDP villages · Number and type of conflict-solving coalitions formed · Number of routines changed as a way of solving conflicts · Literacy levels · Monthly income levels · Size of land household owns · Measure of perceived levels of wealth · Number of livestock household raises · Participation in training events on the rights of low castes and indigenous peoples · Membership in (1) externally orga- nized groups, such as credit or saving group, users' group, women's group, (2) traditional indigenous organizations · Measure of extent of awareness of the candidates who ran for office · Measure of perceived levels of respect with which members of one caste or ethnicity are treated by members of other castes or ethnicities 264 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION A.4.1. State Domain (continued) Empowerment Case country indicator Macro level Degree of Empowermentc a. Agency: measured through endowment of psychological, informational, organizational, material, financial, and human assets. b. Opportunity Structure: measured through presence and operation of informal and formal rules. c. Degree of Empowerment: measured through presence of choice, use of choice (direct or indirect), effectiveness of choice. APPENDIX 4 265 Intermediary level Local level · Measure of perceived degree of security relative to other castes and ethnicities · Measure of perceived levels of opportunity for improvement rela- tive to other castes and ethnicities · Respondent is restricted from enter- ing certain public areas, such as vil- lage district office · Measure of perceived levels of improvements in the status of indigenous people's rights since 1990 · Measure of extent to which the · Measure of extent to which the respondents find it easy to approach respondents find it easy to approach the police or a court the police or a court · Measure of extent to which respon- · Measure of extent to which respon- dents feel that they are treated fairly dents feel that they are treated fairly by the police or by a court by the police or by a court · % cases where district development · % cases where local development agencies met people's requests agencies met people's requests 266 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION A.4.2. Market Domain Empowerment Case country indicator Macro level Brazil Agencya Opportunity Structureb Degree of Empowermentc Ethiopia Agencya Opportunity Structureb Degree of Empowermentc APPENDIX 4 267 Intermediary level Local level · Levels of literacy · Income levels and household income shares · Possession of job-specific skills · % women who have access to and use information from the radio, TV, newspaper, post office, or telephone · % women who participate in a com- munity group · Amount of time passed since joining the group, if any · % women who are members of a credit and savings association or other community group · Laws grant equal rights to men and women · Women can demand equal work conditions · Distance to the nearest market · Extent to which other household members (husband, parents, chil- dren) participate in household chores · Extent to which government provides job-related training · Extent to which cultural restrictions determine which professions women are allowed to use · Measurement of trust in women among lenders regarding debt repayment · Extent to which women choose their · Extent to which women choose their type of employment, negotiate type of employment, negotiate working conditions with their working conditions with their employers, and have access to credit employers, and have access to credit 268 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION A.4.2. Market Domain (continued) Empowerment Case country indicator Macro level Honduras Agencya Opportunity Structureb Degree of Empowermentc Indonesia Agencya Opportunity Structureb Degree of Empowermentc Nepal Agencya Opportunity Structureb Degree of Empowermentc a. Agency: measured through endowment of psychological, informational, organizational, material, financial, and human assets. b. Opportunity Structure: measured through presence and operation of informal and formal rules. c. Degree of Empowerment: measured through presence of choice, use of choice (direct or indirect), effectiveness of choice. APPENDIX 4 269 Intermediary level Local level · % members of disadvantaged · % citizens with access to credit groups employed in public sector · % members of disadvantaged groups who are employed in public sector · % households with access to water and sanitation 270 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION A.4.3. Societal Domain Empowerment Case country indicator Macro level Brazil Agencya Opportunity Structureb Degree of Empowermentc Ethiopia Agencya Opportunity Structureb Degree of Empowermentc APPENDIX 4 271 Intermediary level Local level · Number of registered neighborhood · Number of registered neighborhood associations, NGOs, civil associations, NGOs, civil society groups society groups · Number of registered umbrella · Number of registered umbrella groups groups · Number of members in above groups · Number of members in above groups · % women receiving training on women's rights, female genital mutilation, milk tooth extraction, early marriage · Measure of levels of self-confidence · Extent of awareness of reproductive health issues · Laws protect women from domestic violence · Measure of extent to which women are treated equally under the law in practice · Measure of operation of non-formal courts that discriminate against women · Measure of perception of men and women that domestic violence is acceptable · Measure of extent to which men are punished in courts for committing acts of domestic violence · % women who take action against · Measure of extent to which women harmful traditional practices (female are allowed to participate in com- genital mutilation, milk tooth extrac- munal meetings tion, and so forth) · % women who participate in a com- munity group · Amount if time passed since joining the group, if any · % women who take action against harmful traditional practices 272 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION A.4.3. Societal Domain (continued) Empowerment Case country indicator Macro level Honduras Agencya Opportunity Structureb Degree of Empowermentc Indonesia Agencya Opportunity Structureb Degree of Empowermentc Nepal Agencya APPENDIX 4 273 Intermediary level Local level · % women and poorest members who are literate · % women and poorest members who are aware of when council meetings take place · % women and of the poorest who understand what type of decisions are made during the meetings · % women and of the poorest who have received training regarding the council's functions · Measure of levels of perceived effec- tiveness of training · Women and the poorest are included in public decision making · % women and poorest members who participate in school coun- cil meetings · % women and poorest members who are involved in other, non- school community organizations · Type of position held in other orga- nization, if any · Literacy levels · Monthly income levels · Membership in (1) externally orga- nized groups, such as credit or saving group, users' group, women's group, (2) traditional indigenous organizations · Participation in training events on women's rights · Measure of extent of awareness over women's rights 274 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION A.4.3. Societal Domain (continued) Empowerment Case country indicator Macro level Nepal (cont.) Opportunity Structureb Degree of Empowermentc a. Agency: measured through endowment of psychological, informational, organizational, material, financial, and human assets. b. Opportunity Structure: measured through presence and operation of informal and formal rules. c. Degree of Empowerment: measured through presence of choice, use of choice (direct or indirect), effectiveness of choice. APPENDIX 4 275 Intermediary level Local level · Measure of perceived levels of improvement in the status of women's rights since 1990 · Measure of perceived levels of eco- nomic success of own caste or ethnicity · Measure of perceived levels of improvement in the status of indige- nous people's rights since 1990 · Measure of perceived levels of opportunity for improvement rela- tive to other castes and ethnicities · Measure of perceived degree of security relative to other castes and ethnicities · Measure of frequency with which · Measure of frequency with which respondents were verbally harassed, women are subject to domestic threatened, or physically assaulted violence (verbal and physical by high caste people, police, harassment) other groups · Measure of frequency with which · Measure of respondents' ability to respondents were verbally harassed, get help regarding these incidences threatened, or physically assaulted with police or authorities by high caste people, police, · Perceived levels of usefulness of other groups attempts to get help · Measure of respondents' ability to get help regarding these incidences with police or authorities · Measure of perceived levels of use- fulness of attempts to get help Appendix 5 Empowerment Themes and Strategies in Selected PRSPs PRSP country Empowerment theme Empowerment strategies Albania Governance · Enhancing accountability through public administra- tion reforms and commu- nity participation in local government · Legal and judicial reform · Anticorruption · Decentralization · Public education on decen- tralization and civic rights in local government Ethiopia Good governance, account- · Democratic decentraliza- ability and improved service tion: fiscal federalism and delivery, with a focus on gen- enabling legislation der equality · Strengthen capacity of com- munities to federate and take advantage of opportu- nities for voice afforded by decentralization Community participation · Ethiopian Social Rehabili- tation and Development Fund has during previous four years empowered poor communities through participation in projects The Gambia Empowerment of women · Eliminate gender disparity in primary and secondary education Participation in governance · Decentralization as a key poverty alleviation compo- nent for the empowerment of the poor Community empowerment · Community empower- ment initiatives 277 278 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION PRSP country Empowerment theme Empowerment strategies Ghana Empowering grassroots · Not clear organizations · Ministry of Parliamentary Access to information on gov- Affairs disseminated gov- ernment policies, linked to ernment policies and pro- accountability among public grams to the unit committee office holders and informed levels of the decentralized choices administrative system "Deepened" citizen participa- · Strengthen administrative tion in decision making capacity among District Assemblies (with outcomes including more regular meetings and properly func- tioning committees) · Passage of Local Govern- ment Service Bill (to raise morale on expectation of improved service conditions) · Piloting system of district composite budgets · Expand number of districts and constituencies for administrative and elec- toral functions · Civil service reform: includ- ing performance orientation · Greater cooperation between District Assemblies and Civil Society Organizations · Empower communities to demand accountability (budget advocacy, for example) Empowering women · Funds disbursed through Women's Development Fund · Ministry of Women and Children's Affairs creating new jobs for women nationally · Increase school enrollment for girls nationally Guinea Grassroots empowerment · Participation in the design, implementation, and evalu- ation of development pro- jects by "grassroots entities" APPENDIX 5 279 PRSP country Empowerment theme Empowerment strategies Guyana Local governance and service · Decentralize service delivery provision · Strengthen local government by improving the represen- tation of communities on local government councils, improving accountability mechanisms, and introduc- ing expenditure tracking systems Kenya Farmers' empowerment · Improved legislation and empowerment of farmers' associations Unlocking capacity in local · Not clear governments and communities Lao PDR Participatory community · Improve levels of education; development: empowering higher degree of human the grassroots level to par- and social development ticipate in development dia- logue and decision-making processes Macedonia, FYR Empowerment and ownership · Participation in the PRSP of the PRSP Madagascar Governance and service · Decentralization and delivery empowerment of individ- uals and local communities Malawi Economic empowerment, with · Raise awareness of gender specific reference to gender issues, women's legal rights, imbalances and economic empower- ment of women Empowerment for forest · Capacity building for forest resource management resource management · Introduce regular meetings to discuss and explain changes in forestry policy Local empowerment and · Civic education participation 280 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION PRSP country Empowerment theme Empowerment strategies Nepal Empowerment broadly · Greater access to markets, public services, income gen- erating activities, and oppor- tunities for self-help, security, and lower vulnerability Women's empowerment and · Mainstreaming women's gender equality participation in every aspect of national development · Micro-credit expansion linked to international nongovernmental organizations/NGOs marketing assistance program Empowerment of local users · Not clear for forest management and utilization Nicaragua Governance: Lack of economic · Develop a set of indicators and political power among for participation, dialogue, the poor to influence decision- and consensus-building for making processes that affect good governance their lives · Improve transparency and accountability among pub- lic officials and prevent and punish corruption Rwanda Good governance: Grassroots · Capacity-building programs participation in development at grassroots level and decision making Economic empowerment, with · Creation of training and specific reference to unskilled employment opportunities youth Community empowerment · Community development through participation projects Women's empowerment · Eliminating gender disparity in primary and secondary education · Targeted micro credit programs Senegal Empowerment to manage · Participatory methods natural resources for natural resource management APPENDIX 5 281 PRSP country Empowerment theme Empowerment strategies Tanzania Governance and service · Stakeholder participation in provision sector strategy design and implementation · Information campaigns on local government reform to promote transparency and accountability in public ser- vices delivery · Institutional reforms on gov- ernance, accountability, and transparency · Codes of conduct for coun- cilors and staff at local gov- ernment level Uganda Powerlessness, defined as the · Public information around inability to affect things around entitlements and roles in one (reflected in the Uganda service delivery Poverty Eradication Action · Mechanisms for citizen par- Plan), with specific ticipation in policy making reference to service provider · Reform of public procure- accountability ment regulations · Leadership code for politi- cal leaders Vietnam Governance and service · Decentralization of service delivery delivery, increased partici- pation, and more transpar- ent resource allocation Yemen, All-encompassing approach · Not clear Republic of policy framework that would enable the government to empower the poor Women's low level of partici- · Not clear pation in labor markets, edu- cation systems, decision- making structures as well as in exercising their reproduc- tive health choices and legal rights 282 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION PRSP country Empowerment theme Empowerment strategies Zambia Economic empowerment to Programs such as the Agricul- escape poverty (contrasted tural Sector Investment Pro- with a safety nets approach) gram, which includes developing infrastructure for small-scale farmers, and the Environmental Support Pro- gram, which includes support for sustainable community- based projects. Appendix 6 Draft National Survey Empowerment Module This appendix presents an example of a National Survey Empowerment Mod- ule, including an Institutional Mapping and an Individual Questionnaire. The format of this module is designed to elicit information that, after analysis, will enable researchers to assess the degree of empowerment of various indi- viduals and groups within the three domains and eight subdomains of the empowerment framework. It will also allow analysis of agency and opportu- nity structure, the factors associated with empowerment. The module combines an institutional mapping section and an individ- ual questionnaire. The institutional mapping section is designed to capture data regarding opportunity structure. During analysis, researchers may also need to take into account asset indicators or opportunity structure indicators captured using preexisting data sources and indexes. The module can also be used in combination with other survey instruments that already capture some of the data elicited by the module, such as the Social Capital Assessment Tool, the Integrated Questionnaire for the Measure- ment of Social Capital, and the Living Standards Measurement Survey. When a questionnaire is used in conjunction with other instruments, it must be adapted to ensure that questions are not duplicated. Institutional Mapping The purpose of the institutional mapping section is to contribute to the mea- surement of opportunity structure discussed in part I of this book. The mea- surement of the institutional element of opportunity structure is not easily captured using individual or household surveys and therefore requires a mixed-methods approach, combining various participatory tools and processes in group discussion exercises to generate local in-depth data on the opera- tion of formal and informal institutions together with the national tracking of legislation, regulation, and procedures. Participatory group analysis will enable data to be collected on those indirect indicators where preexisting indexes do not exist, and additionally enable a degree of triangulation on those where they do. 283 284 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION TableA.6.1 of this appendix summarizes the information that can be elicited from both the institutional mapping and from the questionnaire. Institutional mapping can be sequenced with the administration of the questionnaire, pro- viding an opportunity to identify focus groups through subsampling of the questionnaire households (see discussion below). There are various participatory tools that can be used to capture infor- mation not already available through preexisting indexes or sources within different domains and subdomains. These are listed in the final column of table A.6.1. Where there is more than one possible tool, the most appro- priate tool is identified as primary. The selection of tools depends very much on the context and the information being sought. Groups should consist of between five and twelve participants (ideally about eight) and reflect social stratification in any particular context. Key informants with in-depth country knowledge should be able to provide the researchers with the most important social groupings. Researchers should place particular emphasis on ensuring that marginal groups are included in the process. Mixed group interviews can also be conducted to assess levels of consensus, but these should be in addition to separate groups. In participatory research, if the information being generated belongs to the community and can be provided in an unbiased way by key informants, there is no need to select these key informants randomly (Barahona and Levy 2002, 23). Given that this institutional mapping is generating largely interpre- tive qualitative and quantitative data, however, it will be important to try to reduce bias by selecting socially stratified participants from the house- hold survey sample using probability-based (random) methods. It should be noted that within each social stratum there is an ethical trade-off between employing probability-based sampling to offset biases introduced by par- ticipant self-selection and adhering to participatory research principles of including those who wish to be included. Aless objective but more volun- taristic and democratic alternative is to record key social information (for example, sex, age, educational level, religion, and reading ability) about each participant to assess the profile of each socially stratified focus group (Barahona and Levy 2002, 26). Each group should have a moderator and two observers. The moderator facilitates the discussion, probes key issues, elicits comments from all par- ticipants, and focuses the discussion on the issues of interest. This should be done without interrupting or ignoring extraneous comments from partici- pants, but also ensuring that the discussion remains focused as far as possi- ble. The observers take notes on the content of the discussion and process of groupdynamics,noting,forinstance,whotalksthemostortheleast,whodoes not participate at all or defers to others in the group, who tries to dominate the discussion, and so on. The observers will record the discussion on tape, APPENDIX 6 285 and photograph and sketch any visual diagrams from the tools. The facilita- tors will explain the purpose of the discussion and research, and prior consent will be asked from the group for the use of tape recorders. The facilitation and observation of participatory group discussions using some of the tools suggested is not an easy task. Researchers should be well aware not only of how to apply and use the tools, but also of the importance of the manner in which they facilitate and behave when using the tools in a discussion. Training researchers is key to the success and accuracy of partici- patory methodologies in the gathering of data. Each group discussion should last about two to three hours. The discus- sion should be based on the tools suggested in table A.6.1 and around a set of guide questions discussed below. One possible problem with focus group discussions is that too much information may be generated. Key prompts are highlighted in box A.6.1 of this appendix, designed for use by experienced researchers to explore selected key indicators. 286 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Table A.6.1 Summary of Survey Module and Institutional Mapping Themes and Tools Instrument How administered Themes Survey module Random-stratified · Location details sample administered · Respondent details through existing sur- · Indirect indicators of agency vey instrument (can agency (asset ownership) be administered to a · Direct indicators of empow- subsample) erment measuring: (a) oppor- tunity to use influence; (b) use of influence; and (c) effectiveness of use of influence Institutional mapping Socially stratified focus Enactment of rules (processes) groups composed in the following areas: from subsample of questionnaire survey State · Justice systems (primary tool ­ 1; additional tool ­ 5) · Political representation (primary tool ­ 5; addi- tional tool ­ 4) · Access to and quality of services (primary tool ­ 5; additional tools ­ 1, 4) Market · Labor market and employ- ment conditions and choices (primary tools ­ 1, 3) · Asset entitlements and con- sumption (primary tool ­ 4; additional tools ­ 1, 2) · Private services, focusing on credit provision (pri- mary tool ­ 1; additional tools ­ 4, 5) Society · Household and kinship group entitlements, roles, and responsibilities (primary tools ­ 3, 4, 5; additional tool ­1) · Community organization and relationships (primary tools ­ 4, 5; additional tools ­ 1, 3) APPENDIX 6 287 Tools Individual Questionnaire Section 1 for location details Section 2 for respondent details Section 3 for indirect indicators of agency Section 4 for direct indicators of empowerment Participatory tools (1) Preference ranking or scoring ­ This method involves ranking or scoring people's priorities, problems, or preferences. Disaggregation of groups performing the analysis by age, gender, class, ethnic group, and so on, enables comparative analysis and exploration of people's experience, perceptions, priorities, and choices based on criteria identified by them regarding a range of subjects from resource allocation to service provision to choice of employment (for example, how do people rate different justice systems or health services according to effectiveness, cost, accessibility, and so forth?). (2) Well-being (or wealth) ranking ­ This method involves ranking different indi- viduals, households, or communities according to an overall view of well- being. Within the context of the measuring empowerment study, it can be adapted to allow expression of people's own definitions of empowerment and also enable them to identify, using their own criteria, who in their com- munities is more or less empowered. It can be used only within the limita- tions of the shared mutual knowledge of the group carrying out the analysis (detailed knowledge is needed to establish the ranking). Performing such exercises for communities as well as households or individuals can illustrate the significance of factors and assets that affect empowerment at the com- munity or group level (for example, distance from local or regional government offices or road infrastructure). (3) Charts illustrating cyclical change (seasonality, daily activities, and so forth) ­ These methods address the distribution of phenomena over time in more or less predictable cycles (for example, employment options through the year or the distribution of tasks and workload over a woman's day). They enable temporal analysis of, and the trends evident in relation to, selected variables, and can also enable an understanding of the links among variables. (4) Social mapping, modeling and transects ­ These methods enable situational analysis of social structures and services. Representations of spatial distribu- ution and location of resources, social groups, facilities, and so on, can help analyze performance and coverage of existing services and identify services 288 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Table A.6.1 (continued) Instrument How administered Themes Analyzing the qualitative data normally produced using participatory tools can be difficult, especially when trying to incorporate the data into the predominantly statistical analysis of the individual survey. The analy- sis of this data often relies to a certain extent on interpretation and reflec- tion. However, generating numbers from, or quantifying, the qualitative outputs of participatory approaches and tools is possible and can help when trying to combine the analysis with the data from the individual survey. Given that the information being analyzed will be from focus groups span- ning a number of different sites, recoding the information should be sys- tematic and should use consistent predetermined formats and terminology where possible. The analysis of the focus group discussion data will be crosschecked with data from the individual interviews. Crosschecking may produce con- tradictions that will also need to be explained or resolved. Diversity is another important factor to consider when analyzing the data from focus groups and interviews. Diversity in responses can be seen as an indicator of empowerment, with diversity of behavior at the population level being a gross indicator of agency (of the ability to make choices), relative to homo- geneous behavior by the same set of people. Analysis of responses should therefore take into account the range of responses, as well as the average response. For the focus group discussions, this means that accurate docu- mentation of discussions is vital to record all views expressed before any con- sensus is reached by a group. The need to analyze a diversity of qualitative factors for the existence of complex relationships may well benefit from the use of computer-based qualitative data analysis or thinking support software, such as NUD*IST, Creative Thinker, or Visual Concept, to enable alternative relationships to be visualized, documented, and assessed in an effective manner. APPENDIX 6 289 Tools that are needed but are not available. Mapping social structures can help analyze how social differences can affect people's lives (for example, the degree of influence of social differences on political participation and influence). (5) Institutional and Venn diagramming: Diagramming enables a representation and analysis of institutional relationships, linkages, accessibility, significance, and influences affecting local people, households, and communities from within and outside their area. Institutions could include government service providers, the police, or even individuals with significant power. An analysis of institutional impact (whether positive or negative) can also be undertaken. Checklist of Prompts for the Facilitation of Focus Group Discussions in Each Domain and Subdomain The following prompts are designed to help an experienced focus group facilitator to guide discussion and elicit key information in each domain and subdomain. The prompts are not designed to be followed rigidly. State, justice: · Is there a local word for "a right" or "rights," and to what things is it applied? · What rights do people in general have? · What rights do ____________________________________ [describe group membershiporcharacteristics,forexample,youngunmarriedwomen]have? · What can ____________________________________ [describe group mem- bership or characteristics, for example, young unmarried women] do when they feel discriminated against or are the victims of crimes? · How common are crimes rooted in living, customary, or religious law (for example, honor killing, domestic violence, and sexual abuse)? · What are the most preferred local informal justice or dispute resolution systems? What are their advantages and disadvantages compared to other systems? · How protected by legislation do people feel from political and social oppression and from domestic violence? State, political: · Are some people or groups left out of society or excluded from com- munity life or decision making (social exclusion)? If yes, who is left out, why, and how? 290 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION · To what degree are different groups of people (differentiated by social dif- ferences) able to participate in political processes? State, service delivery: · To what extent does the formal justice system ensure that economic, social, and cultural rights are recognized and provided by government institutions? · When faced with a crisis or shock (i.e. unemployment, illness, crop fail- ure, for example), what institutions do people turn to? How are they ranked in terms of preference? · What government and nongovernment safety nets or informal social transfer systems are available to vulnerable people and how are they ranked in terms of preference (that is, in terms of transparency and equity of operation)? · Are there any sources of public credit? · What can ____________________________________ [describe group mem- bership or characteristics, for example, young unmarried women] do when they feel discriminated against or are the victims of crimes? Market, labor: · How aware are people of any legal labor standards that employers should comply with? To what degree do employers comply, and how effective is government in ensuring compliance? · Within the household, how are roles allocated and work divided? How easy is it for different people to change roles? Market, goods: · How have markets, for example, for labor (local, national, and interna- tional), land, water, housing, and produce, and access to markets changed? Over what period? · Are different social groups affected differently by any changes? · To what degree are relationships between product producers and dis- tributors or buyers transparent and accountable? Is there any legislation designed to ensure fair trading conditions and how effectively is it enforced? · To what degree is access to and control over productive assets influ- enced by social characteristics? · Are there any rules regarding inheritance of assets (gendered inheri- tance rules, for example)? How strongly are they upheld and enforced? · Are there any government policies and programs concerned with the redistribution of land (land reform)? · To what degree do different people in households have access to and control over consumption goods and services? APPENDIX 6 291 Market, private services (primarily credit): · Where can you access credit? How do different groups in the commu- nity (differentiated by social differences) rank them in order of accessi- bility, effectiveness, transparency, accountability, and freedom from corruption? · Who has control and access to credit within different households? Society, community: · How many and how diverse are the membership organizations that exist? What are the "rules" of community membership groups? · To what degree does social identity affect membership of community groups and associations? · Is there conflict between different groups in the community? Individual Questionnaire: Guidelines on Application Thisquestionnaireisdesignedtobeadministeredwitharesearcherandrespon- dent present. In some contexts, it may be more difficult to interview some respondents privately (for example, in some households it may be difficult to interview women without other people being present, even when the enu- merator is female). Where this is the case, it is important to indicate this on the questionnaireandrecognizeanypossibleimplicationsintheanalysis.Similarly, where respondents have chosen not to answer a question, the questionnaire provides additional space for the enumerator to write in a reason for a nonre- sponse, although it may be difficult or inappropriate to elicit this information. Researchers should follow the questionnaire's wording exactly to ensure consistency and to allow comparisons across sites. Changes made to the wording of the questionnaire should be fully described and also applied consistently across research sites. Where the questionnaire is being used in conjunction with other survey instruments such as the Social Capital Assessment Tool (SCAT) or the Inte- grated Questionnaire for the Measurement of Social Capital (IQMSC), there may be replications. Those questions that are already found in the SCAT are marked (*) after the question, and those already found in the IQMSC are marked (+). The precise questions used in theLiving Standards Measurement Survey (LSMS) modules were not available at the time of writing. The LSMS Economic Activities, Housing, Food Expenditure, Education, and Health Modules cover areas also covered in the individual questionnaire. If LSMS modules are used in conjunction with the questionnaire, care must be taken to avoid duplication. It should also be noted that the SCAT, IQMSC, and LSMS are household-level instruments, whereas this questionnaire is designed for use at the individual level. 292 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Questionnaire Thank you for agreeing to take part in this survey. We would like to ask you some questions that will help us to understand the situation in which you find yourself in various areas of your life, and how these are connected with how much control you feel you have when you are making decisions and putting your decisions into action. The results of this survey will be completely confidential and no iden- tifying data will be collected. Some of the questions may also be quite per- sonal and we hope this will be OK with you. If, however, you do not feel comfortable answering any questions, please feel free to say so. Section 1: Location Details 1.1 Province/state ______________________________ (*) 1.2 District ____________________________________ (*) 1.3 Sub-district ______________________________ (*) 1.4 Town/village ______________________________ (*) 1.5 Community ______________________________ (*) 1.6 Street ______________________________ (*) 1.7 Type of area: [Observation only] (*) 1 Urban 2 Rural 3 Indigenous 4 Difficult access 1.6 Location: Unit __________________ (*) Number _______________ 1.7 Respondent code number (from list): APPENDIX 6 293 Section 2: Respondent Details First, I would like to ask some questions about yourself. If you do not wish to answer a particular question, please feel free to say. 2.1 Sex of respondent [Observation only] 1 Female 2 Male 2.2 Can you please tell me your age group? 1 Under 16 2 16 ­ 20 3 21 ­ 25 4 26 ­ 35 5 36 ­ 45 6 46 ­ 55 7 56 ­ 65 8 66 or over 2.3 What is your marital status? 1 Married 2 Living with domestic partner 3 Single 4 Separated 5 Widowed 6 Divorced 2.4 How many people do you share your house with? 2.5 What is your religion? [Options and codes to be filled in as locally appropriate] 1 2 3 4 5 294 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 2.6 In terms of your ethnicity, do you consider yourself ...? [Options and codes to be filled in as locally appropriate] 1 2 3 4 5 2.7 (If appropriate) Do you belong to a particular tribe? [Options and codes to be filled in as locally appropriate] 1 2 3 4 5 2.8 (If appropriate) What caste do you belong to? [Options and codes to be filled in as locally appropriate] 1 2 3 4 5 2.9 What educational level have you reached at the moment? (*) 1 Elementary (not completed) 2 Elementary (completed) 3 Secondary (not completed) 4 Secondary (completed) 5 Technical college graduate 6 University graduate 7 Post-graduate 8 Other [Specify and add code:___________________] APPENDIX 6 295 2.10 Please imagine a nine-step ladder where on the bottom, the first step, stand people who are completely without power, and on the highest step, the ninth, stand those who have a lot of power. On which step are you today? 2.11 Who is present during the interview? [Observation only] 1 Respondent and enumerator only 2 Respondent, spouse, and enumerator 3 Respondent, other household member, and enumerator Section 3: Indirect Indicators of Individual Agency The next set of questions we would like to ask you concern your feelings about yourself, the community and society you live in, and the property and assets you own or have access to. If you do not wish to answer a par- ticular question, please feel free to say. If you can tell us why you do not want to answer a particular question, that would be very useful to us, but you should not feel under any obligation to do so. Informational assets 3.1 How long does it take you to reach the nearest working post office? (+) 1 Less than 15 minutes 2 15-30 minutes 3 31-60 minutes 4 More than one hour 5 More than four hours 3.2 How many times in the last month have you read a newspaper or had one read to you? (+) 296 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 3.3 How often do you listen to the radio? (+) 1 Every day 2 A few times a week 3 Once a week 4 Less than once a week 5 Never 3.4 How often do you watch television? (+) 1 Every day 2 A few times a week 3 Once a week 4 Less than once a week 5 Never 3.5 Howlongdoesittakeyoutogettothenearestworkingtelephone?(+) 1 Telephone in the house 2 Less than 15 minutes 3 15-30 minutes 4 31-60 minutes 5 More than 1 hour 6 More than four hours 3.6 In the past month, how many times have you made or received a phone call? (+) 3.7 In general, compared to five years ago [Enumerator: Time period can be clarified by situating it before/after a major event], has your access to information about [specify] improved, deteriorated, or stayed about the same? (+) 1 Improved 2 Deteriorated 3 Stayed about the same APPENDIX 6 297 3.8 Is your house easily accessible by road all year long or only during certain seasons? (+) 1 All year long 2 Only during certain seasons 3 Never easily accessible 3.9 In the last three years, do you feel the roads leading to your com- munity have: 1 Improved 2 Worsened 3 Remained the same 3.10 How many times have you traveled to [Enumerator: In rural areas, specify a neighboring village or town; in urban areas, specify another part of the city] in the past year? (+) Organizational assets 3.11 Are you a member of any organization or group? (*) (+) 1 Yes 2 No [Go to question 3.21] 3.12 Which of the following groups are you a member of? (*) 1 Farmer or fisher group or cooperative 2 Other production group 3 Traders or business association 4 Professional association (doctors, teachers, veterans) 5 Trade union or labor union 6 Neighborhood or village committee 7 Religious or spiritual group (for example, church, mosque, temple, informal religious group, religious study group) 8 Political group or movement 9 Cultural group or association (for example, arts, music, theatre, film) 298 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 10 Burial society or festival society 11 Finance, credit, or savings group 12 Education group (for example, parent-teacher association, school committee) 13 Health group 14 Water and waste management group 15 Sports group 16 Youth group 17 NGO or civic group (for example, Rotary Club, Red Cross) 18 Ethnic-based community group 19 Other groups [Please specify in table below and add code] [Code] [Code] [Enumerator: List all categories of organization/groups] 3.13 Which of these organizations or groups are the most important to you? Please specify up to three. Please rank (1 most important) (*) (+) Org/group 1 Org/group 2 Org/group 3 APPENDIX 6 299 3.14 For each of these three important groups, how effective overall is the group's leadership ? (*) (+) 1 Very effective 2 Fairly effective 3 Not effective Org/group 1 Org/group 2 Org/group 3 3.15 How are leaders in each group selected? (+) 1 By an outside person or entity 2 Each leader chooses his or her successor 3 By a small group of members 4 By decision or vote of all members 5 Other [Specify and add code: ______________________________] 6 Don't know/not sure Org/group 1 Org/group 2 Org/group 3 3.16 How much influence do you think you have when each group chooses its leaders? 1 A lot of influence 2 Some influence 3 A little influence 4 No influence Org/group 1 Org/group 2 Org/group 3 300 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 3.17 How much does being a member of these groups benefit you individually? 1 Greatly 2 Fairly 3 A little 4 Not at all Org/group 1 Org/group 2 Org/group 3 3.18 What is the most important benefit, if any, that you feel you gain from being a member of these groups? [Enumerator: Specify benefit for each group and add code] Org/group 1: [Code] Org/group 2: [Code] Org/group 3: [Code] 3.19 Overall, are the same people members of these three different groups, or is there little overlap in membership? (*) 1 Little overlap 2 Some overlap 3 Much overlap Org/group 1 Org/group 2 Org/group 3 APPENDIX 6 301 3.20 For each group, do the members mostly hold the same political values or belong to the same political party? (*) (+) 1 All with the same political values or belonging to the same polit- ical party 2 Mainly with the same political values or belonging to the same political party 3 With a few different political values or belonging to a few differ- ent political parties in the community 4 With many different political values or belonging to many differ- ent political parties in the community 5 Not applicable in this situation or context Org/group 1 Org/group 2 Org/group 3 Material assets 3.21 Does your household use any land or property (for farming, live- stock, renting out, and the like)? 1 Yes 2 No [Go to question 3.23] 3.22 What is the "ownership status" of this land? 1 Owned 2 Rented 3 Sharecropped 4 Combination 5 Used with no formal agreement 6 Other [Specify and add code: ________________________] 302 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 3.23 Do you personally use any land or property (for farming, livestock, renting out, and the like)? 1 Yes 2 No [Go to question 3.25] 3.24 What is the "ownership status" of this land? 1 Owned 2 Rented 3 Sharecropped 4 Combination 5 Used with no formal agreement 6 Other [Specify and add code: ________________________] 3.25 Is your home... (*) 1 Owned and completely paid for 2 Owned with a mortgage 3 Rented 4 Given in exchange for services 5 Squatter 6 Other [Specify and add code: ________________________] 3.26 How many rooms are used for sleeping only? (*) APPENDIX 6 303 3.27 Type of house [Observation only] (*) 1 Individual house 2 Open roof and patio 3 Apartment 4 Room within a larger house 5 Other [Specify and add code: ________________________] 3.28 What construction material is used for the majority of the exterior walls of the house or building? [Observation only] (*) 1 Cinderblock/brick/stone/concrete/cement 2 Fiberglass 3 Wood 3 Adobe/wattle and daub 4 Cane/straw/sticks 5 No walls 6 Other [Specify and add code: ________________________] 3.29 What is the construction material of most of the roof of this house? [Observation only] (*) 1 Concrete/cement 2 Tiles 3 Metal (zinc, aluminum, etc.) 4 Wood 6 Straw or thatch 7 Other [Specify and add code: ________________________] 304 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 3.30 What is the construction material of most of the floor of this house? [Observation only] (*) 1 Concrete/cement 2 Tiles, brick, granite 3 Wood 4 Vinyl 5 Earth, sand 6 Cane 7 Other (specify) [Specify] 3.31 What type of sanitary services does this household use? (*) 1 Connected to sewage system 2 Connected to septic tank 3 Latrine 4 None 5 Other (specify) [Specify] 3.32 What is the primary source of water for this household? (*) 1 Public piped water system to individual house 2 Private well 3 Public well 4 Shared open tap or faucet 5 River or stream 6 Other [Specify and add code: ________________________] 3.33 What type of lighting does this household use? (*) 1 Electricity (public source) 2 Electricity (private source) 3 Electricity (combination of public and private) 4 Only kerosene, gas, candles 5 Other [Specify and add code: ________________________] APPENDIX 6 305 3.34 In your work or livelihood, do you need to use any particular tools or equipment? 1 Yes 2 No [Go to question 3.37] 3.35 What tools or equipment do you need? [Enumerator: Please specify and add code] Tool A: [Code] Tool B: [Code] Tool C: [Code] 3.36 Which of these tools or equipment do you own (either individu- ally or collectively), rent, borrow, or not have any access to? 1 Own individually 2 Own collectively 3 Rent individually 4 Rent collectively 5 Borrow 6 Do not have any access to 7 Other [Specify and add code: ________________________] Tool A Tool B Tool C 306 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 3.37 Which of the following items do you own, if any? 1 Bicycle 2 Television 3 Radio 4 Refrigerator 5 Motor bike 6 Motor vehicle Financial assets 3.38 What is your main occupation? (*) 1 Farmer 2 Fisherman 3 Trade 4 Manufacturing ­ Artisan 5 Manufacturing ­ Industrial 6 Private sector ­ Unskilled 7 Private sector ­ Skilled 8 Public sector ­ Unskilled 9 Public sector ­ Skilled 10 Other [Specify and add code: ________________________] APPENDIX 6 307 3.39 What is your secondary occupation? 1 Farmer 2 Fisherman 3 Trade 4 Manufacturing ­ Artisan 5 Manufacturing ­ Industrial 6 Private sector ­ Unskilled 7 Private sector ­ Skilled 8 Public sector ­ Unskilled 9 Public sector ­ Skilled 10 Other [Specify and add code: ________________________] 3.40 How would you categorize your employment status? 1 Self-employed 2 Employed on permanent contract 3 Employed on temporary contract 4 Employed but with no contract 5 Casual employee with contract 6 Casual employee without contract 7 Employed on a daily basis 8 Working within the household 9 Unemployed 3.41 How often have you voluntarily changed your employment or occupation in the past? 1 Very often 2 Fairly often 3 Not very often 4 Never 3.42 How often have you involuntarily had to change your employ- ment or occupation in the past? 1 Very often 2 Fairly often 3 Not very often 4 Never 308 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 3.43 How secure do you feel in your present employment or occupation? 1 Very secure 2 Fairly secure 3 Neither secure nor insecure 4 Fairly insecure 5 Very insecure 3.44 Have you ever borrowed money from another person or institution? 1 Yes 2 No 3.45 Are you in debt to anyone at the moment? 1 Yes 2 No [Go to question 3.48] 3.46 How indebted would you say you are at the moment? 1 Extremely indebted 2 Very indebted 3 Fairly indebted 4 A little indebted 3.47 Do you feel you struggle to repay any debts you have? 1 Yes, I struggle greatly 2 Yes, I struggle a little 3 No, I don't struggle at all APPENDIX 6 309 3.48 Can you tell me what proportion of your household expenditure you think is spent on the following in an average month? 1: Food 2: Rent and housing costs 3: Utility bills 4: Clothing 5: Loan repayment 6: Livelihood related expenses 7: Education fees and costs 8: Healthcare expenses 9: Savings 10: Entertainment 3.49 How many illnesses or medical problems that have stopped you working or attending school have you had in the last... A. Month B. 6 months C. Year D. 3 years Psychological assets 3.50 Are there any community activities, such as those organized by the local government, religious organizations, the school, the local development association, etc., in which you think you are not allowed to participate? (+) 1 Yes 2 No, I can participate in all activities [Go to question 3.53] 310 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 3.51 In which activities do you perceive you are not allowed to partic- ipate? (+) [Enumerator: List up to 3 activities and add codes] [Code] [Code] [Code] 3.52 Why do you think you are not allowed to participate? (+) [Enumerator: List up to 2 reasons] 1 Poverty 2 Occupation 3 Lack of education 4 Gender 5 Age 6 Religion 7 Political affiliation 8 Ethnicity or language spoken, race, caste, or tribe 9 Other [Specify and add code: ________________________] 3.53 How often have you met with and talked to people from other social groups outside your home in the last week? (+) 1 Not at all 2 Once 3 Several times 4 Daily 5 Several times a day APPENDIX 6 311 3.54 Are there any people from different social groups that you feel you cannot, or would have difficulty in socializing with? 1 Yes 2 No [Go to question 3.56] 3.55 Why do you feel you cannot socialize with these people? [Enumerator: List up to 2 reasons] 1 Poverty 2 Occupation 3 Lack of education 4 Gender 5 Age 6 Religion 7 Political affiliation 8 Ethnicity or language spoken, race, caste, or tribe 9 Other [Specify and add code: ________________________] 3.56 Is there anything in your life that you would like to change? 1 Yes 2 No [Go to question 3.62] 3.57 What thing(s) would you most like to change? [Enumerator: List up to 3 areas/things and add codes] A: [Code] B: [Code] C: [Code] 312 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 3.58 Do you think these will ever change? 1 Yes 2 No [Go to question 3.62] 3.59 When do you think they will change? 1 Very soon 2 Fairly soon 3 A long time in the future 3.60 Who do you think will contribute most to any change? [Enumerator: list up to 2 reasons] 1 Myself 2 My family 3 Our group [Specify and add code: ________________________] 4 Our community 5 The local government 6 The national government 7 Other [Specify and add code: ________________________] 3.61 What are the main difficulties that you feel might prevent these changes from occurring? [Enumerator: List 1 reason for each area/thing listed in 3.57 and add code] A: [Code] B: [Code] C: [Code] APPENDIX 6 313 3.62 Do you feel that people like yourself can generally change things in your community if they want to? 1 Yes, very easily 2 Yes, fairly easily 3 Yes, but with a little difficulty 4 Yes, but with a great deal of difficulty 5 No, not at all 3.63 What is the one thing you would most like to do in your life? [Enumerator: List and add code] [Code] 3.64 How difficult do you think it will be for you to achieve this? 1 Very difficult 2 Fairly difficult 3 Fairly easy 4 Very easy Section 4: Direct Indicators of Empowerment We would like to ask your opinions about what choices you have in your relationship with the government, service providers, the justice system, and the institutions that have an effect on the lives of people. We would also like to ask you questions about the choices you have in terms of employ- ment, access to credit, and about relationships within your household and community. If you do not wish to answer a question, please feel free to say. 314 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Domain/Subdomain: State/Justice 4.1 To your knowledge, what mechanisms are used in your area and in other parts of the country to achieve justice? [Enumerator: List all systems mentioned and add codes. Codes must dis- tinguish between formal and informal justice systems] A: [Code] B: [Code] C: [Code] D: [Code] E: [Code] 4.2 Have you ever used these systems to seek redress or access justice? 1 Yes 2 No [If none at all, go to question 4.6] A: [Code] B: [Code] C: [Code] D: [Code] E: [Code] 4.3 How many times in the last three years have you used these sys- tems to seek redress or access justice? A: [Code] B: [Code] C: [Code] D: [Code] E: [Code] APPENDIX 6 315 4.4 How happy were you with the outcome? 1 Completely happy 2 Fairly happy 3 Neither happy nor unhappy 4 Fairly unhappy 5 Completely unhappy A: [Code] B: [Code] C: [Code] D: [Code] E: [Code] 4.5 How fairly do you think you were treated? 1 Completely fairly 2 Reasonably fairly 3 Not fairly A: [Code] B: [Code] C: [Code] D: [Code] E: [Code] 4.6 How fairly do you think you would be treated if you were involved in any of these systems of justice in the future? 1 Completely fairly 2 Reasonably fairly 3 Not fairly A: [Code] B: [Code] C: [Code] D: [Code] E: [Code] 316 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 4.7 Do you think women/men [Enumerator: Delete as appropriate, that is, opposite to respondent] get better, equal, or worse treatment in these systems of justice compared to yourself? 1 A lot better 2 A little better 3 Equally 4 A little worse 5 A lot worse A: [Code] B: [Code] C: [Code] D: [Code] E: [Code] 4.8 Do you think other groups of people, for instance __________ , [Enumerator: Insert as appropriate, that is, different group to respon- dent] get better, equal or worse treatment in these systems of justice compared to yourself? 1 A lot better 2 A little better 3 Equally 4 A little worse 5 A lot worse A: [Code] B: [Code] C: [Code] D: [Code] E: [Code] APPENDIX 6 317 4.9 How easy is it for you to seek and access justice using these systems should you need to? 1 Very easy 2 Fairly easy 3 Fairly difficult 4 Very difficult 5 Impossible A: [Code] B: [Code] C: [Code] D: [Code] E: [Code] 4.10 How active are you in complaining about the systems of justice that you mentioned above? 1 Very active 2 Fairly active 3 A little bit active 4 Not active at all A: [Code] B: [Code] C: [Code] D: [Code] E: [Code] 4.11 How effective are your complaints about the systems of justice that you mentioned above? 1 Very effective 2 Fairly effective 3 A little bit effective 4 Not at all effective A: [Code] B: [Code] C: [Code] D: [Code] E: [Code] 318 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 4.12 How independent of government or politicians or other powerful people do you feel the police force is? 1 Very independent 2 Fairly independent 3 Not independent 4 Would rather not say [If possible, specify reason and add code: ________________________] 4.13 How confident do you feel that corrupt people will face justice? 1 Very confident 2 Fairly confident 3 Not confident 4 Would rather not say [If possible, specify reason and add code: ________________________] Domain/Subdomain: State/Political 4.14 How often are elections usually held to choose your local, regional, and national government and administrations? 1 Never 2 Not held on a regular basis at all 3 Every two to three years 4 Every four to five years 5 Every six to seven years 6 At intervals greater than seven years 7 Do not know Local Regional National APPENDIX 6 319 4.15 How interested are you in these different elections? 1 Very interested 2 Fairly interested 3 Slightly interested 4 Not interested at all Local Regional National 4.16 Were you entitled to vote in the last elections that were held at these levels? 1 Yes 2 No [Go to question 4.20] Local Regional National 4.17 Did you vote in the last elections that were held at these levels? (+) 1 Yes 2 No 3 Would rather not say [If possible, specify reason and add code: ________________________] Local Regional National 320 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 4.18 Did you want to vote in the last elections held at these levels? [go to 4.19] 1 Yes 2 No 3 Would rather not say [If possible, specify reason and add code: ________________________] Local Regional National 4.19 If you vote in an election, whom do you decide with when choos- ing which candidate to support at the following levels? 1 I decide by myself 2 I decide with my spouse 3 I decide with another family member [Specify and add code: ________________________] 4 A community leader helps me decide 5 My employer helps me decide 6 A government official helps me decide 7 A member of a political party contacts me 8 Would rather not say [If possible, specify reason and add code: ________________________] Local Regional National APPENDIX 6 321 4.20 Do your tribal, social, or religious leaders ever discuss election can- didates at the following levels with you? 1 Very often 2 Fairly often 3 Sometimes 4 Never Local Regional National 4.21 Do your tribal, social, or religious leaders ever discuss election can- didates at the following levels with others? 1 Very often 2 Fairly often 3 Sometimes 4 Never Local Regional National 4.22 Do they ever tell you who they will vote for in the elections at the following levels? 1 Very often 2 Fairly often 3 Sometimes 4 Never Local Regional National 322 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 4.23 Have you ever changed your mind when you voted at the follow- ing levels because of discussions with other people (such as tribal, social, or religious leaders)? 1 Yes 2 No 3 Would rather not say [If possible, specify reason and add code: ________________________] Local Regional National 4.24 How involved in the political process at these levels do you feel you are at the moment? 1 Very involved 2 Fairly involved 3 Slightly involved 4 Not involved at all 5 Would rather not say [If possible, specify reason and add code: ________________________] Local Regional National APPENDIX 6 323 4.25 Would you like to be more or less involved in the political process than you are at the moment? 1 Much more involved 2 A little more involved 3 Neither more nor less involved 4 A little less involved 5 Much less involved Local Regional National 4.26 How many representatives of national political parties or move- ments have you heard of in you local area? 1 Many 2 Several 3 One 4 None 5 Would rather not say [If possible, specify reason and add code: ________________________] 4.27 How much power or influence do you think your local elected representative at each level has in the political process? 1 A lot of power/influence 2 Some power/influence 3 No power or influence 4 Would rather not say [If possible, specify reason and add code: ________________________] Local Regional National 324 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 4.28 Overall, how fair do you think the electoral process is at each level? 1 Very fair 2 Reasonably fair 3 Not fair 4 Would rather not say [If possible, specify reason and add code: ________________________] Local Regional National 4.29 Have you ever been dissatisfied with the way that your elected representative behaves? 1 Most of the time 2 Some of the time 3 Rarely 4 Never 5 Would rather not say [If possible, specify reason and add code: ________________________] Local Regional National 4.30 Are there ways of holding him or her accountable? 1 Yes 2 No 3 Would rather not say [If possible, specify reason and add code: ________________________] Local Regional National APPENDIX 6 325 4.31 Have you ever used these? 1 Often 2 Sometimes 3 Never [Go to question 4.33] 4 Would rather not say [Go to question 4.33] [If possible, specify reason and add code: ________________________] Local Regional National 4.32 If yes, did they work? 1 Yes 2 Some impact 3 Little impact 4 No impact 5 Would rather not say [If possible, specify reason and add code: ________________________] Local Regional National Domain/Subdomain: State/Service Delivery 4.33 What publicly provided services [Give examples such as education, health, and the like] are generally available to people in your area? 1 Primary schools 2 Secondary schools 3 Medical clinic 4 Hospital 5 Agricultural extension 6 Transportation 7 Water supply 326 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 8 Sanitation services 9 Waste disposal services 10 Electricity supply 11 Other [Specify and add code: ________________________] 4.34 What publicly provided services do you feel you personally can have access to should you need them? 1 Primary schools 2 Secondary schools 3 Medical clinic 4 Hospital 5 Agricultural extension 6 Transportation 7 Water supply 8 Sanitation services 9 Waste disposal services 10 Electricity supply 11 Other [Specify and add code: ________________________] APPENDIX 6 327 4.35 Which publicly provided services listed above do you use? 4.36 How would you rate the general quality of the publicly provided services you use? 1 Very good 2 Fairly good 3 Neither good nor bad 4 Fairly bad 5 Very bad 4.37 What other public services are provided by the state but you do not have access to? 1 Primary schools 2 Secondary schools 3 Medical clinic 4 Hospital 5 Agricultural extension 6 Transportation 7 Water supply 8 Sanitation services 9 Waste disposal services 10 Electricity supply 11 Other [Specify and add code: ________________________] 328 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 4.38 Have you individually ever made a complaint to the authorities regarding the delivery of public services? 1 Yes 2 No 4.39 Have any members of your household ever made a complaint to the authorities regarding the delivery of public services? 1 Yes 2 No [Go to question 4.44] 4.40 How many times have you made a complaint in the last? Month 6 months Year 3 years 4.41 How successfully do you feel your complaint was resolved? 1 Completely successfully 2 Fairly successfully 3 Slightly successfully 4 Not at all successfully 4.42 Do you think that the authorities are more or less effective when addressing other people's needs and concerns compared to yours? 1 Much more effective 2 Slightly more effective 3 Neither more nor less effective 4 Slightly less effective 5 Much less effective APPENDIX 6 329 4.43 Do you feel the way in which the authorities treat people is affected by people's ethnicity or religion (or other social characteristic)? 1 Yes, very much 2 Yes, slightly 3 No, not at all 4 Would rather not say [If possible, specify reason and add code: ________________________] Domain/Subdomain: Market/Labor At the start of the questionnaire you described your main occupation or livelihood as [enter code] _____________________. We would like to ask you some questions about your occupation and work. 4.44 How much choice do you feel you have in deciding your occupation? 1 Complete choice 2 Some choice 3 No choice 4.45 How easy would it be to change your occupation if you wanted to? 1 Very easy 2 Fairly easy 3 Not very easy 4 Impossible to change 4.46 Why would it be easy or not easy [Enumerator: See above and delete as appropriate] to change your occupation? 1 Lack skills 2 No local alternatives 3 Occupation is determined by caste 4 Other [Specify and add code: ________________________] 330 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 4.47 Do you ever do any work within the household? 1 Yes 2 No [Go to question 4.51] 4.48 When you are at home what household work do you do? 1 Childcare 2 Laundry 3 Cooking 4 Cleaning 5 House maintenance and repair 6 Collecting water 7 Collecting firewood or fuel 8 Other [Specify and add code: ________________________] A B C D E 4.49 How often do you do this work? 1 Every day 2 Every few days 3 Every week 4 Once a month 5 Every few months 6 Other [Specify and add code: ________________________] A B C D E APPENDIX 6 331 Chores/ Every few Every Once a Every few Frequency Every day days week month months Other Childcare Laundry Cooking Cleaning House maintenance Collecting water Collecting firewood/fuel Other 4.50 What household work would you never do? 1 Childcare 2 Laundry 3 Cooking 4 Cleaning 5 House maintenance or repair 6 Collecting water 7 Collecting firewood and fuel 8 Other [Specify and add code: ________________________] A B C D E 332 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Domain/Subdomain: Market/Goods Previously you stated that you _________________ [Enumerator: Enter code as applicable, such as own, rent, and the like] land or property. 4.51 Have you ever felt threatened with eviction from this land or prop- erty? 1 Yes, very often 2 Yes, fairly often 3 Yes, occasionally 4 No, never 4.52 How strongly do you feel the authorities would protect you if some- body tried to make you leave your property or land? 1 Very strongly 2 Fairly strongly 3 Not at all 4.53 Are there any restrictions on what you are able to own or rent? 1 Yes 2 No [Go to question 4.55] 4.54 Why do you think there are restrictions on what you can own or rent? 1 Your gender 2 Your ethnicity 3 Your age 4 Your tribe 5 Your caste 6 Your religion 7 Other [Specify and add code: ________________________] APPENDIX 6 333 4.55 Haveyoueverpersonallyinheritedanyland,property,orotheritems? 1 Yes 2 No 4.56 Have your brothers or sisters ever inherited any land, property, or other items? 1 Yes, brothers 2 Yes sisters 3 Yes, brothers and sisters 4 No 4.57 Who is traditionally allowed to inherit land, property, or other assets? 1 All family members 2 Male family members only 3 Female family members only 4 Other [Specify and add code: ________________________] Domain/Subdomain: Market/Private Services (primarily credit but could also refer to some elements of public credit) 4.58 Did you feel the need to borrow goods or money in the past year? 1 Yes, very often 2 Yes, fairly often 3 Yes, sometimes 4 No, not at all 4.59 Did you actually borrow money or goods in the past year? 1 Yes 2 No 334 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 4.60 How many sources of credit do you think you have access to, includ- ing informal sources? 4.61 Which two sources do you most usually borrow from? 1 Bank 2 Credit association 3 Shopkeeper 4 Landlord 5 Family 6 Other [Specify and add code: ________________________] Source A Source B 4.62 Why do you choose to borrow from this or these source(s)? 1 Close location 2 Interest rates 3 Easy requirements and procedures 4 No formal requirements or procedures 6 Other [Specify and add code: ________________________] Source A Source B 4.63 Are there any other sources of credit for people in your area that you feel are not available to you? 1 Yes 2 No [Go to question 4.65] APPENDIX 6 335 4.64 Why are these not accessible by you? 1 Lack of collateral 2 No guarantor 3 Interest rates too high 4 Culturally unacceptable 5 Other [Specify and add code: ________________________] 4.65 Do you have any savings? 1 Yes 2 No [Go to question 4.67] 4.66 How do you decide when the savings will be used and what for? 1 I decide on my own 2 I decide jointly with my spouse 3 My husband/wife decides for me/us 4 Another household member decides 5 Other [Specify and add code: ________________________] 336 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Domain/Subdomain: Society, Household, and Kinship Groups 4.67 When decisions are made regarding the following aspects of house- hold life, who is it that normally makes the decision? 1 Male head of household 2 Adult male household members 3 Female head of household 4 Adult female household members 5 Male and female heads of households 6 All adult members of household 7 All members of household, including children 8 Other [Specify and add code: ________________________] Household Education Political Marriage Religious expenditure and health decisions choices beliefs 4.68 To what degree do you feel you can make your own personal deci- sions regarding these issues if you want to? 1 To a very high degree 2 To a fairly high degree 3 To a small degree 4 Not at all Household Education Political Marriage Religious expenditure and health decisions choices beliefs APPENDIX 6 337 4.69 To what degree do you feel you have control over decisions regard- ing your own personal welfare, health, and body? 1 To a very high degree 2 To a fairly high degree 3 To a small degree 4 Not at all 4.70 Where do you go on your own? 1 Everywhere I want to 2 Most places I want to 3 Some places I want to 4 Nowhere 4.71 How easy do you find it to access health services when you need to? 1 Very easy 2 Fairly easy 3 Fairly difficult 4 Very difficult 5 Impossible 4.72 How easy do you find it to access education or training services when you need to?? 1 Very easy 2 Fairly easy 3 Fairly difficult 4 Very difficult 5 Impossible 338 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 4.73 Looking back over the past year, do you feel more or less has been spent on your personal health care compared to other household members? 1 Much more 2 A little more 3 About the same 4 A little less 5 Much less 6 We all spend whatever is needed for our care 7 Not sure 4.74 Do you think this is generally the case each year? 1 Yes 2 No Domain/Subdomain: Society/Community 4.75 Who makes the main decisions about public services in your community? [Enumerator: List and add code] A: [Code] B: [Code] C: [Code] 4.76 How involved do you feel in these decision-making processes within your community? 1 Very involved 2 Fairly involved 3 Slightly involved 4 Not involved at all APPENDIX 6 339 4.77 How much would you like to be involved in these decision-making processes within your community? 1 Much more involved 2 Slightly more involved 3 Neither more nor less involved 4 Slightly less involved 5 Much less involved 4.78 How much influence do you feel you have in community level decision-making processes? 1 A great deal of influence 2 A reasonable level of influence 3 A low level of influence 4 No influence at all Section 5: End Comments and Feedback Thank you very much for taking part in this survey, which has taken ________ [Enumerator: fill in as appropriate] hours of your time. We would like to ask you some final questions and would appreciate any comments you have about the survey or the way it was conducted. 5.1 What would you have normally been doing at this time? [Enumerator: List and add code] A: [Code] B: [Code] 5.2 Would you be willing to take part in a similar survey in the future? 1 Yes 2 No 3 Possibly 340 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION 5.3 Is there anything you would like to ask me or the research team? [Enumerator: List and add code] A: [Code] B: [Code] 5.4 Do you have any other comments or suggestions you would like to add about the survey and our research? [Enumerator: List and add code] A: [Code] B: [Code] Once again, thank you very much for your time and effort. Bibliography Abers, R. 2000. Inventing Local Democracy: Grassroots Politics in Brazil. Boulder, CO: Lynne Rienner. Acharya, M., and L. Bennett. 1981. "Rural Women of Nepal: An Aggregate Analysis and Summary of 8 Village Studies." In The Status of Women in Nepal, vol. 2, Field Studies, part 9. Kathmandu: Centre for Economic Devel- opment and Administration, Tribhuvan University. Albertyn, R. 2001. "How to Approach and Design a Measuring Exercise." Paper presented at Working Seminar on Methods for Measuring Women's Empowerment in a Southern African Context, Kvinnoforum and Namibia National Association of Women in Business, Windhoek, Namibia, October 17­18. Alkire, S. Forthcoming. "Subjective Quantitative Measures of Human Agency." Social Indicators Research. Alsop, R. 2004a. "Local Organizations in India: Their Roles and Relationships." SouthAsia Environment and Social Unit, World Bank, Washington, DC. ------. 2004b. "Local Organizations in Nepal:AFive Sector Study of District and Sub-district Organizations." South Asia Environment and Social Development Unit, World Bank, Washington, DC. ------. ed. 2005. Power, Rights, and Poverty: Concepts and Connections. Wash- ington, DC: World Bank. Alsop, R., and J. Farrington. 1998. "Nests, Nodes and Niches: ASystem for Process Monitoring, Information Exchange and Decision Making for Multiple Stakeholders." World Development 26 (2): 249­60. Alsop, R., and N. Heinsohn. 2005. "Measuring Empowerment: Structuring Analysis and Framing Indicators." Policy Research Working Paper 3510, World Bank, Washington, DC. Alsop, R., and B. Kurey. 2004. "Empowerment in Ethiopia:AStatus Review." Poverty Reduction Group, Poverty Reduction and Economic Manage- ment Network, World Bank, Washington, DC. Alsop, R., A. Krishna, and D. Sjoblom. 2001. "Inclusion and Local Elected Governments: The Panchayat Raj System in India." Social Development Paper 37, World Bank, Washington, DC. 341 342 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Appadurai,A. 2004. "The Capacity toAspire: Culture and the Terms of Recog- nition." In Culture and Public Action, ed. V. Rao and M. Walton, 59­84. Stanford, CA: Stanford University Press. Atkinson, A. B., and J. E. Stiglitz. 1980. Lectures on Public Economics. New York: McGraw-Hill. Bachrach P., and M. Baratz. 1962. "The Two Faces of Power." American Politi- cal Science Review 56 (4): 947­52. Baiocchi, G., ed. 2005. Radicals in Power: The Workers'Party (PT) and Experi- ments in Urban Democracy in Brazil. London: Zed. Baiocchi, G., P. Heller, S. Chaudhuri, and M. Silva. 2005. "Measuring and Eval- uating Empowerment: Participatory Budgeting in Brazilian Municipal- ities." World Bank, Brasilia. Baker, J. 2000. Evaluating the Impact of Development Projects on Poverty: A Handbook for Practitioners. Directions in Development Series. Washington, DC: World Bank. Bamberger, M., ed. 2000. Integrating Qualitative and Quantitative Research in Development Projects. Directions in Development Series. Washington, DC: World Bank. Barahona, C., and S. Levy. 2002. "How to Generate Statistics and Influence Policy Using Participatory Methods." Working Paper, Statistical Services Centre, University of Reading, Reading, UK. Barron, P., and D. Madden. 2003. "Violent Conflict in `Non-Conflict' Regions: The Case of Lampung, Indonesia." East Asia and Pacific Region, World Bank, Washington, DC. Barron, P., S. Clark, andA. Marwadi. 2004. "The Links between KDPand Local Conflict: Preliminary Results from a Key Informant Survey in East Java and NTT." World Bank, Jakarta. Barron, P., R. Diprose, and M. Woolcock. 2005. "Local Conflict and Commu- nity Development in Indonesia:Assessing the Impact of the Kecamatan Development Program." World Bank, Jakarta. Barron, P., C. Smith, and M. Woolcock. 2004. "Understanding Local Level Con- flict Pathways in Developing Countries: Theory, Evidence, and Implica- tions from Indonesia." Conflict Prevention and Reconstruction Working Paper 19, World Bank, Washington, DC. Barron, P., R. Diprose, D. Madden, C. Smith, and M. Woolcock. 2004. "Do Participatory Development Projects Help Villagers Manage Local Level Conflicts?AMixed MethodsApproach toAssessing the Kecamatan Devel- opmentProject,Indonesia."ConflictPreventionandReconstructionWork- ing Paper 9 (revised edition). Social Development Department, World Bank, Washington, DC. Bartle, P. 2003. "Measuring Community Empowerment." World Bank Empow- ermentLibrary,http://www.scn.org/ip/cds/cmp/modules/mea-int.htm. Bastagli, F.,A. Coudouel, and G. Prennushi. 2004. "Poverty Monitoring Guid- ance Note 1: Selecting Indicators." Poverty Reduction Group, Poverty BIBLIOGRAPHY 343 Reduction and Economic Management Network, World Bank, Washing- ton, DC. Becker, G. 1995. "The Economic Way of Looking at Behavior." In The Essence of Becker,ed.R.FebreroandP.Schwartz,633­58.Stanford,CA:HooverPress. Bennett, L. 2003. "Empowerment and Social Inclusion:ASocial Development Perspective on the Cultural and Institutional Foundations of Poverty Reduction." Social Development Strategy Paper, Social Development Department, World Bank, Washington, DC. ------. 2004. "Nepal: Gender and Social ExclusionAssessment." World Bank, Kathmandu. Bennett, L., and K. Gajural. 2005. "Negotiating Social Change in Rural Nepal: Crosscutting Gender, Caste and Ethnic Dimensions of Empowerment and Social Inclusion." World Bank, Kathmandu. Besley, T. 2004. "The New Political Economy." Paper presented at the Keynes Lecture in Economics, the British Academy, London, October 23. Besley,T.,andR.Burgess.2002."ThePoliticalEconomyofGovernmentRespon- siveness: Theory and Evidence from India." Quarterly Journal of Economics 117 (4): 1415­52. Biggs, S. D., S. M. Gurung, and D. Messerschmidt. 2004. "An Exploratory Study of Gender, Social Inclusion and Empowerment through Develop- ment Groups and Group-Based Organizations in Nepal: Building on the Positive." Report submitted to the Gender and Social Exclusion Assess- ment Study, World Bank and UK Department for International Develop- ment, Kathmandu. Booth, D., and H. Lucas. 2001. "Good Practice in the Development of PRSP Indicators and Monitoring Systems." Working Paper 172, Overseas Devel- opment Institute, London. Booth, D., and X. Nsabagasani. 2005. "Poverty Monitoring Systems:AnAnaly- sis of InstitutionalArrangements in Uganda." Working Paper 246, Over- seas Development Institute, London. Bourdieu, P. 1977. Outline of a Theory of Practice. Cambridge: Cambridge Uni- versity Press. ------. 1986. "The Forms of Capital." In Handbook of Theory and Research for the Sociology of Education, ed. J. G. Richardson, 241­58. New York: Greenwood Press. Bourguignon, F., F. Ferreira, and M. Menendez. 2003. "Inequality of Out- comes and Inequality of Opportunities in Brazil." World Bank Working Paper 3174, World Bank, Washington, DC. Brown, S. 2005. "Applying Q Methodology to Empowerment." In Measuring Empowerment: Cross-Disciplinary Perspectives, ed. D. Narayan 197­215, Washington, DC: World Bank. Carvalho, S., and H. White. 1997. "Combining the Quantitative and Qualita- tiveApproaches to Poverty Measurement andAnalysis." Technical Paper 366, World Bank, Washington, DC. 344 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Central Statistical Authority. 2001. Ethiopia Demographic and Health Survey 2000. Addis Ababa: Central Statistical Authority; Calverton, MD: ORC Macro. Christiaensen, L., and H. Alderman. 2004. "Child Malnutrition in Ethiopia: Can Maternal KnowledgeAugment the Role of Income?" Economic Devel- opment and Cultural Change 52 (2): 287­312. Clegg, I. 1989. Frameworks of Power. London: Sage. Cohen, J., andA.Arato. 1995. Civil Society and Political Theory. Cambridge, MA: MIT Press. Coleman, J. 1990. The Foundations of Social Theory. Cambridge, MA: Harvard University Press. Cox, M., and N. Thornton. 2005. "Poverty Monitoring Systems:AnAnalysis of InstitutionalArrangements."Agulhas Development Consultants, London. Crocker, L. M., and J. Algina. 1986. Introduction to Classical and Modern Test Theory. New York: Holt, Rinehart, and Winston. Csaszar, F. 2004. "Understanding the Concept of Power." In Alsop 2004b, 137­46. Dahl, R. 1961. Who Governs? Democracy and Power in an American City. New Haven: Yale University Press. DFID (Department for International Development). 2000. "Realising Human Rights for People." DFID, London. Di Gropello, E., and N. Heinsohn. 2004. "School Based Management in Cen- tralAmerica: WhatAre the Effects on Community Empowerment?" Latin America and Caribbean Education Sector, World Bank, Washington, DC. Diprose, R. 2004. "East Java Provincial Report." World Bank, Jakarta. Dollar,D.,andA.Kraay.2000."PropertyRights,PoliticalRights,andtheDevel- opmentofPoorCountriesinthePost-ColonialPeriod."DevelopmentEco- nomics Research Group, World Bank, Washington, DC. http://www. worldbank.org/research/growth/pdfiles/GIGFTP3.pdf. ------. 2002. "Growth Is Good for the Poor." Journal of Economic Growth 7 (3): 195­225. Du Toit,A. 2003. "Hunger in the Valley of Fruitfulness: Globalization, Social Exclusion and Chronic Poverty in Ceres, SouthAfrica." Paper presented at the conference on Staying Poor: Chronic Poverty and Development Policy, Institute for Development Policy and Management, University of Manchester, UK. April 2003. Elster, J., ed. 1986. Rational Choice. Oxford: Blackwell. Enterplan and CDS. 2004. "Inception Report: Provision of Consultancy Sup- port to the Ghana Poverty Reduction Strategy Monitoring and Evalua- tion Division." Enterplan, Reading, UK; Centre for Development Studies, Swansea, UK. Evans,A. 1991. "Gender Issues in Rural Household Economics." IDS Bulletin 22 (1): 51­9. BIBLIOGRAPHY 345 Evans, P. 2002. "Collective Capabilities, Culture, andAmartya Sen's Develop- ment as Freedom." Studies in Comparative International Development 37 (2): 54­60. ------. 2004. "Development as Institutional Change: The Pitfalls of Mono- cropping and the Potentials of Deliberation." Studies in Comparative Inter- national Development 38 (4): 30­52. Eyben, R. 2003. The Rise of Rights: Rights-Based Approaches to International Development. IDS Policy Briefing Issue 17, Institute of Development Stud- ies, University of Sussex, Brighton, UK. ------. 2004. "Linking Power and Poverty Reduction." In Alsop 2004b, 15­28. Fischer, M., D. Zeitlyn, J. Bagg, S. Bharwani, and G. Martin. 2004. "Research in Interactive Ethnography. Session on Interactive Research Methods: Cre- ation, Use and Transmission of Ethnoecological Knowledge." Paper pre- sentedatESRCResearchMethodsFestival,St.Catherine'sCollege,Oxford, UK,July1­3.http://www.ccsr.ac.uk/methods/festival2004/programme/ Fri/am/A/. Folbre, N. 1996. "Hearts and Spades: Paradigms of Household Economics." World Development 14 (2): 245­55. Foucault, M. 1980. Power/Knowledge: Selected Interviews and Other Writings 1972­1977. Ed. C. Gordon. Brighton, UK: Harvester Press. Fox, J. 1996. "How Does Civil Society Thicken? The Political Construction of Social Capital in Rural Mexico." World Development 24 (6): 1089­103. Fung, A. 2002. "Collaboration and Countervailing Power: Making Partici- patory Governance Work." Working Paper, Center for Business and Gov- ernment, John F. Kennedy School of Government, Harvard University, Cambridge, MA. www.archonfung.net. Fung, A., and E. Wright. 2003. Deepening Democracy: Institutional Innovations in Empowered Participatory Governance. London: Verso. Geertz, C. 1977. The Interpretation of Cultures. New York: Basic Books. George, A., and A. Bennett. 2005. Case Studies and Theory Development in the Social Sciences. Cambridge, MA: MIT Press. Gibson, C., and M. Woolcock. 2005. "Empowerment and Local Level Conflict Mediation in the Kecamatan Development Project in Indonesia: Con- cepts, Measures and Project Efficacy." Policy Research Working Paper 3292, World Bank, Washington, DC. Giddens,A. 1984. The Constitution of Society: Outline of the Theory of Structura- tion. Cambridge: Polity Press. ------. 1993. "Profiles and Critiques in Social Theory." In The Giddens Reader, ed. P. Cassel, 218­25. London: Macmillan. Goetz,A. M., and R. S. Gupta. 1996. "Who Takes the Credit? Gender, Power, and Control over Loan Use in Rural Credit Programs in Bangladesh." World Development 24 (1): 45­63. 346 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION GovernmentofEthiopia.2002."PovertyProfileofEthiopia."MinistryofFinance and Economic Development, Addis Ababa. Graham, C., and S. Pettinato. 2005. "Subjective Well-Being and Objective Mea- sures: Insecurity and Inequality in Emerging Markets." In Measuring Empowerment: Cross-Disciplinary Perspectives, ed. D. Narayan, 141­75, Washington, DC: World Bank. Gramsci, A. 1971. Selections from the Prison Notebooks. Ed. Q. Hoare and G. Nowell-Smith. London: Lawrence & Wishart. ------. 1978. Antonio Gramsci: Selections from Political Writings (1921­1926). Trans. and ed. Q. Hoare. London: Lawrence & Wishart. Grootaert, C. 2005. "Assessing Empowerment at the National Level in Eastern Europe and Central Asia." In Measuring Empowerment: Cross- Disciplinary Perspectives, ed. D. Narayan, 309­40, Washington, DC: World Bank. GTZ (GermanAgency for Technical Cooperation ). 2004. "National Monitoring of Strategies for Sustainable Poverty Reduction/PRSP," vol. 1. Eschborn, Germany. Guggenheim, S. Forthcoming. "Local Institutions and Local Development in Indonesia." In The Search for Empowerment: Social Capital as Idea and Practice at the World Bank, ed. A. Bebbington, M. Woolcock, S. Guggenheim, and E. Olson. Bloomfield, CT: Kumarian Press. Guyer, J., and P. Peters, eds. 1987. "Conceptualising the Household: Issues of Theory and Policy inAfrica." Special issue, Development and Change 18 (2). Hall, P., and R. Taylor. 1996. "Political Science and the Three New Institu- tionalisms." Discussion Paper 96/6, Max Planck Institute for the Study of Societies (MPIfG), Köln, Germany. Haugaard, M. 2002. Power: AReader. Manchester, UK: Manchester University Press. Heifetz, R. 1994. Leadership Without Easy Answers. Cambridge, MA: Harvard University Press. Heinsohn, N. 2004. "Mexico Lifelong Learning Project." Mission Report: Empowerment Impact Evaluation. Poverty Reduction Group, Poverty Reduction and Economic Management Network, World Bank, Washing- ton, DC. Hirschman, A. 1970. Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States. Cambridge, MA: Harvard University Press. Holland, J., and J. Campbell. 2005. Methods in Development Research: Combin- ing Qualitative and Quantitative Approaches. London: ITDG Publications. Holland, J., S. Noble, A. Norton, and K. Sigrist. 2005. "Monitoring Social Policy Outcomes in Jamaica: Combined Methods, Democratic Research and Institutional Change." In Methods in Development Research: Combining Qualitative and Quantitative Approaches, J. Holland and J. Campbell, 219­26. London: ITDG Publications. BIBLIOGRAPHY 347 Holtom, D. 2004. "Coercion and Consent: The World Bank in Tanzania, 1970­2001." PhD diss., Centre for Development Studies, University of Wales, Swansea, UK. Horowitz, D. 1985. Ethnic Groups in Conflict. Berkeley: University of Cali- fornia Press. Hughes,A., and J. Wheeler, with R. Eyben and P. Scott-Villiers. 2003. "Rights and Power Workshop Report." Institute of Development Studies, University of Sussex, Brighton, UK. Kabeer, N. 1994. Reversed Realities: Gender Hierarchies in Development Thought. London: Verso. ­----. 1998. "`Money Can't Buy Me Love?' Re-evaluating Gender, Credit and Empowerment in Rural Bangladesh." Discussion Paper 363, Insti- tute of Development Studies, University of Sussex, Brighton, UK. ­----. 1999. "Resources,Agency,Achievements: Reflections on the Measure- ment of Women's Empowerment." Development and Change 30: 435­64. Kaufmann, D. 2000. "Unbundling Governance and Corruption: Some New Empirical Findings and CollectiveActionApproaches." Paper presented at Anti-Corruption Summit, World Bank Institute, Washington, DC, September 21. Kaufmann, D., A. Kraay, and M. Mastruzzi. 2004. "Governance Matters III: Governance Indicators for 1996­2002." World Bank Economic Review 18 (2): 253­87. ------. 2005. "Governance Matters IV: Governance Indicators for 1996­2004." World Bank Institute, Washington, DC. http://www.worldbank.org/ wbi/governance/pubs/govmatters4.html. Kaufmann, D., A. Kraay, and P. Zoido-Lobatón. 1999. "Aggregating Gover- nance Indicators." Policy Research Working Paper 2196, World Bank, Washington, DC. KDP(Kecamatan Development Program). 2003. "Indonesia: Kecamatan Devel- opment Program." KDP National Secretariat and National Management Consultants, Jakarta. King, G., and J. Wand. 2004. "Comparing Incomparable Survey Responses: New Tools for Anchoring Vignettes." Harvard University, Cambridge, MA. http://gking.harvard.edu/files/c.pdf. Krebs, V. 2005. "An Introduction to Social NetworkAnalysis." http://www. orgnet.com/sna.html. Krishna, A. 2003. "Mapping Empowerment in a Comparative Perspective." Consultant report for the Empowerment Team, Poverty Reduction Group, Poverty Reduction and Economic Management Network, World Bank, Washington, DC. Kroeker, C. 1995. "Individual, Organizational, and Societal Empowerment: A Study of the Processes in a Nicaraguan Agricultural Cooperative." American Journal of Community Psychology 23 (5): 749­64. 348 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Kurey, B. 2005. "Empowerment in Ethiopia: Issues and Options." Poverty Reduction Group, Poverty Reduction and Economic Management Net- work, World Bank, Washington, DC. Kvinnoforum. 2001a. "Concepts, Methods and Project Activities." Paper presented at Working Seminar on Methods for Measuring Women's Empowerment in a SouthernAfrican Context, Kvinnoforum and Namibia NationalAssociation of Women in Business, Windhoek, Namibia, Octo- ber 17­18. Kvinnoforum. 2001b. "Summary of Workshop Discussions." Paper presented at Working Seminar on Methods for Measuring Women's Empowerment in a Southern African Context, Kvinnoforum and Namibia National Association of Women in Business, Windhoek, Namibia, October 17­18. Legovini,A.2004."DraftReportfromtheQualitativeEvaluationoftheWomen's Development Initiatives Project, Ethiopia." Poverty Reduction Group, Poverty Reduction and Economic Management Network, World Bank, Washington, DC. ------. 2005. "Measuring Women's Empowerment in Ethiopia." World Bank, Addis Ababa. Lokshin, M., and M. Ravallion. 2005. "Self-Rated Power and Welfare in Russia." In Measuring Empowerment: Cross-Disciplinary Perspectives, ed. D. Narayan, 177­95. Washington, DC: World Bank. Lukes, S. 1974. Power: A Radical View. London: Macmillan. Malena, C., and V. F. Heinrich. 2005. "The CIVICUS Civil Society Index." In Measuring Empowerment: Cross-Disciplinary Perspectives, ed. D. Narayan, 341­64, Washington, DC: World Bank. Malhotra,A.,S.R.Schuler,andC.Boender.2002."MeasuringWomen'sEmpow- erment as a Variable in International Development." Gender and Devel- opment Group, World Bank, Washington, DC. Mansuri, G., and V. Rao. 2004. "Community-Based (and Driven) Develop- ment: A Critical Review." World Bank Research Observer 19 (1): 1­39. Mason, K. O., and H. Smith. 2003. "Women's Empowerment and Social Con- text:ResultsfromFiveAsianCountries."GenderandDevelopmentGroup, World Bank, Washington, DC. McMillan, B., P. Florin, J. Stevenson, B. Kerman, and R. Mitchell. 1995. "Empowerment Praxis in Community Coalitions." American Journal of Community Psychology 23 (5): 699­728. Moncrieffe,J.2004."PowerRelations,InequalityandPoverty:AConceptPaper for the World Bank." Overseas Development Institute, London. Moore, M. 2001. "Empowerment at Last?" Journal of International Development 13 (3): 321­29. Moore, M., and J. Putzel. 1999. "Thinking Strategically about Politics and Pov- erty". Working Paper 101, Institute of Development Studies, University of Sussex, Brighton, UK. BIBLIOGRAPHY 349 Mosedale, S. 2005. "Assessing Women's Empowerment: Toward a Con- ceptual Framework." Journal of International Development 17 (2): 243­58. Moser, C. 1998. "The Asset Vulnerability Framework: Reassessing Urban Poverty Reduction Strategies." World Development 26 (1): 1­19. ------. 2005. "Peace, Conflict, and Empowerment: The Colombian Case." In Measuring Empowerment: Cross-Disciplinary Perspectives, ed. D. Narayan, 247­65, Washington, DC: World Bank. Moser, C., and A. Norton. 2001. "To Claim Our Rights: Livelihood Security, Human Rights, and Sustainable Development." Overseas Development Institute, London. Murthi, M., A. Guio, and J. Drèze. 1995 "Mortality, Fertility and Gender Bias in India." Population and Development Review 21: 745­82. Narayan, D., ed. 2002. Empowerment and Poverty Reduction: A Sourcebook. Washington, DC: World Bank. ------, ed. 2005. Measuring Empowerment: Cross-Disciplinary Perspectives. Washington, DC: World Bank. Nelson,N.,andS.Wright,eds.1995. PowerandParticipatoryDevelopment:Theory and Practice. London: Intermediate Technology Publications. North, D. C. 1990. Institutions, Institutional Change and Economic Performance. Cambridge: Cambridge University Press. Nunnally, J. C., and I. H. Bernstein. 1978. Psychometric Theory. New York: McGraw-Hill. Nussbaum, M. 2000. Women and Human Development: The Capabilities Approach. Cambridge: Cambridge University Press. O'Loughline, B. 1999. "In Defense of the Household: Marx, Gender and the Utilitarian Impasse." Working Paper 289, Institute of Development Stud- ies, University of Sussex, Brighton, UK. Parsons, T. 1937. The Structure of Social Action. New York: McGraw-Hill. ------. 1963. "On the Concept of Political Power." Proceedings of the American Philosophical Society 107 (3): 232­62. Pasteur, K. 2003. "Organisational Learning: Challenges andActions for Donor Organisations." Institute of Development Studies, University of Sussex, Brighton, UK. Peters,T.1989.ThrivingonChaos:HandbookforaManagementRevolution.London, Pan Books. Petesch, P., C. Smulovitz, and M. Walton. 2005. "Evaluating Empowerment: A Framework with Cases from Latin America." In Measuring Empowerment: Cross-Disciplinary Perspectives, ed. D. Narayan, 39­67, Washington, DC: World Bank. Putnam, R. D. 1993. Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ: Princeton University Press. Raich, U. 2005. "The Fiscal Determinants of Empowerment." Policy Research Working Paper 3705, World Bank, Washington, DC. 350 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Rajan, R. 2004. "AssumeAnarchy? Why an Orthodox Economic Model May NotBetheBestGuideforPolicy."Finance&Development,September,56­57. Rao, V., andA. M. Ibáñez. 2005. "The Social Impact of Social Funds in Jamaica: A`Participatory Econometric' Analysis of Targeting, Collective Action, and Participation in Community-Driven Development." Journal of Devel- opment Studies 41 (5): 788­838. Rao, V., and M. Walton. 2004. "Culture and Public Action: Relationality, Equality ofAgency, and Development." In Culture and Public Action, ed. V. Rao and M. Walton, 3­36. Stanford, CA: Stanford University Press. Rao, V., and M. Woolcock. 2003. "Integrating Qualitative and Quantitative Approaches in Program Evaluation." In The Impact of Economic Policies on Poverty and Income Distribution: Evaluation Techniques and Tools, ed. F. Bourguignon and L. Pereira da Silva. NewYork: Oxford University Press. Rasyid, I. 2004. "KDP Implementation and Its Spillover Effects on the Lives of the Villagers in Ponorogo: Key Informant Survey Data." World Bank Office, Jakarta. Rennie, J. K., and N. C. Singh. 1999. AGuidebook for Field Projects: Participatory Research for Sustainable Livelihoods. New York: United Nations Develop- ment Programme. Rowlands, J. 1997. Questioning Empowerment: Working with Women in Honduras. Oxford,UK:Oxfam; AtlanticHighlands,NJ: HumanitiesPressInternational. Schuler, S. R., and S. M. Hashemi. 1994. "Credit Programs, Women's Empow- erment, and Contraceptive Use in Rural Bangladesh." Studies in Family Planning 25 (2): 65­76. Sen, A. 1985. Commodities and Capabilities. Amsterdam: Elsevier. ------. 1992. Inequality Reexamined. Cambridge, MA: Harvard University Press. ------. 1997. "Editorial: Human Capital and Human Capability." World Devel- opment 25 (12): 1959­61. ------. 1999a. Development as Freedom. New York: Knopf. ------. 1999b. "The Possibility of Social Choice." Nobel Lecture. American Economic Review 89 (3): 349­78. Senge, P. 1990. The Fifth Discipline: The Art and Practice of the Learning Organi- sation. New York: Doubleday. Spreitzer, G. M. 1995. "An Empirical Test of a Comprehensive Model of Intra- personal Empowerment in the Workplace." American Journal of Commu- nity Psychology 23 (5): 601­29. Stern, N., J.-J. Dethier, and H. Rogers. 2005. Growth and Empowerment: Making Development Happen. Cambridge, MA: MIT Press. Sunstein, C., and E. Ullmann-Margalit. 1998. "Second-Order Decisions." Ethics 110 (1): 5­31. Swift, J. 1989. "WhyAre Rural People So Vulnerable to Famine?" IDS Bulletin 20 (2): 8­15. BIBLIOGRAPHY 351 Tendler, J. 1997. Good Government in the Tropics. Baltimore: Johns Hopkins Uni- versity Press. Terefe, H. 2002. "Violence Against Women from Gender and Cultural Per- spectives." Reflections: Documentation of the Forum on Gender 7. Panos Ethiopia and Heinrich Boll Foundation, Regional Horn of Africa Office, Addis Ababa. UNDP (United Nations Development Programme). 1995. Human Develop- ment Report 1995: Gender and Human Development. New York: Oxford University Press. Uphoff, N. 1986. Local Institutional Development: An Analytical Sourcebook with Cases. West Hartford, CT: Kumarian Press. ------. 2005. "Analytic Issues in Measuring Empowerment at the Commu- nity and Local Levels." In Measuring Empowerment: Cross-Disciplinary Perspectives, ed. D. Narayan, 219­46, Washington, DC: World Bank. Varshney, A. 2002. Ethnic Conflict and Civic Life: Hindus and Muslims in India. New Haven, CT: Yale University Press. ------. 2005. "Democracy and Poverty." In Measuring Empowerment: Cross- Disciplinary Perspectives, ed. D. Narayan, 383­401, Washington, DC: World Bank. Weber, M. 1904. "The Objectivity of the Sociological and Social-Political Knowledge." In Max Weber on the Methodology of the Social Sciences, trans. and ed. E. A. Shils and H. A. Finch. Glencoe, IL: Free Press, 1949. ------. 1905. The Protestant Ethic and the Spirit of Capitalism. Repr., New York: Scribner's, 1958. Weick, K. E. 1984. "Small Wins: Redefining the Scale of Social Problems." American Psychologist 39 (1): 40­9. Weiss, C. 1998. Evaluation: Methods for Studying Programs and Policies. Upper Saddle River, NJ: Prentice Hall. White, H. 2005. "Challenges in Evaluating Development Effectiveness." In Evaluating Development Effectiveness, ed. G. Pitman, O. Feinstein, and G. Ingram, 33­54. New Brunswick, NJ: Transaction Publishers. Whitehead, A. 1981. "I'm Hungry Mum: The Politics of Domestic Budget- ing." In Of Marriage and the Market: Women's Subordination in International Perspective, ed. K. Young, C. Wolkowitz, and R. McCullagh, 88­111. London: CSE Books. Wilson, I. 2005. "Some Practical Sampling Procedures for Development Research." In Methods in Development Research: Combining Qualitative and Quantitative Approaches, J. Holland and J. Campbell, 37­52. London: ITDG Publication. World Bank. 1997. "Rural Women's Development and Empowerment Pro- ject, India." Staff Appraisal Report 16031-IN. South Asia Environment and Social Unit, World Bank, Washington, DC. 352 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION ------. 1999a. "Guinea Village Communities Support Program: Project Appraisal Document." Report 17934-GUI, Africa Region, World Bank, Washington, DC. ------. 1999b. "Integrated Watershed Development Project: ProjectAppraisal Document." SouthAsia Rural Development Unit, World Bank, Washing- ton, DC. ------. 2000. World Development Report 2000/2001: Attacking Poverty. New York: Oxford University Press. ------. 2002a. "Ethiopia: Summary Gender Profile." GenderStats database, World Bank, Washington, DC. ------. 2002b. Social Analysis Sourcebook: Incorporating Social Dimensions into Bank-Supported Projects. Washington, DC: World Bank. http://www. worldbank.org/socialanalysissourcebook/. ------. 2002c. "Social Funds: Assessing Effectiveness." Operations Evalua- tion Department, World Bank, Washington, DC. ------. 2003. World Development Report 2004: Making Services Work for Poor People. New York: Oxford University Press. ------. 2004a. "The Poverty Reduction Strategy Initiative: An Independent Evaluation of the World Bank's Support through 2003." Operations Evaluation Department, World Bank, Washington, DC. ------. 2004b. "Well-Being and Poverty in Ethiopia: The Role ofAgriculture, Agency and Aid." World Bank, Addis Ababa. ------. 2005a. "Social Accountability in the Public Sector: A Conceptual Discussion and Learning Module." World Bank Institute, Washington, DC. ------. 2005b. "Tamil Nadu Empowerment and Poverty Reduction Project: Project Appraisal Document." South Asia Rural Development Unit, World Bank, Washington, DC. ------. 2005c. World Development Report 2006: Equity and Development. New York: Oxford University Press. Index abduction, 151n variables and outcomes, 162 accountability, 3 WDIP, 127 internal, 77 agency-structure, 1, 9, 11, 26n­27n, 46, KDP, 190 58 social, 22­23 CME, 166 voice, 24 framework operations, 86 adaptive conflict management, interaction, 15­18, 84 183­184 PB, 111f adaptive preferences, 12 agents, 11 adaptive problem-solving, 188 Albania AECO. See Asociación Educativa empowerment themes, 277 Communitaria Albertyn (2001) agency, 9, 14, 23tn measurement approach, 220, 221 and power theory, 236­237 analysis, 4, 12, 29, 39, 40 asset base translation, 16­17 cooperation among researchers, 88 asset indicators, 58 data collection, 157 asset interaction, 12 for poverty assessment, 74 civil society capacity, 108­110 KDP, 177, 183­189 contributing factors, 12 PB, 50, 98, 99f definition, 10, 11, 84, 151n policy-level monitoring systems, 74 investing in, 60, 62­64 Proheco, 169 KDP, 186 reliability, 156 market domain indicators for coun- WDIP, 145 try studies, 266­269 anchoring, 44n PB, 50, 97, 110, 122 Andhra Pradesh District Poverty Proheco indicators for community Initiatives Project, 250, 251 members, 158 Asociación Educativa Communitaria school council indicators, 157­158 (AECO), 152, 159 societal domain indicators for coun- community engagement, 157 try studies, 270­275 indigenous people participation, state domain indicators for country 168 studies, 258­263 legal status, 164, 171n 353 354 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION asset endowment, 11 interaction of caste and gender, 209 Ethiopian women, 130­131 SII, 205­206, 210 indicators, 33, 33t WEI indicators, 211 assets, 11, 57 WEI levels by caste and ethnicity, economic outcomes, 139 203f education, 13b Brazil. See also participatory budgeting Ethiopian women, 125­127 citizen participation, 103t imbalances, 86 PB, 49­50, 121 increasing across caste and gender, state domain indicators, 258, 259 214 Brown (2005) indicators for agency, 58­59 measurement approach, 220, 221 information, 139 interaction, 12 Camaragibe, 101­102, 104t, 112 KDP, 188, 190 agency-structure changes, 111f measurement, 11 citizen participation, 103t opportunity structure, 86 civil society measurement, 108t psychological, 55, 62, 151n, 204­205 mode of intermediation, 109 association interaction, 187t capacity to aspire, 27n associationalism, 97, 108t, 109 carok, 181, 182b attitudes, 27n CAS. See Country Assistance Strategy autarchic power, 187t caste, 203­206 CEI levels for women, 209f Bachrach and Baratz empowerment and social inclusion, power as agenda setting, 234 212­215 social theoretical models of power, opportunity structure, 206 238, 239 WEI, 210­212 bargaining, 3 women's empowerment, 208­210 Ethiopian marital relations, causal attribution, 31, 64, 92 141­142 policy-level monitoring, 76 Bartle (2003) CDD projects. See community-driven measurement approach, 220, 221 development projects BCN. See Brahman-Chhetri-Newar CEI. See Composite Empowerment (BCN) group and Inclusion Index Benin Social Fund Project, 254, 255 choice, 1, x binding participation, 104 capacity to use, 18 Biting, 178­179 conceptualization of, 84 Bolsa Familia Project, 244, 245 effective capacity, 10, 15­16 Brahman-Chhetri-Newar (BCN) exercise of, 72 group, 217n habitual and reasoned, 14 caste hierarchies, 203 implicit, 31­32 CEI levels, 202f inhibitions, 75b EMI score versus Dalits, 203­204 measurement, 17, 34 group membership, 214 PB, 104, 106, 112 INDEX 355 political, 16b literacy and numeracy effects, rational, 14, 27n 164 tracking, 29 participation, 168 citizen engagement agency, 157­158 PB, 102­104, 122 analysis, 169 citizen mobilization, 178 analytic focus, 48t citizen participation, 16b, 24 CME, 165­169, 166t binding, 104 data collection, 41, 154­155 civil society, 97, 98, 121 degrees of empowerment, 166 agency, 108­110 empowerment definition, 47t PB, 50, 106 empowerment equation, 154 Clegg government transfers, 164 frameworks of power, 237 hypotheses and indicator develop- social theoretical models of power, ment, 153, 169 240, 241 impact evaluation, 153 clientelism, 97, 108t indicators, 155­156, 157­158, 159 CME. See community's most excluded indigenous people, 167­168 member index legal status, 164, 171n codes of conduct, 141 lessons learned, 170 collaborative power, 232b literacy and numeracy effects, 164 collective action, 174 men's participation, 167 KDP, 178, 191n methodology, 154­155 MESI, 196 Ministry of Education responsibili- collective empowerment, 85 ties, 152 Community Conflict Negotiation opportunity structure, 159­160, 164, Study, 172 170n­171n Community Development and parent associations, 152 Livelihood Improvement "Gemi participation, 168 Diriya" Project, 246, 247 participation context, 167 community empowerment, 53 power relations, 169­170 community member empowerment, between indigenous and non- 156­157 indigenous people, 168 community participation, 143 program administration, 170n empowerment themes, 277, 279 promoters, 159, 164 Community's Most Excluded Member sample and methodology, 155 Index (CME), 157, 165-169 school council empowerment, agency and structure, 166 155­156, 161t by region, 166t agency indicators, 170n Community-Based Education Project outcomes, 160­165, 160t (Proheco), 45, 52­53, 85, 152 SCE, 161 AECO by number of variables, 163t indicators, 155­156 multiple regression, 162t legal status, 164, 171n school council engagement, 153 356 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION school council variables, 161 adaptive conflict management, school management functions, 153 183­184 study conclusions, 169­170 KDP, 54 study framework, 153­154 Kecamatan Proppo, 180 the poorest, 168­169 procedures, 184­185 variables, 156, 169 reasons for lack of, 187­188 women's participation, 167 routines for local conflict manage- community-driven development ment, 187t (CDD) projects, 172, 188 Country Assistance Strategy (CAS), comparability, 32, 93 81n comparative analysis, 4 country studies Composite Empowerment and analytic focus, 48t Inclusion Index (CEI), 199 design and structure, 45­46 caste and ethnicity, 205f, 212­213 framework application, 49 for women by caste and ethnicity, indicators used, 257­275 209f gender, 207f Dahl gender, caste, and ethnicity, 202f power as decision making, 231, 233 composite indexes, 39 social theoretical models of power, conflict case studies, 177­178 238, 239 conflict evolution, 176 Dalits conflict management, 53 CEI levels, 202f adaptive, 183­184 development services, 215 capacity, 174 EMI versus BCN group, 203­204 collaborative routines, 183, 189 WEI levels by caste and ethnicity, disaggregation, 185­186 203f KDP, 176, 177­178 Data, 39, 87 local routines, 187t availability, 73t marginalized groups, 54, 175 collection, 39 traditional Islamic norms, 181 future analysis, 157 conflict pathways, 41, 44n KDP, 176, 190­191 case studies, 175­176 MESI, 197­198 KDP processes, 177 methods, 40­41 conflict resolution, 185 Proheco, 154 conflicts, everyday, 174­175 sequencing, 38 conflictual power, 230, 235 generation, 80 theories, 231, 233­234 monitoring systems, 74­75 consensual power, 230 PB, 114 theories, 230­231 recording, 39 consultative participation decision-making PB, 106­107, 122 empowerment themes, 278 context, 13, 32, 87 income effects, 143 countervailing power, 175, 186 PB, 103, 104t, 105, 106 INDEX 357 power theory, 231, 233 KDP, 54 degrees of empowerment, 4, 15­18, 18, Madhya Pradesh Poverty Initiatives 27n, 59 Project, 65t across domains, 20 market, 36t agency-structure, 26n­27n MESI study, 198 community, 53 PB, 49 direct measurements, 34 RWDEP, 67b indicators, 51 seven of empowerment, 28n market domain indicators for coun- social measurement, 28n try studies, 266­269 society, 37t measurement, 23tn state, 36t PB, 97 subdomains, 23t Proheco, 166 WDIP, 51 societal domain indicators for coun- domestic decisions, 142­143 try studies, 270­275 domestic violence, 28n, 135, 141­142 state domain indicators for country doxa, 14 studies, 258­265 Sumorobangun Dam, 179 econometric approaches, 40 development goals, 2, 29 economic empowerment themes, development outcomes, 22­25, 83 279 indirect indicators, 27n Ecuador Poverty Reduction and Local interventions, 26 Rural Development Project, 248, PB, 112, 113b, 122 249 development performance education PB effects and indicators, 118t­119t civic participation, 16b Diadema, 100, 104t, 112 effect on other assets, 13b agency-structure changes, 111f increase in empowerment levels, citizen participation, 103t 214 civil society measurement, 108t elite capture, 180, 186 consultative participation, 107 EMI. See Empowerment Index participation, 109 employment, 138­139 discrimination empowerment, 1, 2 Ethiopian judicial system, 133­135 approach integration, 87­88 disempowerment CME, by region, 166t women, 11­12 collective, 85 dispositional power, 237 community members, 156­157 divorce, 133 country study definitions, 47t domains, 19­21, 23t, 86 cross-sectoral nature, 31 concept, 27n­28n definition, 1, 10, 83, 84 Ethiopia, 73t KDP, 173 indicators, 35t, 36t­37t RWSS, 44n Integrated Watershed Development WDIP, 151n Project, 63b World Bank, 195­196 358 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION degrees of (See degrees of empow- abduction, 151n erment) data availability, 73t development outcomes, 83 divorce, 133 lack of research, 87 domains and levels, 73t link to power, 237 employment rate, 126 livelihood, 196, 213 empowerment objectives, 75b low levels, 12 empowerment themes, 277 means or end, 59­60 framework application, 72, 74 measurement (See measurement) gender equality, 33­34 MESI study, 194­195 genital mutilation, 14­15 determinants, 212­215 girls, 137­138 process of, 196 indicators of women's lives, 130b outcomes and process, 3­4 informal institutions, 15b participation, 151n information access, 25b relationship between outcomes and judicial discrimination of women, correlates, 10f 133­135, 135t school councils, 155­156 judicial protection for women, stimulating, 58 126­127 themes and strategies, 277­282 marital relations, 141­142 understanding, 25­26 National Committee on Traditional value, 24 Practices, 141 WDIP, 144­145 National Policy on Women, 126 World Bank projects, 243­255 poor services, 138 Empowerment and Poverty Reduction: A poverty reduction, 73t, 75b Sourcebook, 5n PRSP and SDPRP, 72 Empowerment Base Survey Module, rape, 141, 151n 37, 76 state domain indicators, 258, 259 empowerment framework. See frame- traditional practices, 141 work water availability, 138 Empowerment Index (EMI), 42 WDIP, 250, 251 BCN group versus Dalits, 203­204 women MESI study, 198­199 awareness of rights, 137, 142 gender differentials, 207 codes of conduct, 141 gender gap by caste and ethnic- economic empowerment, 140t ity, 208, 208f employment, 138­139 indicators used, 200b empowerment, 52 endogeneity, 31, 43n happiness, 131, 132t definition, 44n illiteracy and information access, econometric approaches, 40 138 episodic power, 237 low asset endowment levels, equality of agency, 191n 130­131 Ethiopia. See also Women's low empowerment levels, Development Initiatives Project 129­130 INDEX 359 participation in groups, 144, 144t sequencing methods, 37 political engagement, 133, 134t summary, 21­22, 23t protection for, 136 Freedom House, 34, 35t reproductive and sexual deci- sions, 142 Gambia, The right to choose spouse, 137t empowerment themes, 277 rights, 51 gender skills, 137 differentials, 207 variables, 130b EMI, 208, 208f ethnicity equality, 33­34 empowerment and social inclusion, inequality, 125­127 212­215 RWSS, 55­56 RWSS, 55 WEI dimensions, 210­212 ethnographic case studies, 41 genital mutilation, 14­15, 141 ethnographies, 175­176 Ghana exclusion, 97, 108t empowerment themes, 278 Poverty Reduction Strategy, 78 face validity indicators, 34 Giddens facilitative power, 237 agency and power theory, 236­237 false consciousness, 234 social theoretical models of power, farmer empowerment, 279 238, 239 fasi, 3 girls, 137­138 fertility, control of, 211 Goetz and Gupta (1996) financial assets, 33t measurement approach, 220, 221 financial markets, 146­147 governance "fit-for-purpose" data systems, 76 empowerment themes, 277, 279 forums, 186 government participation KDP, 174, 175, 188­189 local, 16b Foucault Gramsci power theory, 235 social theoretical models of power, social theoretical models of power, 238, 239 238, 239 grassroots framework, 47, 49, x empowerment themes, 278 application, 71­72, 74 Gravataí, 101, 112 case studies, 46­47, 49 agency-structure changes, 111f enabling environment, 87 binding participation, 104, 104t operationalizing, 79­81 citizen participation, 103t operations and analysis, 85­87 civil society measurement, 108t overview, 10­11 governance and public services, 105 policy formulation, 71 mode of intermediation, 109 project focus, 81n Grootaert (2005) retrospective use, 64 measurement approach, 222, 223 RWDEP, 67b groups, 214, 215 360 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION growth, 113b, 115 indicators Guinea asset endowment, 33t empowerment themes, 278 development outcomes, 27n Guinea Village Communities Support direct, 17, 34, 36t­37t Program, 61b, 252, 253 Ethiopian women, 130b Guyana ethnic group agency, 204­205 empowerment themes, 279 face validity, 34 for country studies by domain, health, 126 258­275 heterogeneity, 100, 122 identifying, 33­37 HIV/AIDS, 126 indirect, 17, 35t Honduras. See also Community-Based measurement, 30­31 Education Project objective or hypothetical, 37 state domain indicators, 260, 261 outcomes, 76 honor killings, 3 PB, 50, 113b Household Budget Survey, 33t development effects, 118t­119t human assets, 33t Proheco, 152, 155­157, 157­158 human development, 113b RWSS, 55 typology, 33 illiteracy, 138 values, 4 index of progress, 115, 124 WDIP, 51, 125, 129, 146, 150 indexes, 39. See also Community WEI, 210 Members Most Excluded Index; WEI for MESI study, 201, 201b, 204b Composite Empowerment welfare, 114 Index; Empowerment Index; indigenous people, 167­168 School Council Empowerment Indonesia, 85 Index; Social Inclusion Index; development project goals, 87 Women's Empowerment Index state domain indicators, 262, 263 CME, 157, 165 INEA. See Institute for Adult cost-effectiveness, 157 Education limitations, 198 inequality, 10, 58 problems, 43n gender, 125­127 RWSS, 42 PB, 113b, 115 school council empowerment, 156, influence, 106, 183 161 women's on community decisions, tracking institutional progress, 34 145t India information, 38­39. See also data Integrated Watershed Development quality generation, 80 Project, 62b information access, 13b local government participation, 16b addition of source, 27n RWDEP, 66b civic participation, 16b Tamil Nadu Empowerment and empowerment themes, 278 Poverty Project, 59 Ethiopia, 25b INDEX 361 WDIP, 126, 138 development services, 215 information flows WEI levels by caste and ethnicity, monitoring, 80 203f nested, 68, 78, 88 João Monlevade, 101, 104t informational assets, 33t agency-structure changes, 111f Institute for Adult Education (INEA), citizen participation, 103t 13b civil society measurement, 108t institutional context, 13 mode of engagement, 109 institutional inequality, 58 judicial institutions institutional policies, 57 fairness to women, 135t institutional surface area, 97 institutional variables, 139­140 Kabeer institutions, 27n measurement approach, 222, 223 formal and informal, 13­14 Kabupatan Pamekasan, 179­180 indicators, 35t, 58­59 Kabupatan Ponorogo, 178­179 interactions, 206 Kecamatan Badegan, 178­179 operational gulf, 33 Kecamatan Development Project participation, 98 (KDP), 45, 53­55, 183 presence and operation of, 86 accountability, 190 resistance to change, 84­85 adaptive challenge, 177­178 tracking progress, 34 adaptive problem-solving, 188 Integrated Questionnaire for the agency challenge, 186 Measurement of Social Capital analysis, 177, 183­189 (IQMSC), 33t analytic focus, 48t Integrated Watershed Development association interaction style, 187t Project, 62b Biting, 178­179 interactive fieldwork, 128 block grants, 172­173 interactive research, 38, 41, 43n CDD projects, 188 interventions, 63 citizen mobilization, 178 development outcomes, 26 collaboration and political power, framework application, 60 183, 190 framing, 58­59 collaborative routines, 175, 185, 189 reduction of caste and ethnicity collective action, 173, 174, 178, 191n influence, 213 Community Conflict Negotiation scope, 57 Study, 172 types, 45 conclusions, 189­190 IQMSC. See Integrated Questionnaire conflict case study types, 177­178 for the Measurement of Social conflict management, 175, 178, 187t Capital capacity, 174 disaggregation, 185­186 Jamaica, 79, 82 conflict pathways, 177 Janajatis context, 173 CEI levels, 202f context-revising routines, 184 362 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION countervailing power, 175, 186, Kvinnoforum (2001a; 2001b) 187­188 measurement approach, 224, 225 adaptive conflict management, 183­184 land ownership, 16b procedures, 184­185 land reform, 57 data collection, 41­42, 176, Lao PDR 190­191 empowerment themes, 279 deliberative skills, 175 law enforcement, 14­15 elite capture, 186 legal disputes, 15b empowerment definition, 47t, 173 levels, 21, 23t empowerment impact, 186­189 definition, 21 ethnographies, 175­176 Ethiopia, 73t everyday conflicts, 174­175 framework operations, 86 flaws, 189­190 Integrated Watershed Development forums, 174, 186, 188­189 Project, 63b invitations to join, 191n Madhya Pradesh Poverty Initiatives lack of assets, 188, 190 Project, 65t marginalized groups battling each RWDEP, 67b other, 182 Liberia Trust Fund: Community mediators, 180­181 Empowerment Project, 244, 245 methodology, 175­177 livelihood empowerment, 196, 213 objectives, 172­173 Living Standards Measurement Study participation, 189 (LSMS), 33t, 35t power vacuums, 186 log frame approach, 64 powerlessness, 174 Lokshin and Ravallion (2005) problem-solving, 185, 189 measurement approach, 224, 225 project components, 173 Lukes propensity score matching, 176 social theoretical models of power, routines, 184 238, 239 spillovers, 176­177 splintering of elite groups, 183 Macedonia, FYR state-owned resources, 178­179 empowerment themes, 279 study framework and hypothesis, Madagascar, Republic of 173­175 A Third Social Fund Project, 252, Sumorobangun Dam, 178­179 253 survey, 191n empowerment themes, 279 Tattangoh, 179­180 Madura, 181, 182b traditional Islamic norms, 181 Madyha Pradesh Poverty Initiatives women's participation, 181, 191n Project, 63, 64, 69, 248, 249 Kecamatan Pasean, 181­182 assets and opportunity structures, Kecamatan Proppo, 179­180, 185 65t Kroeker (1995) Malawi measurement approach, 222, 223 empowerment themes, 279 INDEX 363 Malena and Heinrich (2005) reliability, 174 measurement approach, 226, 227 research, 43 Malhotra, Schuter, and Boender (2002) RWSS, 215­216 measurement approach, 226, 227 Measuring Empowerment and Social Mangarrai, 180­181 Inclusion (MESI) study, 55, 193 marital relations, 141­142 BCN group, 217n market access, 139 caste and ethnicity, 207­208, market domain. See also domains 212­213, 212t Ethiopian women, 130b CEI, 202f indicators for country studies, by caste and ethnicity, 205f 266­269 for women by caste and ethnic- WDIP, 137­140, 146­147, 149­150 ity, 209f market indicators, 146 levels by gender, 207f market policies, 57 collective action capability, 196 marriage, 141 cooperation and inclusion, 216 Mason and Smith (2003) data analysis, 202 measurement approach, 226, 227, data collection, 197­198 228 development interventions, 195, matched-pair analysis, 40 197 PB, 98, 100, 110, 112, 121­122 domains, 198 material assets, 33t EMI, 198­199, 199, 200b Mauá, 100, 104t gender, 207, 208, 208f agency-structure changes, 111f empowerment, 194, 195­196, 196 associationalism, 109 determinants, 212­215 citizen participation, 103t gender differential, 206 civil society measurement, 108t group membership effects, 214 civil society weakens, 112 hierarchies, 203­206 plenary meetings, 106 indexes, 198 McMillan et al. (1995) indicators, 204­205 measurement approach, 224, 225 institutions' interactions, 206 measurement, 17, 21­22 interaction of caste and gender, approaches, 30, 220­229 209 assets, 11 livelihood empowerment, 196, 213 challenges, 19, 30­32, 87­88 methodology, 197­198 choice, 29 mobilization empowerment, 196 civil society, 121 project objectives, 195 cooperation among researchers, 88 psychological assets, 204­205 criteria, 32 research design, 194, 197­198 degrees of empowerment, 34 services' access by gender, 208 indirect, 33 SII for BCN group, 205­206 methodology, 32­42 SII indicators, 201, 201b PB, 112­113, 114 social change dimensions, Q-methodology, 221 196­197 364 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION social inclusion measurement, mode of participation, 102­103 215­216 monitoring social injustice, 199 causal relations, 58 social transformation model, 197 data quality, 80 study hypothesis, 197 local-level, 76 study objective, 217n outcomes, 78­79 WDO, 211 poverty, 82n WEI research, 43 by caste and ethnicity, 203f monitoring & evaluation (M&E), 64, 68 by caste, ethnicity, and gender, "theory of change," 88 210­212 Madhya Pradesh Poverty Initiatives indicators, 204b, 210, 211 Project, 69 women's empowerment by caste monitoring and learning system, 69f, and ethnicity, 208 70t women's empowerment variables, monitoring systems, 68 211­212 policy-level, 74, 76­79 mediators, 180­181 poverty, 76, 77, 77b methodology, 42­43 organizational requirements, 78b challenges, 12 project-level, 64 issues, 92­93 monitoring, information flow, and KDP, 175­177 decision-making (MID), 68 measurement, 32­42 Moser (2005) MESI study, 197­198 measurement approach, 228, 229 research, 40 metric, 44n nahi, 3 microcredit programs, 221 National Committee on Traditional microregions, 105 Practices, 141 MID. See monitoring, information National Policy on Women, 33, 126 flow, and decision-making natural target level, 114 Ministry of Education, 159 negative power, 235, 236 Mirante da Serra, 102 neighborhood associations, 105 agency-structure changes, 111f Nepal, 280. See also Measuring citizen participation, 103t Empowerment and Social civil society measurement, 108t Inclusion study; Rural Water exclusion, 109­110 Supply and Sanitation Project mixed-methods approach, 41 development outcomes, 24 intervention evaluation, 38 state domain indicators, 262, 263 longitudinal, 42 Status of Women study, 208 PB, 121 nested information flows, 68, 78, 88 Proheco, 52, 154 nested monitoring and learning WDIP, 50­51, 127­128 system, 69f mobilization empowerment, 196 New Economics Foundation, 82n mode of engagement, 97 New Haven power study, 233 INDEX 365 Newars. See Brahman-Chhetri-Newar panchayat, 16b group Parbatiya, 217n Nicaragua, 280 Parsons Nusa Tengarra Timor, 180­181 social theoretical models of power, 238, 239 observable phenomena, 34 participation, 97, 104 oligarchial parties, 98 consultative, 106­107, 122 OLS. See ordinary least squares not equal to empowerment, 151n opportunity structure, 13­15, 23tn, participatory budgeting (PB), 24, 28n, 62­63 40, 49­50, 85 assets, 86 agency, 97, 122 caste and ethnicity, 206 agency-structure, 110, 111f community empowerment, 53 analysis frame and sequence, 99f concept, 84 analytic focus, 48t definition, 10, 151n associationalism, 109 enabling environment, 87 Camaragibe and Quixadá, 101­102 indicators, 33 choice, 112 institutions, 58­59 citizen engagement, 102­104, 121, investing in, 60, 62­64 122 KDP, 54 city performance, 113­114 Madhya Pradesh Poverty Initiatives civil society, 97, 108­110 Project, 65t consultative participation, 122 market domain indicators for coun- council, 105 try studies, 266­269 counterfactual construction, 98 PB, 50, 97, 110 data generation, 114 Proheco, 53, 164, 170n­171n decision-making, 104t indirect indicators, 158­160 development RWSS, 56 indicators, 113b societal domain indicators for coun- outcomes, 113b, 122 try studies, 270­275 performance effects, 118t­119t state domain indicators for country effects, 120 studies, 258­263 empowerment, 47t, 121 variables and outcomes, 162 evaluation, 110 WDIP, 51, 127 Gravataí and Sapucaia do Sul, 101 ordinary least squares (OLS) regres- heterogeneity, 100, 122 sional analysis, 44n hypotheses, 49 organizational assets, 33t impact on extremely poor, 115 organizationally negotiated disorder, index of progress, 115, 124n 237 input qualitative measures, 102­103 outcome monitoring, 78­79 João Monlevade and Timóteo, 101 outcomes, 162 matched-pair analysis, 110, 112, relationship with correlates, 10f 121­122 WDIP, 129­133 matching rule, 123n 366 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION Mauá and Diadema, 100 positive power, 235 measuring civil society changes, poverty, 115 108, 108t alleviation, x municipality indicators, 116t­117t assessment, 74, 81n pair selection, 98, 100 exercise of choice, 72 partial, 106­107 measurement, 30 participation, 123n monitoring, 76, 77, 82n PB municipalities versus non-PB organizational requirements, 78b municipalities, 116t­117t purpose and functions, 77b performance measures, 114 outcomes, 112­115 plenaries, 105 reduction, 22 poverty, 112­115, 120, 122 Ethiopia, 73t, 75b PT, 124n information access, 25b regression-discontinuity design, PB, 115, 120, 122 123n "Poverty Profile of Ethiopia," 72 research framework, 97 Poverty Reduction Strategy Paper São Miguel do Guaporé and (PRSP), 81n. Mirante da Serra, 102 poverty monitoring goals, 77 selection bias, 100, 122 power, 2, 31, 84 social activism, 107 agenda setting, 234 social inclusion, 109 and agency theory, 236­237 state-controlled participation, between men and women, 200­201 107­108 competing power structures, 233 study time frame, 114 concept, 10, 84, 230, 235 voice, 105 conflictual, 230, 235 participatory council, 106 consensual theories, 230­231 participatory surveys, 47 episodic, facilitative, and disposi- Partido dos Trabalhadores (PT), 100, 110, tional, 237 124n false consciousness, 234 patronage, 97 KDP, 175, 183 perceptual information, 34 middle-ground theories, 234­237 plenaries, 105 negative and positive, 235, 236 policy, 71, 80 organizationally negotiated disor- monitoring systems, 74, 76­79 der, 237 political engagement, 134t PB, 120 political parties, 97­98 Proheco, 52, 169­170 politics of influence, 106 social theoretical models, 238­241, polyarchy, 233 239 poorest, the typology, 232b­233b adaptive conflict management, power structures, 169 183­184 power vacuums, 186 empowerment objectives, 59 powerlessness, 174 Proheco, 168­169 preferences, adaptive, 12 INDEX 367 problem-solving, 185, 189 research, 38, 43, 49 process formalization, 103 methodologies, 40 process tracing method, 40 positivist-empiricist, 43n project management, 64, 68 resistance, 3 project monitoring system, 68 resources, 240n promoters, 159, 164 routines, 54, 175 propensity score matching, 176, 191n rules, 84, 240n property ownership, 16b, 28n rules of the game, 10­11, 27n PRSP policy matrix rules, decision, 175 indirect indicators, 35t Rural Water Supply and Sanitation psychological assets, 11­12, 33t, 55, 62 (RWSS) Project, 45, 55­56, 85, 193 attitudes, 27n analysis of variables, 216 MESI study, 204­205 analytic focus, 48t WDIP, 151n BCN group, 217n PT. See Partido dos Trabalhadores data collection, 42 public service delivery, 113b development interventions, 195 empowerment definition, 44n Q-methodology, 221 47t qualitative and quantitative informa- evaluation conclusions, 215­216 tion, 38­39 Fund Board, 194 Quixadá, 101­102 project objective, 194 agency-structure changes, 111f Rural Women's Development and citizen participation, 103t Empowerment Project civil society measurement, 108t (RWDEP), 66b, 67b, 254, 255 Rwanda, 280 radio ownership, 25 RWSS. See Rural Water Supply and Rainfed Natural Resource Sanitation Project Management Project, 254, 255 rape, 135, 141, 151n sanitation schemes, 193­194 rational choice, 14 São Miguel do Guaporé, 102, 104t recall, 34, 37 agency-structure changes, 111f regression analysis, 40, 44n citizen participation, 103t influence of caste and ethnicity on civil society measurement, 108t women, 212t engagement, 109 Proheco, 161­162, 162t Sapucaia do Sul, 101, 104t stepwise, 163t agency-structure changes, 111f regression models citizen participation, 103t discontinuity, 115, 120 civil society measurement, 108t WDIP, 129 state-controlled participation, 107 women's awareness of rights, 133 school council empowerment, regression techniques, 112­113 155­156, 161t regression-discontinuity design, 123n agency indicators, 170n reliability, 174 outcomes, 160­165, 160t 368 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION School Council Empowerment Index indicators used, 201b (SCE), 161 RWSS, 42 multiple regression, 162t social injustice, 199 stepwise regression, by number of social justice, 195 variables, 163t social outcomes, 148t school councils. See Asociación social theoretical models of power, 238, Educativa Communitaria 239 SDPRP. See Sustainable Development social theory, 236 and Poverty Reduction Program social transformation, 197 Second Indigenous and Afro- societal domain. See also domains Ecuadorian Peoples Ethiopian women, 130b Development, 246, 247 indicators for country studies, 270­275 secondary source review, 128 WDIP, 141­144, 147, 149, 150 selection bias, 97, 100 spillovers, 176­177 PB, 122 Spreitzer (1995) WDIP, 151n measurement approach, 228, 229 self-help groups (SHGs), 66b state domain. See also domains self-organization, 97, 122 Ethiopian women, 130 self-weighting, 39 indicators Senegal, 280 Brazil and Ethiopia, 258, 259 sequencing methods, 37­42 Honduras, 260, 261 sexual decisions, 142 PB, 49 SII. See Social Inclusion Index WDIP, 133­137, 149 social activism, 107 status, 61bn Social Analysis Sourcebook, 195­196 Status of Women study, 208 social capital, 13b structuration power theory, 236 social change, 196­197 structure-agency. See agency-structure social empowerment, 61b Sumorobangun Dam, 178­179 social identity, 22 surveys, 38, 47, 69 social inclusion Sustainable Development and Poverty definition, 196 Reduction Program (SDPRP), 72 group membership effects, 214 measurement, 215­216 Tamil Nadu Empowerment and MESI study, 193, 207f Poverty Project, 59 caste and ethnicity indicators, Tamil Nadu Empowerment and 207­208 Poverty Reduction, 244, 245 determinants, 212­215 Tanzania, 281 process of, 196 Tarai Middle group, 202f, 203f Nepal, 24 Tattangoh, 179­180 PB, 109 teeth extraction, 141 RWSS, 55­56, 215­216 Third East Timor Community Social Inclusion Index (SII), 199 Empowerment and Local BCN group, 205­206, 210 Governance Project, 248, 249 INDEX 369 Timóteo, 101, 104t cooperation and inclusion effects on agency-structure changes, 111f project effectiveness, 216 citizen participation, 103t decision making, 167 civil society measurement, 108t disempowerment, 11­12 state-controlled participation, 107 domestic rights, 52 toddy, 27n empowerment, 52, 277, 278 trade, 57 by caste and ethnicity, 208­210 traditional harmful practices, 141­142, measurement, 17 147, 148t MESI study variables, 211­212 traditions, 14 Ethiopia assets, 125­127, 139 Uganda, 281 awareness of rights, 136, 137, 142 United Nations Development codes of conduct, 141 Programme divorce, 133 Human Development Report, 228, economic empowerment, 140t 229 employment, 138­139 utility reform, 57 gender equality, 126 group participation, 144, 144t variables, 130b, 139­140 happiness, 131, 132t Ethiopian women's political illiteracy, 138 engagement, 134t indicators, 130b Proheco, 156, 161, 169 information access, 138 RWSS analysis, 216 judicial system, 126­127, WDIP, 136t 133­135, 135, 135t women's empowerment, 211 legal protection, 136t Vietnam, 281 low levels of empowerment, voice 129­130 accountability, 24 marital relations, 141­142 PB, 105 market access, 139 WDIP, 143­144 National Committee on Traditional Practices, 141 water availability, 138 political engagement, 133, 134t water supply schemes, 193­194 rape, 141 WDO. See Women's Development reproductive and sexual deci- Office sions, 142 WEI. See Women's Empowerment and right to choose spouse, 137t Inclusion Index skills, 137 weighting, 39 traditional practices, 141­142 welfare indicators, 114 variables, 130b women household power differentials, CEI levels by caste and ethnicity, 200­201 209f in India, 66b community engagement, 126­127 influence of caste and ethnicity, 212t 370 EMPOWERMENT IN PRACTICE: FROM ANALYSIS TO IMPLEMENTATION judicial system, 34 human and economic assets, 125 KDP participation, 181, 191n illiteracy, 138 microcredit programs, 221 impact evaluation, 129, 145 property rights, 28n income effects, 143 rights, 51 indicators, 125, 129, 150 RWSS, 194 information access, 126, 138 SII indicators, 201, 201b information assets, 139 social inclusion in Nepal, 24 institutional variables effect on Status of Women study, 208 income, 139­140 WDIP interactive fieldwork, 128 community engagement, 143, literacy and political engagement, 145t 134 domestic control, 142­143 market access, 139 economic outcomes, 140, 144 market domain, 137­140, 146­147, friends to rely on, 149t 149­150 group participation, 147 objectives, 50 income effects, 143 outcomes and determinants, participation, 131­132 129­133 social outcomes, 148t participation, 131­132, 144­145 traditional harmful practices, political engagement, 134 147, 148t political subdomain, 51 voice, 143­144 project components, 127 Women's Development Corporations regression models for outcome (WDCs), 66b variables, 129 Women's Development Initiatives reproductive and sexual power, Project (WDIP), 45, 50­52, 85 142 analytic focus, 48t sampling pitfalls, 150 assets and economic outcomes, 139 secondary source review, 128 attribution of causality, 133 social domain, 141­144, 147, 149, community engagement, 126­127, 150 143, 145t social outcomes, 148t community selection, 128 state domain, 133­137, 149 data, 40­41, 132 study sampling frame, 127­129 domestic control, 142­143 survey coverage, 129 economic constraints, 138 terminology definitions, 151n economic outcomes, 140, 144, 146 traditional harmful practices, employment, 126, 138­139 141­142, 147, 148t empowerment, 47t, 144­145 voice, 143­144 evaluation framework, 128f water availability, 138 financial markets, 146­147 women's awareness of rights, 130, friends to rely on, 149t 136, 136t, 142 group participation, 147 women's economic empowerment, HIV/AIDS, 126 140t INDEX 371 Women's Development Office (WDO), development outcomes, 22 211 empowerment projects, 243­255 Women's Empowerment and Inclusion Ethiopia poverty assessment, 72 Index (WEI), 200­201 poverty assessment document, 81n by caste and ethnicity, 203f Social Analysis Sourcebook, 195­196 caste, ethnic, and gender dimen- World Development Report 2000/2001, sions, 210­212 22 indicators, 201, 201b, 204b, 211 Worker's Party, 100 Yemen, Republic of, 281 World Bank Country Policy and Institutional Zambia, 282 Assessment, 35t zero-sum game, 2 This book opens up the important concept of empowerment with enough structure to guide the researcher into important discoveries, but also enough grey zones to make room for the country specific as well as the ideological and methodological preferences of the analyst. --Ishac Diwan Country Director, Ethiopia World Bank This book approaches evidence in a rigorous and systematic way, seeking valid comparisons, but not at the expense of local, social, and historical context. It has put choice and opportunity, often treated as afterthoughts, in the foreground of development, and demonstrated ways of analyzing them systematically yet with flexibility. As such, it will appeal and be useful to a wide audience. --Paul Francis Senior Social Scientist, Africa Region World Bank, Washington, D.C. This document offers an excellent explanation of the concept of empowerment and develops a very useful framework for disentangling and clarifying the concept. It is both practical and well justified on the basis of the literature. The approach and survey instruments developed to measure empowerment are innovative and very useful. Developing an approach to empirically measure empowerment, and to track empowerment indicators over time in a way that is operationally feasible and consistent with World Bank standards is certainly an outstanding achievement. --Regina Birner Senior Research Fellow International Food Policy Research Institute, Washington, D.C. ISBN 0-8213-6450-2