Social Service Delivery in Violent Contexts Achieving Results Against the Odds A Report from Afghanistan, Pakistan, and Nepal Social Service Delivery in Violent Contexts Achieving Results Against the Odds A Report from Afghanistan, Pakistan, and Nepal © 2017 The World Bank Group 1818 H Street, NW Washington, DC 20433 All rights reserved. This volume is a product of the staff of the International Bank for Reconstruction and Development/World Bank. The findings, interpretations, and conclusions expressed in this paper do not necessarily reflect the views of the 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. Copyright Statement The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with com- plete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, http://www.copyright.com/. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights@worldbank.org. Cover photo: Schoolgirls walk past a damaged mini-bus after it was hit by a bomb blast in the Bagrami district of Kabul, Afghanistan, April 11, 2016. Mohammad Ismail/Reuters Contents Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix Abbreviations and Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi Executive Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Part I. Research Approach and Major Findings 1 Research Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.1 Knowledge Gaps and Practical Challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.2 Scope and Framework of Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2 Applying the Framework: Puzzles and Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.1 Afghanistan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2.2 Nepal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 2.3 Pakistan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings 3 Political Settlements Matter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 3.1 National Settlements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 3.2 Sectoral Settlements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 3.3 Central and Subnational Elites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 3.4 Subnational (Secondary) Settlement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 4 The Form of Violence and the Ideology behind It Matter . . . . . . . . . . . . . . . . . . . . . . . 57 4.1 Organization of Violence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 4.2 Ideologies, Incentives, and Motivations behind Violence . . . . . . . . . . . . . . . . . . . . . 61 4.3 Localization of Violence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 Contents | iii 5 Elite Bargaining Matters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 5.1 Forms of Bargaining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 5.2 Trade-Offs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 6 Modes of Delivery Matter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 6.1 Mitigating the Effects of Sectoral Patronage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 6.2 Anticipating and Mitigating Insurgent Resistance . . . . . . . . . . . . . . . . . . . . . . . . . . 89 6.3 Increasing Local Bargaining Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 6.4 Supporting Progress Outside the Sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 Part III. Implications for Policy and Operations 7 Policy Takeaways: A Radical Rethinking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 7.1 Adapt to a New Set of Contextual Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 7.2 Recognize that Some Forms of Violence Are Less Bad than Others for Service Delivery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 7.3 Address the Operational Ways through Which Elites Matter . . . . . . . . . . . . . . . . . 103 7.4 Rethink the Function of Service Delivery: Rents Matter . . . . . . . . . . . . . . . . . . . . 104 7.5 Be Realistic about Good Governance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 8 Practical Directions and Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 8.1 Improve Diagnostics for Measuring the Actual Scope of Progress . . . . . . . . . . . . . 109 8.2 Think and Work Politically . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 8.3 Tailor Approach to Specific Forms of Violence . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 8.4 Take Bargaining Seriously . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 8.5 Reform Donor Rules and Incentives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 8.6 Explore New Frontiers of Delivery and Research . . . . . . . . . . . . . . . . . . . . . . . . . . 122 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 Appendixes A Methodology and Case Study Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 B A Diagnostic Tool: Steps and Question Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142 Reference List and Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 iv | Contents Boxes 1.1 Some research caveats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 3.1 Nepal’s shift toward inclusion shapes social service delivery . . . . . . . . . . . . . . . . . . . . . 46 3.2 Political settlements in Afghanistan’s education and health sectors . . . . . . . . . . . . . . . 48 3.3 Relationships between elites shapes delivery of polio services . . . . . . . . . . . . . . . . . . . 50 3.4 Central-subnational linkages and service delivery in Afghanistan . . . . . . . . . . . . . . . . . 51 3.5 Subnational settlements in Nepal help explain SSD outcomes . . . . . . . . . . . . . . . . . . . 53 4.1 Organization of violence in Badghis and Wardak . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 4.2 Ideology matters? Education in Balochistan versus polio eradication in KP and FATA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 4.3 Wardak, Afghanistan: Education incentives before the Taliban . . . . . . . . . . . . . . . . . . 66 4.4 Localized armed elites and allied insurgents and social service delivery bargains . . . . . 69 5.1 Enlisting the support of local elites in Wardak, Afghanistan . . . . . . . . . . . . . . . . . . . . 73 5.2 Religious elites, fatwas, and polio delivery in Pakistan . . . . . . . . . . . . . . . . . . . . . . . . . 74 5.3 Bargaining with the Maoists in Palpa, Nepal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 5.4 Bargains and skewed health provision in Wardak, Afghanistan . . . . . . . . . . . . . . . . . . 82 6.1 Engaging local elites in the Balochistan Education Support Project: Challenges to scaling up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 6.2 Lack of bundling undermined elite bargains on polio eradication in Pakistan . . . . . . . 90 6.3 Female volunteers deliver health services in violent contexts in Nepal . . . . . . . . . . . . . 92 6.4 The impact of socioeconomic factors outside a sector . . . . . . . . . . . . . . . . . . . . . . . . . 97 7.1 Clarifying expectations: What this study tells policy makers and practitioners . . . . . 101 8.1 What the proposed diagnostic toolkit can and cannot do . . . . . . . . . . . . . . . . . . . . . . 110 8.2 Potential strategies for building pro-SSD coalitions . . . . . . . . . . . . . . . . . . . . . . . . . . 114 8.3 Making delivery more politically astute: Key practical lessons . . . . . . . . . . . . . . . . . . 115 8.4 Key questions for further exploration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126 A.1 Summary of Research Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138 Contents | v Figures ES.1 Proposed framework for strengthening service delivery in violence-affected contexts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xix 1.1 Study framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 1.2 Dimensions of settlements that affect service delivery . . . . . . . . . . . . . . . . . . . . . . . . . 15 1.3 Dimensions of violence that affect social service delivery . . . . . . . . . . . . . . . . . . . . . . . 17 1.4 Overlapping forms of bargaining to deliver services . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 2.1 Access to skilled antenatal care, 2004–12 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 2.2 Cases of polio in Pakistan: The Federally Administered Tribal Areas versus Khyber Pakhtunkhwa, 2001–14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 8.1 A new framework for strengthening service delivery in violence-affected contexts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 B.1 Main diagnostic modules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142 Tables ES.1 Types of political settlement in violence-affected contexts and implications for social service delivery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxi ES.2 Forms of violence and implications for service delivery . . . . . . . . . . . . . . . . . . . . . . xxiii 1.1 Types of political settlements in violence-affected contexts . . . . . . . . . . . . . . . . . . . . . 16 1.2 Six Types of violence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 2.1 Social service delivery in Afghanistan: Framework, puzzles, and findings . . . . . . . . . . 27 2.2 Health indicators, Nepal, 1990–2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 2.3 Health service delivery, Nepal, 1996 and 2006 (by region) . . . . . . . . . . . . . . . . . . . . . 31 2.4 Road networks in Siraha (plains), Palpa (hills), and Bardiya (plains) . . . . . . . . . . . . . . . 33 2.5 Social service delivery in Nepal: Framework, puzzles, and findings . . . . . . . . . . . . . . . 34 2.6 Social service delivery in Pakistan: Framework, puzzles, and findings . . . . . . . . . . . . . 39 7.1 Beyond good governance: Service delivery in violence-affected contexts . . . . . . . . . . 107 8.1 Implications of political settlements for social service delivery in violence-affected contexts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 8.2 Forms of violence and implications for service delivery . . . . . . . . . . . . . . . . . . . . . . . 118 8.3 Operational strategies for bargaining with insurgents . . . . . . . . . . . . . . . . . . . . . . . . . 120 8.4 Ideas for shifting donor lexicon and mind-sets on elites and bargaining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 8.5 Shifts in how aid actors work: Frontiers and challenges . . . . . . . . . . . . . . . . . . . . . . . 123 vi | Contents Preface This study has been a long time in prepara- First, the case selection, and related research tion. We began in 2013 with the intention methodology, does not claim to be represen- of looking at how the delivery of social ser- tative of the sectors or geographical regions in vices was affected by violent conflict in four question. The research questions were devel- countries—Sri Lanka, Nepal, Pakistan, and oped to address important gaps in our under- Afghanistan. We soon had to drop Sri Lanka standing of how progress in service delivery because it proved difficult to get agreement happens in such difficult contexts. We recog- to carry out research in the country. Still, in nize our findings rest on the investigation of the three remaining countries the varying na- a restricted number of sectors and projects, ture of the conflict offered a good laboratory but our hope is that the case studies and new to better understand how differing patterns framework will stimulate further research to of violence and of the “political settlements” corroborate our findings. Indeed, generating that form around conflict had affected or a statistically representative sample in such were still affecting service delivery. We were contexts, given the security and financial im- able to look at how different modes of ser- plications, is a perennial challenge. vice delivery were able to adapt to and often surmount the challenges of violence. Our an- Second, we are not trying to offer an exhaus- alytical framework enabled us to explore how tive view of all the possible factors that af- political settlements at different levels of the fect service delivery. The focus on forms of state and different forms of violence shaped settlement, violence, bargains, and service de- service delivery. We were also able to examine livery modality does not mean that other fac- the role that bargains among elites played in tors (such as beneficiary participation, service the way services were delivered. Finally, we provider incentives, and the technical quality explored how certain modes of delivery were of services, among others) are unimportant. more effective than others in violent contexts. Rather, based on gaps in the literature (which fails to explain observed progress), we are of- Given the sensitive nature of this field of en- fering a different perspective on why things quiry, it is important to understand what this work, or do not. As noted above, this should study is not. stimulate further debate. Preface | vii Third, the study does not attempt to explore Whatever views about armed groups are pre- the impact of social service delivery on broader sented in the report come from key informant political dynamics such as peace-building interviews, focus group discussions, and re- and state-building. Our research focuses on lated literature. What the report does point to, the “how to” of delivering services on the however, is the need to accept that in many ground when violence is present. In that re- situations of violent conflict, “good gover- spect, the report explores the delivery of ser- nance” precepts are often an imperfect tem- vices in their own right (to reduce poverty or plate for providing services, and local bargains for other humanitarian reasons) but does not are sometimes necessary in order to deliver address whether or how delivery may cause humanitarian and development assistance. fragility or contribute to state-building. The link between social service delivery and state We believe the report will play a timely and legitimacy is too indirect and too broad to important role in sparking new research, de- provide granular, day-to-day guidance for bate, and practice in this critical, yet difficult, policy makers and practitioners. area. Finally, this report does not endorse the ac- Richard Hogg tions of the insurgent groups it describes nor Simon O’Meally condone the observed corruption or collusion Jonathan Di John between these groups and service providers. May 2017 viii | Preface Acknowledgments This study was led by Richard Hogg (program Management Sciences), the Balochistan Uni- leader, Afghanistan, World Bank) and Simon versity of Information Technology, Engineer- Carl O’Meally (senior governance specialist, ing and Management Sciences (BUITEMS), World Bank). The lead authors are Jonathan and the Institute of Management Sciences, Di John (lead consultant and senior lecturer Peshawar. in political economy, SOAS, University of London), Simon Carl O’Meally, and Richard In addition, we would like to thank the fol- Hogg. Other contributors are Mithila Desh- lowing for their guidance to the team: Keiko pande (consultant), who provided invaluable Miwa (practice manager, education, World research assistance, and Steven Kennedy and Bank), Rekha Menon (practice manager, Laura Johnson, who edited the final report. health, World Bank), and Karla Gonzalez Carvajal (practice manager, transport, World We would like to thank Alexandre Arrobbio Bank), and their respective education, health, (practice manager, governance, World Bank) and infrastructure teams at the World Bank. In for the guidance he provided to the team Afghanistan, Dr. Sayed Ghulam (senior health throughout the process. We would also like specialist), Abdul Hai Sofizada (senior edu- to thank Tony Verheijen and Maria Correia, cation specialist), and Nathalie Lahire (senior former practice managers in the World Bank’s economist) made invaluable comments. In Pa- South Asia Governance and Social Devel- kistan we would like to thank Sher Shah Khan opment groups, respectively, for their earlier (senior public sector specialist) for his overall support to this research. We would also like to guidance to the team. Thanks also go to Um- thank Aditi Haté and Lauren Keevill (former breen Arif (senior education specialist), Ali consultants). Ansari (economist), Inaam Ul Haq (Pakistan program leader, human development, World The Afghanistan country case study was based Bank), Tayyeb Masud (senior health special- on fieldwork carried out by the Afghani- ist), Ousman Jah (program coordinator), and stan Research and Evaluation Unit (AREU), Uzma Basim (senior country officer). In Nepal, Kabul. The Nepal country case study was Preeti Kudesia (senior health specialist), Dom- based on fieldwork carried out by Social Sci- inic Patella (senior transport specialist), Hi- ence Baha, Kathmandu. The Pakistan coun- ramani Ghimire (former governance adviser, try case study was based on fieldwork carried World Bank), and John Bevan all provided out by Dr. Ejaz Akram (Lahore University of very helpful comments. Martin Rama (chief Acknowledgments | ix economist, South Asia, World Bank) provided specialist, World Bank). Finally, we would invaluable guidance to the team throughout like to thank the World Bank Country Man- the process, for which we are extremely grate- agement Teams in Afghanistan, Pakistan, and ful. The team also benefitted from the peer re- Nepal for their support of this study. view comments of Thomas Parks (governance and fragility specialist and former assistant The study is funded by the World Bank–Aus- director, Australian Department of Foreign tralia Partnership for South Asia Trust Fund Affairs and Trade), James Putzel (professor and the World Bank–Korea Trust Fund to Sup- of development studies, London School of port Economic and Peace Building Transitions. Economics), Alexandre Marc (chief technical We would like to thank both governments for specialist, World Bank), Bernard Harborne their generous assistance. In addition to these (lead social development specialist, World two sources, financial support was also pro- Bank), and Verena Fritz (senior public sector vided directly by the World Bank Group. x | Acknowledgments Abbreviations and Acronyms BEF Balochistan Education Foundation BESP Balochistan Education Support Project EQUIP Education Quality Improvement Project FATA Federally Administered Tribal Areas FCHV female community health volunteer FCV fragile, conflict, and violence-affected FGD focus group discussion KII key informant interview KP Khyber Pakhtunkhwa MDG Millennium Development Goal MP Member of Parliament NGO nongovernmental organization OECD Organisation for Economic Co-operation and Development PEP Polio Eradication Project SDG Sustainable Development Goal SHARP Strengthening Health Activities for the Rural Poor SSD social service delivery US United States USAID United States Agency for International Development Abbreviations and Acronyms | xi Executive Summary This report provides the foundation for a Study Rationale and Objectives new approach to service delivery in vio- lence-affected contexts that is more sensitive About 2 billion people live in countries where to the actual forms of violence, politics, and fragility, conflict, and violence make it diffi- bargaining encountered in many conflict-af- cult to achieve development outcomes. These fected states. The findings unearth sensitive countries have made the slowest progress issues about how development organizations toward the Millennium Development Goals. should approach service delivery in contested On average, the poverty rate of a country that settings. As many countries today are riven by experienced major violence between 1981 conflict and internal division, some familiar and 2005 is 21 percentage points higher than rules of the game may be inadequate to deal that of a country that saw no violence (World with the mounting humanitarian and devel- Bank 2011). By 2030 about 50 percent of the opment challenges posed by complex conflict world’s poor will live in countries affected by situations, particularly where affected people fragility and violence (World Bank 2016b). need access to social services. This raises di- Violent conflict also has major effects that lemmas about the ethical and political judg- spill over national borders, as the refugee cri- ments and trade-offs that development actors sis affecting parts of Europe, the Middle East, frequently have to make. A key challenge is and North Africa shows. whether development actors can adapt their procedures and ways of working to the flu- Delivering basic services in these settings idity, uncertainties, and risk taking that the is extremely challenging. Only one-fifth of new, conflict-riven landscape demands while fragile states are on track to achieve univer- preserving financial accountability, doing no sal primary schooling (compared with nearly harm, and ensuring aid effectiveness. This half of nonfragile developing countries) challenge cannot be ignored, as progress in (OECD 2015); and, only 28 percent are on basic service delivery in violence-affected track to halve the number of their citizens contexts will make or break global efforts to who lack access to safe water (61 percent of end poverty and increase shared prosperity nonfragile countries have reached this target) over the coming decades. (OECD 2015). Executive Summary | xiii In response to these issues, the international but there is limited evidence on how such community has committed to improving dynamics influence daily programming, development assistance in so-called fragile especially in violence-affected contexts. contexts. In September 2015, world leaders adopted the Sustainable Development Goals • Ground-level research in violence-affected (SDGs). Goal 16 is to “promote peaceful areas is rare; most studies are based on sec- and inclusive societies for sustainable devel- ondary literature reviews or, at best, inter- opment, provide access to justice for all and views with key informants in capital cities. build effective, accountable and inclusive in- stitutions at all levels” (UN 2015). They also • There is a consensus that practice on the agreed to SDGs focused on improving access ground urgently needs to become more to and the quality of services. In addition, differentiated, context specific, and sys- the World Bank Group committed to doing tematically tailored to political economy more and better in violence-affected contexts and violence risks. under the International Development Asso- ciation’s 18th funding replenishment, which To address these gaps, the report synthesizes spans the years 2017–20. Finally, there is a and builds on field research in Afghanistan, growing consensus in the international com- Nepal, and Pakistan on the delivery of health, munity—as argued in the 2017 World De- education, and rural infrastructure services. velopment Report (World Bank 2016 )—that This research inductively piloted a new re- governance and institutional factors can make search framework in order to answer the fol- or break development and service delivery lowing questions: progress in violent contexts. • How and why do political economy fac- These ambitious global development goals tors and forms of violence shape the are, however, unlikely to be achieved until implementation and outcomes of social critical knowledge gaps are filled and prac- service delivery (SSD) in violence-affected tice on the ground improves. This report ad- contexts? dresses four critical gaps: • What explains the varying success rates of • It is not clear why some service delivery service delivery within, and between, vio- interventions in violence-affected contexts lence-affected contexts? make progress and others fail. • What lessons can be learned from what • It is widely recognized that political econ- worked, and what did not? omy dynamics matter for service delivery, xiv | Executive Summary Main Findings arrangements related to specific sec- tors (such as health, education, and According to this research, four key interre- infrastructure). lated elements—political settlements, forms of violence, elite bargains, and modalities of • National-subnational linkages are re- service delivery—help explain variations in lationships between national and subna- success and results. tional elites that influence service delivery by shaping the flow of sectoral resources to a locality and influencing the extent to Political Settlements which the state apparatus (and its public Political settlements set the broad, often in- administration) can control insurgents and formal, rules of the game and create con- can influence the delivery of services at straints, spaces, and opportunities for SSD. the subnational level. They provide a map of the ruling coalition: who the power brokers are; the balance of • Subnational and village-level settle- power between them; and what interests ments arguably play the most decisive and norms they are defending or promoting role in explaining subnational variation in in economics, politics, and service delivery. delivery. This relates to the extent to which Settlements take different forms and can op- the local elites’ balance of power is stable erate at different levels of a polity. Four inter- and oriented toward the delivery of social connected dimensions of the settlement are services, the incentives of local elites to- found to matter: ward service delivery, and the local elites’ negotiating power relative to insurgents. • Primary (or national) settlements are power configurations at the central state Forms of and Motivations level. They usually include elites from behind Violence dominant groups that have traditionally held national political power. The stabil- Violence takes different forms and is driven ity/lack of stability of the settlement and by different motivations, which affect the the extent to which power brokers within extent to which insurgents tolerate, seek to the ruling coalition are supportive of some modify, or bargain over a service. In this re- form of SSD influences outcomes. gard, three dimensions of violence are key: • Sectoral settlements are linked to the • The extent to which violent actors are national settlements but involve differing organized, disciplined, and homog- elite incentives, ideas, and power-sharing enous. This shapes whether insurgent Executive Summary | xv leaders can be identified for negotiating This report goes beyond broad statements service delivery bargains, and whether like “elite bargains matter” to identify the ac- such agreements can be upheld. tual processes through which bargains were achieved and reproduced. It identifies four • The ideologies, incentives, and mo- overlapping forms of bargaining: tivations behind violence. Specifically, the prospects for delivering services are • Fostering dialogue and negotiations influenced by the degree of alignment with armed elites and allied insur- between insurgents’ objectives and the gents. This entailed persuading powerful objectives of the service delivery interven- gatekeepers to allow delivery, building tion, the extent to which insurgents (and trust between providers and insurgents, their families and friends) need the service, and, in some cases, the co-delivery of ser- and the extent to which insurgents can ex- vices by providers and armed elites. tract rents from the service. • Enlisting the support of influential • The degree of localization of vio- nonarmed elites. Such elites helped lence. The degree to which insurgents are mediate the relationship between armed embedded in local communities and the elites and their allied insurgents and ser- extent to which they seek to secure local vice providers in order to allow for some support shape their incentives to facilitate degree of delivery. or block service delivery. As a general rule, the more fragmented, mobile, and ideo- • Striking concessions and compro- logically antagonistic the violent actors, mises. Various modifications were made the more difficult it will be to deliver ser- in the way a service is delivered so as to vices and the more mitigation strategies make it more acceptable to insurgents and will be needed. reduce the risk of violent backlash. • Sharing rents and tolerating corrup- Elite Bargains tion. Certain rent-sharing practices were Linked to the political settlement and form knowingly or unknowingly enabled to of violence, elite bargains—sets of rolling appease and buy off opposing armed and negotiations and agreements between pow- unarmed elites to sustain service delivery. erful actors, armed or unarmed—are critical in explaining whether and how services get delivered. xvi | Executive Summary Such bargains, or their absence, are critical in Five Policy Takeaways explaining whether progress occurs. But they involve not insignificant trade-offs, as noted These findings have important implications below. for policy and practice, summarized here. Strategies and Modalities of Delivery 1. Better adapt to a new set of contex- tual factors. Most practitioners recognize Even the broader context of politics, violence, that “context matters” for service deliv- and bargaining does not, however, seem to ery, but there has been limited progress simply determine outcomes. Progress in ser- in understanding how. This report shows vice delivery is also shaped in diverse ways by how contextual factors that remain un- different modalities and strategies of delivery. derexplored—the specific characteristics In terms of modalities, three main ways of of political settlements at various levels, delivering services were explored: state ver- the different dimensions of violence, the sus nonstate, centralized versus decentralized, processes and forms of bargaining and and single sector versus multisector. Context strategies of delivery—can make or break influenced which modalities had greater pros- service delivery. This is especially true at pects of success. In terms of strategy, service the subnational level. Development actors delivery interventions made greater progress have no choice but to take these aspects when they were: (1) violence sensitive: by an- much more seriously. ticipating and mitigating some of the risks of violence; (2) politically sensitive: by reducing 2. Recognize that some forms of vio- the influence of anti-service-delivery political lence are less bad for service deliv- coalitions; and (3) bargain sensitive: by creat- ery. Mainstream approaches tend to view ing space for and facilitating pro-service-de- violence as simply bad for service deliv- livery bargains, especially among subnational ery. But the research shows that different elites, armed and unarmed. These strategies forms of and motivations for violence can helped foster progress in the cases presented be more or less amenable to SSD, and that in the report, but they were not magic solu- these motivations can vary considerably tions and all presented trade-offs. within and between violence-affected contexts. There is therefore a need to go beyond overly simplistic classifications of violence, and to unpack how different forms of violence present different obsta- cles to and opportunities for service deliv- ery, particularly at the subnational level. Executive Summary | xvii Address the operational ways in 3. armed or unarmed, affecting (for good which elites influence service delivery. or ill) the balance of power and politi- Clearly, elites influence development prog- cal order; (2) rents created or constrained ress. Yet, in many ways, this assertion has by a SSD program may affect (for good frustrated service delivery practitioners, as or ill) the level of violence, because vio- it tends to yield a long list of broad recom- lence is often controlled by rent-sharing mendations and to present elites as actors agreements in such contexts; and (3) rent who simply frustrate and subvert service appropriation of SSD resources by armed delivery. Instead, this report suggests that elites can in some cases be considered as a the lens of elite bargaining offers a new “cost of doing business” to ensure peace- frontier for social service policy design ful project implementation (the extent to and implementation in violence-affected which rent appropriation by armed elites contexts. At least three policy insights are is a cost worth bearing is as much a polit- worth highlighting: (1) elites, armed and ical as a technical judgment). These issues unarmed, are often (if not always) inti- need further attention in programming in mately involved in shaping SSD in such violent contexts. fluid, informal, and contested contexts; (2) elite bargains are critical and such bargains 5. Be realistic about good governance. can take a range of different forms, which The report shows how governance, poli- need to be considered in the program de- tics, and elites matter, but not in ways that sign; and (3) elites can play a more or less are necessarily expected. The still-dom- positive role in violence-affected contexts, inant “good governance” paradigm— and the extent to which armed elites facil- centered on best-practice formal rules, itate or undermine a service depends on liberal-democratic norms of transpar- their varied incentives and ideologies. ency and accountability, and technocratic fixes—does not always emerge as a par- 4. Rethink the function of service deliv- ticularly useful guide to understanding ery: rents matter. A rethinking may well progress in violent contexts. Instead, the be needed of the widely held idea that report highlights two major policy im- service delivery is or should be a neutral plications: (1) there are a wider range of intervention, and that rent-sharing must options for progress in the midst of far- or can be entirely avoided in delivering from-perfect forms of governance than the services. This research points to a set of good governance paradigm assumes; and difficult and nuanced policy implications, (2) informal relationships, rent-sharing, which are: (1) service delivery generates far-from-perfect transparency or account- rents, which can strengthen certain elites, ability, and deep politicization of service xviii | Executive Summary delivery—through political parties or “un- bargaining. It includes six mutually reinforc- savory” powerful actors—played critical ing recommendations (see figure ES.1): roles in the change process on the ground. 1. Become better at diagnosing the real scope In short, the report adds to growing calls for progress. for a new paradigm for the governance of service delivery. 2. Think and work more politically in vio- lence-affected contexts. 3. Tailor interventions to different forms of Toward a New Programming violence. Approach 4. Take bargaining dynamics more seriously. Building on the findings and policy impli- 5. Realign internal donor rules and incentives. cations outlined above, the report presents the foundations of an operational framework 6. Explore new models of delivery and new to help better tailor service delivery to con- research in this field. text-specific forms of politics, violence, and Figure ES.1. Proposed framework for strengthening service delivery in violence-affected contexts 1. Improve diagnostics for measuring the actual scope of progress Understand relevant political Map the relevant dimensions Draw lessons from local settlements, who the relevant of violence in the program service delivery and elites are, and how they matter areas bargaining 2. Think and work politically 3. Tailor approach to specific forms of violence 4. Take bargaining seriously 5. Reform donor rules and incentives 6. Explore new frontiers of delivery and research Executive Summary | xix Recommendation 1: Better Diagnose • Adequate organizational resources and in- the Risks and Scope for Progress centives should be dedicated to address- ing, and implementing, the findings. Service delivery practitioners could more sys- tematically assess the issues surfaced by this research. As such, the report provides a new Recommendation 2: Think and diagnostic tool that can be applied in different Work Politically in Violence-Affected contexts, consisting of three main modules. Contexts Module 1 includes submodules on the basic Tailor social service delivery interventions features of the national settlement, the nature to different types of political settlements of the sectoral settlement, the relationship be- Political settlements differ, particularly at the tween central and subnational elites, and the subnational and sectoral level. These differ- subnational settlement. Module 2 focuses on ences have important implications for SSD. the drivers behind and the organization, ide- Table ES.1 briefly outlines the different types ology, and localization of violence. Module of settlements that can be found and their im- 3 draws lessons from the implementation of plications for service delivery programming. different strategies for service delivery and bargaining. Key points to keep in mind in ap- Build coalitions for service delivery in plying this diagnostic include that: very fluid contexts Improving SSD is a technical challenge, but • The tool seeks to complement, not re- it is also a political one. Effective implemen- place, existing diagnostics. Its questions tation requires the building of coalitions— can be adapted and included in existing however limited, fluid, or localized—that analytical tools. enable some degree of service delivery im- plementation. While development actors have • The diagnostic does not need to be exten- limited tools to create and sustain coalitions, sive but does require careful application. three golden rules on coalition-building The process of ensuring organizational emerge from the research: uptake of the findings is perhaps more im- portant than the rigor of the diagnostic. • Expect the unexpected. Fluid and un- stable politics are the name of the game, so • The diagnostic is not a one-time exer- practitioners will need to adapt and keep cise: given the fluid nature of violent con- an ear close to the ground. texts, the diagnostic needs to be regularly updated. xx | Executive Summary Table ES.1. Types of political settlement in violence-affected contexts and implications for social service delivery Type of subnational/sectoral Example Implications for social service delivery (SSD) settlement programming Relatively stable, with ruling coa- Balkh Province, Afghanistan Reasonably good prospects for SSD, as incentives lition that has some incentives to (organized around regional align. Stability in settlement potentially allows for favor SSD strongman with pro-SSD planning over longer time horizon and the building of interests/incentives) a pro-SSD coalition. However, SSD needs to align with the ruling factions’ preferred mode of delivery (given Hill/mountain region of Nepal their relative power in setting the rules of the game). after large-scale violent conflict (organized around pro-SSD political party) Moderately stable, with mixed Wardak Province, Afghanistan, Medium prospects for SSD. Moderate stability allows SSD incentives (contested particularly before 2008 for better planning and better coalition-building between state, disciplined armed around SSD programs, but contestation can under- insurgencies, and powerful tradi- mine delivery depending on the level of contestation tional elites) at the front line of service delivery. Mechanisms of coordination between powerful elites may be needed. Moderately unstable, but Nepal during large-scale Some prospects for SSD, given pro-SSD ideology. presence of pro-SSD incentives violent conflict However, active contestation between state and (contested between state and insurgent groups potentially undermines service de- disciplined armed insurgency Balochistan, Pakistan livery. Efforts, compromises, and bargains likely to be that favors SSD) needed to reach consensus among contesting groups on “ring-fencing” services (to a degree). Moderately to highly unstable, Khyber Pakhtunkhwa, Pakistan SSD is likely to be challenging. Some state pene- with some state penetration in tration offers a modicum of stability that may help areas controlled by relatively with service delivery, but weak incentives as well as disciplined armed groups, with antagonism can undermine service delivery. weak incentives for SSD Highly unstable, with weak state Federally Administered Tribal SSD is likely to be challenging. Weak SSD incentives penetration, less disciplined Areas (Pakistan) as well as antagonism can undermine SSD. armed groups, and weak SSD incentives/ideology Highly unstable, with marked Badghis Province, Afghanistan SSD is likely to be very challenging. The lack of fragmentation of power (and no broadly agreed rules of the game as well as the monopoly of violence) and weak Tarai region of Nepal after presence of dominant ideology/incentives that seek incentives for SSD (weak polit- large-scale violent conflict short-term appropriation of SSD make progress ical organization and multiple difficult. Delivery needs to be flexible; practitioners centers of power controlled by need to learn by doing and (most likely) accept that armed criminal organizations) bargaining strategies must be built in. Executive Summary | xxi • Be realistic about the role of spoils • Small-scale interventions may circum- and rents. Given that rents often play a vent sectoral elite networks and spark a role in sustaining coalitions, controlling slow-burning political demonstration ef- violence, and allowing service provision fect, but scaling can be elusive. to be functional, think carefully before at- tempting to constrain or redirect them. • Bundling delivery into a suite of services— such as combining basic health services • Find “good enough” ways to support with access to other services—may increase (or at least not undermine) pro–service the political feasibility of success by buying delivery coalitions. This may include: off and appeasing opposing elites, though (1) helping pro–service delivery elites co- implementation logistics are challenging. ordinate themselves and act collectively; (2) reducing information asymmetries • Decentralized multistakeholder approaches between factions; (3) helping build trust may create the space for, and facilitate, across groups; and (4) increasing predict- more inclusive delivery bargains at the local ability in such contexts to enable longer level, but this requires strong in-built mech- time horizons. anisms to reduce the prospects that local elites will entirely subvert service delivery. Seek politically savvy modes of delivery Different modes of delivery can help bypass, Recommendation 3: Tailor Service or reduce the power of, forces opposed to Delivery to Different Forms of SSD. Depending on the local context, a multi- Violence tude of potential strategies can make delivery more politically savvy. Whatever the mode of In order to help practitioners better tai- delivery, programs are likely to face trade-offs lor programs to the specificities of violence rather than win-wins; these trade-offs need in their context, table ES.2 offers a stylized to be recognized and managed, not wished breakdown of the types of violence that away. Key lessons include the following: emerged in this research, and the implications for SSD. The rule of thumb is that the more • Semiautonomous delivery mechanisms fragmented, mobile, and ideologically antag- may reduce the power of sectoral cli- onistic the violent actors, the harder it will entelist networks, but sustainability is a be to deliver services and the more mitigation challenge. strategies will be needed. xxii | Executive Summary Table ES.2. Forms of violence and implications for service delivery Form of violence Implications for delivery of social services Well-organized, disciplined armed groups with Bargaining feasible (actors clear, incentives for incentives favoring social service delivery (SSD) service delivery aligned). Good prospects of direct Example: dominant/organized strongman allied state provision given alliance with ruling coalition. with ruling coalition, such as General Atta in Balkh, Resource transfer—from the center to the subna- Afghanistan tional level—may be substantial to maintain alliance. Well-organized, disciplined insurgent armed Bargaining feasible (actors clear, incentive align- groups with incentives favoring SSD ment). But poor relationship with the state means Examples: organized insurgency resisting or trying to strategies are needed to mitigate risks such as capture the state but wanting some services, such as nonstate provision and dialogue with insurgents to Maoists (Nepal), Balochistan (Pakistan) ensure that delivery of social services is not per- ceived as a proxy for state penetration. Somewhat organized/disciplined insurgent armed High risks of violence targeting delivery. Insurgent groups largely opposed to SSD leaders relatively clear in their antiservice ideology Example: relatively organized insurgents who want driven by their poor relationship with the state. Mit- to undermine delivery, for example of polio inoc- igation strategies could include bundling of services ulation in Khyber Pakhtunkhwa and the Federally to increase insurgent incentives not to undermine Administered Tribal Areas of Pakistan delivery. Somewhat organized/disciplined armed groups Complex scenario. Strategies for service delivery competing for regional power and legitimacy; will depend on various factors: local contestation positions favoring service delivery contested may lead to competition to improve services or to Example: large parts of Wardak Province, capture/undermine services. Competition may also Afghanistan be over who is seen to deliver the service. Disorganized/fragmented, mobile/roving A more difficult context (compared to the four pre- insurgent armed groups with some local roots vious ones). Leaders fragmented, power contested, creating incentives for service delivery and violence less predictable, with the result that Examples:fractionalized, uncoordinated armed bargaining is less feasible. Second-best solutions groups with some incentives for service delivery as need to apply, such as using local implementers to in parts of Badghis, Afghanistan increase safety and securitize delivery. Highly disorganized/fragmented, highly mobile/ Most difficult context: leaders fragmented, power roving armed groups with criminal (nonpolitical) contested, and violence less predictable; incentives motives of violent actors are to undermine delivery of social Examples: some parts of Badghis, Afghanistan; to a services. Second-best solutions will be required. lesser extent, Tarai, Nepal, after large-scale violent conflict. Executive Summary | xxiii Recommendation 4: armed elites, or the provided services Take Bargaining Seriously might be accessed by insurgents and their allies, indirectly sustaining them. Building on the previous recommendations, three practical messages emerge from the • Trade-offs of bargain-sensitive program- findings on the role of bargaining in SSD: ming: For example, bargains might help short-term, localized progress of ser- 1. We cannot simply ignore the role vice delivery, but little is known about of bargaining. In all of the cases stud- if and how they might contribute to ied, forms of bargains with armed elites longer-term state or peace-building and allied insurgent groups were critical objectives. Using locally trusted actors in enabling some progress in SSD. There and volunteers to deliver services may was little evidence, however, that project reduce the risks of attack and main- implementers were aware of, or systemat- tain service quantity, but such use may ically considered, such dynamics. Making also hinder efforts to improve service project choices without at least consid- quality (because volunteers may not be ering the role of bargains may put SSD able to deliver complex services, for at greater risk and lead to “shots in the example). Also, service delivery may dark.” Moreover, bargaining and compro- be maintained via bargaining, but re- mises may not be a “first-best” option, but source flows may be diverted to areas they may be the only way to make prog- and groups seen by powerful insur- ress in some contexts (see discussions on gents/elites as more important. road construction in Nepal and education in Pakistan in the report). 3. Think about what it would take to de- velop a “bargain-sensitive” service de- 2. Carefully consider the trade-offs of livery program. As noted, the question both traditional and bargain-sensitive of whether bargaining with armed elites approaches to service delivery. In such and allied insurgents to deliver services is difficult, violent contexts, trade-offs tend better than slower or no progress must be to be prominent across the board: answered case by case. The report offers some preliminary ideas on mainstream- • Trade-offs of traditional programming: For ing bargain-sensitivity in service delivery example, a traditional program (that is, design, which includes mechanisms to: one that ignores the role of bargaining) (1) facilitate dialogue and negotiations might unknowingly create rents that with armed elites and allied insurgents; sustain locally powerful elites and/or (2) enlist the support of locally influential xxiv | Executive Summary nonarmed elites; (3) make deliberate con- This may include setting aside internal cessions and compromises in the way ser- budgets for the type of the diagnostics vices are delivered; and (4) facilitate (or proposed here, and changing the vocab- at least not undermine) some degree of ulary of programming. Indeed, this re- rent-sharing if it might produce more ser- search suggests that bargaining issues are vice delivery benefits than costs. often unknown, brushed under the carpet, or painted in an overly simplistic light. Yet internal management culture may not Recommendation 5: Reform Donor always be conducive to discussing such Incentives and Mind-sets challenges. Many of this study’s implications challenge tra- ditional approaches to the delivery of aid. In this • Mainstream strategies for bargaining regard, the report gives weight to the findings in delivery programs. Yet, fiduciary in the 2017 World Development Report (World rules and financial accountability may Bank 2017), and adds a voice to two contem- run counter to this, and there are various porary narratives around changing aid delivery: trade-offs to manage. (1) advocacy for doing development differently and thinking and working politically,1 which includes • Integrate incentives for adapting pro- adaptive programming;2 and (2) advocacy for a grams to subnational variation and more context-specific approach in fragile set- regularly adapt programs as they tings that systematically considers risks and unfold. This would require improved draws on lessons about what really works on incentives and systems for adapting to the ground (IEG 2016; Marc et al. 2016). subnational variation and lessons learned along the way. Yet adaptation to subna- This report goes further than these narratives tional variation can be costly and com- by offering a more granular framework for plex, and linear results frameworks are addressing politics and violence in SSD op- often the dominant mode of planning and erations. Furthermore, it pinpoints some spe- monitoring programs. cific measures—and internal dilemmas and risks—in order to further align the aid system • Modify human resource practices to with realities on the ground. These measures adapt to these realities. Steps could be include the following: taken to ensure that country offices have deep enough local, political knowledge, • Foster frank internal debate on elites such as by recruiting and retaining knowl- and bargaining to change mind-sets. edgeable local staff, or reducing the high Executive Summary | xxv turnover of international staff who may • Service delivery is rarely politically serve only one or two years. Yet, attracting neutral. The research, by showing how and retaining people in violence-affected SSD is a source of rents that differs by contexts can be extremely difficult. sector, undermines traditional approaches that suggest it can be technical, neutral, or ring-fenced from politics and power. Recommendation 6: Pursue New Frontiers for Delivery and Research • Bargains and trade-offs are part of the Finally, the report offers the seeds of a new new normal. Bargains, trade-offs, and paradigm for SSD. There are different mod- compromises are integral to service deliv- els for service delivery. This research is most ery—an element scarcely treated by other aligned with an approach that views service service delivery paradigms. delivery as profoundly embedded in, and shaped by, power politics. This differs from Finally, the report gives way to a set of mainstream approaches to service delivery in future research questions, which include the the following ways: following: • No one service delivery modality is • How can this report’s typology of vio- privileged. Different models can be more lence be deepened and further tested in or less effective depending on local polit- other geographical areas to strengthen its ical settlements, the form of violence, and generalizability and applicability? the opportunities for bargaining, as out- lined in chapter 6. • How do development interventions create and sustain new bargaining structures? Are • New actors are brought into much there other examples of or lessons from sharper focus. The findings push us to SSD cases in which bargaining was inte- analyze the central role, interests, and in- grated and its risks effectively mitigated? centives of different elites in service deliv- ery, such as landlords or warlords. • How can donor agencies reconcile their fiduciary requirements (accounting and • Political economy is not just about procurement procedures) with the need material power; it is also about ideas. for some flexibility and bargaining in We show the important role of “ideas” SSD? What aid instruments might be most (held by elites, armed or unarmed) in appropriate? shaping the constraints and opportunities for SSD. xxvi | Executive Summary Introduction This report provides the foundation for a new need access to social services. This raises di- approach to service delivery in violence-af- lemmas about the ethical and political judg- fected contexts that is sensitive to the actual ments and trade-offs that development actors forms of violence, politics, and bargaining may have to make. A key challenge is whether encountered in such polities. It raises sensi- development actors can adapt their procedures tive issues about how development organiza- and ways of working to the fluidity, uncertain- tions that work with recognized governments ties, and risk taking that the new, conflict-riven should approach service delivery in contested landscape demands while preserving financial settings. As many countries today are riven by accountability, doing no harm, and ensuring conflict and internal division, some familiar aid effectiveness. This challenge cannot be ig- rules of the game may be inadequate to deal nored, as progress in basic service delivery in with the mounting humanitarian and devel- violence-affected contexts will make or break opment challenges posed by complex conflict global efforts to end poverty and increase situations, particularly where affected people shared prosperity over the coming decades. Community at discussion of water supply and sanitation. Kaski Nepal. Simone D. McCourtie/World Bank Introduction | 1 The report has three parts. The reader is en- Chapter 3 explores how political settlements couraged to read them all, but may find cer- at various levels shape the spaces, constraints, tain parts or chapters of more relevance for and opportunities for different forms of ser- their purposes. Part I presents the overall ap- vice delivery. Chapter 4 focuses on violent proach and key findings from this study. Part actors—that is, armed elites, including insur- II, for the more interested reader, presents a gents and warlords—and explores how the more detailed and rich ethnography of the varied forms, organizations, motivations, ide- main findings presented in Part I, including ologies, and levels of mobility among these some of the stories behind the big messages. actors affect SSD. Chapter 5 examines the Part III draws conclusions and policy im- ways that elite bargains are critical to mak- plications, and identifies avenues for further ing or breaking SSD. The chapter explores thought and research. how bargains occur and how they may per- mit progress, albeit with trade-offs. Chapter Part I has two chapters. Drawing on the crit- 6 concludes with an exploration of how cer- ical gaps in our knowledge about the obsta- tain modes of delivery seem to work better cles to social service delivery (SSD) in areas than others in navigating through regressive beset by violent conflict, chapter 1 presents aspects of local politics and violence. a framework for research organized around four key elements, the interaction of which Part III, in two chapters, presents a set of pol- explains the success or failure of SSD: politi- icy takeaways and recommendations. Chapter cal settlements (mainly at the subnational and 7 urges development practitioners to adapt to local levels but also at the national level); the context, to recognize that not all forms of vi- varied forms of and motivations for violence olence affect service delivery equally, to turn at the subnational level; the potential for into action the knowledge that elites matter, forging elite bargains to allow some degree and to be realistic about good governance. of service delivery; and the extent to which The six recommendations offered in chapter the mode of delivering services is sensitive to 8 echo the policy implications sketched in the the first three concerns. Chapter 2 applies the previous chapter. They are: to regularly mea- framework to conflict-riven areas in Afghan- sure the actual extent of progress, to think and istan, Nepal, and Pakistan; and outlines the work politically, to tailor service approaches main research puzzles and findings from the to the specific forms of prevailing violence, research. to take bargaining seriously, to revise donor rules in light of research and experience, and Part II, which delves into the rich ethno- to be open to new frontiers of service delivery graphic data, is divided into four chapters. and research. 2 | Introduction Part I. Research Approach and Major Findings Afghan children look from the window of their shelter in the former Soviet embassy, a huge compound amid the destruction of west Kabul, November 27, 2001. Damir Sagolj/Reuters Part I. Research Approach and Major Findings | 3 1. Research Approach “Fragility, Conflict and Violence remain among the most pressing global policy issues today and constitute major obstacles to achieving development objectives.” (World Bank 2016b: 44) “We now have to admit that the paradigms and frameworks that have guided our perspective on fragility and conflict may be less relevant than we had hoped.” – Dr. Jim Kim, World Bank President, Fragility Forum opening remarks, March 2016. (World Bank 2016c) Fragility, conflict, and violence are among The impact of fragility and violence spills the most urgent global policy challenges, across borders. Violent conflict has resulted in representing significant barriers to achiev- major displacement and has contributed to the ing global development objectives. About so-called refugee crisis that has recently en- 2 billion people live in countries where de- gulfed parts of the Middle East, North Africa, velopment outcomes are affected by fragil- and Europe. According to the former Secre- ity, conflict, and high levels of violence, and tary General of the United Nations (UN): where there is a clear correlation between fra- gility and poverty. These states have experi- “The world today is witness to higher enced the slowest pace of progress toward any levels of conflict than a decade ago, of the Millennium Development Goals. The and the highest level of forced dis- average poverty rate among countries that ex- placement since the Second World perienced major violence between 1981 and War . . . The drivers of violence and 2005 is 21 points higher than where there instability are more complex and more was no violence (World Bank 2011: 5). Over intractable than ever. Nothing empha- half of the population in these low-income sizes this reality more than the conflicts violence-affected countries lives in extreme in the Middle East and North Africa. poverty, accounting for about 20 percent of To address the scale and the nature of the world’s extreme poor in 2012. By 2030, the conflicts, we need new approaches” an estimated 50 percent of the world’s poor – Ban Ki Moon will live in a country affected by violence (UN Secretary General 2015). (World Bank 2016b). Part I. Research Approach and Major Findings | 5 While the delivery of basic services in fragile health care, education, water and sanitation, and conflict-affected settings is critical, it is and infrastructure. extremely difficult. For example: This report explores the challenge of de- • A mere one-fifth of fragile states are on livering social services in violence-affected track to achieve universal primary school- contexts. ing compared with nearly half of nonfrag- ile developing countries (OECD 2015). Background case studies from Afghanistan, Nepal, and Pakistan are synthesized with the • Only 28 percent of fragile states are on latest global evidence to chart both a concep- track to cut in half the number of citizens tual and operational path forward.3 In chart- who do not have access to safe water com- ing that path, the report attempts to fill some pared with 61 percent of nonfragile coun- important knowledge gaps and to pilot a new tries (OECD 2015). analytical framework to answer unresolved questions, as outlined in the next section. • A person living in a fragile or violent set- ting is almost three times more likely to be undernourished (Bread for the World 1.1. Knowledge Gaps and Institute 2016). Practical Challenges Given that such services play a critical role The Sustainable Development Goals are in reducing poverty, it is no wonder that the unlikely to be achieved unless we address development community is highly concerned. important knowledge gaps and improve practices on the ground. For the purposes The international community has responded of this report, eight interrelated gaps can be with commitments to improve governance highlighted:4 and development assistance in such con- texts. Goal 16 of the post-2015 Sustainable 1. Differential rates of success in deliv- Development Goals is to: “promote peaceful ering social services in violence-affected and inclusive societies for sustainable devel- contexts have been poorly explained. opment, provide access to justice for all and Progress in social service delivery (SSD) has build effective, accountable and inclusive in- been achieved in some violence-affected con- stitutions at all levels” (UN 2015). Many of texts, but the knowledge base is limited in the Sustainable Development Goals also focus terms of explaining: on improving services such as agriculture, 6 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds • Why some SSD interventions succeed Bank 2017) puts governance and politics front while others fail in the same country, and and center. It shows how policy is made and implemented in complex political and social • Why there is variation among violence-af- settings in which individuals and groups of fected countries. unequal power pursue conflicting interests ac- cording to shifting rules. For example, World Bank–supported research has examined the nexus between conflict and 3. While there is a growing consensus development (World Bank 2011); explored the that governance and politics matter for political dimensions of development in frag- service delivery, the devil is in the details: ile contexts (Bain, Porter, and Watts 2015); • The consensus breaks down rather easily. Less unpacked the societal dynamics of fragility agreement can be found regarding pre- (Marc et al. 2013); and shown how institutions cisely how governance matters; what as- take root and services get delivered in difficult pects of governance most affect SSD; and contexts (Barma, Huybens, and Vinuela 2014; what can be done by development actors to Brixi, Lust, and Woolcock 2015). However, influence governance. As one research con- none of these studies have focused systemati- sortium puts it, “Almost everyone agrees cally on SSD or have explained how and why that ‘politics matters’ for development, but SSD progress occurs within and among differ- there is little consensus about what this ac- ent violence-affected contexts. tually means” (ESID 2016). Similarly, it is widely recognized that “context matters,” 2. Prevailing technical approaches to ser- but there is limited clarity on exactly how vice delivery have had limited success. it matters in violence-affected situations. Failure to deliver basic social services in vio- lence-affected countries is often attributed to • The good governance paradigm continues to technical deficits, such as weak capacity, lim- dominate our understanding of what forms of ited resources, and a lack of expertise, but a politics matter. It focuses on promoting for- body of evidence demonstrates that technical mal rules; fostering liberal-democratic and approaches tell only part of the story. Political best practice norms of transparency and economy dynamics also influence the function- accountability; and making technocratic ing of service delivery agencies and underpin fixes. Yet a growing number of observ- institutional capacity constraints (Mcloughlin ers question its utility in explaining the and Batley 2012; Leftwich and Wheeler 2011; forms of governance that drive develop- Levy and Walton 2013; Unsworth 2010; Wild ment progress in a number of countries et al. 2015).5 Most recently, the World Develop- (Andrews, Pritchett, and Woolcock 2012). ment Report 2007: Governance and the Law (World Part I. Research Approach and Major Findings | 7 • Alternative governance paradigms need to be on those that in the formal scheme of fleshed out and mainstreamed. In light of things are supposed to set the insti- the perceived failings of the good gov- tutional and policy framework gov- ernance paradigm, attempts have been erning implementation.” made to develop new paradigms, such as “best fit” or “good enough governance” 4. An analytical approach based on “po- (Albrecht 2013; Levy 2014; World Bank litical settlements” has helped move the 2017: 11–12).6 However, these alternative debate forward, but its full promise has approaches need to be fleshed out, main- not yet been realized. The approach ex- streamed, and operationalized. plores how the underlying balance of political power among contending elites and political • The evidence base is limited, especially in vio- coalitions, and the negotiations and bargains lence-affected contexts. Systematic evidence between them, are critical in explaining de- on the way governance and political velopment outcomes. However, there are at economy factors impact service delivery least two reasons why this approach has not is growing, but it remains quite limited, fulfilled its practical potential. First, the ap- especially in violence-affected contexts.7 proach can often fail to offer concrete policy As Mcloughlin and Batley (2012: 30–31) recommendations. As Kelsall (2016: 1) notes: note: “despite its intuitive appeal, it seems diffi- cult to use in practice.” Indeed, various aid “Significant gaps in our understand- agency officials believe it has strong poten- ing remain . . . [service delivery] tial but have complained that it does not offer research needs to give special im- enough granular detail to inform everyday portance to the point of implemen- programming.8 Second, there have been ex- tation, where formal policies most tremely few attempts to link political settle- often fail and where ‘real’ policies ment thinking to service delivery, especially emerge from the interplay of inter- in violence-affected contexts (Hickey 2013: ests and incentives . . . and adopt an 11; Kelsall, Hart, and Laws 2016: 1). essentially ‘bottom-up’ approach to the field research, working upwards 5. The terms “fragility” and “violence” in order to identify the key political have not been sufficiently defined, ana- factors that underpin [service deliv- lyzed, and applied to service delivery. ery] performance. This would put the focus on those institutions, in- • Mainstream definitions of “fragility” have centives and actors that are effective been found to be lacking. Fragile states are at the point of delivery—rather than generally defined as those states without 8 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds the will or capacity to function in ways • Little is known about how different forms of that reduce poverty, ensure development, violence impact SSD. We also know that “vi- or safeguard human rights. This definition olence” takes different forms, but there are does not distinguish between the particu- no attempts to understand how these dif- lar conditions of “fragility” and the gen- ferent forms might have different impacts eral conditions of “underdevelopment.” on SSD. Indeed, how and why political Indeed, the label “state fragility” has been violence affects SSD is poorly understood applied not only to poor, poverty-stricken (Putzel and Di John 2012). Moreover, countries that experience large-scale po- many studies focus on the negative impact litical violence and warfare, but also to violence has on SSD. Only a few studies poor, poverty-stricken countries that have have partially sought to explain progress achieved long periods of peace (Putzel in SSD in the face of violence (for exam- and Di John 2012). As one piece sums up: ple, Baird 2011; OECD 2015).10 “… fragility is not a static condition, 6. There is a limited understanding of nor is it an all-or-nothing experience. how different strategies and modalities Fragility can best be understood as a of SSD might make a difference in vio- continuum: societies can experience lence-affected contexts. Important debates extreme state failure and violent con- continue around what modes of delivery are flict at one end and varying degrees most appropriate in violence-affected con- of fragility at other points along the texts. Ideas range from those inspired by World continuum.” (Marc et al. 2013: 2) Development Report 2004: Making Services Work for the Poor (World Bank 2003), to commu- • Fragile and violent contexts have too often been nity-driven development (World Bank 2011). conflated. The specific challenges of deliv- Moreover, an important strand of literature ering SSD in the context of large-scale on modes of delivery suggests that different political violence have not been dealt with sector characteristics—such as opportunities explicitly in policy and academic circles. for rents and different ways of delivering One of the reasons for this is that World services (for example, top-down versus bot- Bank classification criteria for “fragile and tom-up)—present different constraints and conflict-affected states” are not based on opportunities for negotiating the politics of an assessment of the prevalence of violent delivery (Harris, Mcloughlin, and Wild 2013). conflict or on estimates of political fragil- Yet this assumption has rarely been tested in ity—that is, the likelihood that the politi- violence-affected contexts. Further, focusing cal order will collapse (Moore 2014).9 on a sector’s characteristics does not explain Part I. Research Approach and Major Findings | 9 why SSD can perform so differently in the • Improvement in policy and operations in fragile same sector within and across less-developed and conflict-affected contexts. These include countries in general and violence-affected mainstreaming enhanced risk monitor- countries in particular. ing and management; further identifying, building on, and integrating lessons about 7. There is a paucity of firsthand, primary what works; and refining operational ef- data from violence-affected contexts. Re- fectiveness and flexibility (World Bank search in violence-affected settings is more re- 2016b: 44–45). stricted, costlier, and riskier. As a result, many studies are at worst based on secondary litera- • A better understanding of the actual mecha- ture reviews or at best on interviews with key nisms through which political settlements affect informants from capital cities. These sources SSD. A better grasp of this could lead to provide only a partial picture of the on-the- insights on what donors might do differ- ground reality and a weak understanding of ently (DFID 2016). the dynamics of front-line delivery; and they fail to include perspectives of front-line service users, providers, and other stakeholders. 1.2. Scope and Framework of Study 8. Improved policy and practice guid- ance are urgently needed. There are global In order to address the above-mentioned calls to develop an improved set of practical gaps, this report addresses the following core tools to address development challenges in questions: these difficult contexts (World Bank 2016b). Recent reports have pointed to a number of • How and why do political economy fac- critical needs: tors and forms of violence shape the im- plementation and outcomes of SSD in • A differentiated approach. The World Bank’s violence-affected contexts? Independent Evaluation Group recom- mends developing differentiated mecha- • What explains the variation of SSD prog- nisms for assessing and classifying fragile, ress within and between violence-affected conflict, and violence-affected contexts; contexts? tailoring to country context more system- atically; and adopting more realistic ap- • What lessons can be taken from this to in- proaches (IEG 2013). form future analysis, policy, and practice? 10 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Inductive Approach, Case Selection, and focus group discussions (FGDs); a sys- and Methodology tematic analysis of interview transcripts; and multistakeholder consultations. Case studies To answer these questions, the study induc- were selected based on research puzzles de- tively developed and piloted a new approach veloped from gaps in the literature (see ap- to thinking about service delivery in vio- pendix A for a more detailed discussion of lence-affected contexts. Inductive reasoning the methodology).12 This, as with all research, is based on learning from experience. Pat- carries certain caveats, as outlined in box 1.1. terns, resemblances, and regularities in ex- perience (premises) are observed in order to reach conclusions (or to generate theory). As The Analytical Framework’s Bernard (2011: 7) puts it, inductive research Four Key Elements “involves the search for pattern from observa- To address the above-mentioned knowledge tion and the development of explanations— gaps and research questions, an analytical theories—for those patterns through series of framework was developed. The basic prem- hypotheses.” As such, an analytical framework ise is that SSD in violence-affected settings was developed and applied inductively—the is shaped by the interaction between political framework is described in the next section.11 settlements, forms of violence, the nature of The research focuses on service delivery in elite bargains, and modes of delivery (see fig- the health and education sectors in Afghan- ure 1.1).13 A brief outline of these terms and istan and Pakistan and the health and rural key typologies follows. infrastructure sectors in Nepal. The research includes six projects from these sectors. Political settlements Subnational field research sites within each A political settlement is a common under- country were selected for in-depth study and standing among elites that their best in- comparative analysis. Each of the subnational terests—in terms of access to property areas are violence-affected, but have different rights and positions within and outside the characteristics, as outlined in chapter 2. state—are served by a specific way of orga- nizing political power. It is the “distribution The study methodology is qualitative and of power between contending social groups combines multiple methods. Main methods . . . on which any state [or social system] is include a systematic meta-review of global based” (Di John and Putzel 2009). Political evidence; a review of the country, sector, and settlements involve a specific combination of intervention-related evidence; semistructured institutions and set of norms as well as power- key informant interviews (KIIs), field visits, sharing agreements that help achieve the Part I. Research Approach and Major Findings | 11 Box 1.1. Some research caveats The case selection and research methodology do not claim to be representative: further research is required. This work can be considered a starting point. Research is based on a restricted number of projects and sectors. The case studies and research questions were developed inductively and aimed to address important gaps in the current understanding of how service delivery progress happens in the case study contexts (see chapter 2). The ultimate goal was to begin addressing these gaps and to identify patterns for further investigation and corroboration (see chapter 8). That said, the findings are triangulated with the national and global evidence base in order to increase the generaliz- ability of the claims made. Indeed, generating a statistically representative (so-called “large N”) sample in such contexts, given the security and financial implications, is a perennial challenge. The report does not claim to offer an exhaustive view of all the factors that affect progress in service delivery in violent contexts. The focus on forms of settlement, violence, bargains, and service delivery does not mean that other factors (such as beneficiary participation, service provider incentives, and the technical quality of services, among others) are unimportant. These four dimensions emerged as important for filling the gaps and inadequacies in the global and country literature. This offers a different perspective on why things work, or do not. As noted above, this should stimulate further debate. The study does not explore the impact of SSD on peace- or state-building, or on the causes of fragility. The focus of this research is the “how to” of delivering services in the presence of vio- lence, and it explores the delivery of services as a goal in its own right. It does not address the question of whether or how service delivery may contribute to, or undermine, state-building, peace-building, or conditions of fragility. As an aside, we think that the link between SSD and state legitimacy is too indirect and broad to provide granular, day-to-day guidance for policy makers and practitioners on the ground. The assumption that delivering basic services contributes to state-building outcomes appears with a striking degree of regularity in the literature even though on-the-ground empirical evidence supporting or undermining this link is very thin (Carpenter, Slater, and Mallett 2012; Haar and Ruben- stein 2012). The report does not endorse the studied armed elites nor does it necessarily condone the observed collusion between these groups and service providers. Whatever views about armed groups are presented in the report come from key informant interviews, focus group discussions, and related literature. What the report does point to, however, is the need to accept that local “unsavory” bargains are sometimes necessary to deliver humanitarian and development assistance in these contexts. 12 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Figure 1.1. Study framework Political Forms of settlements violence Nature Modes of of elite delivery bargains SOCIAL SERVICE DELIVERY Implementation and Outcomes minimum conditions needed for the society’s A political settlement is not a single, clearly economic and political viability (Khan 2010). articulated agreement, such as a financial, They exist at two levels: horizontal negotia- peace, or legal settlement. It is a rolling agree- tions between elites and vertical relations be- ment that is subject to continual renegotiation tween elites and their followers. and contestation: Part I. Research Approach and Major Findings | 13 “… governance, stability, and the qual- Political settlements in low-income and vi- ity and pace of development are viewed olence-affected contexts are likely to cre- as the outcome of struggles and ensuing ate and/or sustain rents, rent-seeking, arrangements among powerful elites. patron-client, and corruption networks. In These struggles largely involve informal contexts characterized by weak and unsta- processes of conflict, negotiation, and ble ruling coalitions and a poorly enforced compromise.” (Parks and Cole 2010: 5) rule of law—common to violence-affected polities—the creation of mostly informal Political settlements are more unstable and rent-sharing agreements among elites as well fluid in violence-affected contexts than in as their maintenance through restrictions on nonviolent low-income contexts. The litera- economic and political competition may be ture on elite bargains and political settlements prominent (Haber, Razo, and Maurer 2003; attests that the outbreak of political violence North, Wallis, and Weingast 2009; Acemoglu represents a crisis—a situation in which at and Johnson 2005; Khan 2010; Devarajan, some level of the polity, the underlying elite Khemani, and Walton 2011). bargain and broader political settlement is unable to manage social order or contain vi- In sum, political settlements set the broad, olent conflict (see Putzel and Di John 2012). often informal, rules of the game and cre- In other words, this represents a breakdown ate constraints, spaces, and opportunities for of widely agreed-on formal and informal SSD. They provide a map of the ruling coali- rules governing the allocation of resources tion: who the power brokers are; the balance and rents and the peaceful settlement of of power between them; and what interests grievances, as well as often a challenge to the and norms they are defending or promoting ruling coalition (North et al. 2007; North, in economics, politics, and service delivery. Wallis, and Weingast 2009).14 In order to explore the relationship between Levels of instability of the political settlement political settlements and SSD, four interre- across regions and sectors may vary within lated “levels” of the settlement are identified violence-affected polities. Civil wars do not in the framework (see figure 1.2):15 always challenge the ruling coalition at every level of the polity (Kalyvas 2003). These vari- 1. National settlement. The national set- ations could affect SSD because the more un- tlement is the configuration of power at stable the political settlement, the shorter the the central state level. It usually includes time horizon that power brokers and ruling elites from dominant groups who tradi- coalitions have for calculating the political tionally have access to political power at and material benefits of any action. the national level. It can shape the overall 14 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Figure 1.2. Dimensions of settlements that affect service delivery Relationship National between Sectoral Subnational (primary) national and settlement settlement settlement subnational elites SERVICE DELIVERY constraints and space for service delivery including the degree of state penetration at the level of policy and implementation. and control over insurgents, the flow of resources to localities, and the autonomy 2. Sectoral settlement. Linked to but dis- and control of subnational elites over ser- tinct from the national settlement, a sec- vice delivery policy and implementation. toral settlement involves elite incentives and ideas related to particular sectors, such 4. Subnational settlement. Relating to the as health, education, and infrastructure, elite balance of power and the incentives, as well as the dominant norms of power ideas, and power base among elites at the sharing between elites within the sectors. subnational (provincial/district and vil- Sectoral settlements may influence SSD in lage) level, this is important. Arrangements various ways, such as elite rent-sharing in among powerful elites who control political the sector or the offer of various opportu- competition and resource allocation at the nities for elite bargaining. subnational level do not necessarily align with national settlement arrangements. The relationship between central 3. and subnational elites. The linkages At the same time, “settlements” at all the and relationships between ruling elites at above-mentioned levels can have different the national and subnational levels may cross-cutting characteristics. These key char- shape service delivery in numerous ways, acteristics are: Part I. Research Approach and Major Findings | 15 Table 1.1. Types of political settlements in violence-affected contexts Type of subnational/sectoral settlement Example from our case material Relatively stable, with ruling coalition that has some Balkh Province, Afghanistan (organized around incentives to favor social service delivery (SSD) regional strongman with pro-SSD interests/ incentives) Hill/mountain region of Nepal after large-scale violent conflict (organized around pro-SSD political party) Moderately stable, with mixed SSD incentives (con- Wardak Province, Afghanistan, particularly before tested between state, disciplined armed insurgen- 2008 cies, and powerful traditional elites) Moderately unstable, but presence of pro-SSD Nepal during large-scale violent conflict incentives (contested between state and disciplined armed insurgency that favors SSD) Balochistan, Pakistan Moderately to highly unstable, with some state pene- Khyber Pakhtunkhwa, Pakistan tration in areas controlled by relatively disciplined armed groups, with weak incentives for SSD Highly unstable, with comparably weaker state pene- Federally Administered Tribal Areas (Pakistan) tration, less disciplined armed groups, and weak SSD incentives/ideology Highly unstable, with marked fragmentation of Badghis Province, Afghanistan power (and no monopoly of violence) and weak incentives for SSD (weak political organization and Tarai region of Nepal after large-scale violent multiple centers of power controlled by armed crimi- conflict nal organizations) • The degree of stability and contestation Taken as a whole, this led us to develop a in the settlement; The degree to which sixfold typology (table 1.1) that relates to our there is a developmental and/or SSD ori- case studies. More is said on this in chapters entation (however limited) as part of the 2 and 4. settlement; Forms of violence • The degree to which the settlement is ex- The nature of a political settlement does not clusionary or inclusive of the range of so- sufficiently explain how and why services are cial or political groups; and delivered in violence-affected contexts; forms of violence need also be considered. Violence • The balance of power between un- can result from, as well as affect, the nature of a armed elites and armed elites and allied political settlement, and its influence is uncer- insurgents.16 tain and variable. As noted above, an outbreak 16 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Figure 1.3. Dimensions of violence that affect social service delivery Motivation, Localization incentives, Organization and and mobility ideology SERVICE DELIVERY of political violence can represent a crisis our research framework suggests that three where contending groups making up a par- key dimensions of violence are important for ticular political settlement are unable to share service providers to consider: (1) its organi- power and resources in a way that secures so- zation; (2) its motivations, incentives, and cial order. That is, factions within or outside ideology; and (3) its localization (figure 1.3). the ruling coalition—“excluded elites”—de- These dimensions are fluid, interrelated, and cide that it is in their interest to challenge the evolving. status quo through violence. This can occur at the national or subnational level.17 Organization of Violence While important work has been done on how People take up arms for different reasons in the nature of the political organizations pur- different contexts, with varying objectives suing violence affects a war’s duration, the and organizational structures and hierarchies. durability of peace agreements, and prospects The literature on violence shows how it needs for postwar democratization (for example, to be dissaggregated. For example, according Wood 2000), little is known about if and to Kalyvas (2003), violent contestations at the how differences in the organization of vi- regional or local level involve numerous dif- olence shape SSD. For the purposes of this fering agendas. Drawing inspiration from this study, the following dimensions of violence and Olson’s (1993) discussion of banditry,18 “organization” emerge as relevant: Part I. Research Approach and Major Findings | 17 • The degree to which violent actors are are resisting any or most types of central state organized, so that leaders are identifiable penetration in peripheral/border territories, and there is some recognised hierarchy of then determining who and how services decision making. should be delivered is a formidable challenge, and one that is not systematically examined • The degree to which which violent actors in the literature.19 In sum, the specific dimen- are disciplined, in the sense that decisions sions of “motivations” for violence that can taken by the leaders can cascade down play a role include the following: and be upheld. • The degree of alignment between the ob- • The degree to which violence is homog- jectives and ideology of the armed elites enous, as opposed to fragmented and and allied insurgents, and the logic, type, contested. scope, and mode of the service being delivered. Ideologies, Incentives, and Motivations behind Violence • The nature of armed elite and allied in- The second key dimension relates to the surgents’ ideological “policy” or position motivations and incentives for and the ide- toward certain services. ologies behind violence. While violence cer- tainly exacts large-scale costs, history is also • Armed elites’ and allied insurgents’ per- replete with examples of political violence ception of service providers, that is, are leading to improved developmental outcomes they perceived as friend or foe? (Tilly 1990) and sometimes even more inclu- sive, pro-poor outcomes (Putzel and Di John • The extent to which armed elites and al- 2012). As such, attention should be given to lied insurgents have an interest in main- the motivations of organized armed insur- taining a service for their own use or for gencies and to specific instances of political rent extraction. violence to determine if they might provide opportunities to reform SSD in more inclu- Localization and Mobility of Violence sive or pro-poor ways. Understanding what The extent to which armed groups are em- is motivating violence could also be import- bedded in local communities may affect the ant when thinking about the political risks prospects of local bargaining around SSD associated with delivering social services. For during times of violent conflict. The degree example, armed groups may allow SSD if it of localization seems to depend on at least aligns with their agenda or if they directly two factors: benefit from it. However, if armed insurgents 18 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Table 1.2. Six Types of violence Types of violence Example from our case material Well-organized, disciplined armed groups with Dominant/organized strongman allied with ruling incentives favoring social service delivery (SSD) coalition, such as General Atta in Balkh, Afghanistan Well-organized, disciplined insurgent armed Organized insurgency resisting or trying to capture groups with incentives favoring SSD the state but wanting some services, such as Maoists (Nepal), Balochistan (Pakistan) Somewhat organized/disciplined insurgent armed Somewhat organized insurgents who want to groups largely opposed to SSD undermine the delivery of particular services, such as polio inoculation in Khyber Pakhtunkhwa and the Federally Admininstered Tribal Areas of Pakistan Somewhat organized/disciplined armed groups Large parts of Wardak Province, Afghanistan competing for regional power and legitimacy; positions favoring service delivery contested Disorganized/fragmented, mobile/roving Fractionalized, uncoordinated armed groups with insurgent armed groups with some local roots some incentives for service delivery as in parts of creating incentives for service delivery Badghis, Afghanistan Highly disorganized/fragmented, highly mobile/ Some parts of Badghis, Afghanistan; Tarai, Nepal roving armed groups with criminal (nonpolitical) after large-scale violent conflict motives • The degree of reliance on local support. When Taken as a whole, the elements of violence armed elites rely on some degree of local outlined above led us to identify, using our support and the local constituents want case studies, at least six main types of violence the service, such elites may be more will- and their relation to SSD. These are outlined ing to enable the service. in table 1.2 and explored more in chapters 2 and 5. • Extent of localization or mobilization of armed elites and their supporting armed insurgents. As a general rule, the more fragmented, mo- Stronger links to local communities may bile, and ideologically antagonistic the vi- better enable local service delivery com- olent actors, the more difficult it will be to pared with more mobile insurgents who deliver services and the more mitigation strat- move frequently across community bor- egies will be needed. ders and geographies. Part I. Research Approach and Major Findings | 19 Elite bargains Bargains take different forms. It is difficult Elite bargains, which are linked to but distinct to know ahead of time what types of elite from political settlements, are important in bargains will matter. At least four overlapping violence-affected contexts. They are the sets forms of bargain emerged as important for of rolling interactions and implicit or explicit this research (figure 1.4): agreements between powerful actors—armed or unarmed—that shape if and how things are • Fostering dialogue and negotiations with done. Specifically, they involve the processes armed elites and allied insurgents, of conflict, negotiation, accommodation, com- promise, and resource sharing that characterize • Enlisting the support of influential local governing (Parks and Cole 2010). Such bar- nonarmed elites, gains are processes through which elite actors (that is, actors with the ability to directly influ- • Striking concessions and compromises, and ence outcomes within a given sector or issue) and the organizations that support them coor- • Sharing rents and tolerating corruption. dinate and commit to one another to determine outcomes (World Bank 2017: 197). Political Modes and strategies of delivery settlements describe the broad structure, but The final piece of the framework aims to ex- elite bargains involve agency, where national plore how different strategies and modes of and local elites make deals according to the SSD might make a difference. While political rules of the game, or they attempt to relax ex- settlements, forms of violence, and the ability isting political constraints by persuading con- to forge bargains are important drivers of or tending elites to change their interests or views barriers to progress, service delivery strategies regarding policies (Rodrik 2014). In this way, and implementation modalities also seem to elite bargains are central to any policy inno- matter (as the above literature review suggests). vations because they allow political agents to design their own strategies and policy spaces. There are different modes for delivering ser- As the 2017 World Development Report ar- vices. Although this list is by no means ex- gues: “Policymaking and policy implementa- haustive, three main modalities emerged in tion involve bargaining among different actors relation to the case material: (and their varying ability to influence others through control over resources, threat of vio- • State versus nonstate, lence, ideational persuasion, through existing rules)” (World Bank 2017: 7). • Centralized versus decentralized, and • Single sector versus multisector. 20 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds In addition to the modality of delivery, ser- • More or less “politically sensitive”: by re- vice delivery agents can undertake different ducing the influence of anti–service deliv- strategies. Such strategies can be: ery political coalitions; and • More or less “violence sensitive”: by antic- • More or less “bargain sensitive”: by creat- ipating and mitigating some of the risks ing space for and facilitating pro–service of violence; delivery bargains. Chapters 2 and 6 explore these issues in more depth. Figure 1.4. Overlapping forms of bargaining to deliver services Fostering dialogue and Enlisting Striking Sharing negotiations support concessions rents and with of influential and tolerating armed elites local elites compromises corruption SERVICE DELIVERY Part I. Research Approach and Major Findings | 21 2. Applying the Framework: Puzzles and Findings Following is an overview of the main find- 2016; World Bank and AREU 2016; World ings. It outlines the main research puzzles as Bank and SSB 2016) provide a more thor- well as a brief summary of how the frame- ough treatment of these issues. For quick ref- work was applied to generate answers. Part erence, tables 2.1, 2.2, and 2.3 sum up the II (chapters 3–6) as well as the background puzzles and findings. country reports (World Bank and Akram Map 1.1. Map of South Asia 22 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds 2.1. Afghanistan enjoying a high degree of political autonomy. This contributed to the emergence of a frag- Afghanistan (see map 2.2) has been a vio- mented political and military landscape with lence-affected country for almost 40 years, power often centered around local warlords. resulting in complex rivalries with a vari- By 1993, the country was sliding toward ety of actors, interests, and informal insti- factional war, with various armed irregular tutions. During the 1978–92 jihad period, forces competing for territorial control (Gi- there was a massive mobilization of actors, ustozzi and Ibrahimi 2013). The Taliban, such as religious and political counterelites, which emerged as a serious contender for who opposed the Soviet regime. With several state control in 1994, came closest to reestab- competing insurgencies involved, the mobi- lishing a monopoly of violence. In 2001, the lization was decentralized, with local fronts Taliban was ousted by the Northern Alliance Map 2.2. Map of Afghanistan Part I. Research Approach and Major Findings | 23 supported by the United States, leading to the EQUIP represents the government’s main ed- return of many of the factions that had bat- ucation program. The development objective tled for control of Kabul in the early 1990s. of EQUIP I (2004–09) was to: The signing of the Bonn Agreement for an “improve educational inputs and pro- interim power-sharing government in late cesses as a foundation for a long-term 2001 created a framework for a new political strategy through: a) the strengthening settlement and a new elite bargain, presided of schools and teacher training initia- over by Hamid Karzai. The new government tives in communities; b) investment in was heterogeneous, ethnically as well as ideo- human resources as well as physical logically, complicating former President Kar- facilities; and c) institutional develop- zai’s goal of unifying the country. From 2002 ment of schools, District Education to 2013, a delicate balance of power was Departments, Provincial Education achieved between strong provincial actors, Departments, and the Ministry of Edu- with the president at the center. The presi- cation.” (IEG 2010: 1) dent used access to major political or govern- ment posts to buy off powerful, provincial, EQUIP II (2008–16) seeks “to increase access and local elites. Endemic corruption and elite to quality basic education, especially for girls, impunity weakened the image of the govern- through school grants, teacher training, and ment and led to a resurgence of the Taliban in strengthened institutional capacity with the parts of the country after 2005. support of communities and private provid- ers” (World Bank and AREU 2016: 1). Progress in the delivery of health and educa- tion services in the country since 2002 has The SHARP program as well as its successor, been mixed. External aid has played a prom- the System Enhancement for Health Action inent role, and two major programs have been in Transition (SEHAT), are the government’s implemented: the Education Quality Improve- main health programs, delivering basic health ment Project (EQUIP) and Strengthening services. SHARP’s development objective was Health Activities for the Rural Poor (SHARP). to support the government in achieving the Both are “on budget” national programs fi- goal of the Health and Nutrition Sector Strat- nanced by a combination of the Afghanistan egy 2009–13 to “contribute to improving the Reconstruction Trust Fund and International health and nutritional status of the people of Development Association grants. Afghanistan, with a greater focus on women and children and underserved areas of the country” (World Bank 2014: 2). 24 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Puzzles no systematic studies to explain this. The re- search explored two key problems. From an aggregate national level, both EQUIP and SHARP have made progress. National Puzzle 1: How did health service delivery (and health and education outcomes have improved education until 2008) make solid progress in since 2002, starting from a low base. One re- the violence-affected, insecure, and clientelistic port on SHARP noted that it “was highly rel- political environment of Wardak Province? evant for the development challenges of the The province of Wardak is characterized by Afghan health sector in 2009 and implemen- high levels of insecurity, and yet it still man- tation was largely successful with minor chal- aged to make notable progress on certain lenges” (World Bank 2014: 19–20). health delivery indicators. The education sec- tor also made some progress, especially up However, some important subnational vari- until 2008. From 2004 to 2012, for example, ations should be noted. The education and the health sector in Wardak was marked by an health data, albeit patchy, suggests notable increase in the number of facilities, services, progress in certain provinces as well as vari- and medicines; and it also made progress on ation between different provinces over time. health outcomes such as child immunizations This progress and variation is counterintuitive and access to skilled antenatal care, although in the context of widespread violence and progress in the latter area slowed after 2008 “dysfunctional” governance, but we found (see figure 2.1). Figure 2.1. Access to skilled antenatal care, 2004–12 100 90 80 Badghis 70 Balkh Percentage 60 Wardak 50 40 30 20 10 0 2004 2008 2012 Year Source: World Bank and AREU 2016. Part I. Research Approach and Major Findings | 25 Puzzle 2: In the face of similarities in the level of insecurity and incidence of “violence,” why was service delivery implementation in general and health-care delivery in particular much more challenging in Badghis compared with Wardak? What was it about Balkh that made service delivery easier to implement and allowed it to achieve relatively good health and education outcomes, especially when compared with either Wardak or Badghis? Both Wardak and Badghis have been re- peatedly affected by violence, yet there was differential performance in service delivery implementation and outcomes. In Badghis, it was more difficult to deliver and sustain ser- vices, and its health outcome scores are no- ticeably lower than those of Wardak. Service delivery in Balkh was more organized than Wardak and Badghis, and its performance on a number of indicators for both sectors is rel- atively better than the others. Portrait of shop owner Abdul, Enjil District, Herat, Afghanistan Graham Crouch/World Bank Findings reliant on local support. Political networks Puzzle 1: How did health service delivery (and and organizations were linked to national education until 2008) make solid progress in and sectoral elites in Kabul. One important the violence-affected, insecure, and clientelistic trade-off to note is that while the local elite political environment of Wardak Province? facilitated social service delivery (SSD) in ed- One key driver of progress in Wardak was ucation; they did so selectively. While boys’ a series of diverse and rolling elite bargains. attendance increased, girls’ attendance in pri- The bargains were struck and upheld because mary and secondary school dropped in the of a more organized form of violence—en- post-2008 period, mostly because of an an- abling bargaining points to be identified— tagonism to girls’ education among the local as well as a relatively strong local elite who elite and armed groups, suggesting that the were supportive of the health (and education) ideology of local elites matters as much as the sectors and where the Taliban were relatively balance of power. 26 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Table 2.1. Social service delivery in Afghanistan: Framework, puzzles, and findings Prospects for Dominant form of pro-SSD elite Service and mode of Site Political settlement violence bargain delivery Puzzles and findings Balkh Relatively stable. Pro- Well-organized, disci- High Education: Centralized 1: How did health service delivery (and education until 2008) SSD incentives exist for plined armed groups state provision make solid progress in the violence-affected, insecure, and clien- ruling coalition. Organized with incentives favor- Health: Decentralized telistic political environment of Wardak Province? and centralized around ing SSD. Dominant/ nonstate provision • Diverse and rolling elite bargains in Wardak served as a key driver a regional strongman organized strongmen of progress. with pro-SSD interests/ allied with ruling • A more organized form of violence and a relatively strong local incentives. coalition (General Atta elite supportive of health and education enabled the identifi- Mohammed Noor). cation of bargaining points and the striking and upholding of Badghis Highly unstable. Situa- Disorganized/frag- Low Education: Centralized bargains. tion fluid; marked frag- mented, mobile/rov- state provision • Political networks and organizations were linked to national and mentation of power and ing insurgent armed Health: Decentralized sectoral elites in Kabul. no monopoly of violence; groups with some nonstate provision weak SSD incentives and local roots creating 2: In the face of similarities in the level of insecurity and inci- political organization; SSD incentives. Frac- dence of “violence,” why was service delivery implementation in multiple centers of power tionalized, uncoordi- general and health delivery in particular much more challenging controlled by armed crimi- nated armed groups. in Badghis compared with Wardak? nal organizations. • Highly fragmented, contested forms of violence and an unstable Wardak Moderately stable. Somewhat organized/ Medium Education: Centralized provincial settlement in Badghis compared with Wardak led to SSD incentives mixed; disciplined armed state provision fewer agreements and bargains and to the targeting of SSD for incentives are contested groups competing for predatory appropriation by roving groups. between state, disciplined regional power, legit- Health: Decentralized • More organized, disciplined form of violence and stable political armed insurgencies, and imacy, and SSD. nonstate provision settlement in Balkh enabled SSD implementation compared with powerful traditional elites the fragmented but relatively organized violence in Wardak and (particularly pre-2008). the highly fragmented and contested form of politics and violence in Badghis. • General Atta Mohammed Noor, a strong ally of the national ruling coalition, and current governor of Balkh Province, likely facili- tated elite bargains around securing resource flows to health and education in Balkh. • Greater economic prosperity and higher median household income in Balkh also contributed to greater progress in SSD. Part I. Research Approach and Major Findings | 27 Puzzle 2: In the face of similarities in the level of 2.2. Nepal insecurity and incidence of “violence,” why was service delivery implementation in general and Since the 1990s, Nepal (see map 2.3) has un- health-care delivery in particular much more dergone significant sociopolitical upheavals. challenging in Badghis compared with Wardak? The most significant was the “People’s War,” What was it about Balkh that made service launched by the Communist Party of Nepal delivery easier to implement and allowed it to (Maoist) in 1996, affecting the country for achieve relatively good health and education a decade. The Maoist insurgency is widely outcomes, especially when compared with either considered to have had a negative impact Wardak or Badghis? on SSD: health-care facilities, schools, com- In Badghis, compared with Wardak, highly munity centers, and local government offices fragmented and contested forms of violence were burned, while transportation within and as well as an unstable provincial settlement, between districts slowed down or was signifi- led to fewer agreements and bargains and to cantly reduced due to frequent checkpoints the targeting of SSD for predatory appropria- and curfews (Thapa and Sijapati 2005). tion by roving groups. A peace settlement was reached in 2006, but In Balkh, the more organized and disciplined the so-called postconflict period was marked form of violence and the stable political set- by fluid and unstable political settlements tlement—concentrated under General Atta and acts of violence. The mobilization of the Mohammed Noor, the governor of Balkh— enabled smoother SSD implementation com- Madhesis, for example, known as the “Mad- pared with Wardak (where the violence was hesi Movement,” called for the fulfillment of more fragmented, although still relatively or- certain rights and greater integration into the ganized) and Badghis (where politics and vio- political mainstream. Elements of popular lence were highly fragmented and contested). movements in the eastern and central Tarai The fact that General Atta was a strong ally of evolved into different armed groups with the national ruling coalition likely facilitated political demands; some resorted to various elite bargains around securing resource flows forms of violence and criminality. to both health care and education in Balkh. The greater economic prosperity and therefore The health sector in Nepal has been sup- higher median household income in Balkh ported by various donors both during and also contributed to greater SSD progress. The since the violent conflict. During the 1990s political settlement, characterized by a domi- and early 2000s, the government of Nepal nant, organized strongman, made governance supported health through the World Bank– more predictable, also likely contributing to supported Population and Family Health economic progress in the province. Project (1994–2000) as well as subsectoral 28 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Map 2.3. Map of Nepal projects supported by various donors. A sec- Rural infrastructure also received support torwide approach was then adopted in the during and since the wide-scale violent con- form of the National Health Sector Project. flict. The World Bank provided a loan for the Phase I, implemented in 2004–10, sought Rural Infrastructure Project in 1999–2003, in “to expand access to, and increase the use of addition to other sectoral investments (World essential health care services, especially by Bank 2005a). The Bank then supported the underserved populations” (World Bank 2004: Rural Access Improvement and Decentral- vii). Phase II, implemented in 2010–15, ization Project, implemented in two phases: aimed to “enable the Government of Nepal 2005–09 and 2010–13. Its objective was to to increase access to essential health care ser- assist “residents of participating districts to vices and their utilization by the underserved utilize improved rural transport infrastructure and the poor” (World Bank 2010: 7). and services in order to have enhanced access to social services and economic opportunities” (World Bank 2005a: 3; World Bank 2009: 5). Part I. Research Approach and Major Findings | 29 Puzzles Despite the adverse context of violence, political instability, and weak governance, Like Afghanistan, an examination of the Nepal Nepal registered progress. The country’s cases reveal impressive yet counterintuitive health sector made progress throughout the progress as well as some notable subnational Maoist insurgency and since (table 2.2). Two variation. Unfortunately, there is a dearth of aspects of this progress are particularly puz- literature to explain this, particularly from a zling. First, improvements occurred regard- political economy and violence perspective: less of the intensity of violent conflict. Table Puzzle 1: What explains the impressive progress 2.3 illustrates how the midwestern hill re- achieved in health delivery—that is, on track gion, which bore the brunt of the fighting, to achieving the Millennium Development did generally as well as the other regions Goals—in the context of large-scale violence and is not even the worst performer. Sec- and supposedly “dysfunctional” governance? ond, the progress occurred in a context that In addition, how were gains made in rural would normally be considered the antithesis infrastructure in this context? to good governance: a public administration Table 2.2. Health indicators, Nepal, 1990–2011 Millennium Development Goal/ Maoist insurgency Target impact indicator 1991 1996 2001 2006 2011 2015 Maternal mortality ratio (per 1,000 live births) 539 539 415 281 170a 134 Total fertility rate 5.3 4.6 4.1 3.1 2.6 2.5 Adolescent fertility rate (15–19 years) per 1,000 NA 127 110 98 81 70 population Contraceptive prevalence rate, modern methods 24 26 35 44 43.2 55 (percent) Under-five mortality rate (per 1,000 live births) 158 118.3 91 61 54 38 Infant mortality rate (per 1,000 live births) 106 78.5 64 48 46 32 Neonatal mortality rate (per 1,000 live births) NA 49.9 43 33 33 16 Percent of underweight children NA 49.2 48.3 38.6 28.8 29 Tuberculosis cases Detection rate (percent) NA 48 70 65 76 85 Successful treatment rate (percent) NA 79 89 89 90 90 Malaria annual parasite incidence per 1,000 NA 0.54 0.40 0.28 0.11 0.06 Source: MoHP 2010; Government of Nepal and UN Country Team of Nepal 2013. a. Further analysis of the 2011 census released in December 2014 showed the maternal mortality rate to be a high of 480 (Central Bureau of Statistics 2014). 30 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Table 2.3. Health service delivery, Nepal, 1996 and 2006 (by region) Vaccinations Demand for contraceptives satisfied a All basic vaccines None (percent) Percent Percent increase,  increase,  Subregion 1996 2006 1996–2006 1996 2006 1996 2006 1996–2006 Eastern Mountain 35.6 83.8 48.2 30.5 6.3 42.7 45.9 3.2 Central Mountain 51.5 54.6 3.1 19.1 8.4 35.6 68.1 32.5 Western Mountain 26.3 70.8 44.5 35.0 3.2 26.3 53.0 26.7 Eastern Hill 42.6 81.4 38.8 24.8 2.9 42.1 65.9 23.8 b Central Hill 67.9 82.1 14.2 13.5 7.2 63.1 72.8 9.7 Western Hill 63.8 88.9 25.1 9.9 0 43 49.5 6.5 Mid-western Hill 34.7 72.4 37.7 30.6 4.7 34.7 56.0 21.3 Far-western Hill 23.8 71.6 47.8 33.3 10.1 32.0 50.2 18.2 Eastern Tarai 48.5 89.5 41.0 6.8 3.0 56.4 71.2 14.8 Central Tarai 26.1 78.1 52.0 19.6 0 42.4 65.7 23.3 Western Tarai 31.7 88.8 57.1 27.5 0 44.4 66.0 21.6 Mid-western Tarai 49.6 90.4 40.8 21.9 0 61.2 73.7 12.5 Far-western Tarai 45.3 94.4 49.1 29.8 0 45.5 82.1 36.6 Source: Pradhan et al. 1997; MoHP, New ERA, and Macro International Inc. 2007. a. BCG tuberculosis vaccine, measles, and three doses each of DPT (diphtheria, whooping cough and tetanus), and polio vaccine. b. The data from this region appear to diverge from the national trend because they also include Kathmandu. The bolded entries denote the worst performers. Part I. Research Approach and Major Findings | 31 Village road. Nepal. Simone D. McCourtie/World Bank characterized by politicized appointments mountain regions of Palpa and Rolpa compared and rapid turnover of staff, the absence of with the Tarai lowlands of Siraha and Bardiya? locally elected governments, and wide- The data point to notable subnational vari- spread corruption and clientelism outside ation between the hill region and the Tarai and within sectors (World Bank and SSB (“lowlands,” “plains”). For example, the Tarai 2016). Progress was less impressive for rural was among the lowest-ranked regions on infrastructure, although some progress was health indicators such as the percentage of made, including at the height of the conflict women giving birth in a health facility, ante- (2004–06). natal care, and nutritional status. Meanwhile, the hill region ranked among the highest na- Puzzle 2: Why, in the post-2006 period, in a tionally on these same health indicators. In context where large-scale violence has ceased, terms of roads, Palpa, in the hill region, has aid has increased, and political settlements are seen rapid expansion in the years since 2006 fluid, has much greater progress been achieved while two districts in the Tarai have regis- in health and rural infrastructure in the hill and tered very limited progress (table 2.4). 32 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Table 2.4. Road networks in Siraha (plains), Palpa (hills), and Bardiya (plains) 2006–07 2011–12 2013–14 Siraha 143.93 144.93 144.93 Palpa 128.59 199.59 235.89 Bardiya 189.32 211.42 211.42 Source: World Bank and SSB 2016. Findings mountain regions of Palpa and Rolpa compared with the Tarai lowlands of Siraha and Bardiya? Puzzle 1: What explains the impressive progress Progress in the health and roads sectors was achieved in health delivery—that is, on track to stronger in the hill region than the Tarai due, achieving Millennium Development Goals—in the in part, to differences in subnational political context of large-scale violence and supposedly settlements, forms of political organization, “dysfunctional” governance? In addition, how and the nature of the violence. The Tarai is were gains made in rural infrastructure in this characterized by more predatory and feu- context? dalistic elite bargains and fragmented roving Progress was made in the health sector, even armed groups (often with criminal motives). in the face of violent conflict, due to a se- This form of violence makes it harder to ries of elite bargains between insurgents, local strike the types of bargains that can protect elites, and service providers. This was possible SSD from appropriation. The hill region is because the insurgency was organized and characterized by programmatic forms of po- disciplined (which made it possible to iden- litical organization—especially the Maoist tify leaders who could engage in bargaining); party, which competes as a political party and had an ideology in support of health provi- advocates for pro-poor service delivery, and sion; needed health services itself; and relied for a more equal land distribution and inclu- on local legitimacy. In addition, rent-sharing sive political settlement. There is evidence agreements were struck and maintained. that links between Maoist party elites and the state administration were closer than those Puzzle 2: Why, in the post-2006 period, in a of landlords and political elites in the Tarai, context where large-scale violence has ceased, which could help explain why elite bargains aid has increased, and political settlements are were more durable and credible around health fluid, has much greater progress been achieved and road provision. in health and rural infrastructure in the hill and Part I. Research Approach and Major Findings | 33 Table 2.5. Social service delivery in Nepal: Framework, puzzles, and findings Prospects Dominant form of for pro-SSD Mode of Site Political settlement violence elite bargain delivery Puzzles and findings Palpa Moderately un- Well-organized, disci- Medium/high Health and Puzzle 1: What explains the impres- (hill district) stable. Pro-SSD plined armed groups roads—central- sive progress achieved in health de- During large- incentives exist but with incentives favoring ized state pro- livery—that is, on track to achieving scale violent are contested among SSD. Organized insurgen- vision (during Millennium Development Goals—in conflict: state and pro-SSD, cies resisting the state but and after large- the context of large-scale violence 1996–2006 disciplined armed desiring of some services. scale violent and supposedly “dysfunctional” insurgencies. conflict) governance? In addition, how were Post large- Relatively stable. Relatively little criminal High Multisector— gains made in rural infrastructure in scale violent Competitive electoral gang activity. simultaneous this context? conflict politics; program- provision • A series of elite bargains (2006–) matic pro-SSD of mutually between insurgents, local elites, political party dom- reinforcing and service providers was inant; Maoist party services (not possible because the insurgency influential. by design) was organized and disciplined Rolpa Moderately unstable. Well-organized, disci- Medium/high Health and so leaders who could engage in (hill district) Pro-SSD incentives plined armed groups roads—central- bargaining were identifiable, During large- existed but were with incentives favoring ized state pro- had an ideology supporting scale violent contested among SSD. Organized insurgen- vision (during health provision, relied on the conflict: state and pro-SSD cies resisting the state but and after large- health service, and needed local 1996–2006 disciplined armed desiring of some services. scale violent legitimacy. insurgencies. conflict) • Rent-sharing agreements were Post large- Relatively stable. Relatively little criminal High Multisector— struck and maintained. scale violent Competitive electoral gang activity. simultaneous Puzzle 2: Why, in the post-2006 pe- conflict politics; program- provision riod, in a context of the cessation of (2006–) matic pro-SSD of mutually large-scale violence and increased political party dom- reinforcing aid and fluid political settlements, inant; Maoist party services (not was much greater progress achieved influential. by design) in health and rural infrastructure Siraha Moderately un- Well-organized, disci- Medium/high Health and in the hill and mountain regions of (the Tarai) stable. Pro-SSD plined armed groups roads—central- Palpa and Rolpa compared with During large- incentives exist but with incentives favoring ized state pro- the Tarai lowlands of Siraha and scale violent are contested among SSD. Organized insurgen- vision (during Bardiya? conflict: state and pro-SSD cies resisting the state but and after large- (Key reasons are differences in 1996–2006 disciplined armed desiring of some services. scale violent subnational political settlements, insurgencies. conflict) forms of political organization, and Post large- Highly unstable. Significant elements of Low/medium Multisector— nature of violence: scale violent Factionalized political highly disorganized/frag- simultaneous conflict parties. No pro-SSD mented, mobile/roving provision (2006–) programmatic po- armed groups with of mutually litical party. Landed criminal (nonpolitical) reinforcing elites dominant. motives. No pro-SSD pro- services (not grammatic political party. by design) (continued) 34 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Table 2.5. Continued Prospects Dominant form of for pro-SSD Mode of Site Political settlement violence elite bargain delivery Puzzles and findings Bardiya Moderately un- Well-organized, disci- Medium/high Centralized • The Tarai is characterized by (the Tarai) stable. Pro-SSD plined armed groups state provision more predatory, feudalistic elite During large- incentives exist but with incentives favoring in health and bargains and fragmented, roving scale violent are contested among SSD. Organized insurgen- roads (during armed groups, making it more conflict: state and pro-SSD cies resisting the state but and after large- difficult to strike the types of 1996–2006 disciplined armed desiring of some services. scale violent bargains that could protect SSD insurgencies. conflict) from appropriation. Post large- Highly unstable. Significant elements of Low/medium Multisector— • The hill region is characterized scale violent Factionalized political highly disorganized/ simultaneous by more programmatic forms of conflict parties. No pro-SSD fragmented, mobile/rov- provision political organization, especially (2006–) programmatic po- ing armed groups with of mutually the Maoist party, which supports litical party. Landed criminal (nonpolitical) reinforcing more equitable land distri- elites dominant. motives. No pro-SSD services (not bution and inclusive political programmatic political by design) settlement. party. • There is evidence that links between the Maoist party elites and the state administration were closer than those of land- lords and political elites in the Tarai, which could help explain why elite bargains were more durable and credible around health and road provision. 2.3. Pakistan multiple, overlapping causes of conflict and violence in the region, including political and In Pakistan (see map 2.4), violent conflict has economic marginalization, Baluch national- characterized Khyber Pakhtunkhwa (KP), the ism and militancy, the role of state security Federally Administered Tribal Areas (FATA), forces, the impact of the long-running con- and Balochistan for many years. This has led flict in Afghanistan since 1979, and violence to heavy loss of life and economic stagnation. from armed Islamist groups (Gazdar, Kaker, FATA is a border region that shares a long, and Khan 2010: 2–3). porous, and mostly unmanned border with Afghanistan. FATA, as well as the Swat Val- Health services, including polio eradication, ley in KP, have both seen the rise of violent face numerous challenges. This part of the re- extremism and significant state military oper- search focuses on efforts to eradicate polio in ations. Balochistan has experienced violence KP and FATA. Ninety percent of all polio cases as Baluch nationalists have battled the Pa- in Pakistan, and the nation’s post-2005 in- kistani state for greater autonomy. There are crease in polio incidence, relate to the Pashtun Part I. Research Approach and Major Findings | 35 Map 2.4. Map of Pakistan areas of KP and FATA. Over 70 percent of “to assist Pakistan’s effort to eradicate polio confirmed polio cases in recent years has been through supply of the Oral Polio Vaccine in FATA alone. Attempts to eradicate polio (OPV) for the country’s supplementary immu- have centered on the Polio Eradication Project, nization activities during 2006–07” (World a national campaign involving federal and pro- Bank 2005b: 5). The Third Partnership for vincial governments and international partners. Polio Eradication (2009–14) aimed to “assist The Second Partnership for Polio Eradication Pakistan in its efforts under its Polio Eradica- (2005–08), a follow-up to the Partnership tion Initiative to eradicate polio from its terri- for Polio Eradication (2003–2005), aimed tory” (World Bank 2015c: viii). 36 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Education is an important area of activity in Puzzles Pakistan. The second part of the research fo- Puzzle 1: How and why did social service cuses on efforts to improve education in Ba- delivery—in the form of polio eradication— lochistan, which lags behind other provinces become politicized and targeted by violent across a range of development indicators, resistance? Why was delivering polio vaccine including educational attainment and liter- services so difficult? Why were outcomes worse acy (World Bank 2006). Challenges include in the Federally Administered Tribal Areas than low primary enrollment rates among boys in Khyber Pakhtunkhwa? and girls, low-quality education, high rates of Polio workers, but not other service providers, teacher absenteeism, poor monitoring and su- have repeatedly been targets of political vio- pervision, and weak institutional capacity in lence in some areas of Pakistan. It is crucial the public sector. The main objective of the to understand the reasons behind this, with a 2006–14 World Bank–supported Baloch- particular view to designing future risk miti- istan Education Support Project (BESP) was gation strategies. Moreover, the data underline “to promote public-private and community that progress toward eradication was greater in partnerships to improve access to quality pri- KP than in FATA, especially since 2007 (figure mary education, in particular for girls” (World 2.2). Polio delivery was also more fraught with Bank 2015a: viii). implementation challenges and more intensely targeted by political violence in FATA than KP. Figure 2.2. Cases of polio in Pakistan: The Federally Administered Tribal Areas versus Khyber Pakhtunkhwa, 2001–14 Federally Administered Tribal Areas (FATA) 250 Khyber Pakhtunkhwa (KP) Pakistan 11 200 Cases of polio 150 144 119 117 103 100 90 89 93 11 71 66 59 58 53 55 50 40 38 28 32 29 20 20 20 24 23 2027 11 13 14 11 10 6 2 2 6 2 3 7 9 3 8 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Year Source: World Bank and Akram 2016. Part I. Research Approach and Major Findings | 37 Puzzle 2: Why was the Balochistan Education Findings Support Project relatively successful in the Puzzle 1: How and why did social service context of a violent insurgency and clientelestic delivery—in the form of polio eradication— politics? become politicized and targeted by violent In Balochistan, progress in education through resistance? Why was delivering polio vaccine BESP was relatively successful. Despite the services so difficult? Why were outcomes worse violent context, four out of five indicators of in the Federally Administered Tribal Areas than the project development objectives exceeded in Khyber Pakhtunkhwa? their targets. Project-supported schools in- Polio eradication became the target of polit- creased overall enrollment in the province by ical violence due to: (1) the nature of violent 5.0 percent and girls’ enrollment by 5.3 per- conflict and antagonism with the central state, cent (World Bank 2015a: 13). How did this (2) the perception that SSD was a vehicle for happen in a historically challenging region state penetration—and the manner in which where violence was widespread? What was it the polio vaccine was delivered did not miti- about the nature of politics and violence that gate this concern, and (3) the fact that many enabled this progress? Umair, 9, a student of Government Comprehensive School in Gujranwala, Pakistan reads from a book in the classroom. Visual News/World Bank 38 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Table 2.6. Social service delivery in Pakistan: Framework, puzzles, and findings Prospects Dominant form of for pro-SSD Mode of Site Political settlement violence elite bargain delivery Puzzles and findings Federally Ad- Highly unstable. Somewhat orga- Low Polio eradica- Puzzle 1: How and why did SSD—in the form ministered Weaker state pene- nized/ disciplined tion: Central- of polio eradication—become politicized Tribal Areas tration and control insurgent armed ized state and targeted by violent resistance? Why (FATA) than in KP; more groups largely provision was delivering polio vaccine services so fragmented armed opposed to SSD. difficult? Why were outcomes worse in the groups and weak SSD Federally Administered Tribal Areas than in incentives/ideology. Khyber Pakhtunkhwa? Polio eradication became the target of Antagonistic relation- political violence due to: (1) the nature of ship with state and violent conflict and antagonism with the limited state penetra- “state,” (2) the perception that SSD was a tion in terms of elites vehicle for state penetration—the manner and administration. in which the polio vaccine was delivered Khyber Pa- Moderately to highly Somewhat orga- Low-medium Polio eradica- did not mitigate this concern, and (3) khtunkhwa unstable. Weak SSD nized/disciplined tion: Central- the fact that many insurgents and local (KP)—Pe- incentives; some insurgent armed ized state elites did not consider polio eradication a shawar and state penetration in groups largely provision priority. Swabi areas controlled by opposed to SSD. Polio eradication efforts were more diffi- relatively disciplined cult in FATA than in KP because the form armed groups. Somewhat disci- of violence in FATA was more fragmented; plined, organized the political settlement—and relations Central state penetra- armed groups between national and local elites—was tion in administrative competing for local less stable; the insurgents were more hos- and military terms power, legitimacy, tile to “state activity,” which was partly greater and more and SSD. caused by drone attacks that took place durable than in FATA. in FATA but not in KP; and violent actors Balochistan— Moderately un- Well-organized, Medium Education: were more mobile. These factors meant Quetta, stable. Pro-SSD disciplined insur- Decentralized, that service providers found it more Turbat, Dera incentives exist but gent armed groups communi- difficult to organize local elite bargains Bugti, and are contested among with incentives ty-based, non- to protect polio vaccine service providers Pishin state and pro-SSD favoring SSD. state provision from violence. disciplined armed Organized/domi- (Balochistan Puzzle 2: Why was the Balochistan Educa- insurgencies with nant insurgencies Education Sup- tion Support Project relatively successful nationalist ideology. resisting state but port Project) in the context of a violent insurgency and desiring of some clientelestic politics? Clientelism/corrup- services. tion widespread in Education progress was accomplished the education sector. under the program due to a more orga- nized insurgency; the small-scale model of delivery, which to some degree managed to circumvent national and regional cli- entelist networks; and the fact that there were incentives for insurgents and local elites to support education delivery be- cause Approach Part I. Research and Major they contributed Findings to their | 39 ideological goals of empowering Balochistan. insurgents and local elites did not consider Puzzle 2: Why was the Balochistan Education polio eradication a development priority. Support Project relatively successful in the context of a violent insurgency and clientelestic politics? Polio eradication efforts were more difficult The BESP managed to increase enrollment in FATA than in KP because the form of vi- rates because of the nature of the insurgency olence in FATA was more fragmented; the in Balochistan; the small-scale delivery model, political settlement—and relations between which to some degree managed to circumvent national and local elites—was less stable; national and regional clientelist networks the insurgents were more hostile to “state in the education sector; the fact that it was activity,” in part because drone attacks took NGO-led, which was reportedly more accept- place in FATA but not in KP; and violent ac- able to an antistate insurgent group than a tors were more mobile. These factors meant state-led mode would have been; and the fact that service providers found it more difficult that some armed elites and nonarmed local to organize local elite bargains in order to elites saw education as contributing to their protect providers of the polio vaccine from nationalist goals of empowering Balochistan. violence. 40 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Soraya gets a check up at the local medical health clinicTajikhan Village, Jabalseraj District Parwan Province, Afghanistan, May 2012. World Bank/Graham Crouch Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 41 3. Political Settlements Matter This chapter explores how the four interre- instability and uncertainty. According to a lated dimensions, and cross-cutting charac- former government official, the main problem teristics, of settlements outlined in chapter 1 with the Bonn Agreement was its exclusion affect service delivery. of the Taliban: “The Bonn Agreement (December 3.1. National Settlements 2001) was a bad start in terms of in- creasing the insurgency in the country. The broad characteristics of the political set- Why would you bring the warlords to tlement at the national level inform what we the Bonn Agreement and use them as consider to be the likely obstacles to and/ ground forces, but deliberately isolate or enabling factors for social service deliv- Hezb-e Islami and the Taliban, who ery (SSD). These include the extent to which were the real opposition? The warlords the settlement is clientelistic/developmental, were not the opposition; they were al- exclusionary/inclusive, or stable/unstable— lies. You should bring peace and secu- characteristics that impact service delivery rity to those whom you are fighting, in a broad sense by shaping the policy space not those who are eating with you.” toward more or less progressive forms of de- (KL: IDI–63) livery, as well as affecting the nature of local implementation arrangements. The new government was ethnically and ideo- logically heterogeneous, complicating former Health and education delivery in Afghanistan, President Karzai’s attempts to unify the coun- for example, occurs in a fluid and contested try. From 2002 to 2013, a delicate balance of context with a precarious set of national po- power was achieved between strong provin- litical settlements. The post-2005 Taliban re- cial actors, with the president at the center. surgence and the subsequent withdrawal of The president bought off elites with access to most international troops at the end of 2014 major political or government posts. Endemic contributed to the insecurity. The post-2001 corruption and elite impunity weakened the Kabul government’s attempts at maintaining a image of the government and led to a grow- fluid, “inclusive” peace agreement to buy off ing Taliban resurgence in the countryside. and/or co-opt regional power brokers while The nature of this settlement broadly impacts excluding the Taliban was a major source of service delivery in two ways: Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 43 A young boy points to the letter F, with his teacher’s help in a classroom in rural Nepal. Aisha Faquir/World Bank 1. The unstable and fragile nature of the set- • In many cases leading to unstable and tlement translates into fragmented, varied, fluid linkages between the center and and unpredictable elite bargains that link the provinces, disrupting SSD; provinces to the center, which makes ser- vice delivery coordination increasingly • Limiting the ability of provincial and politicized and challenging. Provincial central elites to strike and sustain elite elites often entertain disparate or even bargains to deliver services; and antagonistic relationships with elements of the political elite at the national level, • Heightening the incentives for elites to with the aim of accessing greater resources engage in short-term profiteering with from Kabul. regard to the allocation of resources and incentivizing some to allocate service 2. The fluid and uncertain nature of the set- delivery resources toward strengthen- tlement creates incentives for the sectors ing local patronage networks. to engage in short-term profiteering and impacts the space for delivering services in The nature of Nepal’s political settlement af- several ways, including: fects service delivery there as well. Political 44 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds violence during the 10-year large-scale violent postwar period, informal arrangements conflict (hereon referred to as the conflict) in and bargains between contending politi- 1996–2006 and the period of postwar polit- cal elites tended to dominate attempts at ical contestation (post-2006) led to changes reintroducing local government institu- in the political settlement, which in turn con- tions, as local-level transitional authority tributed to shifts in the way services were was “up for grabs” (Byrne and Shrestha delivered. Two overlapping points deserve 2014). These local political elite bargains mention: were shaped, to a large degree, by the na- ture of central political party competition, 1. In the late 2000s, the national politi- which lead to patronage-ridden yet more cal settlement shifted to become a more inclusive SSD at the sector and local levels inclusive elite coalition, which led to an in basic health care and in the roads sector. increasingly inclusive approach to deliver- ing services. The ostensibly pro-inclusion Three important points emerge from the agenda of the Maoist insurgents contrib- Nepal cases: uted to shifts toward a more inclusive cen- tral political settlement and encouraged 1. The politicization of local service delivery traditional elite coalitions to address issues is the norm—not an aberration or dys- of socioeconomic inequalities in SSD. The function, as is commonly suggested in dis- changes at the national level then filtered courses about good governance. It serves down to the sectoral policy and local im- the function of “making things work” in plementation levels (box 3.1). This case the absence of unclear rules and contested illustrates the point that political violence sites of local authority. is not always “development in reverse.” 2. The front line of delivery in the health 2. The nascent, unstable, competitive, and and roads sectors continues to experience clientelistic form of political settlement progress in this highly political environ- at the central level contributed to the for- ment, suggesting that politicization is mation of contested and politicized local not the problem per se. Instead, the form structures for service delivery. The fluidity of politicization seems to matter. Progress and uncertainty at the center, most evi- varies depending on the local nature of dent in the postwar transitional period, politics, including the nature of elites and translated into a lack of clear rules and the bargaining between them. power sharing at the local level. During the Maoist insurgency and during the Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 45 Box 3.1. Nepal’s shift toward inclusion shapes social service delivery An increasingly inclusive form of political settlement emerged during and after the Maoist insurgency. The war and political violence appear to have influenced the motivations of decision makers within the ruling coalition to address inequalities in the polity. This shift created space and mechanisms at the national and local levels for more inclusive service delivery. In this instance, political violence—although it created many negative impacts—cannot be characterized as simply “development in reverse.” The forging of this new, fluid, and unstable political settlement was punctuated by measures and agreements rein- forcing inclusiveness. A series of agreements culminated in declarations in the Comprehensive Peace Accord from both sides of their “determination for a progressive restructuring of the state to resolve existing problems in the country, based on class, caste, region, and sex”a and in the 2007 interim constitution, which recognizes women, Dalits, Janajatis, and Madhesis—in addition to peasants and laborers—as groups that require special attention.b These shifts filtered down into various measures. In 1991, the health ministry introduced the National Health Policy, which focused on outreach to vulnerable groups, women, children, rural populations, the poor, and the marginalized.c There were impacts to local operational mechanisms for public goods delivery that were focused on increasing the involvement of local bodies and communities in the planning, implementation, and management of local-level development activities, such as the 1999 Local Self-Governance Act.d The shift at the center did not solve all of the country’s issues around inclusion but did translate into more inclusive service delivery structures, which then impacted service delivery. A short survey of groups typically excluded from health services, conducted for this study, points to a longer-term shift on the ground (World Bank and SSB 2016): • Seventy-five percent of respondents reported using services provided by the local health-care post and/or pri- mary health-care center; • Ninety-four percent said they used government facilities more often than they had 10 years back; • Fifty-six percent felt health services had improved; and • Fifty-five percent of Dalit respondents felt that the behavior of the health personnel had changed for the better.e a. Unofficial translation of the Comprehensive Peace Agreement concluded between the Government of Nepal and the Communist Party of Nepal (Maoist), November 21, 2006 (www.unmin.org.np). b. See Article 11(3) of the 1990 constitution and 13(3) of the 2007 interim constitution. c. The National Health Policy extended the modern health-care system to the rural population by making the sector more decentralized and regionalized (MoHP 1991). Successive government health plans and policies that focused on outreach to vulnerable groups, the poor, and the marginalized included the Second Long Term Health Plan (1997–2017), Health Sector Strategy: An Agenda for Reform (2003), the Tenth Five-Year Plan (2002–07) and the Three-Year Interim Plan (2007/08–2009/10). d. Other local operational mechanisms included increased emphasis on bringing the poor and marginalized groups into the development mainstream as well as various measures adopted by the government to support more inclusive government, most notably by reserving 45 percent of all government jobs for excluded groups. e. Almost half felt that this was due to education, growing awareness, and the fact that it is now illegal to discriminate on the basis of caste. 46 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds 3. Political party competition driven by 3.2. Sectoral Settlements patronage and collusion can increase in- clusion, upward mobility, and local own- Elite balance-of-power and rent-sharing ership. In the absence of local elections, agreements, like national settlements, affect political parties ensure that the smaller SSD in various sectors. Two aspects of these political parties receive at least some rec- settlements—how elite rent-sharing influ- ognition; they also provide a forum for ences SSD and how different sectors present dialogue, which smooths project imple- different forms of and opportunities for elite mentation. The chairperson of a local road bargaining—are examined here. user committee justified the dominance of political parties in the formation of user committees by noting that including rep- Rents resentatives from all locally active parties Patterns of rent-seeking among political and ensures that people take ownership of a bureaucratic elites shape the allocation of ser- project. Moreover, in the case of the Rural vices across the studied cases in Pakistan. For Access Improvement and Decentralization example, networks of patronage are embed- Project, a local road user committee is re- ded in the polio vaccine delivery chain. Evi- quired to include at least one woman and dence points to instances of corruption, ghost one member of a disadvantaged group, al- teams, and claims of immunizations where though the effort is subordinated to politi- none occurred. One senior donor official ex- cal party logic and affiliation. In the words presses the challenge posed by corruption and of one informant: “People from the in- rent-seeking behaviors by noting that money digenous community are hand-picked by meant for polio eradication lines the pockets political parties to fulfil the quota needs of some officials who dispense jobs as favors, of people and not for actual representa- turning over vaccination posts to their broth- tion. Political interest takes precedence ers or cousins, who pass them on to their over true representation” (G15–KII–BAR). children. The official concedes that his or- Nevertheless, many see the quotas as a ganization is helpless in these circumstances. route, albeit contested and imperfect, to- Regarding the negative aspects of the Polio ward increased inclusion, and political Eradication Project, he remarks, “We created parties serve as a vehicle for the recogni- a monster.” He acknowledges that there are tion of previously excluded demands. instances of ghost campaigns, where money is spent but no actual work takes place. He Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 47 Box 3.2. Political settlements in Afghanistan’s education and health sectors The elite balance of power, interests, and incentives in Afghanistan’s education sector, combined with blurred po- litical and bureaucratic boundaries, influences the selection of school locations, the number of schools built in each province or district, and the hiring of staff. Key findings include the following. Vote bank. Even though the Ministry of Education is the “largest single civilian employer in the country,” it is sometimes treated as a “vote bank.” As one informant expresses it: “Education in particular . . . is very interesting, as it is one of the largest employers in the country. For instance, in election campaigns, it is an extremely useful ministry to have. If you want, you can instruct all teachers to campaign for certain candidates; you can instruct them to tell students to tell their parents . . .” (KL: KII–44) Recruitment. Although the Ministry of Education in general and the Education Quality Improvement Project (EQUIP) in particular are supposed to abide by certain criteria in terms of recruitment, cases of nepotistic recruit- ments are reported (KL: KII–37 and 43). Another national program implementer admits that while hiring people without adequate qualifications compromises the quality of education service delivery, it is nonetheless necessary (KL: KII–37). Patronage pyramids. The existence of patronage pyramids is alleged, where each bureaucratic layer sells positions to the layer below, and the bottom layer regularly demands payments from the beneficiaries (teachers, students, and even parents). According to an international researcher: “Everything is the same everywhere; if you want the job of provincial director of education, you’ll have to pay someone, that’s for sure. You will not get the job without paying . . . How do they get reimbursed? Well, they will make sure that all the people who work for them collect money for them. Through this, they get reimbursed for the investment that they made to get the position of provincial director [of education]. How do these people get money [referring to those below the level of the provincial director]? From the students, from the parents, from the teachers . . .” (KL: KII–45) Diversion to certain areas. Local political elites such as Members of Parliament (MPs) and power brokers play a role in shaping how education services are distributed. One national program implementer, referring to some MPs, remarked, “They have facilitated [the delivery of education services] in the sense that there are now more education service deliveries in their respective areas. They did not equally distribute the education services; more are found in their areas” (KL: KII–37; also, KL: KII–34 to 37). The leverage of politicians and local power brokers is enhanced by the fact that, historically, the Afghan government had difficulties with communities not previously exposed to state education (Giustozzi 2010).  (continued) 48 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Box 3.2. Continued Creating rent opportunities. Interviews with key study participants suggest that the emphasis on school con- struction is driven, at least in part, by the need to maintain and expand patronage networks: construction projects provide greater opportunities for rents than soft investments (KL: KII–45 and 50). EQUIP I’s implementation review hints at this, but does not allude to the political drivers: it notes that infrastructure grants are an area where “financial overcommitments” were made (World Bank 2009: 4). Patronage politics. Similarly, the health sector is not immune to the impact of patronage politics. Actors well con- nected to the ruling elite manipulate the allocation of resources to favor their power bases, resulting in the unequal distribution of health resources between provinces and districts because better-connected politicians and power brokers can lobby more effectively (KL: KII–53; KL: KII–47). One study participant puts it this way: “If you go to some provinces, you will see two clinics in one village, but another village does not have one. In [so-and-so’s, naming a particular Member of Parliament] district, there are many clinics, but in the other districts of the same province, which do not have a good connection with a powerholder, there is not a single clinic.” (KL: KII–53) believes this is why some are skeptical of the the “failure to devolve sufficient authority, ex- polio eradication effort and do not trust the acerbated by local collusion to benefit private polio teams (see Roberts 2012). interests, has prevented the development of participatory and integrated planning in the A number of notable rent-sharing arrange- health sector” (Asia Foundation 2012: 24). ments and patterns were also identified in The growing number of strikes and scandals Afghanistan’s education and health sectors; emerging from prominent hospitals suggests these affect if, how, and where services are that the public administration, including the delivered (box 3.2). health sector, has been embroiled in a proxy war between the three main parties contend- In Nepal, sectoral settlements, revolving ing for power (Sharma 2010). around rent-sharing, shape progress in the health sector. Observers point to the politi- Bargains Differ by Sector cization of health appointments, where elites seek to expand their political base by influ- Interestingly, sectoral characteristics and sec- encing the hiring of staff and the appoint- tor-specific modes of delivery seem to affect ment of key officials (Asia Foundation 2012: the form of and opportunities for rent-shar- 23; Harris et al. 2013; Sharma 2010). This has ing and patronage (KL: KII–44). In Afghan- led to a lack of integrated planning, where istan, there are different and fewer forms of Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 49 elite bargaining and patronage in the health to wider political party patronage, help explain sector compared with the education sector, in why progress in basic preventive health care part because the latter provides more oppor- in Nepal was more advanced than in curative/ tunities for rents in the form of construction diagnostic care. This is because basic preven- projects and employment. Further, extracting tive health care relied on locally embedded rents from the education sector is deemed to female volunteer workers who were dedicated be more straightforward because education is to their jobs and were not appointed because largely delivered through state mechanisms, they had political connections. Indeed, they as opposed to basic health-care delivery, operated outside the clientelist party structures. which is contracted to NGOs, making it more Conversely, curative health care depended on difficult to directly divert resources without skilled physicians and nurses who were often co-opting staff. able to avoid being posted in remote rural areas (generally considered undesirable) be- In Nepal, elite interests in the health sector cause they were part of unions closely allied varied at the subsectoral level (in preventative to political parties. One of the main sources of versus curative health); this in turn had varied political party power was the ability to broker impacts on the progress of delivery. Sectoral deals for members of powerful unions (includ- political settlements in the health sector, linked ing those for skilled health workers). Box 3.3. Relationships between elites shapes delivery of polio services Delivering polio eradication services is more challenging in the Federally Administered Tribal Areas (FATA) than in Khyber Pakhtunkhwa, partly due to the different relationships each region has with the central state. FATA has long had a tenuous relationship with the state, and following a 2002 military intervention, its governance structures were dissolved. A series of central government military operations to destroy local militia and insurgent groups were only partially successful. This in turn has further isolated the region from the rest of Pakistan. Khyber Pakhtunkhwa’s links to the central government are comparatively stronger. The public participates in the local, provincial, and central legislature through elections. There is state patronage, particularly in the roads and electricity sectors, whereas it is virtually nonexistent in FATA. Interviewees suggest that the delivery structures are more contested and complex in FATA because of these differences, which makes it much more challenging to strike or to maintain service delivery bargains. Another difference between the two regions is in the forms of violence and objectives of violent actors (see chapter 4). 50 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds 3.3. Central and National service delivery programs impact Subnational Elites regions differently depending on these rela- tionships, which helps explain why delivering The nature of the relationship between rul- services is more difficult in the Federally Ad- ing elites at the national level and subnational ministered Tribal Areas (FATA) than in Khy- elites at the provincial, district, and village ber Pakhtunkhwa (see box 3.3). level shapes opportunities for accessing SSD resources and support. It impacts the degree of The case of Afghanistan demonstrates the state penetration and control over insurgents, importance of this relationship to service de- the flow of resources to a locality, and the au- livery. As box 3.4 illustrates, the relationship tonomy and control of subnational elites over influences the flow of resources and opportu- service delivery policy and implementation. nities for SSD and therefore affects progress Box 3.4. Central-subnational linkages and service delivery in Afghanistan Diverse networks The diversity among the linkages and relationships between provincial elites and those in Kabul is significant. Each province has a unique set of formal and informal relationships with the central government, which appears to sig- nificantly impact social service delivery. Balkh. This province is characterized by a stable provincial settlement centered on a charismatic and resourceful strongman: Provincial Governor General Atta Mohammed Noor. The general has established an influential power base and positive linkages with the capital, giving the province more privileged access to Kabul-based ministers and senior civil servants. Wardak. Some Wardak elites have obtained prestigious appointments in Kabul, and the main provincial-level networks have been nationally integrated, at least to a degree. The networks operate as quasi-political parties, and they are relatively well organized. They run the province through their local associates who control the machinery of government at the provincial and district levels. Badghis. Widely seen as strategically unimportant, Badghis is economically underdeveloped and subject to fluid secondary political settlements because of the unstable relationships between its various strongmen. As a result, the province is unable to effectively lobby Kabul or form strong alliances with actors who are part of the primary elite bargain. (continued) Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 51 Box 3.4. Continued Impact on service delivery A background country report (World Bank and AREU 2016) illustrates how the relationship between the subna- tional and central actors in each of the three provinces shapes the quality and quantity of its social services. The weak relationship between the provincial settlement in Badghis and the primary (central) one, in addition to its under representation in Kabul, negatively affects its opportunities for social service delivery. Badghis is not consid- ered strategic by the Kabul-based elite. In the words of one key informant: “If you look at it, it is far from Kabul, and unlike Herat and Faryab, it is not in a position to create problems for those in the center. So yes, it is left out. Also, it is not significant in terms of border relations, and there are no connecting roads leading to Kabul ... It is not that politically significant.” (KL: KII–49) Another informant noted that the lack of powerful provincial warlords left them at a disadvantage: “If we had those kinds of powerful warlords, then the central government would have feared them and our province would have received attention” (BS: IDI–33). Interviewees noted that Kabul’s perceived lack of interest in Badghis is one explanation for the poor level of ser- vices in the province, including a higher student-teacher ratio and a shortage of school buildings compared with Wardak. The reported sense of neglect spills into the health sector as well, with some interviewees noting a lack of qualified personnel, insufficient supplies, and poor quality of medicine (BS: IDI–31). Balkh, however, has managed to secure a more significant share of resources and linkages with Kabul, compared to Badghis. Wardak also suffers from a lack of local resources, but because it is closer to Kabul, the province enjoys privileged access to higher education and career opportunities in the capital and elsewhere. across health and education. Indeed, the extent In the post-2006 period, the different types to which provincial power brokers can draw of relationships between Kathmandu-based on the resources and political commitment of elites and the hill region compared with the central state resources and cooperation seems Tarai shaped the ability of each district to bro- to depend on the history of subnational elite ker service delivery deals. The research points linkages to the center, the coherence and sta- to more embedded linkages between political bility of the provincial-level political settle- elites in the hill region and the public admin- ment, and the extent to which the province’s istration (box 3.5). The Maoist political party economic resources affect the political calcu- is more dominant in the hill region, and the lations of national coalition leaders. social background of the Maoist leadership 52 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Box 3.5. Subnational settlements in Nepal help explain SSD outcomes Our research suggests that a combination of organizational and programmatic factors explain why in the postcon- flict period in Nepal, the health and roads sectors in the Tarai performed more poorly than those in the hill and mountain regions. The Maoist hill and mountain areas are very different than the Tarai in terms of the nature of their subnational political settlements, political organizations, and agendas of political party elites (see map 2.3 for ecological zones). Hill and mountain regions The postconflict settlement in the hill and mountain regions is relatively stable, and organized around the program- matic party of the Maoists, who are electorally dominant there. The party’s appeal is its programmatic push for socioeconomic change and pro-poor social service delivery and its capacity to deliver local results better than most political parties can (Bevan 2010). This makes it easier to deliver services and to organize bargains around service delivery. The Tarai­ a Compared with the hill and mountain regions, the postconflict settlement in the Tarai is unstable and contested; it is characterized by highly disorganized and mobile roving armed groups (often with criminal motives); and there is limited state penetration in many parts of the region. Groups that emerged as a result of the Madhesi movement in 2007 included political parties linked to criminal gangs engaged in cross-border crime and extortion. This predatory behavior flourished in the region due to a weak state presence, and there was a higher incidence of criminal activity. Post-2007, the dominant motivations of the political leadership in the Tarai have centered on greater political representation of Madhesi leaders in Kathmandu rather than on forging a social service delivery agenda. Weaker links and less political influence in government has made brokering deals with central government ministries more difficult, resulting in the targeting of service delivery for appropriation and much riskier and more challenging service delivery. a. Historically, the Tarai has experienced far less state extension and a far higher incidence of criminality as local groups have taken advantage of the state’s absence to engage in criminal enterprises and appropriate valuable assets such as land. The absence of state penetration was a central condition to the rise of mafia-like organizations elsewhere (Gambetta 1993). The political landscape in the Tarai is therefore much more predatory and less oriented toward programmatic political projects that advance social welfare than in the hill and mountain regions, which may account for the fact that political instability and uncertainty is also greater in the Tarai. is similar to those that dominate the public government bureaucracy and the political sys- administration, which ostensibly facilitates a tem than the elites in the hill region. Weaker more functional working relationship with links and less political influence with the gov- the state and more stable local-level politi- ernment appear to make coordinating and cal bargaining. Landlords and political lead- brokering deals with government ministries ers in the Tarai have less influence with the more tenuous (KII–03–NEP–GEN).20 Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 53 3.4. Subnational (Secondary) predatory in Badghis (see chapter 4), which Settlement partly explains their difficulty in delivering services. The fourth, and perhaps most important, di- mension is the subnational settlement. The The settlement in Balkh is quite different. It research uncovered various types of settle- has been generally conducive to relatively ments—and forms of violence—at the sub- smooth implementation of SSD. Balkh’s re- national level, each with its own impacts on cent history is dominated by General Atta the prospects for services and the ability to Mohammed Noor, who has successfully cen- forge pro-SSD bargains. Key factors include tralized much of the power and largely disci- the level of stability and contestation in the plined other armed groups since about 2004. subnational settlement; the degree to which a developmental orientation cascades through In Nepal, differences in the subnational po- the settlement, including the ideas and incen- litical settlement and forms of political orga- tives among elites regarding development nization help explain disparate progress in and service delivery; and the local balance of the health and roads sector in the hill region power between traditional, local elites, and versus the Tarai. This case illustrates that who armed elites and their allied insurgents. is in power and what their agenda is—at the subnational level—reveals a great deal about the politics of local-level SSD (see box 3.5). Stability, Inclusiveness, and Development Orientation Power and Incentives of Local The extent to which a subnational settlement “Nonarmed” Elites is stable or contested, inclusive or exclusion- ary, and in turn has some elements of a de- At lower levels of the polity, the nature of in- velopmental orientation, impacts its ability to centives for localized elites and the balance deliver services. of power between traditional local elites and armed elites and allied insurgents are crucial In Afghanistan, subnational settlements dif- to shaping front-line delivery. The points fer widely across the three provinces in terms briefly outlined here are further explored in of their stability and fragmentation, and this chapter 4. shapes service delivery. In Badghis, for exam- ple, the local power brokers have not man- The first case relates to the nature and incen- aged to form a stable or inclusive provincial tives of traditional local elites in Balochistan, settlement, instead remaining roving bandits. Pakistan, with regard to education delivery. Forms of violence are more fragmented and The Balochistan Education Support Project 54 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds (BESP) model focused on local, smaller-scale, “discipline” armed elites and allied insurgents and “community-run” schooling models. As with regard to driving progress in health noted previously, BESP made significant service delivery. Local elders and traditional progress in the delivery of education services leaders play an important role in service de- despite the violent context and clientelistic livery within the province. As one interviewee politics where the hiring of “ghost teachers” puts it: “The traditional leaders are more in- and the tolerance of teacher absenteeism by fluential in the province than those elites who higher-level central and provincial elites (es- are at the national level, because the latter do pecially those linked with the Ministry of not have large families in Wardak” (KL: KII– Education) was commonplace. Progress was 49; WK: IDI–17). These elites are sufficiently possible, at least in part, because the local cohesive and aware of the potential benefits village elites were in favor of education ser- of lobbying provincial elites and keeping vice delivery free from clientelist politics and them on track to deliver services. Interview- enjoyed robust local autonomy from provin- ees describe the interactions among these cial and central elites in terms of service de- elites as generally good, with a fair degree livery decisions. While Baloch and Pashtun of coordination to agree on basic operating villages are organized differently from one guidelines to deliver services (WK: KII–09 another, they tend to be close-knit and are and 13; WK: FGD–18 and 23; WK: KII–01, frequently composed of members of the same 03, and 07).22 In this sense, secondary settle- tribe or close relatives. All villages have an ments demonstrate a level of resilience. One elder—often whoever has the most land, is interviewee notes: the oldest, and/or is the most experienced in local decision-making councils. These elites “Their interaction and relationship are typically aspire to educate their children, but very good, because they have mutual they cannot afford to send their children to and shared benefits. As in the case of school outside Balochistan, unlike many of any construction project, they benefit the higher-level provincial elite—known as from it in terms of giving a contract nawabs or sardars.21 Because these elites had a to construction companies, and then stake in local education they often supported their relatives are hired to work on the BESP and encouraged bargains with Ba- the project. If they face problems, they loch militants to ensure that schools remained solve them as soon as possible.” (WK: open (see chapter 4). KII–07) The case of Wardak, Afghanistan, simi- The coordination of the local elites over ser- larly demonstrates the importance of local vice delivery positively impacted health de- nonarmed elites and their relative power to livery. Their interests, ideas, and incentives Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 55 support local health service delivery, so they point for local elite bargaining that at least have sought to enable it. And because they provided predictable and relatively safe access are relatively influential and organized, they for SSD workers. As the power and author- can often counterbalance and bargain with ity of local elites declined compared with insurgents to ensure that services are deliv- that of the armed elites and allied insurgents, ered (see chapter 4). the prospects for striking elite bargains that would allow access to health workers became The increased power of armed elites and al- much more tenuous.23 lied insurgents relative to traditional local elites in FATA, on the other hand, resulted in In summary, various dimensions of polit- service delivery problems. The breakdown of ical settlements at different levels of the the links between the central and provincial polity shape the space, opportunities, and government in the post-2005 period created constraints for delivering services. The exam- a power vacuum, which was filled by several ination of these dimensions provides a bet- armed groups. This impacted the prospects ter understanding of how the space for more for SSD because traditionally, local elites— or less progressive SSD is created, and what maliks—are instrumental in implementing the prospects are for making local bargains. development activities. Implementing infra- Understanding the nature of the settlement structure construction projects, delivering alone, however, does not adequately explain aid, and facilitating the movement of goods how and why services are delivered in vio- were conducted largely with the involvement lence-affected contexts. There is a need to dig of local elites. While the maliks were known deeper to get a fuller picture of how progress to extract bribes from contractors tasked is made and why there is variation within and with infrastructure projects, such as road between violent contexts. These issues are ex- construction or the establishment of schools plored in the following chapters. and health centers, they also provided a focal 56 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds 4. The Form of Violence and the Ideology behind It Matter “Many of the interviewees [working on service delivery in Afghanistan] did not see violence as a problem in itself for the delivery of services so long as it was possible to identify the key players and negotiate access to the services. It became much more difficult when local strongmen competed against each other or ‘criminal gangs’ were involved without any clear incentive to ensuring services were sustained ….” (World Bank and AREU 2016) “The national … data suggests that [the] Maoist rebellion [in Nepal] has not had the devastating effects on the health and well-being of the population as we can see resulting from similar types of violent conflict across the globe.” (Devkota and van Teijlingen 2015) This chapter explores how various forms of For example, the organization of violence violence may affect service delivery. As out- helps explain variations across provinces lined in chapter 1, the research finds that in Afghanistan (see box 4.1). For example, elites take up arms for multiple reasons in Balkh’s more organized and disciplined form varying contexts, as discussed in detail below. of violence and more stable political settle- ment enabled easier implementation of social service delivery (SSD) than the more frag- 4.1. Organization of Violence mented—but still relatively organized—vio- lence in Wardak, and the highly fragmented The prospects for delivering social services and contested form of politics and violence partly depend on the ability of the govern- in Badghis. (See chapter 3 for a more in- ment and armed and other elites to make depth discussion of the political settlements political bargains that enable—or at least do in Afghanistan.) not disrupt—on-the-ground delivery. A key determinant for whether some form of bar- Another example is the case of Nepal. Prog- gaining is feasible is the extent to which the ress in SSD in the health and roads sectors was insurgency is organized and uniform, and the possible during the conflict partly due to three leaders recognizable and incentivized to offer organizational elements: (1) leaders were points for negotiation and bargaining. recognizable, focal points for bargaining; Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 57 (2) Maoist leaders were disciplined in their period, there was more progress in service de- policies, and so agreements were predict- livery in the hill region of Nepal than in the able and upheld; and (3) the ideology of the Tarai, partly because politics took a more pro- Maoists meant that health delivery aligned, grammatic form in the hills. Indeed, the frag- to some degree, with their goals. The back- mented, competitive, and criminalized forms ground country report explains: of violence in the Tarai made it extremely dif- ficult to deliver services, protect service pro- “The Maoists represented a dominant, viders from violence and appropriation, and disciplined organized armed insur- negotiate and uphold elite bargains. A former gency, and one that organized around World Bank staff member explains: an ostensibly political program of pro- poor service delivery. Our interview “The project in Tarai suffered the pres- evidence suggests that the Maoists sure/threats from various armed groups were organized armed insurgents with making it difficult to continue working, a ‘pro-poor’ agenda and while they whereas, in the Hill districts a condu- were engaging in threats, kidnapping, cive environment was created for proj- displacement of government officials, ect implementation.” (R6–KII–KTH) destruction of infrastructure for po- litical motives (G3–KII–SRH), their A former district development committee leaders were recognizable and some- member in Siraha also noted that “[p]eople of how disciplined around certain policies [the] Hill Region, businessmen, were threat- (G9–KII–SRH) and therefore a focal ened by these armed groups [in the Tarai] point for organized political bargain- without any political motives, but only for ing.” (World Bank and SSB 2016). financial gain” (G3–KII–SRH ). Chapter 5 explores the bargains that were Similarly, in Pakistan, the differences in the struck with the Maoists that enabled the con- organization of violent actors helps explain tinuance of service delivery in those areas why bargains are more difficult to strike in the under their influence. Federally Administered Tribal Areas (FATA) than in the Khyber Pakhtunkhwa (KP). The The importance of the organization of armed background country report suggests that the groups is further underlined when one tries more fragmented, contested, and radicalized to understand the challenges for SSD in the forms of violence in FATA contribute to more Tarai region of Nepal. During the post-2006 difficult and dangerous service delivery. 58 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Box 4.1. Organization of violence in Badghis and Wardak Badghis Province Delivering services in Badghis Province, Afghanistan, is particularly challenging, mainly due to the disjointed and unstable nature of the political settlement (see chapter 3) and the fragmented, divided, and contested forms of vio- lence, which makes it extremely difficult to identify leaders with whom to bargain, predict the behaviors of violent actors, and reach bargains that enable the service delivery. Violent conflict in Badghis is characterized by regular armed clashes between insurgents and government forces and periodic conflicts between rival warlords. The dominant form of violence is therefore characterized within our ty- pology as highly disorganized and fragmented, with roving (and sometimes competing) armed groups. Dominating Badghis is an expensive enterprise that brings few rewards. Unlike Wardak, Badghis has a history of numerous in- fluential warlords trying to assert control over it, both before and after the Taliban’s Islamic Emirate of Afghanistan. Abdul Malik, Abdul Rashid Dostum, and Ismail Khan each had influence over the province at different times, but none could assert full control (see Giustozzi 2009). This form of violence in Badghis affects service delivery in a number of ways, as outlined below. Weak incentives and coordination. Many informants note that the competitive and fragmented scenario results in poor coordination between power holders and weak incentives for social service delivery, with a focus on pred- atory behavior. As one informant puts it: “There is no doubt that the biggest suffering of Badghis comes from these unauthorized armed groups; they will be a big problem in the future. You know that a wrong policy was previously welcomed—to support militias. But this was happily decided by the central government to prevent a build-up of militia groups, since their existence can be a very big threat. But look at Ab Kamari, for example. It used to be the district of peace; now … unauthorized armed groups create problems that the Taliban doesn’t.” (BS: IDI–32) Another informant expresses it this way: “… We can see the security situation and some conflicts inside the center or in the districts of Badghis. This is a sign of powerholders not having good relationships. However, they portray themselves as honest and friends with each other to show the community that they are not against each other and that they are just working for the province. They are not honest; they are not supporting each other. Everyone looks out for his own position and power. If they show themselves to be allies, it is just to keep their own positions. We all know that they are against each other and that they are in different political parties … When there are contracts for some projects or other sources of money that they can benefit from, they compete to gain a greater share than the others …” (BS: IDI–23) (continued) Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 59 Box 4.1. Continued Challenges to bargaining. Many interviewees do not think violence is itself an obstacle to service delivery as long as it is possible to identify key players with whom to negotiate access. The situation is much more difficult when local strongmen compete against each other or when there is criminal gang activity and an absence of clear incentives for ensuring that services are sustained over the long term, which is largely the case in Badghis. Unstable climate. Many leaders and service providers complain that they cannot understand or predict the behav- ior of the strongmen and warlords, and this creates an uncertain and unstable climate. One interviewee exemplifies this point by saying: “I don’t understand how they interact with each other. They are sometimes at war with each other and sometimes at peace” (BS: IDI–22). Another notes: “They [Badghis warlords and strongmen] have their own agreements with each other in terms of what to do. Sometimes, they fight with each other … sometimes, they make deals, especially during the pistachio har- vest … It is good that they come to an agreement on how to share it, so then there is no argument between them.” (BS: KII–15) In this context, even when bargains were struck (see chapter 5), they were particularly difficult to forge and keep. Many local actors hoped for the arrival of a strong leader who could serve as a catalyst for the formation of a more stable provincial settlement. Some, including elders and activists, suggested that the emergence of a strong warlord such as General Atta Mohammed Noor in Balkh could be the best possible outcome for Badghis. Wardak Province Wardak, like Badghis, is often characterized as insecure and violent. Still, many interviewees perceive Wardak as a place where it is easier to deliver services than it is in Badghis, and its outcomes also tend to be notably better, particularly until 2008. Why is that? Insights from the research include the following: • Violence is, to a degree, more organized and less mobile (compared to Baghdis). In the absence of a strong leader to act as a catalyst for collective action, provincial elites did not develop a strong vision of the future or become more organized and centralized in their monopoly over violence, as they did in Balkh. Nevertheless, the more organized form of violence in Wardak enables bargaining, as the local elite is relatively strong (see chapter 3) and supports health and education, and the Taliban are relatively reliant on local support, which contributed to cases of forging bargains around social service delivery between service providers and insurgents (see chapter 5). • The provincial settlement in Wardak requires less rapacious behavior from the provincial elites and is more cohesive. The main power broker is linked to anti-Taliban factions, which later incorporated Hezb-e Islami networks. The strong Hezb-e Islami and Hizb-i Wahdat networks in the Pashtun and Hazara areas, respectively, may have helped bring service delivery to these constituencies, but there is a lack of conclusive evidence. 60 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds 4.2. Ideologies, Incentives, and armed elites and allied insurgents. Prospects Motivations behind Violence of bargaining with health workers were en- hanced because attempts to deliver health “to The second key dimension of violence in- all” aligned with aspects of the Maoist agenda volves the ideologies, incentives, and motiva- focused on expanding equality and inclusion tions behind it. The aspects of this dimension (see chapter 5). Sociologist Chaitanya Mishra are manifest in various ways across the study sums it up this way: sites, and the differences between them had some surprising and disparate impacts on SSD. “Most analysts … have concurred that poverty, illiteracy and low level of ed- ucational attainment, unemployment Ideology of Violence and and underemployment, inter-house- Service Delivery hold economic inequality—primarily The degree of alignment between the objec- in terms of landownership and in- tives and ideology of armed elites (whether come—and caste, ethnic, regional/ leaders of armed insurgencies or established spatial and gender oppression and in- warlords) and the logic and scope of the ser- equality are the primary referents of vice strongly impacted whether the service the Maoist struggle.” (World Bank and was tolerated or became the target of political SSB 2016) violence. This is illustrated by several exam- ples in the case material. This case also demonstrates that motivations for violence are not static. For example, the For instance, ideology’s crucial role is starkly Maoists consciously changed their strategy illustrated in the comparison between polio around the building of roads. At the start of eradication and education delivery in Paki- the insurgency, the Maoists were opposed to stan. Insurgents have engaged in violent at- any kind of government-sponsored develop- tacks over polio eradication delivery in KP ment activity because part of their strategy and FATA, while insurgents in Balochistan was to discredit the state. But by the early have, at times, tolerated and even supported 2000s, the Maoists had changed tactics. An education delivery. Differences in the history assessment from 2002 concludes: of, motivations for, and organization of vi- olence appear to play a contributing role in “The Maoists have recently recognized explaining the disparity (box 4.2). the adverse propaganda effect of their earlier attacks on rural infrastructure The case of health delivery in Nepal also il- and installations that could be regarded lustrates the importance of the motivations of as primarily for development purposes, Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 61 Box 4.2. Ideology matters? Education in Balochistan versus polio eradication in KP and FATA A comparative analysis of education in Balochistan and polio eradication in the Federally Administered Tribal Areas (FATA) and Khyber Pakhtunkhwa (KP) stresses the role of ideology, as well as other dimensions of violence, to explain the disparate social service delivery risks and performance. Balochistan and education. Armed elites and allied insurgents sometimes did not overtly oppose the Balochistan Education Support Project (BESP). In fact, on various occasions, insurgents let the project take its course without any direct interference or threat, which helps to explain why BESP registered progress (see chapter 2). As one in- dividual from Turbat explains: “Extra judicial killing, abduction, and the like are observable everywhere but as far as this school and educa- tional activities are concerned, Sarmachar [a local term for Baloch militants] never interrupted our activities. Rather, they encouraged us to get education.” (FGD–01) Much depended on the case, but why could this be? • Ideology. There was some degree of alignment between the goal of delivering education and the goals of the Baloch insurgency. First, some Baloch armed elites and allied insurgents viewed education as a means to achiev- ing their goals. They viewed a lack of education as an obstacle for the province and therefore believed some forms of education could promote local empowerment. Second, certain militants relied, in part, on local support, which incentivized them to support services that the local communities wanted. Finally, locally influential na- tionalist elites viewed education as a means to pursue their nationalist objectives.a • Organization of violence. In addition, the Baloch insurgency is relatively well organized and disciplined at the local level, making it easier to identify leaders with whom to bargain and negotiate to ensure that social service delivery was protected. • Bargains and compromises. Baloch armed elites and allied insurgents were further incentivized to support local education by modifying and co-opting the education services (see chapter 5); moreover it was a nonstate, as opposed to a state-led, project. Polio eradication in FATA In contrast to the example of education in Balochistan, the history of the Polio Eradication Project (PEP) in FATA is more rife with violent attacks, with some vaccination workers being abducted or killed. In both regions, the in- surgency is resisting the state, but has different motivations. In Balochistan the insurgency is driven by nationalist sentiment; in FATA it is linked to global Jihadi networks keen on imposing their vision of Sharia law on the state. The FATA insurgency has been increasingly radicalized and its relations with the state further antagonized by re- peated drone attacks. Moreover, attacks against PEP workers increased after it came to light that the U.S. Central Intelligence Agency had used a vaccination drive in Abbotabad to gather intelligence against Osama Bin Laden. Resistance to polio became a proxy for resistance to the state as well as a bargaining tool. 62 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds (continued) Box 4.2. Continued Further, in Balochistan, education is viewed as a vehicle for provincial empowerment, and its localized mode of delivery gives insurgents some control over how the services are delivered. In FATA, however, the polio eradication program is seen as a Trojan horse for external intelligence gathering. It is not considered a priority due to the lack of other basic health-care services and due to local demand for other services, such as electricity. Repetitive polio campaigns and poor communication with insurgents have exacerbated these problems (see chapter 6). Polio eradication: FATA versus Khyber Pakhtunkhwa The role of the ideology and organization of violence is further demonstrated by the comparison of polio eradica- tion delivery in KP and FATA. PEP was more problematic to implement, at greater risk of violent attacks, and made less progress toward achieving its outcomes in FATA than in KP for a number of reasons, including: • Ideology. FATA-based armed elites and allied insurgents have a more radicalized ideology and antagonistic view of the state than those in KP, partly due to the drone attacks in the region. Many focus group participants argue that the drone attacks also fueled the widespread perception in FATA that polio campaigns are a cover for U.S. intelligence gathering prior to pinpointing targets for drone attacks. Association with the state can automat- ically make one a target in FATA. • State penetration and organization of violence. In FATA, state penetration and control is deemed to be relatively weaker than it is in KP (see chapter 3), which has led to more fragmented and autonomous forms of violence there. It also means there are fewer opportunities for political and administrative cooperation and co- ordination to deliver services on the ground. a. The president of the Pashtunkhwa Mili Awami Party, the leading Pashtun Nationalist Party of Balochistan, perhaps representing a more radical, nationalist view, claims: “Our children are being kept out of schools intentionally. This policy dates back to British rule of this region. The British had a clear policy for our children that they must not get education. This state, mostly represented by Punjab, is continuing the same policy… they fear that if we get education and awareness, we will demand our rights.” (KII–14) and have promised to leave them alone education impacts whether or not it is vio- from now on.” (Seddon and Hussein lently targeted as well as the types of edu- 2002) cation that will be tolerated. The Taliban is a diverse group, and its ideology can differ Practitioners would do well to regularly visit from province to province, but some interest- the field to track changes in these kinds of ing points emerge from the research: perceptions and local tactics (see chapter 8). The Taliban’s ideology and “policy” partly Another example emerges from Afghanistan, explains why it conducted violent attacks on where the evolving Taliban ideology around schools. The armed opposition of the Taliban Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 63 identified education delivery as a key field various cases (chapter 5), some Taliban were of state activity. It regarded it as an attempt bargaining around co-opting, rather than to indoctrinate and impose Western ideas on closing, schools. children. The resulting campaign of violence in 2006–08 led to the destruction of numer- Taliban ideology was more tolerant of edu- ous schools and the closure of even more. cation for boys than girls, which impacted the nature of bargaining and their violent The evolution of the Taliban policy toward recourse, especially in Wardak. Despite the schools helps explain the reduction in at- growing acceptance of state-sponsored edu- tacks. Gradually, the Taliban’s campaign cation, Taliban ideology against girls being evolved toward an attempt to influence state educated continued. This was particularly schools through local-level negotiations prominent in Wardak where, with the arrival with Ministry of Education officials. By the of the Taliban in 1995, girls stopped attend- end of 2008, none of the schools in Balkh ing school, and most of the schools were and Wardak were inactive, and only 5 per- turned into madrasas. As a result, Wardak cent were in Badghis. It also explains why, in performed relatively well over the period Classes in Physics and Chemistry being conducted at the Female Experimental High School in Herat, Afghanistan. Graham Crouch/World Bank 64 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds 2008–14 in terms of health and boys’ educa- H8–FGD–PAL). Indeed, a number of inter- tion, but poorly in terms of girls’ education, viewees argue that improvements in health with the most significant decline in prima- were possible during the conflict because the ry-school attendance rates among the three Maoists had a strong interest in keeping them provinces studied. In effect, the Taliban made open as a source of services and rents (H21– implicit sectoral bargains to allow boys to KII–PAL; H29–KII–ROL; G10–KII–SRH). attend school at the expense of girls, which might help explain why in Balkh, where the Similarly, in Afghanistan, insurgents’ need Taliban presence and power is negligible, and for services helps explain numerous cases of in Badghis, where their collective power is health and education services being insu- fragmented, opposition to education for girls lated from violence. In Wardak, for example, was less dominant. there were instances of the Taliban allowing and protecting health services because they Incentives needed them. One informant notes: Armed elites and allied insurgents are more supportive of service delivery if there are “Medical staff does not have security prob- incentives for them to do so. Two questions lems, as the Taliban and militants only tar- emerged as particularly important: are the vi- gets military convoys. Ambulances can easily olent actors in need of the service in question, travel around, because the Taliban and mili- and can they extract rents? tants need health services. They are treated in these clinics, and their relatives also come for Do violent actors need the services? treatment to these clinics. Therefore, they do The extent to which insurgents tolerate ser- not prevent the delivery of health services.” vice delivery is shaped by the extent to which (WK: IDI–07) they, their families, and their “friends” need the service. When insurgents require the ser- Interestingly, there are even examples of the vice, they tend to facilitate basic SSD cover- Taliban actively protecting health service de- age, particularly in the health sector. livery from criminal networks: In Nepal, for example, the Maoists main- “The Taliban also has a level of commitment, tained some health facilities because they and because if they wanted to stop the delivery of their families needed the services. One of the health services, they could have already done reasons given at all the research sites for the it. For instance, there are robbers or other Maoists allowing health facilities to remain criminal groups who want to interrupt ser- open was the need for the rebels and their vices or harm our medical staff. Sometimes, families to make use of them (H3–FGD-SRH; the Taliban protect our staff and they do not Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 65 let people or criminals interfere in the health children from the local community needed service delivery. So yes, there is this level of the service (box 4.3). commitment present.” (KL: KII–47) Can violent actors extract rents? Likewise, an interesting example from Af- The ability of violent actors to extract rents ghanistan shows how commanders con- also impacts whether or not they will allow tinued to protect education despite major the delivery of services. Every service offers armed clashes because their children and the a unique opportunity for rent extraction, and Box 4.3. Wardak, Afghanistan: Education incentives before the Taliban In the 1980s and 1990s, the main jihadi party in Wardak was the Islamist Hezb-e Is- lami, which favored scientific and technical education, albeit within an Islamic framework (WK: KII–05 and 06). Some jihadi commanders wanted schooling for their own children and the support of locals. This is why even during armed clashes, many schools were actively kept open. Interviewees told some interesting stories about this: “The main factors that kept the schools open were the jihadi commanders, village elders, ulema, mullah, imams, and maliks. People gave land for school buildings and they main- tained the security of the schools, because on the one hand, their own children were at- tending these schools, and on the other hand, it was a service for their villagers. Although there were conflicts among the jihadi parties, they tried to support schools and they did not plan to close them.” (WK: KII–01) “Tribal and party competition were the highest in this area. This means that each party and commander had his own government in his area. But most of the people agreed with the delivery of education services and supported education. I can give you an example of their support to education. In Chak District, there was a govern- ment school building in the Shir Khana area. In this area, there were different par- ties. The members of each of these parties occupied a room of this school building; they were living there. When the people of the community asked the commanders to leave the school building so that it could be used for teaching and studying, they also ordered their personnel to leave the school buildings. Finally, all of them left the school building and classes started again.” (WK: KII–02) 66 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds across the studied cases, insurgents were more held sway, and the Maoists allowed health fa- willing to support a service if they could ex- cilities to function, with national campaigns tract rent. This was manifest in a range of elite such as polio and measles immunizations, vi- bargains where insurgents would allow or tamin supplementation, and family planning even enable the service to continue provided hardly affected (Devkota and van Teijlingen there was a clear rent-sharing agreement that 2009). A doctor who had served in a rural gave them access to supplies, privileged ser- primary health-care center during the conflict vices, or payments (see chapter 5). explains: “The Maoists also wanted the sup- port of the general mass for their insurgency and since health services are directly related 4.3. Localization of Violence to the people, they did not choose to obstruct it” (H18–KII–PAL). When armed elites are embedded in local communities during times of violent conflict, Afghanistan offers another example. In cer- the prospects for local bargaining around tain areas of Wardak, the Taliban relies on the SSD improve. A number of factors emerge as support of local tribal elites and communities important, including the insurgents’ reliance for local entry and legitimacy. As a result, on local support and how mobile or localized they were predisposed to enabling services they are. and appeasing local elites (see chapter 5). One informant explains: Reliance on Local Support “You see, the strength of the Taliban When armed elites rely on local support and is highly dependent on the support of local constituents want a service, then the the people [in Wardak Province] … elites are more willing to enable delivery. The Taliban approach the leaders of During the conflict in Nepal, for example, the community like the maliks, and this the incumbent government and the Maoist is their entry to the community.” (KL: insurgency grew to realize that health service KII–42) delivery was a vehicle for gaining legitimacy in the eyes of the people. This reduced dis- Similarly, as noted previously, Jihadi com- ruptions to health services and increased the manders were also willing to enable educa- propensity for accommodations and bargain- tion delivery in part because they relied on ing. The government continued to deliver local support. medical supplies to areas where the Maoists Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 67 Localized Versus Mobile Insurgents In conclusion, context-specific dimensions of violence have varying impacts on service The research shows that localized insur- delivery implementation and progress. They gents, who have strong links with the local explain how and why service delivery out- community, are more likely to enable local comes differ at the subnational level and how service delivery than are mobile insurgents. progress is—and is not—made. The various The more localized the armed elites and al- forms of violence, in addition to political set- lied insurgents, the stronger their links to the tlements, expand or reduce the bargaining community, the more responsive they are to space for delivering services. The following local needs, and the more willing they are to chapter explores the issue of bargaining in tolerate local service delivery (see box 4.4). more depth. A girl receives polio vaccine drops at the door step of her family home in 2016. Akhtar Soomro/ Reuters 68 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Box 4.4. Localized armed elites and allied insurgents and social service delivery bargains Across the studied cases, the ethnographic evidence suggests that the more localized the armed elites and allied insurgents, the easier it is to negotiate for and deliver services. Following is a summary of findings. Afghanistan A range of interviewees note that mobile insurgents, usually deployed from neighboring countries during the fight- ing season, do not have links to local communities and do not stay long enough to establish smooth communication channels with local elders. They note that these armed elites and allied insurgents are much less bound by local needs or bargains, and they cite numerous instances of insurgents from outside the province disrupting service de- livery. Some local representatives claim that “those Taliban who are not independent” (out-of-area Taliban coming from Pakistani territory) and the security forces are equally responsible for school attendance disruptions in some areas (WK: IDI–18). A range of informants claim that it is easier to negotiate service delivery with the help of local elders if the Taliban are locally connected. One informant describes an example of local Taliban being so amenable to service delivery bargains that they allowed unofficial schools to be established: “Until last year, girls in grade six did not have any problems. But due to some concerns, for girls in higher grades, between grade six and ten, the Taliban and the local people advised renting a specific house where these girls could attend school until the 10th grade.” (WK: IDI–14) In Wardak, the rise of external violent actors, who are less respectful of local bargains and elders, more ideologi- cally opposed to education, and less dependent on local support, helps explain the weakening sectoral bargains and overall decline of school attendance (particularly among girls). Nepal Evidence suggests that local village Maoists were more understanding of community needs than those from outside the village. The village of Taratal in Bardiya, for example, was relatively unaffected by the conflict because the Maoists in charge understood local needs. It was indeed Maoists from outside the village who obstructed develop- ment work (H16–KII–BAR). And nonlocal Maoists in the village of Madanpokhara in Palpa were reportedly more “forceful,” while local Maoists were more understanding (H22–KII–PAL). Pakistan The comparison between the Polio Eradication Project in Khyber Pakhtunkhwa and the Federally Administered Tribal Areas (FATA) underlines similar findings. Roving forms of violence, fostered by FATA’s porous borders with neighboring Afghanistan, made organizing negotiation focal points difficult. Indeed, when international forces engage in a military offensive against the Taliban in Afghanistan, militants move to Pakistani FATA. If there is an offensive by the Pakistani military, militants move to adjoining areas in Afghanistan. Khyber Pakhtunkhwa com- munities, however, do not offer such a safe haven to militants as those in FATA tend to do. 5. Elite Bargaining Matters “I see less harm in hiring a person who is corrupt but is capable of doing something … and is doing something.” (KL: KII–42) As discussed in previous chapters, con- 5.1. Forms of Bargaining text-specific political settlements and forms of violence create space and opportunities for Many of the bargains considered in this delivering services. But service delivery is not study are rolling agreements. They were in- mechanically determined by these factors. In- formally devised and must be regularly rene- deed, the research indicates that there is room gotiated. Some bargains were more successful for actors to strategize, negotiate, and com- or far-reaching than others. The form that a promise to get things done. bargain took was shaped by the local elite balance of power, the incentives and ideas of This chapter explores how elite bargain- armed and unarmed elites, and the ability of ing affects if and how services are delivered different actors to identify common interests. and progress achieved. Many bargains were sustained even during periods of intense vi- olence. This research goes beyond broad Negotiating with Insurgents statements like “elite bargains matter” to In Nepal, a range of informal negotiations be- identify the actual processes through which tween local actors and Maoists facilitated the bargains are achieved and reproduced. Bar- delivery of health and rural road services. In gaining is a fluid and overlapping process, Palpa and Bardiya, for example, the Maoists but it can be identified in various forms, as obstructed road construction, viewing it as outlined below. The bargains examined in state encroachment. But in several instances, this study facilitated some progress, but often construction resumed after negotiations took came with trade-offs. place between road users, local elites, and the Maoists, suggesting that the Maoists allowed the road construction projects to proceed when they believed that doing so enhanced their legitimacy with local communities. 70 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Another example involved negotiations be- “Before, the Taliban didn’t even let tween a district health officer in Siraha and boys’ schools open, but in many places, armed elites in the Tarai, which facilitated the boys’ schools are now open. We asked delivery of health training programs: them, ‘Will it be better for your son to be educated or illiterate at that time?’ “The [district health officer] was trying We [service providers and users] asked to organize a training program during them that. In this way, we convinced a strike by one of the two Jana Tan- them and opened schools for Mush- trik Mukti Morcha (JTMM) group [an kawani, which is in the Taliban’s terri- ostensible political party engaged in tory.” (BS: IDI–33) armed criminality in the Tarai]. All the offices were closed. But, Dr C did not Sometimes, negotiations required gaining the want to stop the training program re- permission and trust of armed elites and allied lated to OPD (Outpatient Department). insurgents. Local elites and service providers Some JTMM cadres came and threat- regularly informed and engaged with insur- ened Dr C to stop the program. How- gents, which was important because build- ever, he dealt with them and told them ing a level of trust with armed elites fostered that it was a program that came from needed predictability for service delivery. For donor agencies and stopping the train- example, there were significant dialogues and ing program would send out a wrong trust-building efforts between implementing message and such programs would partners, local armed elites, and local elites be stopped in Siraha district. He also for the Balochistan Education Support Proj- told them that the program was four ect (BESP), enabling service delivery. hours long and two hours had already been completed and requested them This form of bargaining sometimes resulted in to let him carry on with the training insurgents actually codelivering the service. In program. The JTMM cadres listened Nepal, for example, interviewees argue that a to him and returned.” (H4–KII–SRH; great deal of heath delivery continued during H1–FGD–SRH) the conflict, partly because health workers regularly liaised with the Maoist leaders and Persuasion earned support for SSD among because the Maoists were informed in ad- some insurgents in Afghanistan as well. Even vance when health programs were coming to in Badghis, where bargains are uncommon their areas, building trust and predictability. and problematic (see chapters 3 and 4), ser- One interviewee recalls: vice providers sometimes convinced the in- surgents to allow services to continue: Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 71 “Such people [auxiliary health work- negotiations, and accommodations with the ers] needed permission from the Mao- Taliban partly explains why, based on a num- ists and that was not so easily available. ber of indicators, Wardak service delivery has But since we knew the local Maoists, performed relatively well, even in the face of someone from the health post would volatile security conditions. Even at the peak accompany the district people. We of the Taliban’s power in the province, sec- would coordinate with the local Mao- toral elite bargains were struck. According ists here while those in the district to a former Wardak government official, the center would coordinate with the se- “coordination and cooperation between el- curity forces. Whenever we held health ders and government in the province is much camps, the dates would be decided better compared with other provinces” (KL: beforehand so that the security forces IDI–63). Box 5.1 offers examples from the were aware where our camps were field research. It is noteworthy that even in being held and they would not come Badghis, bargains were struck regarding on a patrol [so as not to cause conflict].” health and education delivery, facilitated by (H29–KII–ROL) local elites, even though they were somewhat fleeting and unstable. In Rolpa, local health staff coordinated with Maoists who issued passes for them to con- In Pakistan, religious elites played a role in duct national health programs during the enabling polio eradication delivery even in conflict (H25–KII–ROL). One health worker the difficult environments of Khyber Pakh- coordinated with the Maoists to take a group tunkhwa and the Federally Administered of 20 men on a three-hour walk to a vasec- Tribal Areas (FATA). There are instances of tomy clinic (H32–KII–ROL). religious elites helping to convince armed elites and allied insurgents, local elites, and constituents to allow the delivery of polio Enlisting the Support of Elites eradication services. In some cases, however, Another form of bargaining involves enlisting the support of such elites was needed but in- locally influential elites to serve as intermedi- sufficient to ensure successful delivery (see aries to enable service delivery. This involves, box 5.2). but goes beyond, basic dialogue and nego- tiations with armed elites and allied insur- Finally, the case of road construction in Nepal gents. In Afghanistan, these forms of bargains underlines the role of political party elites in were crucial. In Wardak, elders proactively resolving local conflicts and negotiating with supported education, and their bargaining, armed elites and allied insurgents to deliver 72 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Box 5.1. Enlisting the support of local elites in Wardak, Afghanistan Education Local elites in Wardak were enlisted on multiple occasions to ensure that education services were sustained during the insurgency: “When the Taliban closed schools, the Ministry of Education asked the tribal elders [in Wardak] to talk to the Taliban. Then there were discussions between the tribal elders and the Taliban, and some of their requests were accepted. Actually, only the tribal elders could convince the Taliban to keep the schools open.” (WK: IDI–16) “Last year, a girls’ school was blocked … by the Taliban [in Wardak]. They [pointing to the elders in the room] went to some Taliban members and mediated for the opening of the girls’ school. Those problems are not resolved by the government; those issues are solved by these elders.” (WK: IDI–14) Health Local elites and health shuras (consultative councils) in Wardak also enabled health service delivery: “There were security checkpoints near the clinic, which were attacked by [the Taliban]. Due to the fighting between opposite groups and security checkpoints, it was difficult for patients to come for treatment to the clinic … This problem was shared with health shura members, and to some extent, the problem was solved. The health shura and elders talked to the group of insurgents about not disturbing them during the day when women, men, and children go to the clinic for treatment, which they accepted.” (WK: IDI–07) “The … problem was the presence of insurgent groups against the government; that is a threat to everyone’s life. Through the establishment and communication of the health shuras, this problem was solved … The shura talked to insurgent groups about the need for health services … With that, [the insurgents] agreed to service delivery and that they will not interfere with the health sector staff.” (WK: KII–13) “About the interaction of these power holders [local elites] with the insurgents, they agreed to the delivery of health and education services in the area, because of the involvement of the elders … They verbally agreed that they would do no harm to health and education service delivery.” (WK: KII–09; WK: KII–12) Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 73 Box 5.2. Religious elites, fatwas, and polio delivery in Pakistan The example of Pakistan illustrates the important role religious elites can play in social service delivery. According to some practitioners, fatwas from respected clerics helped convince skeptical local religious leaders to accept the vaccination program, and they in turn convinced insurgents and locals (R20–KII–PSH; R15–FGD–PSH). One donor agency official, noting how fatwas helped tackle cases of refusals, cites an instance of a fatwa issued by a madrasa in Multan persuading the religious elite to cooperate: “We had to face refusal from the whole village once, when the imam of the mosque announced to boycott polio drops in a village named ‘Sadri Jadeed’ in Dagi, Swabi. In the year 2012–13, we brought a fatwa from a madrasa in Multan to convince reluctant religious persons.” (R10–KII–PSH) In the Federally Administered Tribal Areas (FATA), internally displaced female community members from North Waziristan engaged with religious and political elites to convince the Taliban of the benefits of polio vaccination (R19–FGD–PSH). However, while fatwas were necessary, they could not ensure effective polio eradication delivery, especially in FATA. Even though numerous respected clerics supported polio immunization, the Pakistani Taliban have consistently refused to allow polio teams into their areas, especially in FATA. The background country report notes that in some cases, ensuring the support of religious leaders was not enough to broker deals around the Polio Eradication Proj- ect. A focus group discussion participant with the FATA polio program technical staff argues: “Due to lack of education, religious guidelines take heavy precedent over everything else—but having a fatwa issued is not enough, you need to convince local imams as well or they will oppose you despite the fatwa from higher level imams.” (R–24–GI–PSH) services. The presence of politicians in local the most adverse circumstances such as user committees made it easier to resolve con- direct negotiations with the Maoists flict (R5–KII–KTH). As one informant put it: during the insurgency as happened in the case of the Arebhanjyang-Ram- “It is … true that having politicians pur Road in Palpa, a role the project on board helped in ensuring that con- could not have undertaken on its own.” struction work was not halted even in (G17–KII–PAL) 74 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds There were other examples cited as well. In Similar concessions were made in Nepal. If Bardiya, road construction was blocked by local health workers agreed to discreetly pro- the Maoists but then resumed after the road vide services to the Maoists, then health ser- user committee got local supporters to con- vices were often maintained. Sometimes, the vince the insurgents that the road was indeed Maoists cultivated personal relationships with in the community interest (G13–KII–BAR); health workers, seeking their help at any time and a locally influential elite successfully con- of the day or night. The health workers had vinced the Maoists to allow health training to be discrete and cultivate trust with the in- programs to be conducted (H7–FGD–BAR). surgents, who did not spare anyone suspected of being an informant (World Bank and SSB In Palpa, local political parties engaged in di- 2016). alogues with the Maoists in order to reduce the violence (G19–KII–PAL). 2. Modifying the service A second aspect of bargaining involves mod- ifying aspects of a service to appease armed Concessions and Compromises elites and allied insurgents. This tactic was These types of concession-making bargains apparent in education delivery in Afghanistan manifested themselves in at least three ways: and Pakistan. In Balochistan, Pakistan, armed (1) providing privileged services to insur- elites and allied insurgents were more predis- gents; (2) modifying the service; and (3) using posed to allowing education services if the services as a bargaining chip. curriculum was modified. The armed elites and allied insurgents did not interfere with 1. Providing privileged services to school operations as long as: (1) the school insurgents did not raise the Pakistani flag; (2) school In Afghanistan, concessions were made with children did not sing the national anthem; the Taliban so that they would not attack and (3) Pakistani history was not taught health services. For example, nongovern- (PIPS 2012). mental organizations and government clinics have been asked by local elders to treat the Similarly, in Afghanistan, the Taliban allowed Taliban, and to allow the Taliban to conduct education services as long as the curriculum inspections of facilities so they could ensure was modified in a way that aligned with their that these were not being used for “spying” ideology. For example, there were reports that purposes. There have been instances of vi- deals were negotiated in Wardak allowing olence and bans on further service delivery schools to stay open in exchange for conces- when service providers refused (Jackson and sions that included changes to the curriculum Giustozzi 2012; Norland 2012). Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 75 (Giustozzi and Franco 2011). As one infor- political bargaining chip to ask for govern- mant describes: ment concessions, such as improved power supplies, a reduction in the cost of power, “In most of the districts, the govern- and increased wages. This is expressed in a ment curriculum is being taught in number of ways: schools. But in some areas, where there is the Taliban, there is a slight change “[P]eople were saying … that if their in the curriculum, which the govern- salaries were not increased they would ment does not favor. … The Taliban boycott Polio Eradication activities. So … keep control over the curriculum of it is becoming a political tool in the education in their controlled areas … hands of people as a bargaining chip.” We have not heard that the Taliban has (R11–KII–PSH) made any problems with either minis- try [the Ministry of Public Health or the “… if they [elites and constituents] Ministry of Education] or their work in need something (roads, electricity) they Wardak Province.” (WK: IDI–18) target polio, knowing that it is prior- ity of the government and through this Another informant made a similar point: “If their voice can be finally heard. This is there were a problem with the curriculum, kind of a weapon that they have to be the Taliban would go to those schools and heard.” (R13–KII–PSH) would advise for those subjects to be taken out and suggest which subjects to include.” “There are three types of refusals, re- (WK: IDI–14). ligious, demand and misconception. Either they are refusing on religious 3. Using services as a bargaining chip grounds; or they demand for some- Armed elites and other elites sometimes use thing and threaten not to vaccinate services as a bargaining chip to achieve other their children unless the demand is met goals. This tactic emerged most prominently with; or they have a misconception in the case of polio eradication delivery in regarding polio drops about it having Pakistan. There were instances of the Tali- dangerous elements such as inducing ban opposing polio vaccination ostensibly in infertility.” (R–24–GI–PSH) an effort to halt the drone attacks in FATA. Local elites and constituents attempted to use Chapter 6 explores these points in greater resistance to polio eradication delivery as a detail. 76 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Sharing Rents and Tolerating In Nepal, a key puzzle is how health service Corruption delivery was sustained during periods of in- tense violent conflict. One part of this answer One form of bargaining involves rent-shar- relates to the incentives of the armed elites ing arrangements that enable progress in ser- and allied insurgents and a series of bargains, vice delivery in the midst of violent conflict. as outlined above. Another dimension of this These bargains emerged across the three stud- bargaining involved rent-sharing agreements. ied countries, some more enabling of service In a number of cases, Maoists allowed health delivery than others. None resembled best services to remain operational in exchange for practice or good governance approaches to rents in the form of donations and levies. The SSD. This section outlines some of these ar- Maoists typically demanded either “dona- rangements at the front line of delivery. (See tions” or a day’s worth of salary per month as a chapter 3 for a more thorough exploration “levy” from health staff in health posts (health of these rent-sharing arrangements at higher assistants, assistant health workers, nurses) levels of the polity.) as well as private medical practitioners.24 In Principal of Sorya High School in West Kabul, Afghanistan. Graham Crouch/World Bank Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 77 some cases, they also demanded a share of the services we had to provide to the Maoists.” supply of medicines, especially during peri- (H30–KII–ROL) ods of intense fighting.25 Because local health workers could not easily find commensurate From a policy perspective, these examples employment elsewhere, many remained at suggest how rent-sharing can enable contin- their posts despite the extortion.26 ued delivery and reduce the risk that insur- gents will target a service (see also box 5.3 There is evidence that government officials and chapter 8). turned a blind eye to these practices, a crit- ical factor enabling continued delivery. Dis- Rent-sharing bargains were also evident in trict-level officials reportedly understood that road construction projects in Nepal. Some- they needed to continue sending medicine to times, the Maoists viewed road projects as a villages in accordance with the government’s source of their own funding during their in- strategy of delivering services and enhancing surgency; the same was later true for many of local legitimacy even if some would fall into the armed elites in the Tarai. There is some rebel hands. They understood that only by evidence that negotiations and rent-sharing heeding the demands of local health work- between Maoists and local supporters allowed ers could district offices ensure their contin- road construction to continue. One key infor- ued presence in the villages, especially when mant notes: the insurgents heightened oversight of their activities (Hart 2001; Seddon and Hussein “BJ [interviewee] shared an incident 2002). Cases of district officials relaxing their in 2003, when the Maoists looted formal oversight of supplies was noted, as a 40,000 Nepali Rupees from [the chair health worker in Rolpa recalls: of the road construction committee] … They blamed him for constructing “We did have to keep a record of all the the roads without their permission and medical supplies that came to our health for the convenience of security officials post. But we were not made to clarify how to travel rather than for the public. In medicines were consumed to the district of- order to continue the project after a fice. The people at the district headquarters halt of some time, the committee sent knew everything. They were well acquainted local supporters of the Maoists to them with the situation at the local level. Even the to explain the importance of the road CDO (Chief District Officer)27 during our to the community and succeeded in review meetings would ask us to adapt to getting their permission to resume the the situation and would not cross-question project.” (G13–KII–BAR) us on either the supplies we received or the 78 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Box 5.3. Bargaining with the Maoists in Palpa, Nepal K.C., a health care assistant at the Tahu primary health care center in the district of Palpa in Nepal, discussed several strategies that were used to keep health services functioning during the Maoist insurgency in Palpa: Rents. In addition to paying a levy of a day’s salary every month to the Maoists, providing supplies to them was crucial to remaining operational. K.C. gave the insurgents anything they requested, though he could not ask them to submit a medicine demand form, as the procedure dictated. On one occasion, insurgents asked for a sphygmo- manometer (a tool that measures blood pressure), but the center did not have one, so K.C. put in a request for two. The district health officer who sent the equipment knew without asking any questions. K.C. believed that the officer understood the sensitive and difficult nature of the conditions under which health centers located in remote areas were operating, and therefore promptly responded to their requests. Supplies were continually replenished, and the district officer never demanded any kind of accounting for them. Discretion. An important survival strategy was to never provide any information about the security forces to the Maoists and vice versa. For example, government forces would often arrive at the primary health care center a few days after a battle and question the staff as to whether injured combatants had been brought in for treatment. The standard response from the staff was to suggest that they had perhaps been taken to Rampur—a nearby market town. Treat on request. Wounded Maoist fighters were discreetly treated on request in a room separate from other pa- tients. Some of the Maoist insurgents visited K.C. at his home as well because he had a private pharmacy. When this happened, K.C. had to provide both health services and accommodation. Source: Based on H17–KII–PAL. According to another informant, adherence appropriate and share rents from the social to a strict policy of neutrality toward all sides service sectors (BS: IDI–23). In Wardak, some was the only way people could continue to in the education sector accept patronage and live in the affected areas (R20–KII–PAL). corruption as a necessary evil in exchange for delivering services. One informant notes that Similarly, in Afghanistan, we find cases of some who are recruited through patronage rent-sharing in both health and education. networks for the wrong reasons may still be In Badghis, for example, armed and un- able to deliver: armed elites colluded in several instances to Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 79 “I see less harm in hiring a person powerholders disrupted delivery and made who is corrupt but is capable of doing life more difficult for SSD workers. something and is doing something. For instance, look at the enthusiasm that The case of polio eradication delivery in Paki- everyone has in the education sector. stan illustrates this point. As noted in part I, a Everyone is looking to what is going key objective of the research is to explain why on in schools, with teacher training polio eradication delivery was so challenging and the curriculum. However, if you and frequently targeted by acts of violence, place someone there who looks at the especially in FATA. Chapters 3 and 4 describe delivery of any services as something how this was partly shaped by the nature of to help him or her gain economic re- the political settlements and the forms of sources or use it to promote a certain and motivations for violence that contribute agenda, but does not work toward its to the absence of bargaining—negotiations, development, that is another thing.” accommodations, and concessions—with in- (KL: KII–42) surgents, in turn leading to highly problem- atic service delivery.29 Insurgents felt their SSD was more developmental in Balkh but demands were not being met even partially, still not immune to rent-sharing bargains as dialogue and trust building with insurgents a way of getting things done. For example, was at best weak, and local elites wanted ser- interviewees in the education sector admit vices other than polio eradication (see chapter that interference and pressure for favors or 6). As a result, delivery of the Polio Eradi- bribes from Members of Parliament (MPs) cation Project became a dangerous business. and strongmen did occasionally occur during Health workers could not travel to FATA and the implementation of the Education Quality some parts of Khyber Pakhtunkhwa, such as Improvement Project (EQUIP).28 See chapter Swabi, because of the heightened security 3 for further discussion about these types of risk.30 There has been no polio eradication arrangements. campaign in North Waziristan since 2010 be- cause of the security risks and lack of access to the region (R13–KII–PSH).31 Security was If A Bargain Cannot Be Struck particularly difficult for female local health The research indicates that the absence of a workers, who were seen as key enablers to bargain complicates SSD. Across all sectors improved vaccination coverage because they and countries studied, an absence of dialogue could enter households to vaccinate children and bargaining with armed elites and other and were familiar with and known to their 80 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds communities. These women were specifically 5.2. Trade-Offs targeted by insurgents in their effort to dis- rupt the operations of the Polio Eradication The research does not suggest that bargain- Project (R15–FGD–PSH). In some communi- ing is the ideal way to deliver services, merely ties, people would not get vaccinated because that it often appears necessary to achieve they feared retribution by insurgents opposed progress. Elite bargains frequently include to the program (R09–KII–PSH; R10–KII– trade-offs, which do not represent technical PSH). In Swabi, Khyber Pakhtunkhwa, fewer failures in SSD design or implementation, people vaccinated their children at the district but instead reflect the political compromises level because the government required proof made to allow or to maintain SSD. A better of address, and families were afraid of being understanding of these trade-offs can help targeted by insurgents (R10–KII–PSH).32 inform strategies for supporting SSD in vio- lence-affected contexts (see chapter 8). Four Limited outreach, dialogue, and trust-build- significant trade-offs emerge from the re- ing with armed elites and allied insurgents search, which are described in turn below. also contributed to challenges in Balochistan. While local elite bargains and the alignment of interests between insurgents, local elders/ 1. Service Delivery Maintained, but elites, and implementing partners facilitated Flow of Resources Skewed the local elite bargains that underpinned the It is not unusual for resources to be diverted progress of the BESP, the absence of such as a result of bargaining. For example, the bargains seemed to have the opposite effect. bargains between local elites and armed elites Security raids led to school closures in some in Wardak, Afghanistan, were crucial to en- areas (KII–11); some city schools such as in abling the continuation of education and Turbat, Panjgor, and Awaran were affected by health services, but at the same time, signif- raids (KII–05); violence in the southeast re- icant evidence exists of skewed investments. gion between different groups blocked access Local elites, despite their positive attitude to- to schools (KII–02); and there were reports ward education, also view it as an opportunity of attacks on project implementers who were to extend their patronage. In some instances, perceived to be representatives of the govern- lobbying by economic and political elites ment (KII–09). Similar incidents occurred in over the location of schools disrupted edu- Nepal, where delivery of health and road ser- cational development in the area, resulting in vices were disrupted when demands for “do- suboptimal sites that left some communities nations” were not met.33 Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 81 very far from a school and others quite close but this also affected their distribution (box (WK: KII–01). Allegations of corruption 5.4). Similarly, local politicians in Nepal who in the education sector are common (WK: played a role in bargaining also influenced IDI–18). In Wardak’s health sector, there was the selection of roads, diverting projects to an alignment of interests between provincial areas they favored.34 and powerful local elites to enable services Box 5.4. Bargains and skewed health provision in Wardak, Afghanistan The experience of the health sector in Wardak demonstrates how bargaining between provincial and more powerful elites can contribute to a skewed provision of services. For example, provincial council members from the two Hazara-dominated districts in Wardak authorized several local clinics, despite these not being in accordance with Ministry of Health guidance. As one interviewee notes about two related events: “They have many recommendations for upgrading a health facility, for downgrad- ing a health facility, for establishing a new health facility. In the two Behsud areas, this is very common. Therefore, we have an extensive number of health facilities in Behsud, because every time they go to the minister, he is not brave enough to face these Members of Parliament (MPs). The minister refers the case to the Grants and Service Contracts Management Unit in the Ministry of Health and then to us. What began as a small clinic eventually became a well-established clinic. We now have 23 clinics in the two Behsud areas; 40 percent of our clinics in Wardak are in Behsud. He [senior health official] forced us to accept what he thought was right in terms of the staff in one of the clinics, where he had some personal interest. That clinic was previously supported by the government and was later taken over by a nongov- ernmental organization. He still had links in the clinic. He wanted to be involved in all of the recruitment and everything that was happening in the clinic. He was frequently supervising it. He forced us to upgrade it from a Comprehensive Health Centre (CHC) to a CHC+, and he succeeded. But he did not stop with the CHC+, as he said that it should be a special CHC+. So he went to the new minister and gained special approval to hire people.” (KL: KII–47) 82 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds 2. Basic Service Coverage Sustained, modules, such as history, from the curriculum. but Quality Compromised Likewise, deals were struck in Afghanistan with the Taliban to keep schools open that In the Tarai region of Nepal, bargains struck required adaptations to the curriculum and regarding road construction helped maintain there are similar questions about the impact service delivery but compromised quality. on learning outcomes of the dramatic expan- Large-scale road contractors—usually based sion in hours dedicated to religious subjects out of Kathmandu—were more frequently to the detriment of math and reading. Wardak targeted by roving armed groups in the post- commanders were willing to allow education, 2006 period because they were seen as major but there is scant evidence suggesting that money-making targets. A resulting adaptation they were incentivized to deal with issues involved the use of local subcontractors who related to quality of education, such as low could get the work done more easily because attendance rates or poor quality of teachers. they were more localized, more trusted by the bandits, able to negotiate on a local level with armed elites, and better equipped to resolve 3. Some Groups Benefit, but local problems as they arose (R1–FGD–SRH). Others Are Excluded However, because of this deal, road construc- tion quality was compromised. Smaller-scale Sometimes, bargains maintain delivery for local contractors were less skilled, and sub- some groups and exclude others. These bar- contracting made monitoring and controlling gains depend on the interests, incentives, and construction less straightforward. Therefore, ideology of armed elites and allied insurgents the problem of substandard quality in road and the extent to which service providers are construction is not merely technical. Rather, willing to compromise certain goals to en- it involves the political economy challenges sure minimum service delivery. The example around securing a feasible bargain in a violent of girls’ education in Afghanistan is a case in context. point. The Taliban tolerated schools in some areas as long as they did not emphasize edu- The education sector in Pakistan and Afghan- cation for girls. Providers were left in a quan- istan require similar trade-offs. In Balochistan, dary: should they adapt to the demands of the bargains were struck with armed elites and local Taliban or go against them and risk the allied insurgents that enabled education de- entire service being targeted or shut down? livery as long as the curriculum was modified This explains why in Wardak, even though in line with Balochi nationalist ideology. This education for girls is a priority for EQUIP, allowed access to education, but questions re- progress has been minimal, marked by a de- main as to how greatly the quality of educa- cline in attendance rates for girls in primary tion was compromised by removing certain schools over the course of the project. Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 83 4. Small-Scale Impacts, but Difficult In short, different forms of bargaining can to Scale Up make or break service delivery in violence-af- fected contexts. These various modes of bar- Bargains made at the local or micro level can gaining help explain how progress is made prove difficult to scale up. Some interventions against the odds, and why the lack of a bargain are focused on small-scale, localized delivery, can make life very difficult for service pro- which enables them to forge bargains with lo- viders. But bargains bring trade-offs, and the cal-level elites and allows them to circumvent same agreement can contain both progressive higher-level predatory or clientelistic elites. and regressive elements. The extent to which Scaling up, however, is challenging because a bargain is regressive or suboptimal depends elite interests and incentives operating at on the nature of the political settlement and higher levels of the polity may be quite dif- the motivations and ideas of local insurgents ferent (see discussion of BESP in Balochistan and elites. Finally, the research suggests that in chapter 6). different modes of delivery can shape the po- tential for more progressive forms of bargain- ing, as the next chapter explores. 84 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds 6. Modes of Delivery Matter This chapter explores the final piece of our with a mandate to support public-private framework: how progress in service delivery and community partnerships in education is shaped by the extent to which a strategy (World Bank 2006).35 Evidence suggests or mode of delivery adapts to local forms of that the BEF contributed to progress under politics, violence, and bargaining. While po- the BESP in part by bypassing the provin- litical settlements, forms of violence, and the cial-level, clientelistic, incumbent education nature of bargains can be important drivers settlement that was centered around the De- of—or barriers to—progress, service delivery partment of Education. At least two factors strategies and implementation modalities also appear to have enabled this. First, because it seem to matter. was a semiautonomous agency overseen by a board of governors and directors, its account- ability structures and bureaucratic procedures 6.1. Mitigating the Effects of were somewhat separate from other education Sectoral Patronage activities under the department. Second, its mandate to work with nonstate and private Some approaches—by luck or by design— sector actors meant that it was more able to seemed to reduce the impact of sectoral work through new coalitions and actors, such politics that constrain social service delivery as nongovernmental organizations (NGOs) (SSD). These modes of delivery manage to and other agencies with direct links to local lessen the influence of existing, clientelistic elites, which brought a broader network of political coalitions. elites into the operation. However, this model was contested and came Establish Semiautonomous with trade-offs. Points of disagreement in- Institutional Delivery Channels cluded the following: An important aspect of the Balochistan Ed- ucation Support Project (BESP) was the es- • There was a rivalry between the BEF and tablishment of a semiautonomous institution the Department of Education, and the lat- to manage and finance education delivery. ter had concerns about setting up a “dual” The Balochistan Education Foundation (BEF) (or parallel) public sector in rural educa- was the main executing body of the BESP tion. Some department officials suggested Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 85 that the BEF was establishing a parallel and sustainability of operations and building overlapping implementation channel, while trust among citizens” (World Bank 2015b). others suggested that the reason the BEF was being criticized was that it was outper- In fact, under the Balochistan follow-up proj- forming the department, and in so doing ect, implementation shifted from a semiauton- was seen as a threat. One official notes: omous body to a government-led operation. An important question is if such changes were “[W]e felt, on basis of various in- due to inefficiencies or to a backlash from in- teractions, that the government felt cumbent political coalitions seeking to regain threatened by the role of Balochistan sectoral control. Education Foundation (BEF). Their feeling was as if BEF was intending Working Small and Local to act as a parallel department to the education department of the Gov- Another aspect of the BESP that was seen as ernment of Balochistan.” (KII–01)36 politically astute was how it worked through local-level elites. Evidence from interviewees • Some claimed that the BEF and nonstate at various levels of the polity suggests that implementers were themselves prone to the BESP model was relatively successful not clientelistic capture by the new elites who only because it was community based but had been incorporated into the sector.37 also because it was able to work directly with These claims are not explored or corrobo- the local-level elite, who were motivated to rated by this research. support education in their communities and reduce capture by provincial-level elites, who • Some thought the nonstate model under were more linked to the mainstream educa- the BEF could undermine trust-building tion settlement. This strategy also aligned efforts. The World Bank’s implementa- with the BEF model of encouraging greater tion, completion, and results report for the local involvement. But it came with trade-offs Promoting Girls Education in Balochistan (see box 6.1). (PGEB) project (2012–15), a follow-up to the BESP, explains that “implementation by the government as opposed to NGOs or other parties greatly contributes to the 86 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Box 6.1. Engaging local elites in the Balochistan Education Support Project: Challenges to scaling up A politically savvy approach One reason that the Balochistan Education Support Project (BESP) made progress was that it was able to work through local-level elites, which was arguably politically savvy in a number of ways, as outlined below. • Direct links to local elite. The project was able to work directly with local elites who had greater incentives and interest in furthering education than higher-level elites (see chapter 3). For example, by only permitting parents whose children were enrolled in BESP schools to be members of the parent education committees, the project design reduced the direct influence of higher-level elites who would typically send their own children to private schools. • Reduced scope for interference. By working more at the local level, with smaller contracts and protected by the Balochistan Education Foundation model, there was less scope and interests for elites (especially ones linked with the Department of Education) that dominated the sectoral political settlement to interfere with or capture the project. • Space for local bargains. Working at the local level also provided space for bargains between local elites and armed elites and allied insurgents (see chapter 5). • Demonstration effect. Although at a small scale, interviewees suggest that the local-level approach created an important political demonstration effect by showing how projects can work in a context of violence. It further contributed to local demand and education which could eventually create new coalitions of elites and pressure higher-level elites to be more responsive. (continued) Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 87 Box 6.1. Continued Trade-offs There are, however, political and developmental trade-offs that accompany this model, as summarized below. • Prospects for scaling up. Evidence suggests that attempts at scaling up would be challenging because doing so would likely challenge the incumbent sectoral political settlement (World Bank 2006).a Issues include: – The Ministry of Education is already thought to be overstaffed, with more than 60,000 school teachers on the government payroll. Many of these teachers are ostensibly recruited on a political basis, and teacher ab- senteeism is high at 15 percent—among the highest rates in Pakistan. Integrating BESP teachers could put additional pressure on the payroll. – The new provincial coalition government is planning to reform its teacher recruitment method by outsourcing it to the National Testing Service, an independent recruiting and testing agency. Because BESP teachers were recruited by traditional methods, the government may be reluctant to regularize them. – The political parties in the new government may have little political capital to gain by supporting the regular- ized recruitment of the nearly 800 BESP teachers—although some argue that they can be considered a po- tential “800 voters.” A donor technical specialist, commenting on the complexities of the teacher recruitment agenda, explains: “As the nearly 800 BESP teachers existed outside the teacher union, many political economy aspects connected to the role of regularized teachers in the system did not apply to them. This included for in- stance the bargaining power of regular teachers through their role in the election and census.” (KII–11) • Limited impact on sectoral transformation. BESP has remained a small part of the total education provision in the province, constituting only 1.7 percent of the government’s education expenditures from 2006 to 2014. • Quality challenges. Efforts to maximize coverage of marginalized communities has come at the expense of teacher salaries, teacher quality, and potentially, education quality. Evidence gleaned from interviews suggests that low teacher salaries makes recruiting and retaining qualified teachers challenging. a. The project was actually intended as a pilot intervention, to be expanded if successful (World Bank 2006: 3). 88 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds 6.2. Anticipating and Mitigating Buying Off Insurgents and Insurgent Resistance Opposing Elites As described in chapter 5, rent-sharing ar- Various strategies emerge from the research rangements with insurgents are critical to as more or less effective in mitigating resis- maintaining service delivery in the face of vi- tance and violence from armed groups. The olent conflict. By contrast, the case of polio research revealed the following types: (1) eradication efforts in Pakistan suggests that a buying off armed elites and allied insurgents singular focus on polio eradication and the and opposing elites; (2) using trusted service lack of service bundling undermined efforts providers; (3) embedding conflict resolution to buy off local insurgents and elites and to mechanisms; and (4) actively reaching out to forge local bargains (see box 6.2). The case armed elites and allied insurgents. demonstrates that it is extremely risky to em- phasize one service without tailoring it to the incentives of the local elites and insurgents. Students are doing practice of their lessons in the laboratory of Mashriqi high school. Imal Hashemi/Taimani Films/World Bank Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 89 Box 6.2. Lack of bundling undermined elite bargains on polio eradication in Pakistan Limited incentives Evidence from interviews suggests that a singular focus and a lack of service bundling undermined the Polio Erad- ication Project’s ability to buy off local insurgents and elites, incentivize local cooperation, and forge local bar- gains. Armed elites and allied insurgents were more interested in other potential services, such as infrastructure and electricity, as well as other forms of bargaining, such as reducing drone attacks. Local elites and their constituents considered polio eradication a low priority, especially when compared with other health and nonhealth services. As a result, in many cases, local elites had limited incentives to protect or promote polio eradication delivery. One donor notes: “Because there have been significant gains in polio reduction throughout the rest of the country, because there are very few polio cases in a population of 180 million, and because 140,000 children die annually in Pakistan from other causes, it is not hard to see why polio is down the list of priorities.” (R–04–KII–PSH) Potential for bundling Clear political economy incentives for allowing social service delivery and bargaining around polio eradication delivery were lacking. Several interviewees suggest that this problem could be solved through bundling. They argue that local elites and households would be more responsive to polio vaccination efforts if they were offered as part of a wider package of health—and other—services. Polio program technical staff describe it like this: “Free medical camps should be used as mediums to encourage polio vaccination rather than relying on door to door visits that ferment resistance and suspicion of [the] polio-specific eradication program … Provision of ‘complete health packages’ (including polio immunization) and facilities can bolster trust of local people. Such activities, moreover, are supported by political, religious and local elites.” (R11–KII–PSH) This again demonstrates how seemingly technical changes in service delivery can have notable implications to the broader acceptance of a particular intervention. Challenges to bundling There are also challenges to putting bundling into practice. One donor official notes that it may be costly to pro- vide other services in tandem with the Polio Eradication Project (R05–KII–PSH), but program designers should nonetheless weigh the costs against the risk of polio eradication efforts being rejected or targeted by violence. In addition, bundling would require coordination within the health ministry and across other ministries, and this would likely be more difficult in the Federally Administered Tribal Areas than in Khyber Pakhtunkhwa, because administrative structures are much weaker there (see chapter 3). 90 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Students study in the library at the Female Experimental High School in Herat, Afghanistan. Graham Crouch/World Bank Using Trusted Service Providers reliance on local shuras (consultative councils) in Afghanistan; or the use of local implement- Multiple cases demonstrate that certain ser- ing partners for BESP (see chapters 4 and 5). vice providers were more or less trusted by Another example relates to the role of local armed elites and allied insurgents and there- female volunteers in the provision of health fore more or less likely to be allowed to con- services in Nepal. Health services, particularly tinue delivering services. Examples include preventative services, were sustained and even the extremely negative consequences of pro- expanded during the violent conflict partly viders being associated with the state in the because the local female volunteers, on whom Polio Eradication Project; the need to use local the service delivery depended, earned the subcontractors for road projects in Nepal; the trust of the Maoists (box 6.3). Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 91 Box 6.3. Female volunteers deliver health services in violent contexts in Nepal Background The use of female community health volunteers (FCHV) in Nepal can be seen as an approach that is sensitive to context-specific politics and violence. It’s particularly politically savvy because it bypasses some of the entrenched human resource patronage in the health sector, which revolves around close linkages between political parties, el- ements of the Ministry of Health, and trade unions (see chapter 3). The strategy was also sensitive to the violent context. Nepal’s Female Community Health Volunteer Program, a partnership between Nepal’s Ministry of Health and Popula- tion and the United States Agency for International Development (USAID), began in 1988 and by 1995 covered all 75 districts.a The program’s goals were to “improve community participation and outreach of health promotion through local women working voluntarily” (Bhandari, Gordon, and Shakya 2011). By 2012 there were around 52,000 female volunteers in Nepal serving as the primary contact between health facilities and communities.b From providing door- to-door services disseminating health information to supplying necessary medications and advice to pregnant women and new mothers, FCHVs guide people to the health facilities and educate them on the services available. Impact of FCHVs One health worker described the volunteers as the “medical officers” of their communities, claiming that without them health information would not have been distributed and the services would not have been as effective (H8– KII–SRH). A World Bank assessment notes that: “Continuation, improvement and strengthening of the FCHV program therefore appear as a key strategic priority for HSS [Health Sector Strategy]. Regarding the importance of first line health workers in rural and remote areas a strengthened FCHV program will act as a powerful booster for the social impact of HSS not least according to gender, indigenous peoples and Dalit communities.” (World Bank 2004) And in a 2004 statement, the minister of health, acknowledges that the then over 49,000 FCHVs served as the main link between the village and the country’s health infrastructure (Nepal Ministry of Health 2004). Trusted providers Even at the peak of the conflict, the volunteers managed to sustain service delivery, largely because they were non- governmental workers and locally based and recruited and so therefore trusted by the Maoist leadership as well as by the local communities. FCHVs tend to remain in their roles for a long time—the program has a very low turnover rate, enabling them to build longer-term relationships and bonds of trust with insurgents, local elites, and community members.c A health worker in Rolpa emphasizes the key role the volunteers play in violence-affected and remote areas: (continued) 92 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Box 6.3. Continued “FCHVs were not harassed by the state nor by the Maoists, and so they were able to carry out their work. Would you think that without the FCHVs family planning measures in their wards and village development committees would have been successful? Would the locals consume medications if there had been no FCHVs? It’s easy to make policies but the crux is its implementation. Even we have not been able to go and raise awareness in the nooks and crannies of each and every VDC. I have worked here for 20 years but I have been able to go to only five of the 22 VDCs under my charge. Do you think that by going to five VDCs there would be changes at the local level in terms of health awareness? The local Maoists have also benefitted from the activities of the FCHVs. If they had a headache, they were able to get paracetamol [acetaminophen] from the FCHVs.” (H25–KII–ROL) Not a panacea FCHVs were not, however, a cure-all for health delivery. They did help drive progress in basic health care, but they were unable to deliver more complex, staff-intensive, curative health care that requires a more skilled workforce, which for political economy reasons cannot be easily deployed in rural areas (see chapter 3). According to one es- timate, there is only one doctor available for every 150,000 people in rural areas, compared with the urban area of Kathmandu where the ratio is one doctor for 850 people. The government relied on lower-skilled staff to deliver preventative health services, enabling it to remain on track in meeting its Millennium Development Goals. For ex- ample, apart from occasional and always brief stints by medical doctors, one primary health-care center in Rolpa had effectively been run by an auxiliary health worker for 20 years.d a. At least one woman serves as a volunteer per ward (the smallest subdivision of local governance). An estimated 40,000 FCHVs were en- gaged by mid-1997 (Fielder 2000), and even during the most intense fighting period (2001–06), their numbers grew by 14 percent. b. Facing a severe shortage of health workers, the World Health Organization promoted the idea of “task-shifting,” that is the “review and subsequent delegation of tasks to the “lowest” category that can perform them successfully” (Lehmann and Sanders 2007: 1). This idea has received renewed interest among community health worker programs. FCHVs are chosen by local mothers’ groups from the community to support activities conducted by local health facilities. Following an 18-day basic training, they can provide an array of services, including distributing vitamin A capsules and deworming tablets to children under five years old, providing health education in terms of family plan- ning, distributing condoms and birth control pills, providing community-based treatment of pneumonia, treating diarrhea, referring out sick newborn babies, providing antenatal counseling to pregnant women, and distributing medication to mothers for improved neonatal cord stump care and to pregnant women to prevent postpartum hemorrhages. Despite these varied responsibilities and the fact that the volunteers do not get paid, the program has a low attrition rate of less than 4 percent per year. c. A focus group discussion participant explains: “During the construction of the Sub-Health Post at Tarataal in Bardiya district, the Maoists had been obstructing the work asking about the source of the funds. Finally, they allowed the construction to proceed provided that the monitoring was done by no one else but an FCHV. Hence, each of the 21 FCHVs active at the time took turns to oversee and monitor the construction work every day, and they were even on duty during important holidays. It was this kind of trust that allowed FCHVs to move around unhindered during the time of the conflict.” (H16–KII–BAR) Even during the Madhesi uprising in 2008 and the subsequent armed activity in the Tarai, during which it became quite perilous for people from the hill region to work there, the fact that the FCHVs were themselves Madhesis allowed them to operate freely (H6–KII–SRH). “There was a time when health officials would come in their vehicles to provide family planning services but the locals would not trust them. But now as FCHVs have been recruited from the local level the people trust us” (H4–FGD–SRH). As members of the community, they could convince others of the benefits of getting vaccinated, as an example (H6–FGD–BAR). d. See “Nepal at a Glance” at http://www.pahs.edu.np/about/about-nepal. Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 93 Embedding Conflict Resolution together; and using traditional customs, Mechanisms language, and culture to resolve issues. Approaches that put a major emphasis on conflict resolution as an in-built function of Actively Reaching Out to Insurgents service delivery also seem to offer promise. The research suggests that proactive and One aspect of BESP was to recruit and train savvy forms of communication and engage- an army of local facilitators who were hired ment with armed elites and allied insurgents by implementing partners—local community can help reduce their interference with SSD. members who would work directly with the Chapter 5 outlines how engaging in dialogue community and build trust during project im- and negotiations can be a form of bargaining. plementation. The implementing partners and An interesting example of what did not work their facilitators were selected partly based on well in this regard is the Polio Eradication their experience with community mobilization Project in Pakistan. Interviewees point to a and engagement. The local facilitators were number of failings. One senior donor official widely considered to be the drivers of progress who is heavily involved in the health sector due to their ability to resolve local conflicts and in Pakistan notes that no serious attempts were made to target the Pakistan Taliban in therefore forge bargains. Aspects contributing a communication campaign to counter their to their effectiveness include the following: resistance to polio vaccinations. The gov- ernment policy of not communicating with • Most came from the community—usu- armed elites and allied insurgents and the lack ally from the same district, which allowed of local communication channels to negotiate them to travel fairly easily, even during with them reduced the prospects for achiev- times of conflict, and help mediate and ing their cooperation. Attempts to engage the resolve disputes between insurgents, elites, military to potentially reduce their resistance and constituents. were also limited (R04–KII–PSH). • Because of their local connections, they In addition, the unsophisticated forms of were also largely trusted by the local pop- communication, which were not sensitive to ulation, enabling them to foster locally the ideology or incentives of violent actors, based collective action and resolve con- backfired. As one informant explains: “Door flicts—violent or not. Training enhanced to door campaigns, high publicity of polio their ability to negotiate bargains with campaigns, advance notice of health work- mechanisms such as community meet- ers’ visits and field campaigns, predictabil- ings to discuss disputes and enable prob- ity of movement of teams, agency surgeons lem solving; bringing disparate groups and stakeholders make them susceptible to 94 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds attacks” (R11–KII–PSH). Another makes the In Nepal, local user committees were estab- point more forcefully: “A life threatening sit- lished for road service delivery. They included uation is invoked in the door to door cam- the village road coordination committees and paigns. When I send my workers on a door to local road user committees. As illustrated door campaign in Bara, it’s like sending them in chapters 3 and 5, these local committees towards their death” (R11–KII–PSH ). became the focal points for political party bargaining. User committee members were There are no easy solutions when an insur- chosen on the basis of their political affilia- gency is aggressively targeting a service, but tion, but the three main positions of chair- field experience suggests that a more sophis- person, vice-chairperson, and secretary were ticated and violence-sensitive approach could usually divvied among the country’s three have been adopted.38 major parties—the Nepali Congress, the Uni- fied Marxist Leninist Party, and the United Communist Party of Nepal (Maoist). The po- 6.3. Increasing Local sitions depended on the relative strength and Bargaining Space the influence of the parties in the area. This strategy contributed to local bargains in three Some of the interventions in this study relied ways: on decentralized and participatory modes of delivery. The overall evidence across the three 1. It provided local focal points and local countries demonstrates that these approaches space for bargaining with the Maoists and created some space for local deliberation and even with the organized groups in the problem solving, and in turn, local-level bar- Tarai during the postwar period. gains and local ownership facilitated service delivery. Yet the on-the-ground unfolding 2. It enabled political parties to “cut up the of these approaches rarely resembled good pie” and follow their imperatives of local governance notions of community-based de- patronage, while still promoting a degree livery. Elites routinely jostled for power; the of progress in road construction. local structures were politicized; a share of re- sources was skimmed; and the bargains could 3.. While local jostling for political repre- work against, or for, SSD (World Bank and sentation was dominated by political par- Akram 2016; World Bank and AREU 2016; ties, it still increased the representation of World Bank and SSB 2016). Examples of this previously excluded groups and increased have been discussed in previous chapters and the local stake in the way the roads were a few are briefly outlined below. managed. Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 95 Similarly, under BESP in Pakistan, the lo- completely functional and very successful.” calized and participatory nature of delivery (KII–01, 3–4) helped enable some services to be sustained even during violent conflict. This model in- Likewise, mechanisms to decentralize deliv- creased the space for local elites, local con- ery and increase local autonomy were found stituents, and at times local armed elites and to enable local service providers to arrive at allied insurgents to negotiate and modify flexible local arrangements with the Maoists services in ways that were locally acceptable. during the conflict. And in Afghanistan, local Moreover, this model fostered local owner- shuras or consultative councils often served ship by encouraging local participation and as vehicles for mediating local conflicts and by incentivizing local communities and elites bringing together conflicting stakeholders to to have a stake in implementation: BESP cri- settle on bargains. teria required that a community request the program for a local school and demonstrate a commitment toward managing the school 6.4. Supporting Progress prior to any physical construction occurring Outside the Sector (KII–11).39 One informant notes: Certain cases in this study show how invest- “Despite the poor law and order situation ments and reforms outside of a sector contrib- in Balochistan, BESP schools would remain uted to progress within it. The investments functional because of the beauty of the model and reforms impacted the demand and access on which this project was based. Under this to services, forms of political organization, model, the community was involved and they and even forms of violence (see box 6.4). Ac- were given the ownership of the schools. An counting for the externalities that affect SSD example of how it worked is in the case of in violence-affected contexts could have im- this school located in Kahaan, District Kohlu. portant implications for how projects are de- Due to armed conflict, it is very hard for signed, implemented, and evaluated, a point outsiders to enter the area, but our school is returned to in chapter 8. 96 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Box 6.4. The impact of socioeconomic factors outside a sector Rising wealth contributed to improved social service delivery (SSD) in Balkh, Afghanistan. Political stability, developmental leadership, and strong linkages with Kabul contributed to significant economic investment and growth in Balkh, which in turn impacted SSD. The net effect was faster development in and around Mazar- i-Sharif, one of the largest commercial centers of Afghanistan. As a result, its poverty rate more than halved from 2007 to 2012 (from 60 to 22 percent). In Wardak, poverty rates dropped by only one-quarter (54 to 40 percent); and in Badghis, the poverty rate barely improved (40 to 39 percent).a Much of the improvements in SSD outcomes in Balkh Province is due the progress made in and around Mazar-i-Sharif, the provincial capital. Neither Badghis nor Wardak had a comparable urban population nor could they match Balkh in terms of its overall wealth. Further, several wealthy businessmen connected to Balkh’s strongman, General Atta Mohammed Noor, invested in the build- ing of schools and clinics in the province. Increased income levels and expanded road networks contributed to health delivery in Nepal. Data sug- gest that factors outside of the health sector contributed to progress in health outcomes over the past two decades, including rising average per capita income levels (there was a 20 percent increase in the 1980s, a 21 percent increase in the 1990s, and a 16 percent increase in the 2000s); a decline in one-dollar-a-day poverty from 68 percent of the population in 1996 to just 25 percent in 2011; increases in household incomes owing to a rise in remittances to Nepal; and improvements in education and advances in empowerment for women. This progress contributed to improvements in health knowledge, access, and outcomes. Furthermore, 33 percent expansion of the road network between 1999 and 2008 improved access of people in remote villages to medical help. Differences in land rights and distribution as well as income equality contributed to varying SSD dy- namics and outcomes in the Tarai versus the hill region of Nepal. There is evidence that the Tarai performed worse than the hill region in terms of SSD, which can be traced back to differences in land and income distribution. The hill and mountain regions, which generally have a lower per capita income than the Tarai, distribute land and income in a more egalitarian way. While a small degree of landlessness still exists in the hill region, 33 percent of eastern and central Tarai households are landless compared with 13 percent in western Tarai. Tarai dalits form a major part of these landless groups, who are the subject of quasi-feudal relations with landlords, especially in the eastern Tarai. These differences have had varied effects on local politics and violence, which in turn impact SSD (see chapters 3 and 4). a. See Islamic Republic of Afghanistan, Ministry of Economy, and World Bank (2015). Part II. Delivering Services in Violent Contexts: An Ethnography of the Findings Presented in Part I | 97 Part III. Implications for Policy and Operations Schoolgirls walk past a damaged mini-bus after it was hit by a bomb blast in the Bagrami district of Kabul, Afghanistan, April 11, 2016. Mohammad Ismail/Reuters Part III. Implications for Policy and Operations | 99 7. Policy Takeaways: A Radical Rethinking This chapter attempts to extract the five over- potentially significant. Scant research has arching and overlapping policy implications been conducted to date that systematically of the above findings. As box 7.1 outlines, the takes into account and addresses the issues implications are preliminary, but nevertheless identified in this study.40 Box 7.1. Clarifying expectations: What this study tells policy makers and practitioners Further research is needed. The policy implications are nuanced and powerful, but also partial for three reasons: 1. The research findings need to be further tested and replicated in other contexts; 2. Given the high level of contextual variation in violence-affected settings, it is not possible to pro- vide an off-the-shelf road map for action; and 3. The extent to which development actors adopt (or interpret) the study implications will depend on a given organization’s objectives, available instruments, and appetite for risk, among other factors. The focus is on the “how to” of delivering services.. The implications focus on the delivery of services and progress toward achieving service delivery objectives over the short-to-medium term in violence-affected contexts. Issues such as the relationship between social service delivery (SSD) and drivers of violence or whether external actors should be supporting SSD in such contexts are import- ant, but not the subject of this research. The report touches on sensitive issues.. The findings highlight the prevalence of corruption, rent-sharing, and collaboration with armed elites in the “real world” of SSD—and the following policy implications do not endorse, but do not also side-step, these thorny issues. Recognizing the role of such unpalatable elements in the cases is a reflection of what the field evidence revealed, but this does not mean that the authors of this report condone these practices. The study does, however, recognize that these issues require systematic and realistic treatment in policy making and program design. Part III. Implications for Policy and Operations | 101 7.1. Adapt to a New Set of modes of delivery, play important roles in Contextual Factors making or breaking service delivery in vio- lence-affected contexts. Development actors Most development practitioners recognize need to take these aspects much more seri- that “context matters,” but this rather broad ously. Chapter 8 explores how these factors adage provides limited practical or granular shape opportunities for and constraints to so- guidance. Moreover, progress has been lim- cial service delivery (SSD) programs. ited in terms of understanding how context matters to driving and constraining service Adaptation to Subnational Contextual delivery progress in violence-affected settings. Variation Is an Absolute Must Subnational contextual variation is particu- Programs Must Better Adapt larly decisive. It helps explain major differ- to Context ences in SSD implementation and outcomes Contextual factors that remain underexplored, in the same violence-affected country. Such including forms of political settlement, forms adaptation is not straightforward, however of violence, processes of elite bargaining, and (see chapter 8). Female health care workers in rural Nepal. Aisha Faquir/World Bank 102 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds 7.2. Recognize that Some Forms whether nonstate armed actors have an inter- of Violence Are Less Bad than est in more effective service delivery. If they Others for Service Delivery do, policy makers should try to understand how this interest can be tapped; if they do Mainstream approaches tend to view violence not, policy makers may need to think about as simply bad for service delivery. But the how such actors might develop such an in- research shows that different forms of and terest. Chapter 8 offers some preliminary motivations for violence can be more or less guidance. amenable to SSD. 7.3. Address the Operational National Notions of Violence Ways through Which Elites Are Not Always Helpful Matter When classifying contexts as fragile or vio- lence-affected, policy makers too often focus Clearly, politics and elites influence devel- on the national level. As our and other re- opment progress. Yet, in many ways, this search demonstrates, such a focus risks gloss- assertion has frustrated service delivery prac- ing over important subnational variations in titioners. It tends to yield a long list of broad violence. Policy makers must move beyond recommendations, rarely specific to the me- national definitions of fragility to focus more chanics of front-line service delivery. More- on subnational specifics. over, elites are too often framed in a negative light, as actors who capture and subvert SSD and whose role should thus be minimized or Unpack “Violence” for the True mitigated. Scope for Progress to Be Revealed The various dimensions of violence—orga- The Technical Is Political nization, ideology, and mobility—need to be unpacked to see how they offer diverse The report reinforces the view that technical opportunities for and constraints to service aspects tell only part of the service delivery delivery. Of course, violence harms SSD, story. Political economy factors repeatedly but opportunities for progress vary within undercut attempts to foster best practices and and between violence-affected contexts to a influence outcomes.41 Even technical aspects much greater extent than previously thought. such as the selection of the mode of delivery Put differently, we need to better understand have profound political implications. Part III. Implications for Policy and Operations | 103 The Mechanics of Elite Bargaining 7.4. Rethink the Function of Must Be Put Front and Center Service Delivery: Rents Matter The lens of elite bargaining provides a more Mainstream approaches tend to view service granular—and thus operationally relevant— delivery as being largely politically neutral. understanding of how politics matter to SSD They also make significant efforts to en- in violence-affected contexts, including at the sure this “neutrality” is maintained and safe- front line of delivery. Specifically, three new guarded throughout project implementation. policy insights emerge: But this research raises questions about the validity of this line of thinking. 1. Elites, armed and unarmed, are often (if not always) intimately involved in shaping SSD in such fluid, informal, and contested Rents Can Be Generated by contexts. Service Delivery On the difficult issue of rents and SSD, the 2. Elite bargains are critical to explaining findings give rise to three interrelated policy SSD progress—or the lack of it—and insights: such bargains can take different concrete forms, which have different implications 1. SSD is not politically neutral. SSD for SSD programming. can generate important rents in vio- lence-affected contexts, and such rents 3. The extent to which elites facilitate or can strengthen certain elites, armed or un- undermine a service in a given context armed, which can in turn impact the bal- depends on the political settlement, es- ance of power and the political order. pecially at the sectoral and subnational levels; the incentives and ideology of the 2. SSD can sustain, spark, or mitigate vi- elites; and the extent to which the mode olence. Rents created or constrained by of delivery creates some space for bargains an SSD program can impact the dynam- that support service delivery. ics of violence because in such settings violence is often controlled by (fragile) rent-sharing agreements. 104 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds 3. Rent-sharing may sometimes be a nec- 7.5. Be Realistic about essary evil. Rent-seeking and rent appro- Good Governance priation of SSD resources by armed elites can, in some cases, be the “cost of doing Overall, the findings—in line with the 2017 business” in terms of ensuring peaceful World Development Report—raise import- project implementation. In some cases, ant questions about the good governance rent-sharing agreements may satisfy key paradigm, which has dominated mainstream elite gatekeepers and help sustain pro-SSD governance-for-development discourse over coalitions. In other cases, some rent-shar- the past two decades. The paradigm pro- ing agreements may totally undermine ser- motes formal rules, liberal democratic norms vice provision. The extent to which partial of transparency and accountability, techno- rent appropriation by armed elites is a cost cratic fixes, and institutional models derived worth absorbing is as much a political as a from best practices often borrowed from the technical judgement—more an art than a Organisation for Economic Co-operation science. and Development (OECD) member and up- per-middle-income countries. Modes of Delivery Affect Rent Access and Elite Bargaining Governance Matters, but Not Sectors and modes of delivery matter by of- Necessarily as Expected fering different opportunities regarding if and The findings do not suggest that governance how rents are accessed and bargained over by does not matter or that “bad” governance elites and their constituents.42 In fact, the rents should be promoted. It is just that many of generated by SSD can be most clearly iden- the governance characteristics in our cases tified through a sector-by-sector analysis of did not always resemble the “good gover- political settlements and elite bargains These nance” paradigm (see table 7.1). Moreover, aspects should be considered when working the findings suggest that there are a wider in specific sectors, and when selecting modes range of options for progress in the midst of of delivery (see chapter 8). far-from-perfect forms of governance than the good governance paradigm assumes. Al- ternative paradigms, such as “good enough governance” or “neo-patrimonial develop- mentalism” can help flesh this out.43 However, such paradigms have yet to be systematically Part III. Implications for Policy and Operations | 105 adopted and operationalized in mainstream Narrow the Disconnect between practice. Furthermore, this aspect of our ar- Donor Discourse and Ground Reality gument is not new: there is now a burgeoning The gap between donor discourse and the body of evidence on these points.44 Yet, the on-the-ground reality needs to narrow. The good governance approach remains widely analysis of the cases in this report reveal a dominant.45 disconnect between the way development ac- tors express what is driving change and what Some Cases of Progress May Not is actually unfolding on the ground. Indeed, Resemble Good Governance many of the factors that the research found decisive to project outcomes were hardly Related to the last point, there is a need to ever systematically considered in the projects rethink how progress happens. Informal re- examined.46 In other words, our dominant lationships, rent-sharing, far-from-perfect mind-sets may be distracting us from what transparency or accountability, and deep really matters. politicization of service delivery—through political parties or “unsavory” powerful ac- tors—can underpin change and progress in Toward a New Paradigm for the SSD. In fact, the categories of “communi- Governance of Service Delivery ties,” “civil society,” and “citizens” were less “History … shows that very significant gains apparent in explaining what happens com- in economic transformation and human pared with “elites,” “warlords,” and “political well-being can be achieved within highly society.” dysfunctional systems. Reform initiatives should surely aim to repeat those gains by This suggests that progress in SSD is not just whatever means are to hand” (Booth 2015b). about, or mainly about, preventing elite cap- ture, but about how and why local elites can Taken as a whole, these findings prompt a actively become part of pro-SSD coalitions, more nuanced view of the forms of gover- even though some rent appropriation and nance and politics that shape SSD progress. corruption may occur. While this may seem Table 7.1 offers a stylized summary of where difficult for some to acknowledge (especially our research diverges from conventional good donors with public constituencies and media governance discourse. suspicious of the benefits of overseas devel- opment assistance in general), there is a grow- ing recognition that this type of scenario is not uncommon, especially in less developed countries (World Bank 2017). 106 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Table 7.1. Beyond good governance: Service delivery in violence-affected contexts Good governance paradigm Our research Rent-sharing Corruption, rent-sharing, and patronage are bad Scope for progress under “patrimonial” governance is larger than often for social service delivery (SSD). These aberra- perceived. SSD progress can take place in a context of, and even be tions need to be eradicated. facilitated by, rent-sharing and patronage-based relationships. Rather than aberrations, these arrangements are the norm and have wide so- cial and political legitimacy. Rent-sharing and patronage are subopti- mal, and measures can be taken to eradicate them if doing so does not trigger violence or major disequilibrium. But it is probably not feasible to ring-fence delivery from these influences all the time. Technicality SSD can be politically neutral. SSD programs SSD in violent contexts can often be a vehicle for rent-seeking. A and neutrality can be approached with a purely technical feedback loop in which SSD programs influence political dynamics is mind-set. likely. Therefore, in addition to thinking technically, there is a need to think politically. Role of elites Elites (where recognized) are largely seen as Elites can play a positive or negative role. Elite capture can subvert a capturing or subverting good programs and un- program, but it can also enable progress. Elite incentives and interests dermining the poor. Elite incentives, interests, must be fully understood, and bargaining must be taken more seriously and bargaining dynamics are rarely unpacked. in SSD design and implementation. Bargaining with elites is rarely encouraged. Violence Violence is harmful to service delivery. Different forms of violence are more or less harmful. This must be recognized and SSD tailored accordingly. Formal and Formal institutions are crucial. They need to be Informal institutions can often prove decisive. Formal institutions informal put in place and be formalized to the greatest are unlikely to function in the absence of the widespread social and institutions extent possible. The form of such institutions, political legitimacy that is often tied to “informal” practices. These usually based on the Organisation for Economic informal practices can be seen as part of the solution, not as just a Co-operation and Development (OECD) models, problem; they should be better understood and built on. Approaches are often the focal point. Informal institutions that actively combine formal and informal institutions, such as prac- are problems to be bypassed or ignored. tical hybrids or institutional bricolage, may be critical.47 Institutional function is what matters. Citizens, Communities, civil society, and citizens tend to These actors can and do matter, but they require more nuanced communities, be framed as homogenous forces for good. treatment. They are not homogenous; they comprise contesting and and civil undemocratic elites and power relations between different groups with society disparate incentives toward or tolerance for broad-based SSD. Context Best practices can be more easily transferred “Good enough” or “best fit” approaches, with substantial contextual across contexts. Win-win scenarios are adaption, are more likely to work. Trade-offs are more common than common. win-wins. Important Prominence is given to the “usual suspects,” Systematic attention is also given to “unsavory” actors, such as insur- actors such as the state, service users/citizens, and gents, the military, landlords, and political parties. service providers. How change Citizens, armed with information, can hold ser- Citizen pressures and top-down state pressures matter, but change happens vice providers and the state accountable; top- is also driven by dynamics between pro- and anti-SSD networks and down state incentives can also drive change. coalitions that include elites from the state, economy, and society. Part III. Implications for Policy and Operations | 107 8. Practical Directions and Recommendations The policy implications presented in chapter Six mutually reinforcing recommendations 7 contain nuanced, potentially difficult, op- emerge from the research (see figure 8.1): erational implications. This chapter identifies some operational starting points that address 1. Improve diagnostics for measuring the ac- the implications of this research and relates key tual scope of progress; points to the global evidence base to make the 2. Think and work politically; implications more widely generalizable. There are no simple solutions, but the following 3. Tailor approach to specific forms of violence; discussion suggests that we can take steps to 4. Take bargaining seriously; develop a more realistic and grounded treat- ment of political and violence dynamics in 5. Revise internal donor rules and incentives; social service delivery (SSD) programming, in and turn enabling us to take more calculated risks. 6. Explore new frontiers of delivery and Moreover, parts of the discussion may raise as research. many questions as they answer (see box 7.1), but this does not make the issues any less wor- thy of significant thought and attention. Zahir Shahe high school library. Imal Hashemi/Taimani Films/World Bank 108 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Figure 8.1. A new framework for strengthening service delivery in violence-affected contexts 1. Improve diagnostics for measuring the actual scope of progress Understand relevant political Map the relevant dimensions Draw lessons from local settlements, who the relevant of violence in the program service delivery and elites are, and how they matter areas bargaining 2. Think and work politically 3. Tailor approach to specific forms of violence 4. Take bargaining seriously 5. Reform donor rules and incentives 6. Explore new frontiers of delivery and research 8.1. Improve Diagnostics and relevant local stakeholders to discuss for Measuring the Actual focal sector(s), the scope of the diagnostic, Scope of Progress and the sectoral problem(s) that need further analysis.48 Next, answers could be sought to In order to improve context-specific diagnos- the questions in each module. tics to inform program design, we propose integrating a new diagnostic toolkit. This Module 1: Understand relevant political has three interrelated modules (box 8.1 out- settlements, who the relevant elites are, and lines the scope of the proposed toolkit and how they matter appendix B has guiding questions for each Identifying the elites and power brokers who module).The first step is to identify the sec- are relevant to the sector and locality as well toral problems to diagnose. This will involve as their ideologies and incentives is crucial. bringing together management, project staff, If feasible, identifying power brokers that are Part III. Implications for Policy and Operations | 109 Box 8.1. What the proposed diagnostic toolkit can and cannot do The proposed diagnostic seeks to complement, not replace, existing diagnostics. The suggested questions can be adapted and included in existing analytical tools that support the strategic planning of development actors. This diagnostic offers tools in the form of modules and questions for actors specifically concerned with supporting social service delivery in contexts affected by active violence. The diagnostic does not need to be extensive but does require careful applica- tion. Its scope can be tailored to available resources and urgency. Information can be gathered through one-hour conversations with experts; one-day workshops and consul- tations; rapid desk reviews; or the commissioning of a report. Care should be taken that the tools be implemented by people who are in tune with the political dynamics of the relevant country and who have a solid understanding of the specific context. The diagnostic is not a one-time exercise. Given the fluid nature of fragile, conflict, and violence-affected contexts, the diagnostic needs to be regularly updated. Perhaps more important than the rigor of the diagnostic is the process of ensuring organizational uptake of its findings. Adequate resources and incentives must be dedicated to discussing and following up on the findings.a There are unlikely to be magic bullet solutions. The diagnostic is unlikely to offer unambiguous solutions; it should, however, provide the basis for better planning and more informed risk taking. a. For other useful resources on performing quality diagnostics in the context of donor programming, see Hudson, Marquette, and Waldock (2016); Yanguas (2015); and DFID (2016). broadly for or against SSD in a given sector Module 2: Map the relevant dimensions of can help with coalition-building. At a min- violence in the program areas imum, this module should provide project Understanding how specific dimensions of teams with a firmer grasp of the power bro- violence in a given context are likely to pres- kers at the central, sectoral, and local levels ent their own risks and opportunities for the of delivery. focal sector is critical. The focus would be on 110 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds the key dimensions unearthed in this research: key operational messages are proposed: (1) organization, ideology, and localization. This SSD design should be tailored to the specific would enable project teams to come up with type of political settlement; (2) the delivery typologies of violence in their context and to of services should be rethought as a func- consider their different implications for their tion of coalition-building; and (3) politically program (see recommendation 3). Because savvy modes of delivery exist and should be forms of violence can differ at the subnational actively pursued. level, project teams need to be highly sensi- tive to subnational variation. Tailor design to political settlements Political settlements—particularly at the sub- Module 3: Draw lessons from local service national and sectoral levels—exist along a delivery and bargaining spectrum and can differ within and between Project teams can also collect lessons on what violence-affected contexts. As the report has modes of service delivery work and what do shown, such differences can have important not, as well as if and how forms of bargain- practical implications for SSD. Based on the ing take place in their context. These lessons typology developed for this report, table 8.1 could derive from the experiences of multiple attempts to tease out some of the initial, prac- stakeholders, and especially those actors op- tical implications. Note that this is just a start- erating at the grassroots level and at the front ing point, and is by no means exhaustive.49 line of delivery. Donor knowledge on such local dynamics can sometimes be limited (see Build service delivery coalitions in also recommendation 5). fluid contexts SSD is able to operate more effectively when it is supported by coalitions of supportive ac- 8.2. Think and Work Politically tors. A key implication is as follows: improv- ing SSD is a technical challenge but it is also “We can’t go around it [politics], we a political one insofar as effective implemen- can’t go over it […], we’ll have to go tation requires the building of coalitions— through it” (Evans, paraphrased in however limited, fluid, or localized—that can Menocal 2014: 11). to some degree enable services to be delivered and can protect some aspects of delivery from Our study amply demonstrates the critical factions that wish to appropriate it. Context role of political economy factors in making matters, but three “rules” on coalition-build- or breaking service delivery. In this area, three ing can be derived from the research: Part III. Implications for Policy and Operations | 111 Table 8.1. Implications of political settlements for social service delivery in violence-affected contexts Subnational/sectoral settlement Implications for social service delivery (SSD) 1: Relatively stable Reasonably good prospects for SSD, as incentives Pro-SSD incentives exist for ruling coalition. align. Stability in settlement potentially allows for • Organized and centralized around a regional planning over longer time horizon and the building strongman/warlord with pro-SSD interests/ of a pro-SSD coalition. However, SSD needs to align incentives. with ruling factions’ preferred mode of delivery • Organized around pro-SSD programmatic (given the ruling factions’ relative power in setting political party. the rules of the game). 2: Moderately stable Medium prospects for SSD. Moderate stability allows SSD incentives mixed; incentives are contested for better planning and better coalition-building between state, disciplined armed insurgencies, and around SSD programs, but contestation can under- powerful traditional elites. mine delivery depending on the level of contestation at the front line of service delivery. Mechanisms of coordination between powerful elites may be needed. 3: Moderately unstable Some prospects for SSD, given pro-SSD ideology. Pro-SSD incentives exist but are contested among However, active contestation between state and state and pro-SSD disciplined armed insurgencies. insurgent groups potentially undermines service delivery. Efforts, compromises, and bargains likely to be needed to reach consensus among contesting groups on “ring-fencing” (to a degree) services. 4: Moderately to highly unstable SSD is likely to be challenging. Some state pene- Weak SSD incentives; some state penetration in tration offers a modicum of stability that may help areas controlled by relatively disciplined armed with service delivery, but weak incentives as well as groups. antagonism can undermine service delivery. 5: Highly unstable (type a) SSD is likely to be challenging. Weak SSD incentives Weaker state penetration; less disciplined armed as well as antagonism can undermine SSD. groups and weak SSD incentives/ideology. 6: Highly unstable (type b) SSD is likely to be very challenging. The lack of Situation fluid; marked fragmentation of power and broadly agreed rules of the game as well as the no monopoly of violence; weak SSD incentives and presence of dominant ideology/incentives that seek political organization; multiple centers of power short-term appropriation of SSD make progress controlled by armed criminal organizations. difficult. Delivery needs to be flexible; practitioners need to learn by doing and (most likely) accept that bargaining strategies must be built in. 112 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Rule 1: Expect the unexpected: fluid and 1. Systematic consideration must be given to unstable politics are the norm the way an SSD program provides rents, In violence-affected contexts, the rules of the and whether it maintains or undermines game are fluid and everchanging. Even bar- specific coalitions; and gains that allow a modicum of stability are often made under the constant threat or exis- 2. Overly aggressive attempts to curb rents in tence of violence. They tend to change over a sector could result in sabotage or even an time and are subject to regular renegotiation increase in violence.50 because the dominant coalition often strug- gles to maintain its supremacy in the face of There are no magic bullet solutions, but these internal and external pressures (Kelsall 2016). points require attention. Practitioners should thus expect the unex- pected, and recognize that an adaptive ap- Rule 3: Find “good enough” ways to support— proach—involving regularly changing tactics or at least not undermine—coalitions that and strategies—will most likely be needed. support service delivery Even during the best of times, external actors Rule 2: Be realistic about the role of “spoils” play a limited role in supporting or maintain- and rents ing coalitions. In violence-affected contexts, This report suggests that, in violent contexts, elites find it particularly difficult to act collec- the relationship between SSD programs and tively due to weak coordination mechanisms, rents needs to be considered very seriously major information asymmetries, extremely and very carefully. As argued earlier, SSD low levels of trust, limited consensus, power programs can provide rents, and rent-sharing jostling, a “winner-takes-all” mentality, and in violent contexts often plays a role in sta- unpredictability (North et al. 2011). bilizing coalitions, controlling violence, and allowing functional service provision. If an Drawing on coalition-building lessons, do- SSD program disturbs these arrangements, nors could support mechanisms to: (1) help factions could increase their demands and pro-SSD elites coordinate themselves and act increase acts of violence. A case in point is collectively; (2) reduce information asymme- the Nepal health sector during the Maoist tries between factions; (3) build trust across conflict: evidence from interviews suggests groups; and (4) increase predictability to en- that attempts to circumscribe rent-sharing able longer time horizons.51 Yanguas (2016) at the local level would have stalled service takes this further by suggesting broad strate- delivery and would have increased the risk gies for coalition-building that target pro- and of violence to service facilities and staff. Spe- antireform factions, which could eventually cifically, two operational implications emerge be applied to violence-affected settings, how- as critical: ever challenging this may be (see box 8.2). Part III. Implications for Policy and Operations | 113 Box 8.2. Potential strategies for building pro-SSD coalitions It is often unclear in fragile, conflict, and violence-affected contexts who the pro- and anti-SSD actors are. This is partly because actors frequently change sides. As outlined below, Yanguas (2016) offers some strategies for building coalitions that could be useful in violence-affected contexts. Aid Recipient Strategy Incumbent: Anti-SSD Challenger: Pro-SSD Diffusion Adaptation Contestation Certification Legitimization Delegitimization Brokerage Consolidation Disruption Diffusion. The flow of ideas across borders and policy domains in which the aid indus- try plays a role: • Adaptation of ideas occurs when diffusion is aimed at supporting incumbents to achieve their goals through incremental change. • Contestation of ideas occurs when the diffusion of ideas by aid agencies is aimed at strengthening challenges to existing ideas held by the incumbents. Certification. International validation signaling recognition of the existing power distribution: • Legitimization. Foreign aid could serve to legitimize the incumbents, bolster elite priorities, and signal political backing from the international community. • Delegitimization. Aid and dialogue is directed away from powerful elites or incumbents. Brokerage. Using donor resources and legitimacy to facilitate bargaining and trust-build- ing to establish new links between actors and policy fields and to coordinate the forma- tion of new coalitions. • Consolidation represents brokerage influences that ultimately strengthen the incumbent coalition. • Disruption entails brokering new relationships outside the incumbent coalition. Source: Adapted from Yanguas (2016). 114 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Design modes of delivery that are on local context, there are myriad potential politically savvy strategies for making delivery more politi- Building on the previous point, different cally astute, however difficult. Key practical modes of delivery may be able to reduce lessons from this research are summarized in the power of anti-SSD forces. Depending box 8.3. Box 8.3. Making delivery more politically astute: Key practical lessons 1. Semiautonomous delivery mechanisms can reduce the power of sectoral clientelist networks, but sustainability and creation of parallel delivery systems is a challenge Examples The establishment of semiautonomous management and oversight institutions, such as the Balochistan Education Foundation (BEF) in Balochistan. Using alternative human resource approaches or bypassing existing systems to staff service delivery, such as the case of health service volunteers in Nepal. Why is this politically savvy? The BEF experience demonstrates that a semiautonomous agency can reduce the impact of clientelist networks on a sector by setting up new procedures that are to some degree ring-fenced from interference. Moreover, the semiau- tonomous institution may, if designed correctly, have credibility with and the backing of reform-minded elites. In other words, “new” institutions must be built on new elite pacts. Using volunteers or alternative human resources systems can help sustain or expand service delivery by reducing the influence of patronage-based appointments and the power of incumbents, such as unions or political parties, in civil service transfers and postings. What are the trade-offs? The creation of a semiautonomous institution generates risks: (1) it could become a donor-driven entity with lim- ited social and political legitimacy; (2) it could perpetuate dual public sectors and create parallel structures; and (3) it could be subverted by the old elites or used by the new elite grouping for unintended purposes. Checks and balances are crucial. Indeed, various interviewees argued that the Balochistan Education Foundation became less effective over time and was subject to capture. (continued) Part III. Implications for Policy and Operations | 115 Box 8.3. Continued The trade-offs for using volunteers or alternative human resources systems include: (1) volunteers might be able to fulfill simple short-term functions but not deliver the more complex services (and over the medium term, human re- sources systems would have to be reformed); (2) the role of unions and political parties would need to be addressed on a case-by-case basis: these actors might be more or less supportive of SSD depending on the context. The global evidence suggests that their relevance to the development process should not be taken lightly. 2. Small-scale interventions can circumvent sectoral elite networks and spark a slow-burning political demonstration effect, but efforts to scale up can be elusive Examples Implementation of smaller-scale and localized interventions, such as the Balochistan Education Support Project (BESP) in Pakistan, road user committees in Nepal, and local health shuras in Afghanistan. Why is this politically savvy? BESP’s small-scale and localized model reduced appropriation by higher-level elites. Central- and provincial-level rent opportunities were reduced, and local elites—who had greater incentives for local education delivery—had more control over resource allocation and implementation. What are the trade-offs? Small-scale or localized successes can spur incremental and indirect political demonstration effects; that is, they can increase political pressure by constituents on higher-level elites and open up reform spaces by demonstrating what is possible, potentially bolstering reform-minded actors and shifting the mind-set and incentives of elites. However, local successes can be difficult to scale up, and new forms of local elite subversion can emerge. Rather than aiming for a province-wide rollout, programs should seek to deepen local successes, harness reform-minded impulses, and broadly disseminate communication about achievements to strengthen the demonstration effect. Further, efforts should be scaled up incrementally and have built-in strategies for compensating losers and antagonists to reform. 3. Bundling delivery into a suite of services can increase the political feasibility of success, but implementation logistics are challenging Examples Bundling services into a suite of selected services, such as providing a range of health services in Nepal or combin- ing polio with other services in Pakistan. Why is this politically savvy? While multisectoral delivery is often framed as a technical and capacity challenge, this research suggests that this approach can help address some of the political barriers to delivery. For example, as the case of Pakistan illustrates, (continued) 116 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Box 8.3. Continued combining pro-poor services—such as basic vaccinations—with other services that powerful opponents and/or local constituents really want—such as other health services or electricity—can reduce resistance and impact the incentives of power brokers to support or block the smooth implementation of SSD. Combining basic SSD with measures to address income or land inequalities and/or infrastructure investments, such as road connectivity, can increase user demand, and can thus have spillover effects on the sector, as the Nepal case illustrates. What are the trade-offs? Before putting together a suite of services, a strong understanding is needed of what local elites want to ensure that the plan is technically feasible and politically palatable. Bundling requires strong multisectoral coordination and logistics and includes associated costs. 4. A decentralized, multistakeholder approach can facilitate more inclusive delivery bargains, but serious mitigation mechanisms are necessary Examples Decentralized, multistakeholder approaches to delivery that set up local structures for decision making around service delivery, such as the road user committees in Nepal and the local school and community development com- mittees in Afghanistan. Why is this politically savvy? This approach can create space for local bargaining; absorb the transaction costs of enabling local collective action; and create opportunities for previously excluded groups to increase their bargaining power with local elites, thus securing a better resource allocation deal for themselves at the local level. In short, it can shift the local balance of power. What are the trade-offs? While a decentralized, multistakeholder approach can create bargaining space, it can also enable local elite capture because local structures are invariably dominated by powerful local elites—political, social, and religious. The focus should be on how to make these structures improve SSD rather than expecting to instantly eliminate their role. In addition, it is difficult for the excluded to bargain effectively with powerful local elites, and their inclusion could be met with backlash and even increased violence. On-going political risk management in project management teams will be needed to manage these trade-offs. 8.3. Tailor Approach to Specific types of violence and their implications for Forms of Violence SSD. As a general rule, the more fragmented, mobile, and ideologically antagonistic the vi- In addition to tailoring SSD to political real- olent actors, the more difficult it will be to ities, it should also take into account the spe- deliver services and the more mitigation strat- cific forms of violence in the context. Table egies will be needed. 8.2 offers a stylized depiction of the different Part III. Implications for Policy and Operations | 117 Table 8.2. Forms of violence and implications for service delivery Form of violence Implications for delivery of social services Well-organized, disciplined armed groups with Bargaining feasible (actors clear, incentives for incentives favoring social service delivery (SSD) service delivery aligned). Good prospects of direct Example: dominant/organized strongman allied state provision given alliance with ruling coalition. with ruling coalition, such as General Atta in Balkh, Resource transfer—from the center to the subna- Afghanistan tional level—may be substantial to maintain alliance. Well-organized, disciplined insurgent armed Bargaining feasible (actors clear, incentive align- groups with incentives favoring SSD ment). But poor relationship with the state means Examples: organized insurgency resisting or trying to strategies are needed to mitigate risks such as capture the state but wanting some services, such as nonstate provision and dialogue with insurgents to Maoists (Nepal), Balochistan (Pakistan) ensure that delivery of social services is not per- ceived as a proxy for state penetration. Somewhat organized/disciplined insurgent armed High risks of violence targeting delivery. Insurgent groups largely opposed to SSD leaders relatively clear in their antiservice ideology Example: relatively organized insurgents who want driven by their poor relationship with the state. Mit- to undermine delivery, for example of polio inoc- igation strategies could include bundling of services ulation in Khyber Pakhtunkhwa and the Federally to increase insurgent incentives not to undermine Administered Tribal Areas of Pakistan delivery. Somewhat organized/disciplined armed groups Complex scenario. Strategies for service delivery competing for regional power and legitimacy; will depend on various factors: local contestation positions favoring service delivery contested may lead to competition to improve services or to Example: large parts of Wardak Province, capture/undermine services. Competition may also Afghanistan be over who is seen to deliver the service. Disorganized/fragmented, mobile/roving A more difficult context (compared to the four pre- insurgent armed groups with some local roots vious ones). Leaders fragmented, power contested, creating incentives for service delivery and violence less predictable, with the result that Examples:fractionalized, uncoordinated armed bargaining is less feasible. Second-best solutions groups with some incentives for service delivery as need to apply, such as using local implementers to in parts of Badghis, Afghanistan increase safety and securitize delivery. Highly disorganized/fragmented, highly mobile/ Most difficult context: leaders fragmented, power roving armed groups with criminal (nonpolitical) contested, and violence less predictable; incentives motives of violent actors are to undermine delivery of social Examples: some parts of Badghis, Afghanistan; to a services. Second-best solutions will be required. lesser extent, Tarai, Nepal, after large-scale violent conflict. 118 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds 8.4. Take Bargaining Seriously accounting of resource flows for health in Nepal, it likely would have disrupted del- Bargaining is, as outlined at length, critical in icate local bargains, increased acts of vio- making or breaking SSD in violence-affected lence, and negatively impacted progress in contexts. The implications of these bargains health delivery (see part II). are ambiguous and fraught with ethical di- lemmas, but some key, preliminary practical Carefully weigh trade-offs tips are outlined below. It is also important to put the issue of trade- offs front and center in this debate. Trade-offs Consider the dynamics of bargaining exist for both traditional and “bargaining-sen- At minimum, the research suggests that pro- sitive” approaches to SSD in violence-affected gram designers and implementers should take settings. The difficult question is: what should the dynamics of bargaining seriously. Indeed, be prioritized? if SSD programs are not tailored to local poli- tics and bargaining, they risk being a “shot in 1. Trade-offs with traditional program- the dark.”52 Two points deserve highlighting: ming. For example, a traditional pro- gram (that is, one that ignores the role of 1. Intended or not, some form of bar- bargaining) might unknowingly create gaining is likely to occur. Indeed, elite rents that sustain locally powerful elites; bargains emerged as fundamental in all the it might circumscribe rents that fuel fac- studied cases, but we found scant evidence tional violence; or the provided services that project implementers are aware of or might be accessed by insurgents and their systematically consider them.53 allies, indirectly sustaining them.54 2. Making decisions about projects 2. Trade-offs with bargain-sensitive pro- without considering the role of bar- gramming. Bargains with insurgents can gains could put SSD at increased risk. enable SSD progress in the short term, but Engaging in bargains and compromises they present challenges that must be con- may not be ideal, but might be the only sidered and managed: way to make SSD progress in some vio- lence-affected contexts. For example, if – Bargains might help short-term, lo- education delivery in Afghanistan had calized delivery progress, but little is entirely circumscribed the space for mod- known about if and how they might ifications to the curriculum, the Taliban contribute to longer-term state- or might have targeted such services. If peace-building objectives. higher-level officials had demanded an Part III. Implications for Policy and Operations | 119 Table 8.3. Operational strategies for bargaining with insurgents Bargaining method Operational options Build SSD program • Create an enabling environment for local service providers to reach SSD bargaining capacity bargains with local armed elites and allied insurgents, for example, by providing front-line staff with some degree of autonomy and incentives to reach out to elites and insurgents and find compromise. • Provide negotiation and training in bargaining to all relevant staff. • Employ staff with expertise in negotiation and conflict resolution. Facilitate dialogue and • Develop a tailored communication strategy to negotiate with armed elites negotiations and allied insurgents and persuade them to tolerate the service. • Absorb transaction costs of bringing together key stakeholders and armed elites to negotiate bargains; use mediation experts to facilitate. Inform and obtain • Inform and build trust with armed elites and allied insurgents and comanage “permission” from delivery with them so delivery is seen as less threatening. insurgents • Use actors trusted by armed elites and allied insurgents to deliver the service. If insurgents have an antagonistic relationship with the state, implement services through nonstate actors. Enlist influential local • Identify local pro-SSD elites who could be allies in delivering services. elites • Engage influential, pro-SSD elites to encourage armed elites and allied insurgents (and local constituents) to allow the service. Make concessions and • Identify acceptable areas for modifying the scope of delivery to make it more compromises tolerable to armed elites and allied insurgents, such as allowing curriculum changes in Balochistan and developing realistic plans for girl’s education in parts of Afghanistan. Engage in rent-sharing • Dedicate some of the budget to compensating losers, opponents, and/or agreements armed elites and allied insurgents with “sweeteners” or incentives, such as bundling or allowing insurgents to use the services. • Consider tolerating some leakage if it is necessary to reduce violence and enable delivery. Note: Selection of activities should depend on organizational objectives, comparative advantages, and the results of the diagnostic. – Using locally trusted actors and volun- – Service delivery might be maintained teers to deliver services might reduce through bargaining, but resource flows the risk of attacks and maintain service could be diverted to areas and groups quantity, but doing so makes it difficult considered more important by pow- to improve service quality. For exam- erful insurgents and elites, and de- ple, volunteers might be unable to de- pending on the group or area being liver complex services. excluded, this can increase violence. 120 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds – Achieving small-scale impacts through 8.5. Reform Donor Rules and local bargaining might be possible, but Incentives scaling up could prove difficult. Many of this study’s implications upset the – Rent-sharing might be costly and un- traditional thinking of the mainstream aid palatable to donor agencies, but these system. The findings of this study build on concerns must be weighed against the at least two important contemporary narra- risk of violence if the insurgents’ rents tives around changing aid delivery: (1) ad- dry up. vocacy for doing development differently and thinking and working politically55 in the aid In short, the question of whether bargain- system, which includes adaptive and iterative ing with armed elites and their allied insur- programming;56 and (2) advocacy for a more gent groups to deliver services is better than context-specific approach in violent contexts slower progress or no progress at all depends that considers the political economy and risk on the particulars of a case. of violence and draws on lessons about what really works on the ground (IEG 2016; Marc What would a bargain-sensitive delivery et al. 2016). program look like? Let us suppose that, after deliberation, a proj- This report builds on these narratives in two ect team decides to consciously integrate main ways: (1) by offering a more granular some form of bargaining into a given SSD framework for addressing politics and vio- program. What might this look like? First, lence in SSD operations based on bottom-up options could be explored for “doing no research (above); and (2) by outlining further harm” to pro-SSD bargains, and for reduc- ideas, and associated challenges, on how to ing the space for anti-SSD bargains. Second, reform traditional ways of delivering aid table 8.3 offers some initial ideas to provoke (below). thought and debate among project teams. In- deed, it is recognized that in any given con- Encourage honest debate on bargains: avoid text “bargain-sensitive SSD programs” may “shots in the dark” not be formally possible as it may undermine One challenge is changing mind-sets. For in- an agency’s relationship with the government stance, interviews undertaken for this research and articles of association that allow it to op- suggest that such issues are currently glossed erate in the country. over or painted in an overly simplistic light in internal debates (see also the discussion on Part III. Implications for Policy and Operations | 121 Table 8.4. Ideas for shifting donor lexicon and mind-sets on elites and bargaining Recommendation Example Budget for bargaining Fund type of diagnostics proposed in chapter 8; pilot and integrate bargaining strategies into design and implementation. Change program Talk about power and elites at every opportunity. Balance terms such as commu- vocabulary nity-driven development and citizen engagement with terms like local-elite-driven development and engaging development elites. Involve the “un-usual Talk to insurgents; to military, political, and religious elites; to tribal elders; and suspects” union representatives—or encourage governments and implementers to do so. Engage in frank internal Encourage project teams and senior management to openly discuss politics, communication trade-offs, and “leakage” in program design. “good governance” in chapter 7). Thus one 8.6. Explore New Frontiers of recommendation is to actively foster frank in- Delivery and Research ternal debate (table 8.4) on politics, violence, and bargaining to start changing organiza- Finally, what does this all mean for a for- tional mind-sets and cultures. ward-looking agenda on models of SSD, as well as research in this field? In what follows What needs to be done differently and why are some initial considerations and thoughts, is it difficult? which are meant to encourage further think- Taken together, the implications of our re- ing and debate among academics and policy search and global evidence underline the makers. need to shift, to differing degrees, the way aid is delivered in violent contexts.57 Table Adding to contemporary debates 8.5 offers a stylized summary of the main During the course of this study, an import- differences between traditional approaches ant and challenging question was posed: how and what may be required based on this (and does the approach emerging here differ from other) research. It also highlights some of the more standard approaches to SSD? This is an operational risks and policy dilemmas that expansive question. This report was not in- would result from trying to adapt to reform in tended to provide a new SSD model, even if this direction. Indeed, the powerful political it contains a number of seeds of a new model. economy of the aid system and donor agen- To explore these points, a literature review cies presents legitimate dilemmas, which can- was undertaken (Deshpande 2016), and the not be wished away. main headlines are outlined below. 122 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds Table 8.5. Shifts in how aid actors work: Frontiers and challenges Considerations and risks in Mainstream aid priority What this research suggests reforming the way of working Emphasize success; under- Emphasis should be on what does Few—if any—agencies seek to line win-wins and underplay and does not work as well as why. highlight their mistakes. Taxpayer difficult trade-offs. Difficult trade-offs and political and donor tolerance for failure and and ethical dilemmas should be trade-offs is likely limited. acknowledged. Leakage/rent-sharing unac- A degree of leakage can accom- It is extremely challenging for ceptable. Donor fiduciary pany progress and in some cases is donors to encourage or endorse rules and requirements necessary to reduce violence and leakage. Fiduciary rules need require zero leakage. allow social service delivery (SSD) to be revised to include built-in to continue. exceptions. National and/or “one-size- Significant adaptation to and Significant subnational variation fits-all” programs. Limited incentives for subnational and local can be costly, very complex, put internal incentives for staff to variation is necessary. This includes additional pressure on overstretched significantly adapt to subna- staff rewards and time and budget staff, and undermine efforts to scale tional variation. Pressure to allocation for this. up. Consideration should be given prepare projects quickly. to how much local calibration is practically achievable. Weak incentives to undertake Project teams should be incentiv- Diagnostics rarely present simple and follow-up on analytical ized to undertake and ensure uptake solutions. Expectations should be diagnostics of project design of the proposed diagnostics. adjusted accordingly. and implementation. “One-size-fits-all” project Greater mitigation and bargaining A bigger budget does not necessarily preparation and implementa- strategies and operational support mean more success. Consideration tion support budgets. are needed in subnational areas should be given to how aid ac- where the politics and forms of tors should best allocate limited violence are strongly opposed to resources in violence-affected SSD. Budgets should be increased contexts. relative to the challenges of local politics and violence. Relatively limited ability of Fluidity of the context requires Challenges include building adaptive and incentives for projects to strong adaptive capacity and the capacity in a violent environment regularly adapt to changes on ability to monitor, learn, and experi- with low institutional capacity, a the ground. Linear log-frames ment over time, including setting up weak public sector, and patchy po- with output focus tend to systems for more regular monitoring litical coverage; making large-scale dominate. at the subnational level. programs adaptive to local programs without creating huge transaction costs; and rectifying iterative approaches with modern rules of accountability and budgeting. (continued) Part III. Implications for Policy and Operations | 123 Table 8.5. Continued Development and SSD prob- Projects should enable and absorb Challenges include engaging with lems and programs defined the costs of gathering multiple local stakeholders in extremely vio- by a tight-knit group of stakeholders to identify locally lent contexts without putting donor experts. Support of top-down defined problems and locally driven staff at great risk and ensuring local implementation, especially in solutions. Teams should reach people also benefit from external fragile contexts where people beyond the normal interlocutors— experiences without being dictated are sometimes stuck in com- government and other donors—to by them. pounds and cut off from life include journalists, political parties, outside the compound of the and others. capital city. Sectoral silos. Poor integra- Build knowledge-practice alliances Classic challenges remain regarding tion of conflict and political between various specialists and sector coordination, resourcing, and economy specialists in ensure a multidisciplinary mix on incentivizing of multidisciplinary project teams. project teams. teams. High turnover of staff among Revise human resource practices Attracting and retaining people many donors. Short tours by retaining local staff with strong in violence-affected contexts is are the norm for most in local knowledge and reducing high difficult. The psychological effects of violence-affected contexts. levels of staff turnover. Engage staff such postings can take a major toll International staff are regu- in longer-term postings with greater on a person. larly rotated. emphasis on country knowledge and experience. Sources: Authors; key informant interviews conducted for this study; Akmeemana et al. n.d.; Devarajan and Khemani 2016. Note: This is a stylized summary; aid agencies differ and exist on a spectrum. The idea that “governance matters” in ex- (1) those inspired by the World Development plaining SSD is now widely recognized, Report 2004; (2) the Community-Driven De- thanks to the groundbreaking 2004 World velopment (CDD) approach and its later vari- Development Report (World Bank 2003). ants; (3) combinations of (1) and (2) rooted There are, however, different SSD models in in good governance norms; and (4) a more both the theoretical and practitioner litera- recent generation of approaches that attempt ture. One of the main issues that differentiates to systematically relate service delivery to the these models are their different understand- political economy context.58 The model that ings of governance and the way it matters for emerges here is most aligned with the fourth service delivery. SSD models include the fol- type of model and the recently published lowing (sometimes overlapping) approaches: 2017 World Development Report. 124 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds In sum, this report’s implications differ from a • Bargains and trade-offs are part of the number of the mainstream approaches in the new normal. Bargains, trade-offs, and following ways: compromises are integral to service deliv- ery—an element scarcely treated by other • No one model is privileged. Different models. models can be more or less effective de- pending on local political settlements, the • New actors are brought into sharper form of violence, and the context (see, for focus. A more concerted effort is made to instance, chapter 6). analyze the central role, interests, and in- centives of key SSD actors, such as elites, • It brings new drivers of progress to the landlords, or warlords, who have not re- forefront of the debate. These include ceived much, if any, attention in the past. informal institutions, elite bargaining, and far-from-perfect modes of governing. • SSD is understood through a lens of political feasibility rather than tech- • It suggests that violence needs to be nical soundness. Through this lens, we unpacked and systematically related reach slightly different conclusions: to service delivery. The varying impacts – Semiautonomous institutional delivery of different forms of violence on SSD are mechanisms are effective because they central preoccupations. can bypass powerful sectoral clien- telistic networks, not because they are • Political economy is not just about more rational-legal, as standard models material power; it is also about ideas. suggest. We show the important role of “ideas” (held by elites and insurgents) in shaping – Small-scale interventions are effective the constraints and opportunities for SSD. because the model is more appealing to local elites and because small grants • Service delivery is rarely politically offer fewer incentives and opportuni- neutral. The research, by showing how ties for rent capture at higher levels SSD is a source of rents that differs by of governance, not just because they sector, undermines traditional approaches involve “local people” and “citizen that suggest it can be technical, neutral, or participation.” ring-fenced from politics and power. Part III. Implications for Policy and Operations | 125 Box 8.4. Key questions for further exploration • How can the typology of violence be deepened and further tested in other geographic areas to strengthen its general applicability? • How do the ideas, values, and ideologies held by different elites and armed actors shape the way in which basic services are delivered in violence-affected contexts? Beyond standard information campaigns, how can develop- ment actors shape these ideas toward better development outcomes?a • Given the risks and trade-offs inherent in engaging with informal institutions and nonstate actors and in pursu- ing bargaining for social service delivery (SSD), are there lessons to be learned from any cases where bargaining was integrated and risks effectively mitigated? • How can we better understand projects in which elite bargains are struck but major progress is not achieved? Or the opposite: are there cases where elite bargains were not struck yet success was achieved? • What role do constituents and project beneficiaries, including the very poor, play in shaping the way elites and armed actors approach a given service? In other words, how can we better understand how pro-SSD alliances and coalitions between elites and nonelites (especially the poor) form? • How precisely do development interventions create and sustain bargaining structures? How can interventions operate within elite bargaining dynamics while creating incentives for the bargains to be more supportive of SSD? • How do gender dynamics fit into this elite bargaining approach in violence-affected contexts? • How can donor agencies square the circle between their fiduciary requirements, such as accounting and procure- ment, and the need for flexibility and bargaining for SSD? What aid instruments would be most appropriate for this? • Given the important role it plays in creating and sustaining rents and in shaping elite bargaining around deliv- ery, how can SSD’s influence on political outcomes—such as strengthening political parties or informal insti- tutions—be more systematically assessed? What is the measure for whether or not these political outcomes are good or bad for development? • How can we better understand if bargaining in the short term to enable the delivery of services endangers or advances the long-term goals of reducing the power of the insurgency, bringing about peace, and ensuring the legitimacy of the state? a. See Rodrik (2014). 126 | Social Service Delivery in Violent Contexts: Achieving Results Against the Odds – Bundling can be effective not just game” may be inadequate to deal with the because of the social and economic mounting humanitarian and development benefits of combining complementary challenges posed by complex conflict situ- services, but also because it means that ations, particularly where SSD access is re- powerful opponents can be compen- quired. This raises complex dilemmas about sated for enabling SSD to continue. the ethical and political judgments and trade- offs that development actors frequently have – Decentralized, multistakeholder ap- to make. Some trade-offs will be especially proaches are effective because they cre- difficult to consider due to the nature of the ate space for local elites to bargain over compromises required, but it is at least time to delivery, not just because local citizens consider and discuss them as part of a broader are “empowered.” debate about how to tackle the growing hu- manitarian and development consequences of Areas for further research fragility, conflict, and violence in the world. Box 8.4 provides a summary of areas inspired by this research that should be explored At minimum, the report sets out to question further. traditional ways of thinking about delivering services in highly fluid, contested, and risky In conclusion, this report provides the foun- contexts. A key challenge is whether develop- dation for developing an approach to service ment actors can adapt their procedures and delivery in violence-affected contexts that is ways of working to the uncertainty of this more sensitive to the actual forms of violence, new agenda and its demands for fluidity and politics, and bargaining that are found on the risk taking, while preserving the sometimes ground in many conflict-affected states today. fragile support for development assistance It raises particularly sensitive issues about in the donor country constituencies: “do no how to approach service delivery in contested harm” and provide effective aid. This agenda settings for development organizations such cannot be ignored. Progress in basic service as the World Bank which work with recog- delivery in violence-affected contexts will nized governments. As many countries today make or break global efforts to end poverty are increasingly riven by violent conflict and and increase shared prosperity over the com- internal division, the traditional “rules of the ing decades. Part III. Implications for Policy and Operations | 127 Notes 1. There are synergies between this report and the thinking and Key principles of the emerging adaptive approach to social working politically (TWP) approach. In our view, this report service delivery (SSD) programs in fragile and violent contexts makes a contribution to further unpacking and operational- are as follows (adapted from Andrews, Pritchett, and Woolcock izing the approach. Thinking politically means understand- [2012] and Valters [2016]): ing that both political and technical dimensions are central • Focus on processes, behavior changes, and intermediate out- to developmental outcomes. Working politically in develop- comes rather than just on activities or outputs. ment means supporting, brokering, facilitating, and aiding the emergence and practices of (public or private) developmental • Focus on solving locally nominated defined problems rather or reform leaderships, organizations, networks, and coalitions than transplanting preconceived ideas of what local priori- at any level, and across all sectors. It means helping them to ties should be or how things should work. respond to, and work with, initiatives and requests from local • Actively engage broad sets of agents to ensure that reforms individuals and groups. are viable, legitimate, relevant, and supportable, rather than Below are emerging findings on what thinking and working having a narrow set of external experts promote a top-down politically means (Leftwich 2011: 9–11): diffusion of innovations. • Prior and existing networks often facilitate the formation of • Allow flexibility and positive deviance by creating an au- developmental coalitions in unfolding situations that pres- thorizing environment for decision making that encourages ent opportunities for working politically (for example, in experimentation rather than designing programs and then the context of democratization or decentralization). Prior requiring agents to implement them exactly as designed. knowledge and experience with working politically extends • Prioritize ongoing learning and adaptation by ensuring that and enriches the tactical and strategic repertoire of such experimentation is embedded in tight feedback loops that leaders and elites. Well-educated and sometimes well-con- facilitate learning rather than having long lag times of ex- nected individuals play an important part in directing and post evaluations. driving coalitions for reform. • Think compass, not map. Set broad directions for change • The character and conditionality of funding by donors or rather than laying out microdetails in advance. supporters can make or break a coalition, for example, if tight conditions are applied or funding arrangements are Some unanswered practical questions remain, such as: how will transparent. adaptive capacities be built in environments with low insti- tutional capacities, a weak public sector, and patchy political • The internal organization of a coalition committed to re- coverage, as is often the case in violence-affected contexts? form is important, such as transparency of procedures and And how can development agencies operate with such high financial arrangements, agreement and clarity on roles and levels of uncertainty? responsibilities, size of a coalition and potential trade-offs between inclusiveness and effectiveness, and navigating be- 3. Underpinning this synthesis are three unpublished coun- tween formal and informal institutions and environments. try reports and global literature reviews (see World Bank and Akram 2016; World Bank and AREU 2016; World Bank and • Coalitions can help drive the endogenous politics of de- SSB 2016). velopmental reforms by achieving a specific policy goal, opening up debate on a previously taboo issue, deepening 4. Note that these gaps were identified through a rigorous re- and strengthening the coalition’s internal organization and view of the literature alongside consultations with a range of relationships for future purposes, and increasing the capac- practitioners in these fields. ity of constituent organizations. 5. As Booth (2013: 2) puts it: “The evidence is now over- For more details on what the thinking and working politically ap- whelming that most of the binding constraints in development proach entails, see Leftwich (2011: 3–9). For details on the are about institutions and institutional change.” principles and characteristics of a politically smart program, as 6. There is little evidence that what matters for developmental per the TWP approach, see Whaites et al. (2015: 44). outcomes depends on best practice or good governance (Khan 2. Given the especially fluid, nonlinear, and unpredictable na- 2010; Rodrik 2014). “Best fit” or “good enough governance” ture of change in violence-affected contexts, adaptive and real- refers to attempts to design development projects in the con- istic approaches to supporting change are probably necessary. text of actually existing governance structures, which often Indeed, our research found important changes on the ground involve navigating the realities of clientelism, corruption, and over time, but limited evidence of major or real-time adap- patronage-ridden politics. tation to such changes by the programs under investigation. Notes | 129 7. There is scant but growing evidence regarding how the po- 13. A great deal of literature is available on these subjects, litical economy impacts SSD in low-income countries (for ex- but it is not the purpose of this study to engage in theoretical ample, Mcloughlin and Batley 2012; Harris, McLoughlin, and debates. Wild 2013). Significant gaps remain: (1) the existing research 14. A challenge to state authority through violent means is tends to be conceptual and not strongly informed by firsthand synonymous with a challenge to the ruling coalition itself, ethnographic field research; (2) even studies that recognize which is the principal reason why political violence creates a the importance of politics treat the subject superficially (for dynamic or fluid situation for a political settlement. example, broadly referring to a lack of “political will”); and (3) the issue has not been applied to SSD in violence-affected 15. This categorization of political settlements does not imply contexts. that the national, provincial, and sectoral levels are discon- nected from one other in terms of elite bargains, contestations, 8. Based on informal consultations with the authors at the and their effects. Moreover, because political settlements are World Bank Fragility Forum, March 2016, Washington, DC historically specific, broad diversity and complexity is likely to (http://www.dlprog.org/opinions/what-do-we-do-on-mon- be found in political settlements at different levels of the polity. day-political-settlements-in-theory-and-practice.php). 16. There are sound reasons why a large variation in subna- 9. For a country to be included on the World Bank’s list of tional political settlements might be expected. Kalyvas (2003) fragile and conflict-affected states, it only needs to score low argued that the main driver of violence at the local and re- on the Country Policy and Institutional Assessment (CPIA) in- gional level is not usually linked to drivers of violence at the dicators, which are as much a measure of “underdevelopment” national level. Master cleavages often fail to account for the as of a particular condition of “fragility.” These indicators, nature of the conflict and its violence or if the violence is unre- drawn up by the World Bank (Moore 2014), are tools that lated or incompletely related to the dominant discourse of the allow a lender to rate the creditworthiness of a potential bor- war. Civil wars are imperfect and fluid aggregations of multi- rower. Countries are given scores based on 16 criteria grouped ple, more or less overlapping civil wars that are smaller, diverse, in 4 clusters: (1) economic management, (2) structural policies, and localized, and that entail Byzantine complexity and splin- (3) policies for social inclusion and equity, and (4) public sector tering authority into “thousands of fragments and micropowers management and institutions. of local character. A great deal of the action in a civil war is 10. Even when studies examine SSD in violence-affected con- therefore simultaneously decentralized and linked to the wider texts (for example, World Bank 2011), there is no attempt to conflict, including violence, which can be simultaneously po- explore if and how diverse forms of violence matters. Indeed, litical and private. A fine-grained analysis is required that takes what is missing is a systematic discussion of how and why into account intracommunity dynamics as well as those of the political violence is likely to alter the context of SSD delivery. civil war to determine when this is the case and who is allied with whom. The prospects for success for any SSD effort de- 11. Because it pilots a new innovative research framework in pend on the identification of the main local elites, including difficult contexts, the inductive approach seemed to be the armed groups, as well as an understanding of what motivated most appropriate. In other words, rigid theories do not apply the violence in the first place. at the beginning of the research, and the researcher is free to alter the direction of the study after the research process 17. This all results in a political settlement that is unstable has commenced (http://www.antropocaos.com.ar/Russel-Re- and fluid, generates uncertainty about the “rules of the game,” search-Method-in-Anthropology.pdf ). The study seeks to ex- and tends to create short-lived ruling coalitions and complex amine and refine in an iterative process patterns emerging from shifting alliances between elites at the national and subnational the field research and assess how and if these patterns fit with levels. Crises of political settlements need not occur only at reigning analytical explanations of the politics around SSD. To the national level. One strand of research focuses on violence build an analytical framework that could explain SSD imple- and breakdowns in political settlements at the subnational level mentation and outcomes, we drew on and attempt to integrate (Parks, Colletta, and Oppenheim 2013). insights from varied and rich literature on the political econ- 18. Olson’s (1993) “bandit theory” offers some insights into omy of institutions in less developed and violence-affected how two different organizations of violence can affect devel- contexts and varying options of SSD strategies and modes. opmental outcomes. In the model, a “stationary bandit” re- 12. The aim was to develop analytical narratives focused on fers to a leader that has eliminated rival armed groups and common patterns and processes of pro-poor service delivery has established a monopoly of violence and has established a rather than provide systematic comparative analysis of sector disciplined chain of command over its armed groups. Because specificities. Focal projects for each sector were selected based stationary bandits are relatively secure in power, they have in- on whether the life of a project spanned a certain period of centives to limit confiscation of property, avoid overburden- violent conflict; if the project and sector presented interesting some taxation, and provide public goods to maximize the rate puzzles of successes, challenges, or divergent outcomes; and if of economic growth because this will maximize the resources the Bank project team was interested in the study remit, which accruing to the leader in the long run. In short, stationary would improve the chances of getting access to data and en- bandits have a long-run time horizon and an “encompassing” courage future uptake. interest in promoting economic development. This contrasts 130 | Notes with a situation of “roving banditry” where there are either of political party contestation (see boxes 3.1 and 3.2). Almost competing regional warlords or where the ruling coalition’s everywhere the leading political figures that the field research hold on power is fragile. In this case, leaders have a short-run team encountered, especially in the rural parts of the districts, time horizon and they have incentives to pillage the economy had a political pedigree that had been passed on to them from as fast as they can which has obvious negative effects on eco- their family. A few exceptions could be seen in Rolpa, where nomic activity. While it has been pointed out that this model new leaders have emerged from among the ranks of the Mao- overlooks the reality that all states are organizations which ists, as well as in Siraha, where the Madhesi movement has emerge from the internal dynamics of relationships among empowered Madhesis to take leadership roles. Overall, how- elites (North, Wallis, and Weingast 2009: 17), it does suggest ever, there was a continuity of elites at the local level. This may that the organization of violence matters to SSD and other likely facilitate the political bargaining required to allow the developmental outcomes. implementation of the projects, even though many of these were of substandard quality. 19. There is a great deal of literature about how ideas and in- terests affect development, how ideas can alter power balances, 24. Based on H5–KII–SRH; H6–KII–SRH; H7–KII–SRH; and even how the powerful can alter their interests. See, for H10–KII–SRH; H17–KII–PAL; H20–KII–PAL; H21–KII– example, Rodrik (2014). PAL; H1–FGD–SRH; G8–KII–SRH; and G10–KII–SRH. 20. The weaker links may also be one of the reasons why un- 25. Based on H17–KII–PAL; H25–KII–PAL; H29–KII–ROL; foreseen government official transfers and the general rate of H30–KII–ROL; H32–KII–ROL; and H8–FGD–PAL. turnover is more common in the Tarai than in the hill region. 26. Health facilities turned out to generate several privileges: The greater degree of turnover and politicization of transfers (1) they were a source of supplies and services for the Maoist clearly affects the planning and the continuity of policy im- rebel forces (H21–KII–PAL; H29–KII–ROL; G10–KII–SRH); plementation among the three main types of local officials: (2) health workers were found paying “levy to the Maoists” government agents, local development officers, and district (H13–KII–BAR) and private medical practitioners had to planning officers. pay donations to Maoists (ibid.); and (3) there was evidence 21. The higher-level elites are the nawabs (Indian-Muslim word that Maoists sent letters to locals asking for rice and wheat for “head of the landed nobility”) and the sardars (Persian word (H14–KII–BAR). for “chief ”), who can serve as the figurehead or de facto leader 27. The chief district officer is the principal district administra- of the wider tribe. These groups are seen as highly influen- tor who also serves as the head of the district security commit- tial—the Baloch nawabs and sardars even more so than their tee along with the district chief of police and the commander Pashtun counterparts. These elites often maintain two resi- of the local military unit. dences because living conditions and the social life in their home districts are generally poor. Families and children of 28. Informants acknowledged that there is no way to avoid this the elite live in the major cities of Pakistan (such as Karachi, kind of interference, that it could only be “managed.” They in- Islamabad, Quetta, or DG Khan) or abroad, and they main- dicated a need for compromise or vaguely referred to solutions tain another home in their ancestral homeland. Consequently, being found, without elaboration. while this group has provided leadership for issues that can 29. Based on KP: R09–KII–PSH; KP: R10–KII–PSH; KP: affect a whole tribe, such as war or disaster, their perpetual R13–KII–PSH; FATA: R07–KII–PSH; FATA: R05–KII–PSH; absenteeism risks them losing their land and legitimacy. These FATA: R11–KII–PSH; FATA: R13–KII–PSH; FATA: R12–KII– elites are not likely to send their children to a BESP school. PSH; and FATA: R24–GI–PSH. In addition to lineage and being conversant in local customs, a prerequisite for entering the upper elite is a certain amount 30. Based on R18–FGD–PSH; and R19–FGD–NWA. As of “Westernization,” and therefore the children of elites typi- recently as November 2015, a polio district coordinator was cally attend one of the top Pakistani schools and then pursue a gunned downed in KP’s Swabi district (Express Tribune 2015). higher education abroad. 31. In another interview, Hafiz Gul Bahadur declared in June 22. Elders are sometimes organized into tribal shuras (consulta- 2012 that no polio campaign would take place in North Wa- tive councils), although only a few tribes have all-Wardak shu- ziristan (R05–KII–PSH). A recent exception to this trend is in ras. The Pashtuns of Wardak have stronger tribal representative Bannu, where there were five campaigns in the bordering areas structures than do those in Badghis, although neither com- of Lakki Marwat, Tank (R13–KII–PSH). pares with the eastern tribes (WK: IDI–14 and 18). The exis- 32. It was reported that in FATA, even nongovernmental orga- tence of shuras tends to magnify the influence of elders, which nizations (NGOs) (which are usually more welcome than gov- makes them more attractive partners in secondary settlements. ernment workers) cannot access areas for the polio program 23. Similarly, one of the factors that contributed to the pros- because of the fierce insurgent resistance to it. One interviewee pects for elite bargains in Nepal was that the power of tradi- claims that the polio program was considered a “taboo” topic tional elites were often maintained at the local level, despite (R05–KII–PSH). shifts in the national political settlement and the fluid nature Notes | 131 33. A failure to provide these “donations” often led to the 39. See also the following interview transcripts: KII–01; KII– disruption of construction work. Government officials in the 02; KII–03; KII–04; KII–09; FGD–01. Tarai were threatened, and as a result, rarely went out into the 40. The main points outlined in chapters 7 and 8 were de- field. Monitoring and supervision was often conducted with rived from the authors’ analysis and from suggestions received the help of security forces. In Bardiya, during the years of con- from a number of experts and development practitioners flict, the Maoists had even asked for up to 50 percent of the related to an earlier draft of this chapter (as noted in the project cost, accusing the road user committee of colluding acknowledgements). with the government to build the road so security forces could use it. The Maoists also demanded money from contractors 41. This does not mean, however, that technical aspects do not who risked the disruption of construction work if they did matter. They just need to be complemented by an attempt to not comply (based on G17–KII–PAL; G3–KII–SRH; R5–KII– understand and adapt to the realities of the political economy, KTH; G13–KII–BAR; R8–KII–SRH). as is widely recognized in the political economy literature. 34. An assessment of local governance finds that pet projects of 42. See also Harris and Wild (2013), which acknowledges this political leaders were given priority and that those of ordinary point. people received attention only when they corresponded with 43. See Booth and Cammack (2013), Unsworth (2015),G and the interests of the politicians (Inlogos 2009). Politicians were Kelsall (2016) for a discussion of the policy implications of sometimes even able to influence the selection of roads under these approaches, and the latest World Development Report the Rural Access Improvement and Decentralization Project. (World Bank 2017). In fact, 35 percent of the village development committee sec- retaries admitted that recommendations from politicians was 44. See, for example, DiJohn and Putzel 2009; Grindle 2007; the clinching factor when making decisions about the proj- Khan 2010; Unsworth 2010; Booth et al. 2015a; Goloo- ects (Based on G2–KII–SRH; R8–KII–SRH; R9–KII–SRH; ba-Mutebi and Hickey 2013; Hickey 2013; World Bank 2017. R8–KII–SRH). 45. See World Bank (2016b). 35. The Balochistan Education Foundation (BEF) was set up 46. This observation is not just limited to our cases—there is in 1994 by the government of Balochistan to strengthen the limited global evidence to suggest that aid agencies are system- private sector provision of education and to raise the standards atically assessing and tailoring their approaches to the issues for education in the province. The government restructured addressed here. The reasons behind this disconnect could be the foundation and brought about significant reforms in its myriad, including knowledge gaps, ideological blind spots, governance, management, and operating procedures through and organizational cultures that prefer to avoid acknowledging amendments to the BEF Act in 2004 and 2006. such realities, but whatever they are, evidence suggests that we 36. Several other informants expressed similar ideas (based on need to get to the root it. KII–01; KII–08, 43; KII–12). The problems between the two 47. At least three main lessons should be taken into account parties emerged in the later stages of BESP when relevant peo- when delivering or implementing SSD. First, in practice, de- ple in the bureaucracy and top management were transferred livery is often facilitated by informal and ad hoc arrangements to other departments. across blurred public-private boundaries (Mcloughlin and Bat- 37. Numerous interviewees suggested that BEF and the NGOs ley 2012: 30–31). Second, informal practices and institutions were equally subject to corruption and patronage. A key in- are often pervasive and are not going away soon, whether we formant interview with a donor office (anonymized, KII–01– like it or not (Mohmand 2012). Third, informal institutions— PAK–BAL) gave the following information: (1) participants at a frequently framed as governance problems—should instead meeting remarked they did not want to work through NGOs as be seen as part of the solution (Moore and Unsworth 2010). implementing partners due to corruption, lack of capacity, and Two notable thinkers have tried to make these insights more difficulties in reaching out to them; (2) participants noted that practical: BEF was allegedly corrupt, with one person receiving the sal- • Practical hybrids. Booth (2012) describes “practical hybrids” ary of 100 people (based on hearsay); and (3) the interviewee as institutional arrangements that combine modern bureau- was of the opinion that local NGOs do not have greater capac- cratic standards and approaches to development with lo- ity than local people—donors/development partners should cally accepted cultures and practices. Specifically, they are work directly with communities by supporting them with new institutional arrangements that are: (1) locally anchored in technologies and adding an economic element/component. that they are locally driven initiatives making use of local 38. For instance, the increased use of the military and police to resources and deriving their legitimacy from “local cultural escort health workers enabled them to reach areas where there repertoires” and “local views on what is important and how was violent political opposition to the Polio Eradication Proj- to get things done” (Booth 2012: 19); and (2) “problem ect, but this also made health workers more easily identifiable solving,” insofar as the arrangements address problems that (R16–FGD–SWB). are perceived as important in the local context, build on 132 | Notes currently existing arrangements, and facilitate collective ac- for future purposes; and (4) increasing the capacity of constit- tion to address the problems. uent organizations. • Institutional bricolage. Cleaver (2001) finds multiple cases 52. We are grateful to Professor James Putzel for the term “shot of “institutional bricolage” where “people consciously and in the dark” in this context. This term was communicated in his unconsciously draw on existing social and cultural ar- comments on an earlier draft. rangements to shape institutions in response to changing 53. There are two potential reasons for this: (1) development situations.” Institutions are formed through the uneven practitioners are detached from how things really work on the “patching together” of old practices and norms with new ground, or (2) they were aware of these issues but were not arrangements. In practice, this requires the constant rene- comfortable with officially documenting them. Our research gotiation of norms, the reinvention of tradition, the identi- suggests the former. fication of legitimate forms of authority, the facilitation of cross-cultural borrowing and multipurpose institutions, and 54. These debates are also captured in the humanitarian liter- the fostering of mutual cooperation and respect. See Cleaver ature on the issue of SSD neutrality and the trade-off between (2012) for examples. It must be noted that bricolage is not treating insurgents and sustaining the conflict. always benign and can reproduce inequalities. 55. See note 1. 48. This is in line with “problem-driven” approaches to devel- 56. See note 2 opment (Fritz, Levy, and Ort 2014). 57. It is nonetheless recognized that the aid system is not ho- 49. There is scant practical guidance on how to deliver services mogenous and embodies different approaches. under different political settlements (Parks and Cole 2010; Kelsall 2016). 58. The World Development Report (World Bank 2004) argu- ment is that when services remain undelivered, break down, or 50. As North et al. (2011: 1) put it, “when development pol- are substandard, it is due to failures in the chain of accountabil- icy advice threatens the logic of stability in limited access or- ity that links service providers to beneficiaries. It introduced ders, these societies often resist or sabotage the recommended the now influential “accountability triangle,” which links poor measures.” people to service providers either through policy makers (that 51. There are a number of approaches for how aid actors can is, the “long route” of accountability) or directly (that is, the support local networks and coalitions that can be drawn on to “short route”). It therefore claims that SSD works when the inform programming (O’Meally 2013). For example, Leftwich incentives of providers and beneficiaries are aligned (Kelsall (2011) notes that prior or existing networks often facilitate the 2016). The second major SSD paradigm among donors is formation of developmental coalitions in unfolding situations community-driven development, an approach that emphasizes of this kind. Well-educated and sometimes well-connected community control over planning decisions and investment re- individuals regularly play an important role in directing and sources (see Wong 2012; Marc et al. 2013; Bennett and D’On- driving coalitions for reform. The internal organization of a ofrio 2015). The idea behind it is that involving communities coalition for reform is important, such as transparency of pro- in local development decisions is not only the inherent right cedures and financial arrangements, agreement and clarity on of every citizen but that participation can often lead to a bet- roles and responsibilities, size of the coalition, potential trade- ter use of resources, geared toward meeting community needs. offs between inclusiveness and effectiveness, and navigating However, there is scarce systematic evidence or analysis exam- between formal and informal institutions and environments. ining why and how community-driven development projects Coalitions can help drive the endogenous politics of develop- work in violence-affected contexts. mental reform by: (1) achieving a specific policy goal; (2) open- ing up the debate on a previously taboo subject; (3) deepening and strengthening its internal organization and relationships Notes | 133 Appendixes Three women train to be midwives. Tajikhan Village, Afghanistan, Graham Crouch/World Bank Appendixes | 135 Appendix A. Methodology and Case Study Background Methodology Gathering Information South Asia is home to more than 1.6 billion The study methodology is qualitative and people—about 24 percent of the world’s pop- combines multiple methods, summarized in ulation and about half of the world’s poor. box A.1. The country-based field research Progress toward ending extreme poverty was conducted by local partners, with tech- will be therefore largely influenced by prog- nical backstopping provided by the research ress made there. The region has experienced task team.1 The methodology was inductive various forms of violent conflict, which have in that we began with broad questions and negatively impacted its development prog- key gaps and iteratively narrowed the focus ress. This study focuses on service delivery in based on the empirical findings that emerged. Afghanistan, Nepal, and Pakistan—countries This process included multistakeholder work- selected to span a range of forms of violence. shops, in-country field visits to deliberate The research focuses on the health and edu- with local partners, refinement of the findings cation sectors in Afghanistan, the health and through expert input, and literature reviews rural infrastructure sectors in Nepal, and the of emerging topics. health and education sectors in Pakistan. The research covers but is not limited to six proj- Regarding the interview method, we made ects in these focal sectors. every attempt to triangulate responses from individual interviews with information from Field research sites were selected within each focus group discussions and with primary and country for in-depth study and comparative secondary data. Much of the information in analysis. Each of the subnational areas was the study regarding the views, interests, and affected by violence, but had different char- behavior of armed insurgent actors are based acteristics in terms of location, forms of vi- on third-party intermediaries because insur- olence, and composition of society. Further gent groups are often extremely difficult to information about the subnational research access and highly fragmented. We triangu- sites is available in the respective background lated the views of the intermediaries when country reports (World Bank and Akram possible. Further research could include inter- 2016; World Bank and AREU 2016; World views with insurgents. Bank and SSB 2016). Appendixes | 137 Box A.1. Summary of Research Methods The findings were generated and triangulated using the following methods: • Systematic metareview of global evidence. The review included existing knowl- edge on social service delivery (SSD) in violence-affected contexts, political economy and violence conceptual frameworks, existing policy guidance on low-income and vi- olence-affected contexts, and knowledge on delivering health and educaton in fragile contexts. • Review of country-, sector-, and intervention-related evidence. The review included the full range of academic-, policy-, donor-, and project-related documenta- tion, both internal and external to the World Bank. • Semistructured key informant interviews (KIIs). Over 220 interviews were con- ducted in Islamabad, Kabul, and Kathmandu—capital cities of the three countries selected—and at field sites with a range of informants, including current and former World Bank staff, project team leaders, sector specialists, government officials, service providers, subject experts, donor agencies, implementing partners, and elites and lead- ers at various levels. • Field visits and focus group discusssion (FGDs). These visits and discussions were conducted to triangulate and validate the findings, observe the on-the-ground power dynamics, enrich the findings through rolling group discussions, and ensure represen- tation from users of the services. • Systematic interview transcript analysis. Based on transcript analysis methodolo- gies, the key informant interviews and focus group discussions were coded, clustered, and analyzed to draw out common findings and issues. • Expert input and consultations. Sectoral, country-specific, and global experts pro- vided comments and input at various points in the process. • Multistakeholder consultations. Consultations were held with local partners, ex- perts, and donor officials. 138 | Appendixes Why We Chose These Methods: other service delivery areas. All the regions Case Selection and Limitations chosen had experienced political violence. The case studies were selected based on re- Case study sites were not chosen randomly, search puzzles developed from gaps in the and no attempt was made to set up a random- literature (see background country reports). ized controlled trial (RCT). Because political Pakistan’s Khyber Pakhtunkhwa and the Fed- contexts and dynamics vary so widely across erally Administered Tribal Areas, post-2006 regions affected by large-scale political vio- Nepal’s Hill Region and the Tarai, and the lence (Kalyvas 2003), we determined that the regions in Afghanistan were chosen based on RCT method—often not feasible in fragile, variations in outcomes of service delivery;2 conflict, and violence-affected areas—would the Balochistan Education Support Project be largely inappropriate as a method of case of (BESP) was chosen for being an outlier of selection. While RCTs are useful for describ- progress in the province; and the period of ing what works, they are much less useful for large-scale violent conflict in Nepal (1996– drawing inferences or demonstrating causal- 2006) was chosen for its counterintuitive na- ity (Basu 2014). In addition, lessons learned tionwide progress in social service delivery from RCTs translate poorly because contexts (SSD) during a period of political violence. differ within and across countries (Hausmann We then attempted to discover what political 2016). Finally, a more fine-grained analysis economy dynamics and forms of violence are with a “small N” sample permits us to trace associated with—if not the cause of—such and explain how a combination of factors— variations. The aim was to construct analyt- local political settlements, elite bargaining, ical narratives to suggest potential explana- forms and motivations of violence, and mode tions for the divergences because the current of delivery—generate divergent outcomes in theory and practice literature could not, at ways that the RCT method could not (Basu least in our cases. Thus, the cases were chosen 2014; Hausmann 2016). In this sense, the purposefully but were not necessarily meant case selection method was used to develop a to be representative of country context. For more general theoretical framework for how example, the polio campaign may not be em- and why these four dimensions matter to SSD blematic of service delivery in Khyber Pakh- outcomes (see Thomas 1998). tunkhwa or the Federally Administered Tribal Areas. It was a discrete activity born out of Sectors and associated projects were selected unique circumstances, with highly specific for several reasons. The sectors chosen in each characteristics. Insights drawn from the polio country focused on basic social services that campaign may not be readily extrapolated to emphasized serving the poor. The intention Appendixes | 139 was not to provide systematic comparative developed hunches and working hypotheses analysis of sector specificities but rather to based on the literature, and we chose study focus on common patterns and processes of sites with divergent outcomes in SSD and pro-poor service delivery. The focal projects examined what differed in terms of political for each sector were selected based on: if the economy and violence patterns. We wanted project spanned some of the period of violent to develop an analytical framework that was conflict; if the project and sector presented in- focused enough to identify patterns that had teresting puzzles about successes, challenges, not yet been illuminated in the literature. For or divergent outcomes; and if the World Bank a similarly focused approach that attempts to project team was interested in the study remit, explain patterns of political order, see North, which would improve the changes of getting Wallis, and Weingast (2009) regarding lim- access to data and encourage future uptake. ited access orders. In short, the study focuses on selected sec- tors with key projects as focal points. Note, These four dimensions emerged as import- however, that the report does not focus on ant contributors to implementation and out- attributing impacts to specific projects. It cov- comes; and they have been underemphasized ers selected sectoral projects but is not limited in the available literature. However, while we to them or does not attempt to separate them focus on these areas, we do not suggest that from other processes in the sectors.3 In many other issues do not matter. For example, there cases, we did not attempt to distinguish them is a well-established notion that beneficiary from broader sectoral or government pro- participation and service provider incentives cesses or to examine the impact of additional are central drivers of service delivery (World resources compared with none. Bank 2004). Our study did not focus on these factors because interview and other evi- We focus on four dimensions in the analytical dence did not suggest that they were decisive framework—local political settlements, elite in addressing the research puzzles. However, bargaining, forms and motivations of vio- we do say something about beneficiaries and lence, and mode of delivery—due to several service providers, even if they tend to be elite factors. Like the case study selection, these dominated, as discussed in the report.4 Our dimensions were chosen as the organizing framework should be seen as an attempt to ideas of the study after a number of iterations specify how politics and forms of violence of the research framework based on what the affect more mainstream models of service empirical material was telling us, what the delivery, such as the accountability triangle gaps in the literature were, and how we could approach from the 2004 World Development explain the puzzles in each case. As previously Report (World Bank 2003) or communi- noted, this was an inductive process: we had ty-driven development. 140 | Appendixes The puzzles and framework can be tested The study does address the extent to which with further research. In this sense, our work SSD can become a target for political vio- is a starting point. Research in other contexts lence. In this respect, we aim to understand would be desirable to further test the findings. why and how this occurs, and provide donors Nevertheless, our findings were triangulated guidance on how to “do no harm” at various with the national and global evidence base points in this study. to increase their wider applicability.5 Further research could similarly incorporate the opin- ions of beneficiaries regarding services, their Notes levels of participation, and the motivation of 1. Technical backstopping included regular two-way commu- providers to understand how the strategies of nications as well regular field visits from the task team to de- the elites take their views into account. We bate and refine emerging findings and fill gaps. For example, the task team provided detailed guidance notes on the ana- hope this work encourages further research lytical framework, field research methodology, key informant interview questions, and background report drafts. into such important dynamics. 2. See King, Keohane, and Verba (1994) on the importance of variation in outcomes when designing research. This study does not attempt to address the 3. Note that the report also focuses on how services are de- impact of SSD on broader political dynam- livered in violence-affected contexts. It does not focus on the relationship between SSD and state-building or state le- ics, such as peace-building and state-build- gitimacy (Mcloughlin 2014); the relationship between SSD and peace-building (Denny, Mallett, and Mazurana 2015); or ing. While this an important area of research, whether SSD contributes to changes in the political settlement we think that the link is indirect and overly (Barrientos and Pellissery 2012). Moreover, it is difficult to es- tablish linear causal relationships between political economy broad to provide granular, day-to-day guid- and service delivery outcomes (Harris and Wild 2013: 4). ance for policy makers and practitioners on 4. For example, while there were very few systematic surveys the ground. The assumption that delivering available on the opinions of beneficiaries, we conducted a small survey of marginalized groups in Nepal; and the scorecard that social protection and basic services contrib- formed the puzzles on Afghanistan embedded beneficiary opinions in the index. utes to state-building outcomes appears with 5. The findings are complementary to rather than a substitute a striking degree of regularity in the litera- for other technical analyses. The political economy analysis ture, and yet empirical evidence supporting or should be viewed as a complement to help solve complex de- velopment problems (Booth, Harris, and Wild 2016). undermining this link is very thin (Carpenter, Slater, and Mallett 2012; Haar and Ruben- stein 2012). Appendixes | 141 Appendix B. A Diagnostic Tool: Steps and Question Set Chapter 8 describes the rationale and modules to these aspects. Figure B.1 summarizes the for this diagnostic tool. Its core purpose is to main diagnostic modules, followed by sample help practitioners gain a better understand- sets of questions. Note that this is merely a ing of the important dimensions of political starting point—additional questions can be settlements, violence, and service delivery developed depending on the context and in- bargaining in a given context to help them terests of practitioners. support interventions that are more sensitive Figure B.1. Main diagnostic modules Module 1: Understand relevant political settlements, who the relevant elites are, and how they matter Relations between Basic features of Sectoral Subnational central and national settlement settlement settlement subnational elites Module 2: Map the relevant dimensions of violence in the program areas Ideology and Localization Organization objectives and mobility Module 3: Draw lessons from local service delivery and bargaining Delivery Lessons in Developments strategies bargaining outside the sector 142 | Appendixes Module 1: Guiding Questions Basic features at the center of power • Who are the key power holders in the ruling coalition? What is the relationship between the ruling elites and violent actors? • What mechanisms help sustain the settlement? To what degree can elites act collectively to deliver goods? • To what extent is the settlement clientelistic/developmental, exclusionary/inclusive, or stable/unstable? • How do these features affect the space for delivery in the given sector? Sectoral dimensions • What is the balance of power and nature of rent-sharing agreements in the sector? What are the interests, incentives, and ideas of eliteswith regard to the focal sector? • What are the opportunities for rents in the sector, and how does this influence the way ser- vice delivery resources are allocated? What are the dominant modes of delivery in the sector, and how does this impact on rent-sharing (for example, state-led versus contracting out, preventative versus curative health care delivery)? Relationship between central and local elites • What is the nature of the relationship between ruling elites at the center (often in the capital) and powerful elites in relevant subnational areas? • What is the degree of state penetration over the subnational territory? • How does the above shape the amount of, and channels for, service delivery resources to the area? How does it shape the degree of autonomy of subnational elites to deliver/manage/ allocate resources? Appendixes | 143 Subnational dimensions • Who are the powerful elites (armed and unarmed) in the subnational area? • What is the nature of the subnational settlement, and how is it sustained and reproduced? • What is the relationship and relative balance of power between subnational elites and armed elites? • To what extent do key elites have ideas, interests, and incentives in favor of broader-based service delivery? Module 2: Guiding Questions Drivers • What are the key drivers of violence? • How does this relate to the dynamics associated with the forms of political settlements? Organization • To what degree is violence organized so that leaders for bargaining are identifiable? • To what degree are violent actors disciplined and homogenous so that bargains can be struck and upheld? Ideology, objectives, and incentives • What are the objectives and ideology of the armed elites? To what degree does the logic, scope, and mode of delivery align with such objectives? • Do armed elites have a “policy” toward the services in question? Does this cascade through their ranks? • Do armed elites have incentives to enable the service (for example, rents, need to use service, helps them achieve goals)? • Are different modes of delivery likely to be more acceptable to armed elites? For example, which actors would they trust to deliver the services? Would they be more comfortable with decentralized delivery? 144 | Appendixes Localization and mobility • To what degree are the armed elites locally anchored as opposed to more mobile/roving across borders? • Do armed elites need/want local support and do locals want the service? • Do local elites have relative power to bargain with/constrain local insurgents? Module 3: Guiding Questions Experience and strategies for delivery bargaining • What are the prior lessons of front-line service delivery and bargaining in the violence-af- fected area? Has dialogue been established with insurgents and bargains struck? Was this formalized? • Are there opportunities for bargaining? If not, how could services be delivered in those areas? • Given steps 1, 2, and 3, which types of bargains might be feasible? What are the likely trade-offs? Service delivery strategies • How are services currently being delivered in the sector? What are the institutional channels for delivery and political economy challenges for implementation? • Do the sectors have certain policies/strategies for delivering services in violence-affected areas? How well are they working, and have any lessons been learned? • What is the feasibility and opportunities for making delivery more sensitive to local politics and violence? Developments outside the sector • What is the nature of income and land distribution in the area? How might this affect service delivery, citizen demand, and forms of political organization? • What other projects are in the focal region (for example, roads and livelihoods)? Are there opportunities for multisectoral approaches? Appendixes | 145 Reference List and Bibliography The following in-text citations refer to the interview transcript codes used for informants and focus group participants: Afghanistan: KL: IDI–63, KL: KII–34, KL: KII–35, KL: KII–36, KL: KII–37, KL: KII–42, KL: KII–43, KL: KII–44, KL: KII–45, KL: KII–47, KL: KII–49, KL: KII–50, KL: KII–53, BS: IDI–22, BS: IDI–23, BS: IDI–31, BS: IDI–32, BS: IDI–33, BS: KII–15, WK: IDI–07, WK: IDI–09, WK: IDI–13, WK: IDI–14, WK: IDI–16, WK: IDI–17, WK: IDI–18, WK: FGD–18, WK: KII–01, WK: KII–02, WK: KII–03, WK: KII–05, WK: KII–06, WK: KII–07, WK: KII–09, WK: KII–12, WK: KII–13, WK: KII–18 Nepal: KII–03–NEP–GEN, G2–KII–SRH, G3–KII–SRH, G8–KII–SRH; G9–KII–SRH, G10–KII–SRH G13–KII–BAR G15–KII– BAR, G17–KII–PAL, G19–KII–PAL, R5–KII–KTH, R6–KII–KTH, R8–KII–SRH, R9–KII–SRH, H3–FGD–SRH, H8–FGD– PAL, H4–KII–SRH, H5–KII–SRH, H6–KII–SRH; H7–KII–SRH, H8–KII–SRH, H10–KII–SRH, H13–KII–BAR, H14–KII–BAR, H16–KII–BAR, H17–KII–PAL, H18–KII–PAL, H20–KII–PAL, H21–KII–PAL, H22–KII–PAL, H25–KII–PAL, H29–KII–ROL, H30–KII–ROL, H32–KII–ROL, H1–FGD–SRH, H4–FGD–SRH, H6–FGD–BAR, H7–FGD–BAR, H8–FGD–PAL, R1–FGD– SRH, R5–KII–KTH, R20–KII–PAL Pakistan: KII–01, KII–02, KII–03, KII–04, KII–05, KII–08, KII–09, KII–11, KII–12, KII–14, KII–15, FGD–01, R–04–KII–PSH, R–05–KII–PSH, R07–KII–PSH, R09–KII–PSH, R10–KII–PSH, R11–KII–PSH, R12–KII–PSH, R13–KII–PSH, R15–FGD– PSH, R18–FGD–PSH, R19–FGD–PSH, R–19–FGD–NWA, R20–FGD–PSH, R–24–GI–PSH, KII01–PAK–BAL Acemoglu, D., and S. Johnson. 2005. “Unbundling Institu- Bain, K., D. Porter, and M. Watts. 2015. “Institutional Change, tions.” Journal of Political Economy 113 (5). http://economics. Political Economy, and State Capabilities: Learning from mit.edu/files/4467. Edo State, Nigeria.” Justice, Security and Development Series Working Paper 97600, World Bank, Washington, DC. Akmeemana, S., D. Booth, D. Brown, et al. n.d. “The Case http://www-wds.worldbank.org/external/default/WDS for Thinking and Working Politically: The Implications ContentServer/WDSP/IB/2015/07/08/090224b082f of ‘Doing Development Differently.’” http://ecdpm.org/ dd09c/2_0/Rendered/PDF/Institutional00m0Edo0stat- wp-content/uploads/Case-Thinking-Working-Politically. e00Nigeria.pdf. pdf. Baird, M. 2011. “Service Delivery in Fragile and Conflict-Af- Albrecht, P. 2013. “Building on What Works—Service Deliv- fected States.” World Development Report 2011 Background ery in Fragile Situations.” DIIS Policy Brief, Danish Institute Paper, World Bank, Washington, DC. https://openknowl- for International Studies, Copenhagen, Denmark. http:// edge.worldbank.org/bitstream/handle/10986/9203/ pure.diis.dk/ws/files/57341/DIIS_PB_RECOM_service_ WDR2011_0003.pdf ?sequence=1&isAllowed=y. delivery_fragile_situations_PAA_web.pdf. Barma, N., E. Huybens, and L. Vinuela. 2014. Institutions Taking Andrews, M., L. Pritchett, and M. Woolcock. 2012. “Escaping Root: Building State Capacity in Challenging Contexts. New Fron- Capability Traps through Problem-Driven Iterative Adapta- tiers of Social Policy. Washington, DC: World Bank. https:// tion (PDIA).” CID Working Paper No. 240, Center for Inter- openknowledge.worldbank.org/handle/10986/20030. national Development at Harvard University. https://www. hks.harvard.edu/content/download/70633/1255342/ Barrientos, A., and S. Pellissery. 2012. “Delivering Social Assis- version/2/file/240_Andrews%2C+Pritchett%2C+Wool- tance: Does Politics Matter?” ESID Working Paper No. 9, cock_BeyondCapabilityTraps_PDIA_FINAL.pdf. Effective States and Inclusive Development Research Cen- ter, University of Manchester, Manchester, United Kingdom. Asia Foundation. 2012. Political Economy Analysis of Local Gov- http://www.effective-states.org/wp-content/uploads/ ernance in Nepal with Special Reference to Education and Health working_papers/final-pdfs/esid_wp09_Barrientos- Sectors. Kathmandu: The Asia Foundation. Pellissery.pdf. Asian Development Bank, UK Department for International Basnett, Y. 2009. “From Politicization of Grievance to Political Development (DFID), and World Bank. 2012. Sectoral Per- Violence: An Analysis of the Maoist Movement in Nepal.” spectives on Gender and Social Inclusion: Health. Kathmandu, Working Paper Series No.07-78, Development Studies Nepal: Asian Development Bank, DFID, and World Bank. Institute (DESTIN) London School of Economics and Political Science, London, United Kingdom. http://www. lse.ac.uk/internationalDevelopment/pdf/WP/WP78.pdf. Reference List and Bibliography | 147 Basu, K. 2014. “Randomisation, Causality and the Role of Development Institute. http://www.institutions-africa. Reasoned Intuition.” Oxford Development Studies 42 (4): 455– org/filestream/20150216-developmental-regimes-in-afri- 72. http://dx.doi.org/10.1080/13600818.2014.961414. ca-synthesis-report. Bennett, S., and A. D’Onofrio. 2015. “Community-Driven Bread for the World Institute. 2016. Fragile Environments, Resil- Development in Conflict-Affected Contexts: Revisiting ient Communities. 2017 Hunger Report. Washington, DC: Concepts, Functions and Fundamentals.” International Journal Bread for the World Institute. http://www.bread.org/sites/ of Security & Development 4 (1): 19, pp. 1–18. https://assets. default/files/2017-hunger-report-full.pdf. publishing.service.gov.uk/media/57a08983ed915d622c- Brixi, H., E. Lust, and M. Woolcock. 2015. Trust, Voice, 00025d/61244-CDD_in_Conflict-Affected_Contexts_ and Incentives: Learning from Local Success Stories in Service Bennett_DOnofrio_Stability_360-1719-1-PB.pdf. Delivery in the Middle East and North Africa. Washington, Bernard, R. 2011. Research Methods in Anthropology (Fourth Edi- DC: World Bank. https://openknowledge.worldbank.org/ tion). United Kingdom: Altamira Press. handle/10986/21607. Bevan, J. 2010. “National Political Economy and Conflict Stud- Byrne, S., and G. Shrestha. 2014. “A Compromising Consen- ies on the Emerging Constitutional Framework, Decentral- sus? Legitimising Local Government in Post-Conflict Nepal.” ization and Local Government.” Submitted to the New International Development Planning Review 36 (4): 435–53. Emerging Towns Project, World Bank, Washington, DC. Carpenter, S., R. Slater, and R. Mallett. 2012. “Social Protec- Bhandari, A., M. Gordon, and G. Shakya. 2011. “Reducing tion and Basic Services in Fragile and Conflict-Affected Sit- Maternal Mortality in Nepal.” BJOG: An International Jour- uations.” SLRC Working Paper No. 8, Secure Livelihoods nal of Obstetrics and Gynaecology 118 (Suppl. 2): 26–30. Research Consortium, London, United Kingdom. https:// doi:10.1111/j.1471­0528.2011.03109.x. www.odi.org/sites/odi.org.uk/files/odi-assets/publica- tions-opinion-files/7859.pdf. Booth, D. 2012. Development as a Collective Action Problem. Africa Power and Politics Programme Synthesis Report. CBS (Central Bureau of Statistics). 2014. Population Monograph London: Overseas Development Institute. http://www. of Nepal. Volume I (Population Dynamics). Kathmandu: CBS. institutions-africa.org/filestream/20121024-appp-synthe- Central Statistics Organization. 2009. Afghanistan Statistical sis-report-development-as-a-collective-action-problem. Yearbook 2008–09. Kabul: Central Statistics Organization. ———. 2013. Facilitating Development: An Arm’s Length Approach Cleaver, F. 2001. “Institutional Bricolage, Conflict and Coop- to Aid. London: Overseas Development Institute. https:// eration in Usangu, Tanzania.” IDS Bulletin 32 (4): 26–35. www.odi.org/publications/7376-facilitating-develop- ment-arms-length-approach-aid. ———. 2012. Development through Bricolage: Rethinking Institu- tions for Natural Resource Management. New York: Routledge. ———. 2015a. “Doing Development Differently: What Does It Mean in the Roads Sector?” World Bank Blog, October Dany, J., E. Davin, , S. De Silva, Sayed G. Dastagir, A. Soler, 16, World Bank, Washington, DC. http://blogs.worldbank. and R. Hogg. 2014. Critical Administrative Constraints to Ser- org/governance/doing-development-differently-what- vice Delivery: Improving Public Services in Afghanistan’s Trans- does-it-mean-roads-sector. formational Decade. Washington, DC: World Bank Group. http://documents.worldbank.org/curated/en/2014/05/ ———. 2015b. “Five Myths about Governance and 20346330/critical-administrative-constraints-service- Development.” ODI Blog, February 16, Overseas delivery-improving-public-services-afghanistans- Development Institute, London. http://www.odi.org/com- transformational-decade. ment/9274-five-myths-about-governance-development. Denny, L., R. Mallett, and D. Mazurana. 2015. “Peacebuild- Booth, D., and D. Cammack. 2013. Governance for Develop- ing and Service Delivery.” Review of the Peacebuilding ment in Africa: Solving Collective Action Problems. London: Zed Architecture, United Nations University Center for Policy Books. Research. http://r4d.dfid.gov.uk/pdf/outputs/slrc/Peace- Booth, D., and S. Unsworth. 2014. “Politically Smart, Locally building-and-Service-Delivery.pdf. Led Development.” ODI Discussion Paper, Overseas Devel- Department of Roads, Nepal. 2012. “Statistics of Strategic opment Institute, London. https://www.odi.org/sites/odi. Road Network: SSRN 2011/12.” HMIS Unit, Department org.uk/files/odi-assets/publications-opinion-files/9158.pdf. of Roads, Government of Nepal, Kathmandu. Booth, David, Daniel Harris, and Leni Wild. 2016. From Politi- Deshpande, M. 2016. “Comparing and Contrasting Modes of cal Economy Analysis to Doing Development Differently: A Learning Service Delivery.” Unpublished Literature Review. Experience. Overseas Development Institute. https://www. odi.org/publications/10235-political-economy-analysis- Devarajan, S., and S. Khemani. 2016. “If Politics is the Prob- doing-development-differently-learning-experience. lem, How Can External Actors Be Part of the Solution?” Policy Research Working Paper No. 7761, World Bank, Booth, D., T. Dietz, F. Golooba-Mutebi, A. Fuady, D. Henley, Washington, DC. http://documents.worldbank.org/ T. Kelsall, A. Leliveld, and J. Kees van Donge. 2015. Devel- curated/en/906091469456194816/pdf/WPS7761.pdf. opmental Regimes in Africa: Synthesis Report. London: Overseas 148 | Reference List and Bibliography Devarajan, S., S. Khemani, and M. Walton. 2011. “Civil Case Study No. 2, The Asia Foundation and Overseas Society, Public Action and Accountability in Africa.” Pol- Development Institute, London. https://www.odi.org/ icy Research Working Paper No. 5733, World Bank, publications/9118-development-entrepreneurship. Washington, DC. http://elibrary.worldbank.org/doi/ Fielder, J. L. 2000. “The Nepal National Vitamin a Program: abs/10.1596/1813-9450-5733. Prototype to Emulate or Donor Enclave?” Health Policy and Devkota, B., and E. R. Van Teijlingen. 2009. “Politicians in Planning 15 (2): 145–56. Apron: Case Study of Rebel Health Services in Nepal.” Fritz, V., B. Levy, and R. Ort. 2014. Problem-Driven Political Asia-Pacific Journal of Public Health 21 (4): 377–84. Economy Analysis: The World Bank’s Experience. Directions ———. 2015. “Exploring Rebel Health Services during the in Development—Public Sector Governance. Washington, Maoist People’s War in Nepal.” In The Dynamics of Health DC: World Bank. https://openknowledge.worldbank.org/ in Nepal, edited by P. S. Sharada Prasad Wasti. Kathmandu: handle/10986/16389. Himal Books and Social Science Baha. Gambetta, D. 1993. The Sicilian Mafia: The Business of Private DFID (UK Department for International Development) and Protection. Cambridge, Massachusetts: Harvard University World Bank. 2006. “The Institutional Context for Address- Press. ing Exclusion and Establishing Equal Rights for All Citizens Gazdar, H., S. Kaker, and I. Khan. 2010. “Buffer Zone, Colo- of Nepal.” Nepal Gender and Social Exclusion Assessment, nial Enclave or Urban Hub? Quetta: Between Four Regions Unpublished Report. and Two Wars.” Crisis States Working Paper No. 69, Cri- ———. 2009. “Political Economy Analysis How to Note.” sis States Research Center, London. http://eprints.lse. DFID Practice Paper, DFID, London. https://www.odi. ac.uk/28472/1/WP69.2.pdf. org/sites/odi.org.uk/files/odi-assets/publications-opin- Giustozzi, A. 2009. Empires of Mud: War and Warlords of ion-files/5866.pdf. Afghanistan. New York: Columbia University Press. ———. 2010. “Politics of Poverty Elites, Citizens and ———. 2010. Nation-Building is Not for All: The Politics of Edu- States—Findings from Ten Years of DFID Funded Research cation in Afghanistan. Kabul/Berlin: Afghanistan Analysts on Governance and Fragile States 2001–2010: A Synthesis Network. Paper.” DFID, London. https://www.gov.uk/government/ uploads/system/uploads/attachment_data/file/67679/ Giustozzi, A., and C. Franco. 2011. The Battle for the Schools: plcy-pltcs-dfid-rsch-synth-ppr.pdf. The Taleban and State Education. Kabul/Berlin: Afghanistan Analysts Network. ———. 2016. Is DFID Getting Real about Politics? A Stocktake of How DFID Has Adopted a Politically Informed Approach (2010– ———. 2013. The Ongoing Battle for Education: Uprisings, Nego- 2015). London: DFID. tiations and Taleban Tactics. Kabul/Berlin: Afghanistan Ana- lysts Network. Di John, J., and J. Putzel. 2009. “Political Settlements: Issues Paper.” Issues paper, Governance and Social Development Giustozzi, A., and N. Ibrahimi. 2013. “From New Dawn Resource Center. http://www.gsdrc.org/docs/open/ to Quicksand: The Political Economy of Statebuilding EIRS7.pdf. in Afghanistan.” In The Political Economy of Post-Conflict Statebuilding: Power after Peace, edited by Mats Berdal and Dixit, H. 2005. Quest for Health. Kathmandu: Educational Pub- Dominik Zaum. London: Routledge. lishing House. Golooba-Mutebi, F., and S. Hickey. 2013. “Investigating the Engel, J., J. Glennie, S. R. Adhikari, S. W. Bhattarai, D. P. Links between Political Settlements and Inclusive Devel- Prasai, and F. Samuels. 2013. Nepal’s Story: Understanding opment in Uganda: Towards a Research Agenda.” ESID Improvements in Maternal Health. Case Study Report. Lon- Working Paper No. 20, Effective States and Inclusive don: Overseas Development Institute. https://www.odi. Development Research Centre (ESID), Manchester, United org/sites/odi.org.uk/files/odi-assets/publications-opin- Kingdom. http://www.gsdrc.org/document-library/inves- ion-files/8625.pdf. tigating-the-links-between-political-settlements-and-inclu- ESID (Effective States and Inclusive Development). 2016. sive-development-in-uganda-towards-a-research-agenda/. “Research by Programme.” ESID research Centre, Man- Gonzalez De Asis, M., and M. Woolcock. 2015. Operationaliz- chester, United Kingdom. http://www.effective-states.org/ ing the Science of Delivery Agenda to Enhance Development Results. research-by-programme/. Washington, DC: World Bank Group. http://documents. Express Tribune. 2015. “Swabi Polio District Coordinator worldbank.org/curated/en/270011467995372873/ Shot Dead.” Express Tribune, November 30. https://tribune. Operationalizing-the-science-of-delivery-agenda-to-en- com.pk/story/1000980/swabi-polio-district-coordinator- hance-development-results. shot-dead/. Government of Nepal. 2006. Unofficial Translation of the Faustino, J., and D. Booth. 2014. “Development Entrepre- Comprehensive Peace Agreement concluded between the neurship: How Donors and Leaders Can Foster Institu- Government of Nepal and the Communist Party of Nepal tional Change.” Working Politically in Practice Series (Maoist), November 21, 2006. http://www.unmin.org.np. Reference List and Bibliography | 149 Government of Nepal and UN (United Nations) Country Team Hickey, S. 2013. “Thinking about the Politics of Inclusive of Nepal. 2013. Nepal Millennium Development Goals Progress Development: Towards a Relational Approach.” ESID Report 2013. http://www.np.undp.org/content/nepal/en/ Working Paper No. 1, Effective States and Inclusive Devel- home/library/mdg/mdg-progress-report-2013.html. opment Research Centre, University of Manchester, Man- chester, United Kingdom. http://www.effective-states. Grindle, M. 2007. “Good Enough Governance Revisited.” org/wp-content/uploads/working_papers/final-pdfs/ Development Policy Review 25 (5): 553–74. esid_wp_01_hickey.pdf. Haar, R., and L. Rubenstein. 2012. Health in Post-Conflict and Hudson, A. 2016. “Politics Matters, So What? Time for Bigger Fragile States. Special Report, United States Institute of Peace, Bets (and More Learning) on Adaptive Programming.” Global Washington, DC. http://www.usip.org/sites/default/files/ Integrity, July 26. http://www.globalintegrity.org/2016/07/ SR_301.pdf. politics-matters-time-bigger-bets-learning-adaptive-pro- Haber, S., A. Razo, and N. Maurer. 2003. The Politics of Property gramming/. Rights: Political Instability, Credible Commitments, and Economic Hudson, D., H. Marquette, and S. Waldock. 2016. “Every- Growth in Mexico, 1876–929. Cambridge: Cambridge Uni- day Political Analysis.” Development Leadership Program versity Press. (DLP), University of Birmingham, United Kingdom. Halloran, B. 2014. “Studying the Politics of Accountability.” http://publications.dlprog.org/EPA.pdf. Politics, Governance and Development. https://politics- Human Rights Watch. 2010. “Their Future Is at Stake: Attacks governancedevelopment.wordpress.com/2014/09/04/ on Teachers and Schools in Pakistan’s Balochistan Province.” studying-the-politics-of-accountability/. Human Rights Watch, December 13. https://www.hrw. Harris, D., and D. Booth. 2013. “Applied Political Economy org/report/2010/12/13/their-future-stake/attacks-teach- Analysis: Five Practical Issues.” Overseas Development ers-and-schools-pakistans-balochistan-province. Institute, London. https://www.odi.org/sites/odi.org.uk/ IEG (Independent Evaluation Group). 2010. ICR Review— files/odi-assets/publications-opinion-files/8193.pdf. Afghanistan Education Quality Improvement Program. Report Harris, D., and L. Wild. 2013. Finding Solutions: Making Sense Number: ICRR13264. Washington, DC: World Bank. of the Politics of Service Delivery. London: Overseas Devel- ———. 2013. World Bank Group Assistance to Low-Income Frag- opment Institute. https://www.odi.org/sites/odi.org.uk/ ile and Conflict-Affected States. Main Report. Washington, files/odi-assets/publications-opinion-files/8331.pdf. DC: World Bank Group. https://ieg.worldbankgroup.org/ Harris, D., C. Mcloughlin, and L. Wild. 2013. The Technical is Data/reports/chapters/fcs_eval.pdf. Political—Why Understanding the Political Implications of Tech- ———. 2016. “World Bank Group Engagement in Situ- nical Characteristics Can Help Improve Service Delivery. Lon- ations of Fragility, Conflict, and Violence—An Inde- don: Overseas Development Institute. https://www.odi. pendent Evaluation.” World Bank Group, Washington, org/sites/odi.org.uk/files/odi-assets/publications-opin- DC. https://ieg.worldbankgroup.org/evaluations/ ion-files/8328.pdf. fragility-conflict-violence. Harris, D., J. Wales, H. Jones, T. Rana, and R. Lal Chitrakar. IMF (International Monetary Fund). 2003. Nepal: Poverty 2013. “Human Resources for Health in Nepal: The Politics Reduction Strategy Paper 2002–2007. IMF Country Report of Access in Remote Areas.” Overseas Development Insti- No. 03/305. Washington, DC: IMF. http://www.imf.org/ tute, London. https://www.odi.org/sites/odi.org.uk/files/ external/pubs/ft/scr/2003/cr03305.pdf. odi-assets/publications-opinion-files/8329.pdf. Inlogos. 2009. Assessment of Village Development Committee Gover- Hart, J. 2001. Conflict in Nepal and Its Impact on Children. Oxford: nance and the Use of Block Grants. Kathmandu: Ministry of Local Refugee Studies Centre, Oxford University. Development and United Nations Development Programme. Hausmann, R. 2016. “The Problem with Evidence-Based Pol- http://www.np.undp.org/content/dam/nepal/docs/ icies.” Project Syndicate, February 25. https://www.project- reports/governance/UNDP_NP_Report%20on%20 syndicate.org/commentary/evidence-based-policy-prob- Assessment%20of%20Village%20Development%20 lems-by-ricardo-hausmann-2016-02?barrier=accessreg. Committee%20Governance%20and%20the%20Use% 20of%20the%20Block%20Grants.pdf. HEART (Health and Education Advice and Resource Team). 2013. “Nepal Health Sector Programme II (NHSP II): Islamic Republic of Afghanistan, Ministry of Economy, and Mid-Term Review.” http://www.heart-resources.org/ World Bank. 2015. “Provincial Briefs.” World Bank, Wash- assignment/nepal-health-sector-programme-ii-nhsp-ii- ington, DC. mid-term-review/. Jackson, A., and A. Giustozzi. 2012. The Other Side: Humanitar- Health Sector Reform Unit. 2007. “Second Long Term Health ian Engagement with the Taliban in Afghanistan. London: Over- Plan (SLTHP) 1997–2017.” Health Sector Reform Unit, seas Development Institute. Ministry of Health and Population, Government of Nepal, Kalyvas, S. 2003. “The Ontology of Political Violence: Action Kathmandu, Nepal. http://www.mohp.gov.np/images/ and Identity in Civil Wars.” Perspectives on Politics 1 (3): pdf/policy/Second-Long-Term-Health-Plan.pdf. 475–94. 150 | Reference List and Bibliography Kelsall, T. 2016. “Thinking and Working with Political Settle- Leftwich, A., and C. Wheeler. 2011. Politics, Leadership and ments.” Briefing, Overseas Development Institute, London. Coalitions in Development: Findings, Insights and Guidance from https://www.odi.org/sites/odi.org.uk/files/odi-assets/ the DLP’s first Research and Policy Workshop. Frankfurt 10–11 publications-opinion-files/10185.pdf. March 2011. A Research and Policy Workshop Report. Frankfurt, Germany: Development Leadership Program. Kelsall, T., T. Hart, and E. Laws. 2016. “Political Settlements http://publications.dlprog.org/Politics,%20Leader- and Pathways to Universal Health Coverage.” Working ship%20and%20Coalitions%20in%20Development%20 Paper No. 432, Overseas Development Institute, London. -%20Findings,%20insights%20and%20guidance.pdf. https://www.odi.org/sites/odi.org.uk/files/resource-doc- uments/10382.pdf. Lehmann, U., and D. Sanders. 2007. Community Health Workers: What Do We Know about Them? The State of the Evidence on Pro- Khan, M. 2010. “Political Settlements and the Governance grammes, Activities, Costs an Impact on Health Outcomes of Using of Growth-Enhancing Institutions.” Research Paper Series Community Health Workers. Geneva: World Health Organiza- on Governance for Growth, School of Oriental and Afri- tion. http://www.hrhresourcecenter.org/node/1587. can Studies, University of London. http://eprints.soas. ac.uk/9968/1/Political_Settlements_internet.pdf. Levy, B. 2014. Working with the Grain: Integrating Governance and Growth in Development Strategies. Oxford: Oxford University King, G., R. Keohane, and S. Verba. 1994. Designing Social Press. Inquiry: Scientific Inference in Qualitative Research. Princeton, NJ: Princeton University Press. Levy, B., and M. Walton. 2013. “Institutions, Incentives and Service Provision: Bringing Politics Back In.” ESID Kosack, S. 2012. The Education of Nations: How the Political Orga- Working Paper No. 18, Effective States and Inclusive nization of the Poor, Not Democracy, Led Governments to Invest in Development Research Centre (ESID), University of Mass Education. New York: Oxford University Press. https:// Manchester, Manchester, United Kingdom. http://ssrn. global.oup.com/academic/product/the-education-of-na- com/abstract=2386655; http://dx.doi.org/10.2139/ tions-9780199841677?cc=ca&lang=en&. ssrn.2386655. Lahire, N. 2016. Afghanistan—Second Education Quality Improve- Marc, A., A. Willman, G. Aslam, M. Rebosio, and K. Balasur- ment Program: P106259— Implementation Status Results Report: iya. 2013. Societal Dynamics and Fragility: Engaging Societies in Sequence 17. Washington, DC: World Bank. http://docu- Responding to Fragile Situations. New Frontiers of Social Policy. ments.worldbank.org/curated/en/2016/01/25731837/ Washington, DC: World Bank. https://openknowledge. afghanistan-afghanistan-second-education-quality-im- worldbank.org/handle/10986/12222. provement-program-p106259-implementation-status-re- sults-report-sequence-1. Marc et al. 2016. “Addressing Violence with Development Interventions; Selected Lessons from Africa and MENA.” Lasse C. N., and S. M. Shrestha. 2004. Social Assessment of the Project concept note review of unpublished internal con- Nepal Health Sector Reform. Report. Washington, DC: World cept note, World Bank, Washington, DC. Bank. McGuinness, K., ed. 2012. “Local First: Development for Laws, E. 2013. “Governance, Politics and Development in the the Twenty-First Century.” Peace Direct, February 2. https:// Pacific.” Background Paper No. 14, Development Lead- www.peacedirect.org/us/publications/local-first- ership Program (DLP), Birmingham, United Kingdom. development-twenty-first-century/. http://www.dlprog.org/publications/governance-poli- tics-and-development-in-the-pacific-a-literature-review. Mcloughlin, C. 2014. “When Does Service Delivery Improve php. the Legitimacy of a Fragile or Conflict-Affected State?” Governance 28 (3): 341–56. http://www.gsdrc.org/ ———. 2015. “What Do We Do on Monday? Political Set- document-library/when-does-service-delivery-improve- tlements in Theory and Practice.” Opinion, Development the-legitimacy-of-a-fragile-or-conflict-affected-state/. Leadership Program (DLP), Birmingham, United Kingdom. http://www.dlprog.org/opinions/what-do-we-do-on- Mcloughlin, C., and R. Batley. 2012. “The Politics of What monday-political-settlements-in-theory-and-practice.php. Works in Service Delivery: An Evidence-Based Review.” ESID Working Paper No. 6, Effective States and Inclusive Leftwich, A. 2011. “Thinking and Working Politically: What Development Research Centre, University of Manchester, Does It Mean, Why Is It Important and How Do You Do Manchester, United Kingdom. http://www.effective-states. It?” Workshop paper, Politics, Leadership and Coalitions in org/wp-content/uploads/working_papers/final-pdfs/ Development: Policy Implications of the DLP Research Evi- esid_wp_06_mcloughlin-batley.pdf. dence, DLP Research and Policy Workshop, Development Leadership Program, Frankfurt, Germany. http://www. Menocal, A. R. 2014. Getting Real About Politics: From Thinking gsdrc.org/document-library/thinking-and-working-politi- Politically to Working Differently. London: Overseas Devel- cally-what-does-it-mean-why-is-it-important-and-how-do- opment Institute. https://www.odi.org/sites/odi.org.uk/ you-do-it/. files/odi-assets/publications-opinion-files/8887.pdf. Reference List and Bibliography | 151 Mercy Corps. 2015. “Study of Humanitarian Access.” Unpub- Nepal Ministry of Health. 2004. “Nepal Health Sector Pro- lished study, Kabul, Afghanistan. gramme—Implementation Plan, 2004–2009.” Govern- ment of Nepal, Kathmandu, Nepal. http://dohs.gov.np/ Ministry of Education (Kabul). 2013. EQUIP Semi-Annual wp-content/uploads/2014/04/NHSP_IP.pdf. Report: July-December 2013. Kabul, Afghanistan: Ministry of Education. NHSP (Nepal Health Sector Programme). 2008. “Aide Mem- oire.” Eighth Joint Annual Review (JAR), Nepal Health Sec- Ministry of Health (Nepal). 2004. “Health Sector Strategy: An tor Programme. Agenda for Reform, 2003.” Ministry of Health, Govern- ment of Nepal, Kathmandu, Nepal. http://www.ministeri- Norland, R. 2012. “After Years of Decline, Polio Cases in al-leadership.org/sites/default/files/resources_and_tools/ Afghanistan Triple in a Year.” The New York Times, Janu- Health%20Sector%20Strategy_An%20Agenda%20for%20 ary 17. http://www.nytimes.com/2012/01/18/world/ Reform.pdf. asia/after-years-of-decline-polio-cases-in-afghanistan-rise. html?_r=0. Mohmand, S. 2012. “Ignore Informal Institutions at Your Own Risk.” IDS Governance and Development (blog), Janu- North, D., J. Wallis, and B. Weingast. 2009. Violence and Social ary 12, Institute of Development Studies, United King- Orders: A Conceptual Framework for Interpreting Recorded Human dom. http://www.governanceanddevelopment.com/2012/ History. New York: Cambridge University Press. 06/ignore-informal-institutions-at-your.html?utm_ North, D., J. Wallis, S. Webb, and B. Weingast. 2007. “Limited source=feedburner&utm_medium=feed&utm_campaign Access Orders in the Developing World: A New Approach =Feed%3A+IDSGovblog+%28Governance+IDS+ to the Problems of Development.” Policy Research Work- research+cluster+blog%29. ing Paper, World Bank, Washington, DC. http://dx.doi. MOHP (Ministry of Health and Population), Nepal. 1991. org/10.1596/1813-9450-4359. “National Health Policy.” Government of Nepal, Kath- ———. 2011. “Limited Access Orders in the Developing mandu, Nepal. World: Rethinking the Problems of Development and ———. 2010. “Nepal Health Sector Program Implementation Violence.” https://web.stanford.edu/group/mcnollgast/ Plan II (NHSP-IP 2) 2010–2015.” Government of Nepal, cgi-bin/wordpress/wp-content/uploads/2013/10/Lim- Kathmandu, Nepal. ited_Access_Orders_in_DW_-II_-2011.0125.submis- sion-version.pdf. MoHP (Ministry of Health and Population, Nepal), New ERA, and Macro International Inc. 2007. “Nepal Demographic OECD (Organisation for Economic Co-operation and and Health Survey 2006.” Ministry of Health and Popula- Development). 2015. States of Fragility 2015: Meeting Post- tion (Nepal), New ERA (Nepal), and Macro International 2015 Ambitions. Paris: OECD Publishing. http://dx.doi. Inc. (United States). http://dhsprogram.com/pubs/pdf/ org/10.1787/9789264227699-en. fr191/fr191.pdf. Olson, M. 1993. “Dictatorship, Democracy, and Development.” Moore, M. 2014. “How Well Does the World Bank Serve The American Political Science Review 87 (3): 567–76. www. Fragile and Conflict-Affected States?” Fragile States Resource jstor.org/stable/2938736. Center. http://www.fragilestates.org/2014/02/05/well- O’Meally, S. C. 2013. “Mapping Context for Social Account- world-bank-serve-fragile-conflict-affected-states/. ability: A Resource Paper.” Social Development Depart- Moore, M., and S. Unsworth. 2010. “An Upside-Down View ment, World Bank, Washington, DC. http://documents. of Governance.” Open Democracy, March 31. https://www. worldbank.org/curated/en/2013/01/17985823/ o p e n d e m o c ra c y. n e t / m i ck - m o o re - s u e - u n swo r t h / mapping-context-social-accountability-resource-paper. upside-down-view-of-governance. Parks, T., and W. Cole. 2010. “Political Settlements: Implica- National Planning Commission. 2003. “Tenth Five-Year Plan tions for International Development Policy and Practice.” (2002–07).” Government of Nepal, Kathmandu, Nepal. Occasional Paper No. 2, The Asia Foundation, San Fran- http://siteresources.worldbank.org/INTPRS1/Resources/ cisco, CA. http://asiafoundation.org/resources/pdfs/Polit- Country-Papers-and-JSAs/Nepal_PRSP.pdf. icalSettlementsFINAL.pdf. ———. 2007. “Three-Year Interim Plan (2007/08– Parks, T., N. Colletta, and B. Oppenheim. 2013. The Contested 2009/10).” Government of Nepal, Kathmandu, Nepal. Corners of Asia: Subnational Conflict and International Devel- http://planipolis.iiep.unesco.org/upload/Nepal/Nepal_ opment Assistance. San Fransisco, CA: The Asia Foundation. TYIPEnglish.pdf. http://asiafoundation.org/resources/pdfs/Contested- CornersOfAsia.pdf. NEAT (Nepal Evaluation and Assessment Team) and RECPHEC (Resource Center for Primary Health Care). PIPS (Pak Institute for Peace Studies). 2012. Conflict and Insecu- 2011. Field Study on Essential Health Care and Free Maternity rity in Balochistan. Islamabad: Narratives Publication. Services in Nepal: Final Study Report. Kathmandu: NEAT and RECPHEC. 152 | Reference List and Bibliography Pradhan, A., R. H. Aryal, G. Regmi, B. Ban, and P. Govin- Unsworth, S., ed. 2010. An Upside Down View of Governance. dasamy. 1997. Nepal Family Health Survey 1996. Kathmandu: Brighton: Centre for the Future State, Institute of Devel- Ministry of Health and Population, New ERA, and Macro opment Studies. http://www2.ids.ac.uk/futurestate/pdfs/ International Inc. AnUpside-downViewofGovernance.pdf. Putzel, J., and J. Di John. 2012. Meeting the Challenges of Cri- ——. 2015. “It’s the Politics! Can Donors Rise to the Chal- sis States. Crisis States Research Centre Report. London, United lenge” in A Governance Practitioner’s Notebook: Alternative Kingdom: London School of Economics and Political Sci- Ideas and Approaches.” OECD. https://www.oecd.org/dac/ ence. http://eprints.lse.ac.uk/45936/. accountable-effective-institutions/Governance%20Note- book%201.2%20Unsworth.pdf. Roberts, L. 2012. “Fighting Polio in Pakistan.” Science 337 (6094): 517–21. http://science.sciencemag.org/content/ Valters, C. 2016. “Learning and Adaptation: 6 Pitfalls to 337/6094/517. Avoid.” Devex, April, 20. https://www.devex.com/news/ learning-and-adaptation-6-pitfalls-to-avoid-88032. Rodrik, D. 2014. “When Ideas Trump Interests: Preferences, Worldviews, and Policy Innovations.” Journal of Economic Whaites, A. et al. 2015. A Governance Practitioner’s Notebook: Alter- Perspectives 28 (1): 189–208. https://www.aeaweb.org/ native Ideas and Approaches. Paris: OECD. http://www.oecd. articles?id=10.1257/jep.28.1.189. org/dac/governance-peace/governance/governance-practi- tioners-notebook.htm. Schmidt, A. 2009. “Health Aid Effectiveness in Nepal: Paris, Accra, Civil Society and the Poor.” Action for Global Wild, L., D. Booth, C. Cummings, M. Foresti, and J. Wales. Health. http://www.actionforglobalhealth.eu/uploads/ 2015. Adapting Development: Improving Services to the Poor. media/Aid_effectiveness_in_Nepal_Final_01.pdf. London: Overseas Development Institute. https://www.odi. org/sites/odi.org.uk/files/odi-assets/publications-opin- Seddon, D., and K. Hussein. 2002. The Consequences of Con- ion-files/9437.pdf. flict: Livelihoods and Development in Nepal. London: Overseas Development Institute. Wood, E. 2000. Forging Democracy from Below: Insurgent Tran- sitions in South Africa and El Salvador. New York: Cambridge Sen, A. 1999. Development as Freedom. New York: Knopf. University Press. Sharma, Sharan P. 2010. “Politics and Corruption Mar Health Wong, S. 2012. “What Have Been the Impacts of World Bank Care in Nepal.” The Lancet 375 (9731): 2063–64. Community-Driven Development Programs? CDD Impact Slater, R. 2016. “Service Delivery and State Legitimacy.” World Evaluation Review and Operational and Research Implica- Bank Fragility Forum Presentation, March 2, 2016, Over- tions.” Working Paper, World Bank, Washington, DC. http:// seas Development Institute, SLRC. documents.worldbank.org/curated/en/967431468161087 566/What-have-been-the-impacts-of-World-Bank- Thapa, D. 2011. World Development Report 2011, Background Case Community-Driven-Development-Programs-CDD- Study: Nepal. Washington, DC: World Bank. impact-evaluation-review-and-operational-and-research- Thapa, D., and B. Sijapati. 2005. A Kingdom under Siege: Nepal’s implications. Maoist Insurgency, 1996–2004. London: Zed Books. World Bank. 2003. World Development Report 2004: Making Tilly, C. 1990. Coercion, Capital and European States, AD 990– Services Work for Poor People—Overview. Washington, DC: 1992. Cambridge, MA and Oxford, United Kingdom: World Bank. http://documents.worldbank.org/curated/ Blackwell. en/527371468166770790/World-Development-Report- 2004-Making-services-work-for-poor-people-Overview. Thomas, A. 1998. “Challenging Cases.” In Finding Out Fast: Investigative Skills for Policy and Development, edited by A. ———. 2004. “Nepal—Health Sector Program Proj- Thomas, J. Chataway, and M. Wuyts, 307–32. London: Sage ect.” World Bank, Washington, DC. http://documents. Publications. worldbank.org/curated/en/2004/08/5057016/ nepal-health-sector-program-project. UN (United Nations). 2015. “Sustainable Development Knowledge Platform.” https://sustainabledevelopment. ———. 2005a. “Nepal—Rural Access Improvement and Decen- un.org/?menu=1300. tralization Project.” World Bank, Washington, DC. http://doc- uments.worldbank.org/curated/en/2005/05/5841647/ UN Secretary General. 2015. “Secretary-General’s Remarks nepal-rural-access-improvement-decentralization-project. at International Stakeholders’ Roundtable Meeting for the Middle East and North Africa Region.” October 10, ———. 2005b. “Pakistan—Second Partnership for Polio Erad- Lima, Peru. https://www.un.org/sg/en/content/sg/state- ication Project.” World Bank, Washington, DC. http://doc- ment/2015-10-10/secretary-general%E2%80%99s-re- uments.worldbank.org/curated/en/2005/12/6519233/ marks-international-stakeholders%E2%80%99-roundtable. pakistan-second-partnership-polio-eradication-project. Reference List and Bibliography | 153 ———. 2006. “Pakistan—Balochistan Education Support ———. 2016b. “Setting the Agenda for IDA 18: Strategic Project.” World Bank, Washington, DC. http://docu- Directions.” IDA 18, IDA Resource Mobilization Depart- ments.worldbank.org/curated/en/2006/05/6837694/ ment, International Development Association, World Bank pakistan-balochistan-education-support-project. Group, Washington, DC. http://www-wds.worldbank. org/external/default/WDSContentServer/WDSP/IB/201 ———. 2009. “Nepal—Additional Financing for and 6/03/08/090224b0841e212f/1_0/Rendered/PDF/Set- Restructuring of Rural Access Improvement and Decentral- ting0the0ag0strategic0directions.pdf. ization Project.” World Bank, Washington, DC. http://doc- uments.worldbank.org/curated/en/2009/11/11415532/ ———. 2016c. “Remarks by World Bank Group President nepal-additional-financing-restructuring-rural-access- Jim Yong Kim at the Fragility Forum 2016.” World Bank, improvement-decentralization-project. Washington, DC. http://www.worldbank.org/en/news/ speech/2016/03/01/remarks-world-bank-group-presi- ———. 2010. “Nepal—Second Health Nutrition and Popu- dent-jim-yong-kim-fragility-forum-2016. lation (HNP) and HIV/AIDS Project.” World Bank, Wash- ington, DC. http://documents.worldbank.org/curated/ ———. 2016d. International Development Association (IDA) 18: en/2010/03/12058938/nepal-second-health-nutri- Special Theme—Jobs and Economic Transformation. Washing- tion-population-hnp-hivaids-project. ton, DC: World Bank. http://documents.worldbank.org/ curated/en/500371467996716841/International-Devel- ———. 2011. World Development Report 2011: Conflict, Security opment-Association-IDA-18-special-theme-jobs-and-eco- and Development. Washington, DC: World Bank. https:// nomic-transformation. openknowledge.worldbank.org/handle/10986/4389. ———. 2017. World Development Report 2017: Governance and ———. 2014. “Afghanistan—Strengthening Health Activ- the Law. Washington, DC: World Bank. https://openknowl- ities for the Rural Poor Project.” World Bank, Washing- edge.worldbank.org/handle/10986/25880. ton, DC. http://documents.worldbank.org/curated/ en/2014/02/19391833/afghanistan-strengthen- World Bank and Akram 2016. “The Political Economy of ing-health-activities-rural-poor-project. Social Service Delivery in Violence-Affected Regions of Pakistan.” Unpublished Background Country Report, ———. 2015a. “Pakistan—Balochistan Education Support World Bank, Washington, DC. Project.” World Bank, Washington, DC. http://docu- ments.worldbank.org/curated/en/2015/01/23994480/ World Bank and AREU (Afghanistan Research and Evalua- pakistan-balochistan-education-support-project. tion Unit). 2016. “Beyond Capacity and Technical Inputs: The Politics of Education and Health Service Delivery in ———. 2015b. Pakistan—“Promoting Girls’ Education Afghanistan.” Unpublished Background Country Report, in Balochistan Project.” World Bank, Washington, DC. World Bank, Washington, DC. http://documents.worldbank.org/curated/en/2015/ 12/25737501/pakistan-promoting-girls-education- World Bank and SSB (Social Science Baha). 2016. “The Polit- balochistan-project. ical Economy of Basic Health and Rural Infrastructure Delivery in Nepal.” Unpublished Background Country ———. 2015c. “Pakistan—Third Partnership for Polio Eradi- Report, World Bank, Washington, DC. cation Project.” World Bank, Washington, DC. http://doc- uments.worldbank.org/curated/en/2015/09/24992011/ Yanguas,P.2015.“WhatComesAfterAlltheShoutingAboutPolitics pakistan-third-partnership-polio-eradication-project. and Aid?” (Blog). Effective States and Inclusive Development Research Centre (ESID), University of Manchester, Man- ]———. 2015d. “Afghanistan—Support to Basic Package chester, United Kingdom. http://www.effective-states.org/ of Health Services (BPHS) Project.” World Bank, Wash- what-comes-after-all-the-shouting-about-politics-and-aid/. ington, DC. http://documents.worldbank.org/curated/ en/332281468190763772/Afghanistan-Support-to-Ba- ———. 2016. “The Role and Responsibility of Foreign Aid in sic-Package-of-Health-Services-BPHS-Project. Recipient Political Settlements.” ESID Working Paper No. 56, Effective States and Inclusive Development Research ———. 2016a. “New Joint Initiative to Mobilize Additional Centre, University of Manchester, Manchester, United King- Support for Refugees, Recovery, and Reconstruction in dom. http://www.effective-states.org/working-paper-56/. the Middle East and North Africa.” Press Release, Lima, October 10, 2015. http://www.worldbank.org/en/news/ press-release/2015/10/10/new-joint-initiative-to-mobi- lize-additional-support-for-refugees-in-the-middle-east- and-north-africa. 154 | Reference List and Bibliography