WPS8435 Policy Research Working Paper 8435 Community-Driven Development Myths and Realities Susan Wong Scott Guggenheim Social, Urban, Rural and Resilience Global Practice May 2018 Policy Research Working Paper 8435 Abstract Community-driven development is an approach to devel- for improving the lives of the poor. The empirical evidence opment that emphasizes community control over planning from evaluations confirms that community-driven develop- decisions and investment resources. Over the past decade, ment programs provide much needed productive economic it has become a key operational strategy for many national infrastructure and services at large scale, reasonable cost, governments, as well as for international aid agencies, with and high quality. They also provide villagers, especially the the World Bank alone currently supporting more than disadvantaged, with a voice in how development funds are 190 active community-driven development projects in 78 used to improve their welfare. Second, community-driven countries. Community-driven development programs have development programs are not a homogeneous category, proven to be particularly useful where government institu- and it is important to acknowledge the differences between tions are weak or under stress. This paper examines what the national, on-budget, multi-year programs, and off-budget evidence shows about the utility of community-driven devel- programs. And finally, community-driven development opment programs for helping governments improve the works best and achieves the greatest results when it is part lives and futures of the poor. The paper also addresses recent of a broader development strategy that includes reforms to critiques of the community-driven development approach. governance, investments in productivity, and integration The paper makes three main arguments. First, communi- with efforts to improve the quality of public service delivery. ty-driven development offers governments a useful new tool This paper is a product of the Social, Urban, Rural and Resilience Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/research. The authors may be contacted at swong1@worldbank.org and guggenheim1955@gmail.com. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Community-Driven Development: Myths and Realities Susan Wong and Scott Guggenheim JEL Classification: C93, D71,D74, H41, I31 Keywords: community-driven development, local governance, citizen engagement, public service delivery, rural infrastructure, impact evaluations, poverty reduction Community-Driven Development: Myths and Realities Susan Wong and Scott Guggenheim1 I. Introduction Community-driven development (CDD) is an approach to development that emphasizes community control over planning decisions and investment resources. Over the past decade it has become a key operational strategy for many national governments, as well as for international aid agencies, and involves many individual programs that cover thousands of villages and has cost billions of dollars. CDD programs can be found working across a broad spectrum of developing country environments, from emergency response programs that follow on from natural disasters and armed conflicts, to programs in middle-income countries that are used to close gaps in basic, small-scale infrastructure and that target national programs of social assistance. CDD is not a new concept. In fact, the concept of community as a self-governing political entity can be traced back directly to Aristotle’s Politics (see for example, Uphoff et al. 1972; Schartz 1978). However, even without straining the historical genealogy, recent articles have separately traced the current interest in CDD back to Gandhi’s independence and swaraj movements in India, Magsaysay’s counter-insurgency program in the Philippines, and various local development program ideas supported by the Rockefeller and Ford Foundations in Latin America. However, the modern roots of CDD for international development agencies lie, first, in the writings of social scientists such as Robert Putnam, James Coleman, and Pierre Bourdieu (Putnam 1993, 2000; Coleman, 1988; Bourdieu 1986), who showed how the historical development of social and cultural institutions could explain patterns of 1 Susan Wong is the CDD Global Lead Specialist/Lead Social Development Specialist at the World Bank. Scott Guggenheim is the CDD/Poverty Advisor with the Government of Afghanistan. The views expressed herein are those of the authors and should not be attributed to the Government of Afghanistan, the World Bank, its executive directors, or the countries they represent. The authors are grateful to J. Victor Bottini, Sean Bradley, Grant Follett, Mai Linh Huynh, Nik Myint, and Robert Wrobel for their insightful comments on this paper. 2    cooperation and development; and second in the World Bank’s influential 2004 World Development Report (WDR) “Making Services Work for Poor People.” The 2004 WDR proposed that in addition to traditional development strategies that relied on political (elections) and managerial mechanisms to hold government service-delivery agencies accountable – what it called “the long route of accountability” -- there was also a “short route” that could generate accountability for service delivery by giving client groups much greater control over information, resources, and choice. Several of the WDR’s examples came from the CDD literature, which in turn then began to draw on the WDR model to provide a theoretical foundation to justify its continued expansion. As even its authors agreed at the time, the 2004 report rested on very little empirical evidence. Since the time of its publication, however, the number of case examples and the corpus of rigorous evaluations have grown considerably. Not surprisingly, not all of the model has worked out as neatly as originally expressed, nor does all of the evidence point in a unilinear direction. Nevertheless, many of its original hypotheses about the role that information, choice, and programs can play in improving local-level service delivery have proven to be useful and also are able to be cast in operational terms. Our goal in this paper is to examine what the evidence shows us about the utility of CDD programs for helping governments improve the lives and futures of the poor. We also address recent critiques of the CDD approach. As both proponents and managers of national community development programs, we are firm believers that CDD offers governments a useful new tool for improving the lives of the poor. However, CDD is not a fix-all tool that can be applied indiscriminately to all contexts, nor does it in any way replace the need for the kinds of structural transformations in developing countries that will create new industrial and service jobs, provide technical services that lie well beyond the communities’ purview, and promote other large-scale improvements to human welfare. CDD is a useful tool in a people-centric development strategy. The challenge is to avoid putting the cart before the horse: the key insight from the CDD experience is that poor people’s agency can drive development much more than it currently does, not that CDD should replace sectoral or transformational programs. But in contexts where more traditional approaches have not been able to reach the poor, having a new approach that developing country governments can use to engage communities that are poor and often hard to reach, and in ways that are popular, sustainable, and effective, is already a valuable contribution. 3    II. Community-Driven Development’s Conceptual Origins Large-scale community development programs usually rest on some combination of two general arguments. The first is an efficiency argument. In many if not most developing countries, the need for basic infrastructure, such as clean water, farm-to-market roads, or irrigation distribution channels, is still extremely large. The poverty benefits from transferring funds directly to large numbers of communities will result in larger quantities of basic development infrastructure being built at lesser cost and at greater speed than would occur using more traditional routes. This argument justifies the use of public investment funds by international and national development agencies. The second argument is that community development programs are very popular with communities, allowing them to have greater participation in decision-making which affects their lives. Engaging poor people as subjects rather than objects of development will build trust in government’s ability to deliver much needed services. Not surprisingly, this argument resonates best with political leaders. But the growing concern with the political and economic fallout from rising rates of global inequality and distrust in state institutions is generating increasing interest among a broad range of players in finding new ways to generate inclusive development, rather than trickle down development. If large-scale public support for community development derives from its overall impact in reducing poverty and providing political benefits, there are also three pragmatic reasons why community development is increasingly popular in development circles. The first is that communities almost always have more knowledge about local circumstances and therefore can set priorities and produce more appropriate designs than can centralized planners. Second, communities have more incentives and a greater ability to use funds efficiently and to stop corruption. Third, most developing country governments have such large development needs that enlisting thousands of communities as “partners” who can work in parallel will let them close infrastructure and service delivery gaps more efficiently than by straining to build comprehensive government delivery systems that attempt to do it all. The first practical argument for CDD states simply that communities are the best judges of how their lives and livelihoods can be improved through localized public investments. If given adequate support, resources, and access to information, communities can organize themselves to provide for their immediate needs. This is because communities will have the most knowledge about local conditions, and as long as the technologies involved are not particularly complicated, the communities can obtain most needed resources from local suppliers and 4    sources of expertise. Most disputes can also be resolved locally provided that local leaders have been properly engaged and support the program. As compared with centrally-led, traditional top-down approaches to local development, community-driven approaches should be more effective because they are suitable to local realities. Unfortunately, just as communities have the practical knowledge and livelihood- driven incentives for building small-scale infrastructure, large ministries and their contractors will often have the wrong incentives. The development evaluation literature is replete with examples of large, potentially beneficial infrastructure investments, such as irrigation dams, that never managed to complete the distribution works needed so that poor farmers at the tail end of the canals could also benefit from the water. All too often, finishing small channels and forming water-user groups was just an extra expense to contractors that would go unnoticed if left out of planning or execution. By contrast, communities will usually have the greatest incentives to use funds properly as well as having the greatest ability to monitor their use. Formal government administrations famously suffer from “principal-agent” problems – the fact that while in theory civil servants carry out policies and programs on behalf of society, in actual practice government officials have their own interests, incentives, and ability to divert funds away from intended uses. Furthermore, all too often, introducing layers of monitoring to check that funds went in the right direction, or that contractors built that “last mile” needed to reach the poorest communities, just adds another layer of an “agent” who can similarly demand a share and will in any case have few incentives to report theft or unfinished work by a fellow civil servant. Communities will not be immune to such problems. However, the fact that community members are themselves usually the end beneficiaries not only gives them incentives to invest well, but also the incentive to monitor that their neighbors are not diverting funds to other uses. Social pressure, local norms, and community leadership become ways to respond to the knowledge that someone is shirking their responsibilities by then applying sanctions and pressures that can bring about change much more quickly than would waiting for action on formal monitoring or audit findings. The second argument for CDD programs is that mobilizing communities to be the active agents of development programs rather than just their “targets" or “beneficiaries” reduces the burden on government institutions, freeing them up to concentrate on larger, technically, and financially more challenging investments. As scholars such as Pritchett, 5    Woolcock, and Andrews have argued (Pritchett et al. 2010), most developing country governments do not yet benefit from strong, well-integrated, smoothly functioning administrative institutions. The history of development projects is a history of “premature load-bearing” — the tendency to task fragile institutions with administratively complex responsibilities and then complaining that they failed to deliver. CDD projects reduce that load. They rely on partnerships between governments and communities. By treating the primary obligation of the government as the delivery of a grant to each participating village, community driven projects can cover very large areas, relying in part on skills and resources available in communities, along with some external facilitation assistance. With communities doing much of the planning, purchasing, and management that otherwise would be the responsibility of ministries and local governments, these projects do not require the mobilization of sophisticated technical expertise. In addition, because they are basically modular in that each community or cluster of communities is its own management unit, the projects are usually more robust in the face of local breakdowns than are large infrastructure or social service delivery programs, where a single procurement delay or logistical breakdown can put whole projects at risk. Even if some communities fall behind for one reason or another, others can continue uninterrupted. This makes CDD projects particularly useful for governments trying to cope with conflict or disaster recovery, which is when government services are particularly weak and vulnerable to disruption. Proponents of CDD therefore often argue that community development is not just an important addition to the toolkit that can help policy makers reduce poverty, but that transferring investment funds to communities marks a conceptually different approach to local development. CDD projects are built around “partnership” rather than being technocratically-directed projects that require a large delivery armature of administrators, technical specialists, monitors, and complex management. In that sense, CDD fits within a family of administratively simplified programs, such as cash transfers or direct giving, that rely on end-user management rather than top-down controls, although because CDD involves work with whole communities rather than individuals or families, there are still some important differences between them. Finally, building on a tradition that goes back at least to the 18th century discussions of Alexis de Tocqueville on the ways that giving community members local-level political responsibilities is the way that democratic culture is embedded into a country’s national 6    political culture (Tocqueville 2000), a third argument for CDD is that providing resources to sustain and enrich a community’s associational life will provide a way to smooth the disruptions caused by urbanization and the breakdown of traditional social institutions. Engaged communities are “schools” where citizens learn to take charge of local affairs themselves, to resolve disputes without violence, and to replace locally chosen leaders who do not deliver on their promises. These are important lessons, de Tocqueville argued, for building a foundation on which stable, democratic orders could rest. Had the newly- independent American colonies not had such a strong base in a democratic community culture that demanded accountability of its leaders, de Tocqueville argued, the federal government could very easily have become just another forum for national elites to centralize control. Strengthening community institutions by giving them statutory recognition and providing them with bureaucratic legitimacy improves the quality of their engagement with higher level parts of government. By reducing poor people’s alienation from state institutions, it provides the foundation for a more democratized development than what would otherwise emerge from a strict regime of individualized market transactions. The political argument for community development, pace de Tocqueville, is that new democracies will need to emerge from bottom-up processes that create a democratic base as much as they will benefit from the higher level (and also necessary) features of a democracy, such as elections, parliaments, and advances in the rule of law. This argument about making democracy inclusive is particularly important for discussions with national leaders hoping to stabilize fragile or conflict-affected states, including states where civil unrest is driven by ethnic, religious, and historical factors or by other concerns of state neglect or discrimination. III. Critiques of Community Development Critics have challenged claims that CDD marks an improvement either in terms of improving the quality of outcomes for the poor, or in terms of increasing trust in government. Summing up all of the evaluation evidence rather than just choosing the examples of success, they say, provides a much more mixed picture of overall performance than triumphalist CDD literature would suggest. Critics argue that communities are themselves heterogeneous, and that despite claims of a participatory process, CDD projects are often captured by village 7    elites, and in the end, provide little more space for participation by women or marginal groups than standard projects do. Critics also maintain that a government’s willingness to transfer resources directly to communities does not automatically translate either into more trust in government or to pressure for other domains of government to become more accountable and responsive to community needs. Finally, specialists in public administration and public finance tend to be particularly incensed by a perception that because community development projects usually transfer funds directly from the national treasury to villages rather than through the inter-governmental transfer system, they build up a structure that is separate and apart from normal government systems. This parallel structure can create confusion within government or even undermine efforts to improve the performance of local administrations. This debate is far from being an academic argument. Development agencies such as the World Bank have become strong supporters of community development approaches. The Bank is currently supporting 190 active CDD projects valued at USD 19.2 billion in 78 countries around the world. Over the past ten years, the Bank has lent on average USD 2.6 billion annually towards CDD programs, representing 5 to 10 percent of overall Bank lending each year. Nor is this wave showing any sign of cresting. Countries such as Indonesia or the Philippines have acted to scale up donor-assisted community projects to cover their entire national territories and they now finance them from their national budgets, with Indonesia’s parliament going so far as to embed annual community transfers to its 73,000 villages in its recurrent budget through a village law that currently transfers up to USD 7.3 billion per year. 8    Figure 1: World Bank Supported CDD Programs Fragile and conflict-affected countries such as Afghanistan and Myanmar have also undertaken CDD projects, with the National Solidarity Project in Afghanistan claiming as much as 25 percent of the government’s budget from the Afghanistan Reconstruction Trust Fund to reach the country’s more than 35,000 villages. Myanmar has similarly scaled-up its first-generation program to cover every state and region in the country. Rather than continuing to wrangle over whether CDD is or is not “the answer” to development and poverty reduction, our goal in this paper is to pick through some of these controversies to examine what the evidence tells us about the utility of CDD programs for helping governments improve the lives and futures of the poor. Virtually all of the first rigorous evaluations of CDD were completed within the past five to ten years,2 which means that lessons learned about effectiveness will only now be entering a second generation of designs. 2  As described in the following section, we argue that the design model of Social Funds begun in the early 1990s is quite different from the latest generation of CDD programs begun in the late 1990s.  9    IV. Defining Concepts and Terms The community development literature both benefits and suffers from the fact that many types of CDD projects can look superficially similar but in fact involve very different histories, designs, and objectives. Mansuri and Rao (2012), in their review “Localizing Development: Does Participation Work?” took the eclectic approach and combined as broad a range as possible of analysis of projects making a claim about popular participation. Their sample included community development projects that ranged from one-time, small-scale projects launched and managed by international and national nongovernmental organizations, to large-scale, long-term, nationally financed and managed programs that extend over a country’s entire national territory. This paper takes the opposite tack. While that Mansuri and Rao review produced important insights about the claims being made for participatory approaches as a whole, it paid a methodological cost in terms of drawing clear conclusions based on an analysis of equivalent experiences. The challenge of constructing comparable samples affects any large- scale evaluation, but the literature evaluating CDD programs suffers deeply from the lack of like-to-like comparisons. Rather than examining the entire range of community development, community-based, and participatory programs, this review introduces a number of criteria that narrow the range to a much more limited selection of programs. These will now be explained. First, this review will only cover CDD projects that are sponsored by governments. The reasons for the filter are both theoretical and practical. CDD is by itself hardly new. Nongovernmental organizations such as Oxfam, CARE, Aga Khan, and others have been working directly with poor communities for decades, long before the World Bank ever started thinking about its first community development program. What is new about CDD in this latest phase is not just the scale of these modern programs compared to historical efforts, but the fact that they are sponsored and managed by national governments using their national budgets. This distinction between governmental and nongovernmental sponsorship matters because national government sponsorship embeds these programs in a range of political decisions, bureaucratic functions, and normative relationships between social actors 10    that are qualitatively different from the operating environment for programs that function outside of public administration. Second, CDD projects consist of a direct financial resource transfer to administratively-defined communities that they can use for productive investments. Defining them this way distinguishes CDD programs from Social Funds, which are basically proposal funding windows that require community proposals to be reviewed by higher level authorities before they are approved, and from cash transfer programs that give money directly to individual households. In CDD programs, the money is transferred to a bank account owned or managed by a locally representative community council. A third starting point, although one not intrinsic to the definition of a CDD program, is that CDD projects are usually facilitated by an external agent, such as a nongovernmental organization, a specialist under contract, or an assigned government officer. The scope and role of these facilitators can vary, but their initial purpose follows from the fact that CDD projects are national government programs embedded within the national fiduciary system. Both local officials and communities themselves need to know the “rules of the game” for how government money is to be transferred, managed, and accounted for. Facilitators themselves cannot control how the money is used. Their role is to provide information and to encourage a participatory process of community decision-making. That function can be construed narrowly or broadly, but in most CDD projects, facilitators not only explain the government’s fiduciary rules for the program, but they also serve to help communities access technical expertise, to monitor performance and to help handle complaints. The facilitators also share information with higher levels of government. Having an external facilitator under contract to a higher level of government to some extent also provides a check on potential monopsonistic practices of local governments by opening up a second channel for passing information up the administrative hierarchy. If the conceptual starting point of this discussion is that CDD programs are part of national development strategies and need to be understood as part of how governments promote development, certain distinctions and implications become important that are not always captured in the literature or by practitioners. As Jonathan Fox (Fox 2014) has noted, it matters whether community development projects are seen to be free-standing transfer programs with no links to other programs, as opposed to whether they have been conceived as part of a broader suite of programs intended to reduce poverty, increase political 11    engagements, support decentralization, and so on. Thus, for instance: the Philippines national program was intended to complement its conditional cash transfer programs; Indonesia’s new village law is framed as the last stage of its decades long program of decentralization; Afghanistan’s Citizen’s Charter is meant to include ministry services such as school construction, primary health care, and rural electrification; and so on. While for impact evaluation specialists the critical task is to break projects down into activities that can be individually measured for attribution purposes, for policy makers it is just as important to understand if cross-program complementarities and spatial linkages are contributing to the overall policy goal. Fourth, placing CDD projects within the context of national development strategies means that evaluations should assess the effectiveness of community programs not just against what program officials claim they will be doing, as most evaluations do, but also against what alternative instruments government policy makers have to achieve equivalent results. Numerous studies of development documents point out there are often many incentives for project proponents to claim results beyond what they can realistically deliver. CDD projects are no exception to this pattern and they deserve to be criticized whenever they exaggerate. However, for a government policy maker, the most relevant question is not just whether a project performs as well as its proponents initially claim it will be able to, but how well it performs compared to the other instruments that are available. Thus, it is not entirely surprising that almost 60 percent of the countries in which Bank-supported CDD programs operate are fragile and affected by the contexts of conflict. In such difficult circumstances, where governments may face a legacy of weak capacity and legitimacy, states have limited options to engage communities and deliver reconstruction effectively, and thus often turn to CDD approaches. Fifth, internal factors, such as what are the total and per capita amounts of investment and whether transfers will be one-time grants as opposed to programs that will be predictable and cumulative, are important considerations for how evaluations get designed and interpreted. Very few community projects transfer more than USD 10/capita/year, but even within that constraint there are orders of magnitude differences in both quantity and frequency. To the extent that CDD projects achieve poverty effects by efficiently building economically useful infrastructure, transfer amounts have to be of a size sufficient to build something that matters, even at the village level. Similarly, programs that are predictable can support a village planning process that involves negotiations, prioritization, sequencing, and 12    longer-term partnerships with outside actors such as local governments or nongovernmental organizations (NGO) in a way that one-time or unpredictable grant transfers cannot. V. CDD Trends to Date The previous two sections lay out the conceptual terms and definitions behind community driven development. We now turn to what we see as some of the current trends and evidence that have been accumulated over the past 20 years of implementation. National Scaling-Up of the CDD Approach. In countries as diverse as India, Timor-Leste, and Morocco, there have been national scale-ups of the CDD approach, moving from discreet projects of modest geographical coverage, to national programs that, in some cases, institutionalize CDD principles into the regulatory frameworks and policies under decentralization. These expansions have occurred for example in East Asia (Cambodia, Indonesia, the Lao People’s Democratic Republic, Mongolia, Myanmar, Papua New Guinea, the Philippines, Timor-Leste, and Vietnam); South Asia (Afghanistan, India, Nepal, Pakistan, and Sri Lanka); the Middle East (Iraq, Morocco, and the Republic of Yemen); Eastern and Central Asia (Azerbaijan); and Africa (Benin, Cameroon, Mali, Nigeria, and South Sudan). In addition to World Bank financing, the Bank estimates that an additional USD 13.4 billion or more was raised as complementary financing from governments and other donors on these projects. Why have governments chosen to scale up these programs? Leaders often cite the ability of these programs to deliver much needed services in the form of roads, clean water, schools, microfinance, and other economic infrastructure to large numbers of villages and poor communities in a transparent manner. Especially in fragile and conflict-affected situations, these officials see CDD as a way to build local-level democracy, allowing communities to have a say in the types of services that are most beneficial to their needs. Just as importantly, governments monitor popular responses to approaches that reduce the number of middlemen and increase local-level flexibility. Governments Contributing More Resources. With this expansion into larger-scale programs, governments are migrating from a situation whereby donor funds (for example, World Bank and Asian Development Bank loans) make up the bulk of the funding to a 13    situation now where governments at the national and local levels are contributing the vast majority of financial resources from their own budgets. In Indonesia, for example, what started in 1998 as the Kecamatan Development Program, a USD 273 million, four-year program financed through a USD 225 million World Bank loan (that is, 82 percent of total funding), by 2012 had become a USD 1.5 billion annual program, with 90 percent of its funding coming from national and local governments. And by 2017, the program expanded to become a national Village Law program with an annual budget of approximately USD 7.3 billion funded almost exclusively by the Government of Indonesia. In the Philippines, the Kalahi-CIDSS program has matured from a USD 182 million seven-year project in 2002 that had a 45 percent government contribution, to becoming a USD 3 billion operation. Furthermore, that program supports other sectoral programs in adopting a CDD approach. Morocco’s National Initiative for Human Development Program II is a USD 1.1 billion program, with 70 percent co-financing by the national government. Institutionalization of CDD Approaches. There have also been several countries whereby CDD programs have directly affected the government’s formulation of key legislation or regulations that institutionalize community-based participatory approaches. A recent example comes from Indonesiaʼs Kecamatan Development Program and its successor, the National Program for Community Empowerment. For the past seven-plus years, the Indonesian government institutionalized these CDD programs into its budget and fiscal transfer systems. The new Village Law (Law No.6/2014), which is an outgrowth of these programs, was approved by the Indonesian Parliament and ratified by the president in January 2014.3 The law provides a stronger legal status for villages and strengthens the delegation of authority and decision-making power to individual villages. The Village Law has incorporated a number of key CDD principles, including participatory village planning, community implementation of village-level projects, inter-village collaboration, and improved accountability mechanisms. While the Village Law faces numerous implementation challenges, it also provides an unprecedented legal platform for village development. Other examples where CDD programs have directly influenced the way governments approach local level planning and financing include: Cambodia’s Organic Laws of 2001 and 3  Dewan Perwakilan Rakyat: DPR in December 2013. 14    2008 and the Strategic Framework for Decentralization and Deconcentration Reforms in 2005; Mongolia’s 2013 Integrated Budget Law; and Afghanistan’s draft Policy for Improving Governance and Development in Districts and Villages. The Afghanistan policy, which is currently being updated for presidential approval, granted official recognition to the Community Development Councils (CDCs) that were created under the National Solidarity Program. It encouraged the use of these community councils as the primary vehicle for development and governance at the district and village levels until formal District Councils and Village Councils were elected, as mandated by the Constitution. There are also country examples whereby CDD programs have been used to add meaning to laws which are “on the books” but have not, in fact, been implemented. CDD programs have been used as a channel to add meaning to citizen engagement in such cases as the Philippines Local Government code of 1991, Vietnam’s Grassroots Democracy Decrees of 1998 and 2003, and Ghana’s 2010 Decentralization Policy Framework. Instrument of Choice to Respond to Disasters. In countries where natural disasters have struck, (for instance, tsunamis, earthquakes, and typhoons) CDD programs and their community networks have been used as a means of quickly channeling funds to communities and serving as a platform for reconstruction. A further, less quantified but widely observed benefit of a CDD approach is the early involvement of displaced households in the rebuilding process, from community negotiations over redrawn land boundaries to the physical work of reconstruction. Examples include: Indonesia’s Aceh in the aftermath of the 2006 tsunami, where the governmentʼs community-based approach enabled Indonesia to rebuild over 100,000 houses in less than three years; a similar approach following that tsunami in Sri Lanka whereby the government, with UN-Habitat assistance, reconstructed 10,300 houses in 6 coastal districts; Pakistan’s 2005 earthquake and 2010 floods; and in 2013, the Philippineʼs Typhoon Yolanda, where the CDD program was able to assist over five million people living in cyclone-affected communities. In all the aforementioned cases, the CDD programs already had a network of facilitators and/or village committees present in those areas and thus were able to use the CDD “plumbing” to channel resources quickly and flexibly to meet emergency needs, whether it was washed-out roads, other connective infrastructure, or housing. 15    Countries Opening Up. Over the last two decades, newly formed countries such as Timor-Leste and South Sudan, as well as countries undergoing dramatic political transformations, for instance, Myanmar and Afghanistan, have begun using CDD approaches. Several of these countries have had years of conflict and fragility, and are now attempting to reverse decades of poor governance and ineffective centralization. The governments recognize that effective local government delivery systems are still lacking or nascent. Partnering with communities and nongovernment organizations to deliver services directly and quickly to poor citizens is an efficient way to use resources and to gain legitimacy. VI. What Has Worked, What Hasn’t and What Don’t We Know? There is a growing body of evidence documenting the results and impacts of CDD interventions, especially in terms of its main twin goals of delivering community-level infrastructure and services and allowing villagers the opportunity to decide upon the use of public funds to improve their lives. Findings from CDD evaluations may vary in terms of longer-term impacts, but with only a few exceptions, there is ample evidence to show what has been achieved on several fronts: Delivering Cost-effective Tertiary Infrastructure and Services. In many of the CDD infrastructure-related programs, there remains little doubt that CDD programs have been able to provide large amounts of much-needed village infrastructure, particularly in less developed areas. Impact evaluations show that some of the largest government-run CDD programs (Indonesia, the Philippines, India, Nepal, and Morocco) are able to improve access to all-weather roads, clean water, sanitation, education, and health. Evidence across the programs is generally strong regarding positive impacts on access to and use of services, especially involving health, education, and drinking water (Beatty et al. 2017; Casey 2017; Voss 2016; Wong 2012). Several independent technical audits and studies have also shown that infrastructure and public works are built at comparatively lower costs than other forms of service delivery, without sacrificing technical quality. Studies from the Philippines, Indonesia, Nepal, Burkina Faso, and Malawi, for example, have demonstrated 15 percent to 40 percent lower costs, depending on the type of investments (World Bank 2007, Wong 2012). These cost savings come primarily from factoring in community contribution of labor and materials and/or eliminating the middleman or contractor overhead. Given the size of 16    government-run CDD programs, these savings can be significant. For example, in the big Indonesia National Community Empowerment Program, a 25 percent savings in a billion dollar annual appropriation translated into USD 250 million more becoming available to the government for its poverty work. Studies on Economic Rates of Return (ERR) of these CDD programs also show levels above the country’s hurdle rate for acceptable investments. Weighted ERR levels for the Philippines, Indonesia, and Burkina Faso, for example, demonstrated relatively high levels of economic impact ranging from 21 to 86 percent. The India livelihood projects were showing internal rates of return for various livelihood activities of 40 percent and higher. For roads, benefits are seen through improved connectivity, savings in travel time, more frequent market trips, more motorized traffic, increased passenger movement, increased labor and agriculture machinery movement, and reduced unit costs of transport. Large benefits accrued from entirely new economic activities and access to markets that were made possible by transport infrastructure. For water projects, the returns come primarily from time savings. Furthermore, CDD programs in Asia, Africa, and Latin America have provided hundreds of millions of poor villagers and self-help groups, primarily poor women, with much-needed credit services, albeit with ongoing controversies over their sustainability. More advanced programs, particularly in South Asia, have coupled microcredit access with small business management training and advances in marketing, insurance, and village-level procurement. Going to Scale and Replicability. As mentioned in the earlier section, countries have succeeded in scaling-up these programs in a significant way. The CDD model is not limited by geographical scope, and its modular and flexible design allows for regional variation and “franchising”. While a CDD program can take many shapes and forms adapted to the local context, the more successful examples keep community demand and poor people as the center of program design. National Ownership. CDD programs often turn out to be enormously popular. For national governments, they allow tangible, visible public goods and services to be delivered to its citizens quickly and transparently. For local governments that often provide counterpart contributions, they also serve to meet the local needs of its citizenry in tangible ways. And for local communities, after years if not decades of receiving few if any services, 17    these programs give villagers a voice in how development financing is used in their areas. Citizens give high marks to the governmentʼs service delivery through CDD programs, ranging for example from 82 percent satisfaction levels in Myanmar to over 90 percent in Lao PDR. The popularity of these programs goes to the heart of the main conceptual underpinnings of CDD, namely that local communities are best placed to perform local development and that involving them as partners and agents of development reduces the burden on governments. Giving Villagers a Voice in Decision-Making. One of the objectives of CDD programs is to give villagers the opportunity to plan and choose public goods that meet their local needs. Generally, participation in community meetings can range from 40 to 70 percent of households in a village, with women representing anywhere from 30 to 70 percent of attendees. Villagers also democratically elect their own committees and councils to prepare plans, implement projects, and oversee operations and maintenance. Women’s Participation. One of the highlights derived from the evaluation literature is the extent to which CDD programs succeed in increasing women’s direct participation in community decision-making forums. In most CDD operations, women’s participation is mandated under program rules, usually as a minimum percentage of development council membership as well as attendance at meetings. As compared to traditional male-dominated councils in many of these countries, seating women on elected local councils and project committees is a monumental feat in and of itself. The quality of women’s participation varies widely, but in nearly all cases reviewed, the trend is broadly positive with women’s participation in CDD operations being exponentially higher than it is in traditional councils or in sectoral community-wide programs. According to several independent monitoring reports, Afghanistan’s Community Development Councils provide a space for women to gather together, form bonds with fellow women, share problems that they are confronted with, and identify possible solutions to these problems. 4 One report (Beath et al. 2013) found that women’s participation in local governance had broader impacts with regard to girls’ school attendance, women’s access to medical services, and improved economic perceptions and 4 The quality and acceptability of women’s participation also seems to rise over time in CDD programs. In Afghanistan’s National Solidarity Project, program rules required a 30 percent female representation on community councils. Following a national congress of community council members to design its successor program, the Citizen’s Charter, the quota was removed. Six months into the new Citizen’s Charter, women’s representation on the reelected councils had risen to 48 percent. (See: http://www.ccnpp.org/)   18    optimism among women. (There is little evidence to support changes in other behaviors, such as women working or exerting influence over household decisions.) It should be noted that a key design variable that recurs in the literature is the importance of insuring that efforts to involve women must be an exercise of a “whole of village” rather than one that could be seen to divide communities and undermine local authority. Poverty Reduction Impacts. A meta-analysis of CDD programs found that seven out of nine projects had statistically significant positive impacts on household consumption and living standards (Wong 2012). These included programs in Indonesia (two projects), India, Nepal, Senegal and Sierra Leone. In these countries, the programs allocated sufficient resources and per capita allocations over several years to reduce poverty levels. One common mistake, seen especially in smaller NGO and donor projects, is low per capita allocations which do not allow for sufficient investments to raise productivity. For example, the Democratic Republic of Congo (DRC)’s Tuungane Project, which was supported by the United Kingdom government and implemented by the International Rescue Committee and CARE International, covered a relatively large geographical area, but the project only allocated USD 1 per capita. This would be insufficient to lift living standards, especially in such a fragile and complex context as the DRC (Humphreys et al. 2012). Projects with such low per capita allocations for investment are not likely to have significant welfare returns, and it is no surprise that evaluations did not find one. To have any impact on income or non-income poverty levels, it is important to ensure that there is an adequate level of investment occurring over multiple years. Poverty Targeting. Overall, most of the larger CDD programs did well on geographical poverty targeting and in selecting poor areas in which to operate. Programs that made use of poverty maps and the latest national statistical data were the most effective at geographical targeting. For example, the impact evaluation of Indonesia’s PNPM Generasi program, which focused on health and education, showed that the poor accrued a greater incidence of program benefits than the nonpoor (Olken et al. 2014). Overall, at the household level, most CDD programs invested in public goods—such as roads, bridges, schools, and health centers—that benefited the broader poor and nonpoor community. However, programs were generally found to benefit more poor than nonpoor households and individuals. Results are mixed based on certain socioeconomic characteristics (for example, urban versus rural) and types of investments. 19    VII. What Didn’t Work So Well? While CDD has proven to be an economically efficient way to build large quantities of economic infrastructure across a broad range of national contexts, other aspects of the original model have not fared as well as originally hypothesized. In some cases, the original theory was just wrong. In other cases, it might be possible to salvage the original model by further specifying which variables are the critical ones. A summary of the most salient areas where practice has belied expectations follows. Social Capital Impacts. Rather surprisingly, CDD projects have not been found to have much impact on increasing social capital. Social Capital is broadly defined by the World Bank as “the norms and networks that enable collective action.” The theory is that through communities working together using a CDD approach, the project builds trust, networks, and collective action. However, there is little evidence to support that claim. Why? One hypothesis is that levels of social capital are already high in some contexts. Casey (2017) points out that mean levels of trust in control communities are already high: for example, fully 95 percent of respondents in Sierra Leone, 93 percent in the DRC, and 85 percent in Afghanistan would entrust another community member with financial transactions. Similarly, at baseline in the Philippines 76 percent of respondents “trust that others are willing to help if needed.” Thus, it could be that there is already sufficient social capital to facilitate collective action within the community. A second hypothesis is that many of the activities in CDD programs are not designed with the explicit purpose of improving social capital or social cohesion. If one were to focus on improving social capital, one could argue that it would take more than a few large village meetings during the year to build trust and networks. The third hypothesis is that we do not have the correct tools to capture fully concepts as complex as trust and association. Many of the impact evaluations to measure trust and collective action use as proxies behavioral games which in fact are fraught with concerns regarding external validity, relevance within cultural contexts, and general limitations of laboratory experimentation compared to field settings (Levitt and List 2007; Galizzi and Navarro-Martínez 2017). 20    Governance impacts. CDD programs aim to improve local governance, by changing the way that government interacts with its citizens. As proxies for “improved local governance”, impact evaluations often measure the following factors: peopleʼs attitudes toward various levels of government; participation in public assemblies or meetings inside and outside of the project domain; awareness of project information and other local civic activities, and spillover effects in terms of the way government officials and citizens approach and manage other development programs and civic activities. Overall, what scant evidence there is shows positive to mixed results for participation in other village assemblies, awareness of project information and other local civic activities, and changing the way in which other development programs are managed. On the positive front, many of these programs are scaling-up nationally, which means that the government has expanded the CDD approach nationwide. In some countries, sectoral ministries such as the Ministry of Education increasingly provide more block grants for early childhood development and educational development to schools and parent-teacher committees that engage communities, publish budgets, and issue post-performance reports. In cases such as the Philippines and Sierra Leone, there is also evidence of governance spillover effects in terms of: increases in attendance in village assemblies; more households who were aware of the finances of their village local government unit; or in the case of Sierra Leone, more local government leaders who were particularly active in the planning, construction, and oversight of local public goods, leading to increased citizen confidence in their local representatives (Casey et al, 2013). In Lao PDR, positive governance spill-overs were found for the perception of communities, especially the poor and women, to influence village affairs and decision-making. In CDD treatment areas there also was increased attendance at general village meetings for communities as a whole as well as for poor women (Voss 2016). But in other programs, there is more of a “project bubble” or ring-fenced effect, with little spill-over into other governance norms outside of the project. In other words, outside of the program, there is no quick demand from villagers that non-CDD government programs begin using the same tools for participatory planning, financial transparency, or community feedback that their CDD programs introduced. This is one area where experts got 21    it wrong, or at a minimum, under-estimated how long it would take for the political effects of CDD interventions to spill-over into other spheres.5 What might have happened?   One hypothesis is that thus far, these CDD interventions remain isolated in vertical programs of specific line ministries while not fully being integrated into other sectors. CDD is thus only one of many interventions occurring at the local level which influences community norms, networks, and behaviors. For example, the education ministry will not use community councils to discuss school maintenance issues, and the agricultural ministry will not make use of village development plans or locally organized community councils to guide agricultural user groups. If the hypothesis is correct that we see few spillovers because government bureaucracies project their own ministry’s rule systems and defend them against intrusions from other ministries, then a corollary is that as we see national CDD programs becoming legitimized through village laws and sectoral programs (like education) we should now see both top-down and bottom-up pushes to spread participatory planning more broadly. That is, the reason for the lack of spillover to date is not because CDD did not generate “enough” social capital, but because the barrier to spillover was government norms, not community values.6  There are other hypotheses for why we are not yet seeing more governance spill- overs. Some governance specialists believe that shifts in local governance and social transformation take much longer to materialize and involve more complex causality than the timeframes and measurement models that most development actors are comfortable with. These changes in norms and attitudes take generations to shift and are likely to involve other 5 Caution about premature conclusions is warranted in this area. Problems arise not just from the very small sample of programs that have tried to measure governance spillovers, but from other methodological problems, such as limited sampling and high geographical and social variance across communities. Thus, for example, while Beath’s randomized control trial of Afghanistan’s NSP program found few spillovers in the central region of the country where most of his measurements were concentrated, working in the northeast, Gozstoni and Kohler found that “CDCs have become a firmly established feature of village-level governance in the region.” Accordingly, 72 percent of respondents assessed the head of the shura (CDC) as the most powerful person in the village, while 72 percent also believed that the decisions of the CDC shura were in the public interest (and not in the interest of a few powerful households). Lastly, CDC conflict resolution was evaluated rather positively: 33 percent and 32 percent of respondents respectively believed that the CDC always or often resolved conflicts in a just way. The results for the perceptions of state-provided district-level governance are less positive. 6  This is the hypothesis behind Afghanistan’s 2016 Citizen’s Charter, a cross-cutting presidential initiative that aligns all major service delivery ministries behind the community councils created by the National Solidarity CDD project. 22    changes to the institutional and economic environment than just spillovers from a CDD program. Handshake between the Supply Side and Community Demand. While many CDD interventions built schools to improve children’s enrollment and attendance or health centers for improved access to health care services, it was far beyond the capabilities of village-level CDD interventions to improve teacher quality or the educational curricula. Similarly, it was beyond the CDD project’s ability to mandate that health practitioners go into the clinics or to ensure there were enough trained midwives available to staff the clinics or appear during clinic hours. The supply side had to deliver and, in many cases, it did not. Linking supply and demand is hardly a problem unique to CDD. Finding the institutional incentive to deliver services effectively and on-scale is still a hurdle on many development fronts. It is common for CDD projects to experience difficulties in coordinating with various line ministries for coordinated area planning, service delivery, technical inputs, and operations and maintenance. As several of the CDD national programs grow and expand, the agenda in this area needs to evolve as well. How can governments use CDD programs to hold local governments and technical ministries more accountable to communities for service delivery? Should CDD programs start providing more support, incentives and/or capacity building for the supply side and for technical quality, for example, teacher and health worker training or financial services? Are there social accountability tools which can work effectively at scale to hold local governments more accountable? Or should they bridge gaps between the supply side and community groups, such as a community liaison person within banks or technical agencies? What are the potential trade-offs in expanding the scope of these CDD programs while keeping them simple in design? However, in many of these countries, the success of CDDs can in part be attributed to the fact that secondary infrastructure, national or provincial government services, and the like already exist. What CDD is providing is an efficient way to “finish the job” of bringing access down to the village and completing “the last mile.” By contrast, in countries such as Afghanistan, where in many areas very little ended up being built aside from the National Solidarity Program’s localized investments, household welfare returns were relatively less noticeable because, presumably, the CDD-built infrastructure did not link into a larger network of infrastructure (Beath et al. 2013). A much broader evidence base on poverty reduction impacts will be needed in the future. Such an analysis should also explore the size, frequency and usage of block grants needed to have measurable poverty impacts. 23    Pooling – In principle, larger village infrastructure investments should produce more benefits than smaller ones will. Similarly, under a system of popular voting, communities that unite around a common project should be able to out-vote atomized groups that only vote for their own small program. However, we see no trends towards building larger projects or of proponent groups banding together to negotiate strategies. If anything, over time, social and political pressures to ensure that everybody gets something will grow stronger. While this is not an issue unique to CDD programs per se, it does raise challenges regarding economies of scale and the ability to plan in a more integrated fashion. VIII. What Don’t We Know? Sustainability of impacts. As mentioned earlier, it will be important to conduct more evaluations looking at program impacts over time. Impact evaluations covering longer time periods will also be important to assess the sustainability of welfare, social capital, and governance impacts. Unfortunately, the number of long-term impact evaluations is so small as to make any generalization premature. The longest panel survey we are aware of for a CDD program is the Philippines’ KALAHI-CIDSS program, which measures impacts over seven years.7 This dearth of longitudinal studies is in no way unique to CDD, but findings over time will certainly be valuable additions to our knowledge base. Positive impacts may attenuate when initial high returns from low baseline situations do not continue and investments yield diminishing returns. On the other hand, positive impacts could also increase as project actors become more used to the processes, and as repeated treatments add greater expertise in planning and managing resources. As many of these programs continue, it will be important to measure whether program impacts attenuate or change, and what design variations are feasible. Because the majority of CDD projects are relatively new, little is known about how they evolve over time. As CDD’s critics of direct disbursements have noted, there is in principle an inherent tension between directly transferring funds to communities on the one hand, and relying on local governments that have the mandate and responsibility to manage public development investments. Over time, a maturing decentralized or deconcentrated local 7  The fourth wave of evaluation for Indonesia’s PNPM Generasi program is forthcoming, measuring results over a nine-year time horizon (2007-2016). 24    government will usually want to exert more control over community investment planning. Against this background, the popularity of direct community management means that communities can make political alliances by providing support to government officials and political parties who can push back. Whether CDD programs turn out to be transitional strategies that eventually get folded into normal local government operations, whether they become the basis for populist party politics, whether communities and local governments can reach an accommodation, or whether they can continue as the extreme application of subsidiarity principles is a question waiting to be answered. Unpacking the Black Box of Decision Making. While many impact evaluations and final reports focus on outcomes and outputs—roads built, school enrollments, and so forth— there has been little discussion or documentation about how decisions are made at the community and local government levels. In general, CDD-sponsored research has concentrated on the social dynamics of the village. But the CDD model is about local government-community interactions, not just community ethnography. There is an urgent need for operational research to focus on the dynamics of this interaction and how it evolves as the rules of community-run management become routinized within the local government machinery. Several studies point to community satisfaction with CDD programs and their services. Others have recorded alignments and discrepancies between whether or not the subprojects match community and local officials’ preferences. But much more qualitative work should be done to understand the decision-making process in the allocation of resources. Is there full, informed participation regarding these decisions, or are the projects merely reinforcing patronage systems and political patterns of largesse? Why are some subprojects chosen and not others? How do we ensure that decision-making bodies and local committees remain accountable and are not captured by preexisting local elites? How do traditional power-holders in the community—such as the village chief or tribal leaders—view these CDD programs and their mechanisms for decision making? How are marginalized groups—ethnic minorities, persons with disabilities, and so forth—involved in the decision making, if at all? Some more qualitative and ethnographic work in this area would be useful to complement the quantitative evaluations. Addressing Community Heterogeneity. CDD projects are to some extent built around the notion that there is an organic community that has the institutions and cultural 25    legitimacy to negotiate priorities for investment. Increasingly, however, community boundaries and forms of social control are more realistically described as subsets of spatial and social hierarchies that cut across territorial boundaries. Job markets, economic value chains, long distance migration, and other changes mean that for many community members, attending a village meeting is just an activity they do in a much larger social world. To what extent are these cross-cutting relationships reinforcing or undermining local forms of collective action and the types of decision-making that people make? Reaching Marginalized Groups. Some CDD programs have been more successful than others at reaching marginalized and excluded groups. Social inclusion and inclusive growth is becoming a major policy issue. Within countries, lagging regions, growing income inequalities, and exclusion of vulnerable groups are some of the major challenges for government policy makers in developed and developing countries alike. Several of the CDD programs—such as in Nepal and India—are worth exploring further for comparative work on their success in improving household targeting, information outreach, capacity-building efforts, and other implementation modalities for reaching marginalized and excluded groups. CDD and Local Economic Growth. We know that CDD translates into short-term job employment for labor-intensive infrastructure works and that roads can often lead to development of businesses. We also know that several of the CDD programs such as those in Latin America and South Asia have helped to organize farmers and self-help groups to access credit and build small-scale enterprises and businesses. More and more governments are requesting assistance for job creation and livelihoods support, especially but not exclusively to cope with the “youth bulge”. However, the use of CDD to promote local economic growth and territorial development requires more experimentation, more clarity on the differences (and complementarities) between CDD and private-sector approaches, and multidisciplinary technical assistance. For the most part, CDD projects have tried to complement infrastructure with various types of financial inclusion such as microcredit, small-business loans and the like. However, realistically, access to labor markets through seasonal or permanent migration, rural enterprise development, or access to off-farm industrial employment are in many areas the dominant pathway out of poverty. CDD projects would seem to be ideally poised to help villagers acquire skills, information, and the resources they need to tap into those opportunities, but to date very few CDD programs have made a systematic effort to develop these types of programs. Expanding these 26    programs may also require potential trade-offs in keeping programs simple and easy to manage. More experimentation with value chain development and inclusive economic development models will be needed in both rural and urban areas. Negative Impacts of CDD. What are some of the negative impacts of CDD programs? For example, does introducing funds into areas wrought with conflict only lead to more violence, crime, and conflict? Is CDD just being used by governments and local elites to spread the largess for political or personal gain? Evaluations in Afghanistan, the Philippines, Indonesia, and Sierra Leone examined the CDD programs’ impact on conflict. There was no impact on macro levels of violent conflict, except in the case of the MILF rebel group in Mindanao, the Philippines (Arcand et al. 2011). These findings should not be surprising, given that many of these violent conflicts are exogenous to communities and beyond their control. Nevertheless, development programs in general—not only CDD interventions—must operate cautiously in conflict and fragile situations because they can cause severe harm or have negative impacts. CDD programs can, by bringing more development funds into a community, attract conflict by introducing competition for funds, exacerbating existing social cleavages, or reinforcing political patronage systems of largess. On the other hand, if designed well, they can also address long-standing grievances of exclusion and of a non-responsive state, for example MILF in the Philippines, or can introduce community mechanisms for mediating burgeoning conflicts, for example, in Indonesia. IX. Recapitulating the Findings and Their Implications It is perhaps unsurprising that two proponents of CDD conclude that a review of the available evaluation literature confirms that national community development programs offer developing countries a new tool for reducing poverty and for engaging positively with large numbers of poor citizens. However, while the conclusion that CDD programs can build economically useful small infrastructure, fill essential infrastructure gaps, and cover large areas quickly and efficiently is supported by the overall evaluation literature, in this concluding section, we introduce a number of important qualifications that are relevant to the discussion of a second-generation CDD agenda. 27    Perhaps the first and most important qualification is that how a CDD program is designed and managed makes a big difference in the likelihood of its success. CDD programs in principle follow standard economic principles that more information, more choice, and direct negotiations are the key instruments for achieving optimal outcomes under given conditions of knowledge and resources. However, putting these principles into practice is often difficult. Government officials are often reluctant to hand over full control, and there are legitimate, core requirements to ensure that public funds are spent well and reported on accurately. But as Scott (1998) has argued in his seminal Seeing Like a State, often the administrative formats and review systems that purport to improve development end up disempowering communities and reverting control back to state agencies simply because they are too complicated for poor people to understand. Successful projects are simple projects. Second, this review has shown that claims that community projects will strengthen social capital and that strengthened social capital will translate into greater government responsiveness are not backed up by the existing evaluation literature. We have argued that the right analytical framework for approaching CDD projects is not just to focus on communities, but to treat communities as one-half of an interaction that also involves government bureaucracies and their incentive structures. Third, despite all the discussion about participatory processes, CDD programs by themselves cannot solve the problem of community heterogeneity and the resultant problems of marginality and capture by the elite. Rich or powerful villagers may continue to dominate village councils. Even with affirmative action requirements, women may still not receive a proportionate voice and representation in decision-making. It should also be noted that these issues of inequality and inadequate representation and voice are not phenomena unique to developing countries. To some extent, it can be argued that local-level capture is still better than local-level capture plus higher-level capture on top of that, but in actual fact it is often top-down interventions that drive reform in areas such as women’s participation, legal redress, and social mobility. Project strategies have attempted to solve this problem in different ways. In South Asia, which has a large, sophisticated body of nongovernmental organizations and a political history of governments cooperating with them, high-level political sponsors in government set the operating rules for the CDD program, but the actual program execution is through nongovernmental partners who have the staff and experience needed to deal with issues of 28    social inclusion. In Indonesia, which lacks the experience of civil society partnerships, community investment decision-making is limited to decisions about public and semi-public goods, rather than private goods, precisely to avoid the risk of elite capture and the exclusion of the poor from benefits. Last, the jury is still out on how to adapt CDD approaches to urban areas. On the one hand, urban areas often lack the normative and integrative institutions that make community negotiations and the enforcement of sanctions possible. Their populations are fluid and urban “neighborhoods” can be little more than places to sleep. Furthermore, unlike in rural areas where people make their living from the same place where they live, in urban areas the primary development issues are often issues of access to jobs, housing, and transportation. On the other hand, in many cities, urban communities do have effective systems of local governance and community organization. Participatory planning can help reduce crime, and small local investments in items such as electrification or water purification can produce significant returns to poor people. While in theory proponents say that such services can be provided by municipal authorities, in practice, in many areas those municipal services have not been provided. There is a problem that in view of issues of tenure uncertainty in municipalities, these services often will not be part of municipal service delivery plans. X. CDD 2.0: The Way Forward The fact that CDD programs have taken root and become national programs in so many parts of the world bodes well for the future of CDD, but it also suggests that it is time to revisit CDD’s basic assumptions. In this review, we have concentrated on the national programs that have successfully delivered on their promise. We have not dwelt on poorly designed or managed CDD programs that began with high hopes but that then fizzled out. CDD programs are no more immune from development pathologies of corruption, mismanagement, inertia, or simply running out of steam than other development programs are. Instead, our argument has been that the concept is sufficiently well advanced that it is worth taking a step back to objectively assess both what is known about how to make a CDD program deliver on its promise, and what must still be improved or researched further. In this concluding note we would like to point to a few promising avenues of operational innovation in CDD programs that are of interest to the larger development 29    community. The first and in many ways most interesting development is not in the further expansion of CDD per se, but in the adoption of CDD principles to improve the performance of other types of poverty programs. It has been a struggle to avoid having social protection programs not think of each other as competitors for the same pot of poverty funds. But in a growing number of cases, governments are starting to use the principles of community targeting for household transfer programs. In Indonesia, for example, Alatas, et al. (2012) have shown through a randomized control trial that using community targeting systems produces nearly as accurate a targeting of results as highly sophisticated statistical systems do, but with much higher rates of local legitimacy and satisfaction due to the abilities of communities to articulate why some people receive benefits over others. Afghanistan’s new Citizen’s Charter program uses the CDD foundation built up under the 2002-2014 National Solidarity Program to coordinate ministry delivery of agriculture, health, education, and rural energy. A second area that is only now getting underway is expanding the use of social media to reduce CDD’s management and transaction costs and to intensify the dialogue between community members and government policy makers. Because of its national coverage, CDD generates very large data sets on topics as diverse as local disputes and conflict to large-scale survey data on village-level choice making. Capitalizing upon these large data sets and making use of more savvy information technology forms of monitoring and communications opens up a large potential area of research and citizen’s feedback. Now that a growing number of countries are operating national CDD programs and financing them from their own budget, expertise from donor agencies is increasingly being replaced by homegrown specialists. This trend can be nurtured. Developing countries are increasingly building their own networks of universities, civil society groups, think tanks, and cadres of officials who understand and can disseminate CDD approaches. In East Asia, the World Bank and Australia’s Department of Foreign Affairs and Trade (DFAT) have launched a regional community that brings together policy makers and specialists to discuss shared needs and experiences, and this approach can be launched in other regions as well. Finally, while CDD projects by their very nature are concerned with community-level decision making, it is far more realistic to view communities as one end of a spectrum that extends into national and regional networks. CDD projects are only now starting to come to grips with how CDD programs can help improve poor people’s access to higher-paying labor 30    markets, whether by providing for community childcare centers so that working mothers can take on better jobs in industry, or whether CDD programs can provide local-level insurance so that the very poorest have the confidence to seek urban work during the down season. Linking the community level with higher levels of service delivery and broader economic growth models is the next frontier. This paper has made three key arguments. The first is that CDD constitutes an important new tool for policy makers. The empirical evidence from evaluations confirms that CDD programs produce large amounts of badly needed, productive economic infrastructure at reasonable cost and quality. They also provide villagers, especially the disadvantaged, with a voice in how development funds are used to improve their welfare. CDD programs have proven to be particularly useful where government institutions are weak or under stress, both because CDD programs can act quickly over large areas, and because they can place relatively few demands on over-stretched government institutions. The second argument is that CDD programs are not a homogeneous category. There are important distinctions to be made between national, on-budget, multi-year programs, and off-budget programs. Each may have their place, but they are not the same thing. The third argument is that CDD works best and achieves the greatest results when it is part of a broader development strategy that includes reforms to governance, investments in productivity, and integration with efforts to improve the quality of public service delivery. 31    REFERENCES Abueva, Jose. 1972. “Bridging the Gap between Elites and the People in the Philippines.” In The Political Economy of Development, edited by Norman T. Uphoff and Warren F. Ilchman: Berkeley, CA: University of California Press. Alatas, Vivi, Abhijit Banerjee, Rema Hanna, Ben Olken, and Julia Tobias. 2012. “Targeting the Poor: Evidence from a Field Experiment in Indonesia.” American Economic Review 104 (2) (October): 1206-1240. Arcand, Jean-Louis, Adama Bah, and Julien Labonne. 2011. Conflict, Ideology and Foreign Aid. Geneva: The Graduate Institute; CERDI Universite d'Auvergne: and Oxford University. Arthapedia, “Decentralization and Rural Governments in India” http://www.arthapedia.in/index.php?title=Decentralization_and_Rural_Governments_inIndia Beath, Andrew, Fotini Christia, and Ruben Enikolopov. 2013. Randomized Impact Evaluation of Afghanistan’s National Solidarity Program: Endline Report. World Bank, Washington DC. Beatty, Amanda, Ariel Ben Yishay, Elisabeth King, Aniceto Orbeta, and Menno Pradhan. 2018. Impact Evaluation of KALAHI-CIDSS: Third Round Report. Draft. Bebbington, Anthony, Michael Woolcock, Scott Guggenheim, and Elizabeth Olsen. 2006. The Search for Empowerment: Social Capital as Idea and Practice at the World Bank. Connecticut: Kumarian Press. Bourdieu, Pierre. 1977. Outline of a Theory of Practice. Cambridge, UK: Cambridge University Press. Bourdieu, Pierre. 1986. “The Forms of Capital.” In Handbook of Theory and Research for the Sociology of Education, Edited by John G. Richardson, 241-258. Westport, CT: Greenwood Press. Casey, Katherine. 2017. Radical Decentralization: Does Commnity-Driven Development Work? Manuscript. Casey, Katherine, Rachel Glennerster, and Edward Miguel. 2013. The GoBifo Project Evaluation Report: Assessing the Impacts of Community Driven Develop ment in Sierra Leone: Final Report. Washington, DC: International Initiative for Impact Evaluation. Coleman, James S. "Social Capital in the Creation of Human Capital." American Journal of Sociology 94 (1988): S95-120. http://www.jstor.org/stable/2780243. De Regt, Jacomina, S. Majumdar, and J. Singh. 2013. Designing Community-Driven Development Operations in Fragile and Conflict-Affected Situations: Lessons from a Stocktaking. Social Development Department Report. Washington, DC: World Bank. Fox, Jonathan. 2014. “Social Accountability: What Does the Evidence Really Say?” Global Partnership for Social Accountability Working Paper No. 1. Washington, DC: World Bank. 32    Galizzi, Matteo M. and Navarro-Martínez, Daniel (2017) On the external validity of social preference games: a systematic lab-field study. Management Science Garg, Teevrat, Caterina Gennaioli, Stefania Lovo, and Gregor Singer. 2017. “Competition and Conflict: Evidence from a Community Driven Development Program in Indonesia.” Draft. Humphreys, Macartan, Raul Sanchez de la Sierra, and Peter van der Windt. 2012. Social and Economic Impacts of Tuungane: Final Report on the Effects of a Community Driven Reconstruction Program in Eastern Democratic Republic of Congo. New York: Columbia University. King, Elizabeth. 2013. A Critical Review of Community-driven Development Programmes in Conflict-affected Contexts. London: DFID / International Rescue Committee. Levitt, Steven, D., and John A. List. 2007. "What Do Laboratory Experiments Measuring Social Preferences Reveal About the Real World?" Journal of Economic Perspectives, 21 (2): 153-174. G. Mansuri, and V. Rao. 2012. Localizing Development: Does Participation Work?  Development Economics Research Department Policy Research Project. Washington, DC: World Bank. Olken, Benjamin, J. Onishi, and S. Wong. 2014. “Should Aid Reward Performance? Evidence from a Field Experiment on Health and Education in Indonesia.” American Economic Journal: Applied Economics 6(4): 1-34. Pritchett, Lant, M. Woolcock, and M. Andrews. 2010. “Capability Traps: the Mechanisms of Persistent Implementation Failure.” Working Paper 234, Washington, DC: The Center for Global Development. Putnam, Robert. 1993. Making Democracy Work, Princeton, NJ: Princeton University Press. Putnam, Robert. 2000. Bowling Alone: The Collapse and Revival of American Community. New York: Simon and Schuster. Scott, James C. 1998. Seeing Like a State. New Haven, CT: Yale University Press. Shartz, Norman B. 1978. “Community Development and Culture Change in Latin America,” Annual Reviews of Anthropology 7: 235-261. Tax, Sol. 1952. Heritage of Conquest: The Ethnology of Middle America. New York: Free Press. Tocqueville, Alexis de. 2000. Democracy in America: The Complete and Unabridged Volumes I and II. New York: (Bantam Classics) Mass Market Paperback. Uphoff, Norman, and Warren F. Ilchman. 1972. The Political Economy of Development. Berkeley: University of California Press. Voss, John. 2008. Impact Evaluation of the Second Phase of the Kecamatan Development Program Jakarta. Washington, DC: World Bank. 33    Voss, John. 2016. Lao Poverty Reduction Fund Phase II Impact Evaluation: Final Report. Laos: World Bank. White, H, Menon, R and Waddington, H, 2018. Community-driven development: does it build social cohesion or infrastructure? A mixed-method evidence synthesis, 3ie Working Paper 30. New Delhi: International Initiative for Impact Evaluation (3ie). Wong, Susan. 2012. “What Have Been the Impacts of World Bank Community Driven Development Programs? CDD Impact Evaluation Review and Operational and Research Implications.” Social Development Department Report. Washington, DC: World Bank. World Bank. 2004 World Development Report 2004: Making Services Work for Poor People. Washington, DC: World Bank. ------------. 2011. World Development Report 2011: Conflict, Security and Development. Washington, DC: World Bank. 34