Community-Based Targeting Mechanisms for Social Safety Nets: A Critical Review
Introduction
Social safety nets can serve an important role in alleviating poverty and in promoting long-term growth by providing households with the protection that markets and informal networks may not supply. A social safety net may redistribute resources toward disadvantaged groups, or sustain political coalitions to support critical structural reforms. Unfortunately, the growing awareness of the importance of social safety nets in developing countries has not been translated into effective action because of the failure of traditional social welfare ministries to effectively reach and engage the poor. This has led to experimentation with new bottom-up service delivery options and poverty alleviation mechanisms that more actively involve the poor and their communities in program design, implementation and monitoring. Examples include reforms that decentralize the delivery of public services to local governments, community management of forests and other natural resources, and group-based microcredit programs. Demand-driven social funds that aim by design to elicit community involvement have become increasingly popular with governments and donors, and international organizations such as the World Bank now make community participation an explicit criterion for funding approval for a growing list of projects (World Bank, 1996; World Food Program, 1998). In the United States, the Bush administration's faith-based initiative program has been offered, at least in part, as a mechanism to use organized religious communities to better target and channel federal resources for the poor.
Common sense and substantial evidence suggest that community participation can lead to improved project performance and better targeting (Baland & Platteau, 1996; Isham Kaufmann & Pritchett, 1995; La Ferrara, 1999; Narayan & Pritchett, 1999; Wade, 1988). For example, a study of India's Integrated Rural Development Project found that Indian states that employed village councils to select beneficiaries had a much smaller proportion of nonpoor participating households (Copesake, 1992). A recent survey of dozens of country experiences with social safety nets conducted by Subbarao et al. (1997, p. 87) for the World Bank contends that programs that involve communities, local groups, and nongovernmental organizations (NGOs) can achieve better targeting outcomes. Perhaps the simplest but most persuasive illustration of the potential benefit of community-based targeting is the increasingly prevalent practice in many poverty-alleviation programs of delivering family and child assistance via a female parent. Many empirical studies now confirm that assistance delivered via a female parent leads to a larger positive impact on child welfare and household investments in health, nutrition and education than the same resources delivered via a father (Beegle, Frankenberg, & Thomas, 2001; Lundberg & Pollak, 1996; Pitt & Khandker, 1998). Thus, the use of categorical targeting of women may be thought of as a form of community-based targeting where children are the ultimate target beneficiaries, and mothers are the chosen intermediary agents within the household community.
The purpose of this paper is to review evidence and propose a framework for thinking about the community-based targeting mechanisms to deliver privately captured benefits, i.e., mechanisms that target welfare or relief.1 For the purposes of this essay, we define community-based targeting as a state policy of contracting with community groups or intermediary agents to have them carry out one or more of the following activities: (a) identify recipients for cash or in-kind benefits, (b) monitor the delivery of those benefits, and/or (c) engage in some part of the delivery process.
Community agents can be social or religious groups, single-purpose NGOs, or local elected officials or governing bodies. The extent to which an agent qualifies or not as a community agent depends on that agent's level of embeddedness in local community affairs. By this we mean the degree of involvement of the group or individual in other functions and activities that imbricate them in poor subcommunities, or their degree of involvement in the day-to-day community life of the poor (through residence, private business, or social activities). Throughout the essay we will often treat community groups and intermediary agents as coterminous, and apply the single label “community agents.”
There are numerous examples, both contemporary and historical, of purposeful large-scale community-based targeting for social safety nets. Arguably, one of the earliest and most studied examples of a community-based targeting mechanism was the English system of poor relief. For several hundred years until the reforms of 1834, the English Poor Laws implemented a highly decentralized system of poor relief administered and financed by local parishes. Although the parish began as a local church institution, by the 16th century the estimated 12,000–15,000 parishes in England had assumed many of the functions of local civil government including the administration of poor relief. Each parish was responsible for deciding who was unable to work and deserving of relief, and for financing and delivering assistance. The following account from the diary of Thomas Turner (1754–65) describes how poor relief decisions were made in his day (Mencher, 1967):
The parishioners were accustomed to meet once a week at the parish workhouse, at which meetings all applicants for relief were received and where all laborers belonging to the parish, who had not in the preceding week been in constant employment, attended to give an account of their earnings and received such sums as, with the earnings, should amount to a sum deemed competent to their maintenance in proportion to their children.
The significant externalities that naturally arose in a locally financed system of poor relief meant that the system became increasingly difficult to manage, especially in periods of rising poverty and through the upheavals associated with the industrial revolution and the enclosure movement. This then fostered incentives for restrictions on population movement; no parish wanted an influx of poor persons.2 Eventually, a rate-payers revolt and changing political tides led to the Poor Law Amendment Act of 1834 which greatly reduced relief, imposed the onerous workhouse test, and led to a much more centrally administered system based on uniform rules (Quigley, 1999).
Another historical example of community-based targeting comes from colonial experience. The language of Indirect Rule—the use of “Native Authorities” by the French and British to administer their African (and other) colonies—is replete with references to the advantages of community intermediation. Indirect Rulers (chiefs, sheikhs and emirs) were supposed to be more accountable to their “subjects,” to know more about their needs, and were definitely cheaper than expatriate administrators. Delegation of responsibility for targeting benefits was common. For example, during scarcity-situations in WW II the British used Native Authorities in western Sudan to ration sugar, tea, petrol and other commodities.
In recent times, the Uzbek government began an experiment in 1994 to involve quasi-official, quasi-religious community groups known as mahallas in the decentralized targeting of child benefits and other types of social assistance to low-income families. The mahallas traditionally acted to mediate community problems and conflicts. A unique aspect of the program is that the State has given local mahallas considerable discretion in deciding whether a family should receive assistance and the amount. External reviews of the program suggest benefits were targeted relatively well (Coudouel, Marnie, & Micklewright, 1998).
Albania, faced with massive unemployment and poverty in a difficult transition period in the early 1990s, introduced the Ndihme Ekonomika (Economic Support) safety net. The central government at first administered grants bureaucratically through local ministry offices but found that this formula provided little incentive for local officials to verify eligibility requirements. The program was then devolved to local communes through block grants. Using data from a recent household survey, Alderman, 1998, Alderman, 1999 found local targeting effectiveness compared favorably with safety net programs in other low-income countries although he notes that overall targeting performance could be improved via better targeting of block grants across localities.
Community-based targeting mechanisms sometimes emerge at the initiative of communities themselves, in response to a particular need or funding formula. In Armenia, where chronic public sector financing problems has meant that health and education had become de facto fee-for-service programs, parents of children in public schools have been paying for food, for instruction that falls outside the core curriculum, and textbook use fees. Social assessments suggest that this type of fee acts as a barrier to access for the poor (World Bank, 1999, p. 67). A government-established school textbook waiver program allocates to each school a fixed amount sufficient to waive annual textbook rental fees for only 10% of students leaving the remaining 90% of funds to be raised locally by charging parents a rental fee of approximately US$1 for each textbook their child uses. A study of the program found local schools setting up mechanisms to exempt particular students from the fees, with the decision as to which students would be exempt being made in some cases by the school principal, and in others by the school parent-teacher association (World Bank, 1999, pp. 67–68).
This paper, an interpretative review of the literature, will explore just how well communities in developing countries might use local information and social capital to allocate new program resources toward the poor and vulnerable. We begin with an outline of many of the most important effects and tradeoffs in constructing a community-based targeting scheme. Loosely, these can be classified into three broad, related categories of question. First, will community-based targeting “increase the size of the cake”? In other words, will community mechanisms lower the costs of delivering benefits to a target population? Second, “what size slice of cake” will the poor obtain? That is, what kinds of distributions are likely to result when community-based targeting is employed after taking into account the need to provide incentives, program leakage, and the rents that intermediary agents might potentially capture? An important consideration here is that under devolution local targeting preferences, determined through local political processes, might differ substantially from national preferences or those of a donor. Finally, what will determine the quantities and types of “cake-making ingredients” available to community agents for disbursal to eligible recipients? Here we focus on the national-level political economy and program design issues that arise while implementing a decentralized policy of community-based targeting.
We conclude with some observations about how to design a community-based targeting scheme. Current experiments have tended to use homogeneous community agents across the country to implement targeting—for example, local town mayors in every poor community. But agents that would empower the poor and be responsive to poor constituencies are unlikely to be the same in all localities, and so a more demand-driven approach to community-based targeting may be recommended in some cases, at least in the short term. We also believe that in many instances the best community-based targeting schemes will be hybrid mechanisms where the center defines and monitors targeting categories, rather than unconditional devolution to community groups with little basis for evaluation or control.
Section snippets
Schematics of the political economy of community-based targeting
Several advantages might be expected from community-based targeting. There may be lower costs of administration simply because community agents, typically living in low-cost rural areas, do not need to be paid as much as educated bureaucrats. Involving community groups as stakeholders may lead to better screening, monitoring and accountability. Community groups may have better information for identification of needs, and households may in turn have less incentive or opportunity to provide false
Increasing the size of the cake: how cost-effective is CBT?
A growing literature has established that community involvement can lead to improved project performance in social funds, microcredit projects, natural resources management, public health, and in local public goods provision. But community involvement is not always and everywhere the optimal policy. In managing small irrigation facilities in India, for instance, the pendulum has swung from state management to local management and in some cases back again to state management (Baker, 1997). There
Slicing the cake: what kind of distribution might result?
Community agents may be in a position to employ more socially desirable, or locally adapted, criteria for assessing need. On the other hand, the community agent may be in a better position to “capture” the program and direct resources away from intended beneficiaries. The distribution of rents, and intended benefits, will vary greatly across communities because of variation in the distribution of local preferences and ability of local groups to influence local political processes.
Getting the ingredients for making cake: national-level issues
The end result of a program to target resources through decentralized community-based targeting is influenced not just by the cost advantages and the nature of local preferences and capture, but also by national choices about funding. National funding is in turn determined by at least three processes. First, decentralized democratic political processes may tend to be less favorable to narrower targeting (the “paradox of targeting”) and community targeting may result in cooptation from above and
Designing a community-based targeting system
To empower marginalized groups community-based targeting may need to incorporate some of the lessons of the social fund approach, and adopt a demand-driven, menu approach that broadens the diversity of participating community agents, perhaps using detailed and transparent public consultations and analysis before decisions are taken regarding contracts with given community groups.
Concluding remarks
We have highlighted four key points in discussing community-based targeting: (a) communities vary in their ability to mobilize information and implement effective monitoring systems, and this will influence whether community-based targeting leads to cost saving advantages or just opens up more opportunities for local-level corruption and costly rent-seeking; (b) local communities will vary in their willingness to target the poor; (c) national political economy effects are likely to be complex
Acknowledgements
We would like to extend our appreciation to Margaret Grosh, Harold Alderman, Pranab Bardhan, John Blomquist, Henry Bruton, Michael Cernea, Nora Dudwick, John Giles, William Jack, Emmanuela Galasso, Eric Hanson, Simon Harrigan, Daniel Klein, Alexandre Marc, Jessica Mott, Akbar Noman, Albert Park, Tamara Perkins, Jean-Philippe Platteau, Ashok Rai, Vivajendra Rao, Gustav Ranis, Martin Ravallion, Mary E. Schmidt, Endre Stiansen, William Sundstrom, Michael Woolcock and participants at workshops at
References (126)
- et al.
Socioeconomic stratification by wealth ranking: Is it valid?
World Development
(1997) - et al.
Enchantment and disenchantment: The role of community in natural resource conservation
World Development
(1999) - et al.
Poverty reduction through geographic targeting—how well does it work
World Development
(1994) New states, new NGOs? Crises and transitions among rural development NGOs in the Andean region
World Development
(1997)- et al.
How reliable are group informant ratings?: A test of food security ratings in Honduras
World Development
(1998) - et al.
Assistance to the poor in a federal system
Journal of Public Economics
(1987) - et al.
Participation, social capital, and intersectoral problem solving: African and Asian cases
World Development
(1996) - et al.
Political sustainability and the design of social insurance
Journal of Public Economics
(2000) - et al.
In or out—centralization by majority vote
European Economic Review
(1996) The emergence of local capacity: lessons from Colombia
World Development
(1997)