|Aug 2006, Caridad Araujo, Francisco H.G. Ferreira, Peter Lanjouw, and Berk Ozler
Decentralization of spending authority from central to local governments has become a common objective of policymakers throughout the developing world. In addition, a whole new breed of anti-poverty programs in developing countries are designed as highly-decentralized, demand driven initiatives, where decisions on the type of expenditures and investments are meant to be made by the local beneficiaries themselves, in a participatory manner. In 2003, the World Bank alone lent $2 billion dollars to such “Community Driven Development” programs.1
Yet, the theoretical literature remains ambiguous in its assessment of the rationale for decentralization. This ambiguity arises from the trade-off between the local government’s advantage in terms of access to superior information at lower cost, and the possibility that the risk of capture of decision-making by special interest groups is higher at the local level than at the national level. This possibility is the so-called “Madisonian presumption” that “the lower the level of government, the greater is the extent of capture by vested interests, and the less protected minorities and the poor tend to be.”2
What is needed most is better empirical analysis for specific programs, with particular attention to whether programs might be diverted from their original mandate, as a result of the influence of local elites. Such evidence is quite rare.3
Policy research working paper 3997 uses a unique combination of data-sets from Ecuador to investigate whether there is elite capture in the pattern of project choice by communities that were awarded Social Investment Fund grants. Our data include reliable poverty and inequality estimates at the community level, obtained by combining high-quality information from a household survey with census data. They also include administrative data on project approvals and expenditures by type and by community, for a period of almost three years.
Social Investment Funds typically provide beneficiary communities with a menu of projects to choose from, most of which produce public goods, and all of which are intended to benefit the poor. The paper exploits the fact that the menu offered by the Ecuadorian Social Fund included basically two types of projects: local public goods (whose valuation may vary across individuals, but which are accessible to all) and excludable (private) goods. In Ecuador, by far the most important private good provided were latrines, which were built on land plots belonging to poor community members with no previous access to toilet facilities.
The paper constructs a theoretical model of project choice between public and private goods, under the assumption that political power is positively correlated with socio-economic status. The model predicts that, controlling for inequality, poorer communities will select latrine projects (the excludable projects mostly needed by the poor) more often than better off ones. It also predicts that, controlling for poverty, more unequal communities would choose latrine projects less often, as a result of a concentration of power in the hands of richer people, who do not need latrines.
Both predictions are borne out empirically, as can be seen in Figures 1 and 2.4 Controlling for infrastructure need and a set of geographic and demographic variables, the poverty headcount is associated with a greater probability that the community receives a latrine project. With the same controls, inequality (measured by the expenditure share of the top 1%, 3%, 5%, etc. of the population) reduces the likelihood that latrine projects are chosen. This effect of inequality on project choice gets smaller as we define the group of elite to be larger and becomes zero when the more commonly used Gini index is employed as the measure of local inequality. That the expenditure share of the top 1% in a community is strongly correlated with the type of project selected may indicate that decisions about project choice are very heavily influenced by a few dominant actors in each community.
These results are consistent with the paper’s model of project choice under political inequality. The results suggest that even programs that are targeted to the poor, and only offer projects from a menu that is designed with poverty-reduction in mind, are vulnerable to capture by local elites. While channeling funds away from latrine construction towards school-building may not appear as a grave distortion, the point is that elites are capable of affecting the outcomes of participatory processes, even when they are reasonably carefully designed. In their essence, these results are similar to those found by Galasso and Ravallion (2005) for the Food for Education program in Bangladesh, where greater land inequality was associated with worse targeting outcomes.
As evidence of the kind presented in this paper mounts, there may be more general implications for the design of Social Funds and other community-driven development programs currently in operation or preparation around the developing world. These implications are not necessarily that such programs should be abolished or that they should be centrally administered. However, there may be a need to develop clearer rules for the manner in which decisions must be taken within each community, with a view to making it harder for the more powerful to exercise an unduly large amount of influence.
Bardhan, P. and Mookherjee, D., 2000. Capture and Governance at the Local and National Levels. AEA Papers and Proceedings 90 (2), 135-139.
Bardhan, P. and Mookherjee, D., 2006. Pro-poor targeting and accountability of local governments in West Bengal . Journal of Development Economics 79(2), 303-327.
Galasso, E. and Ravallion, M., 2005. Decentralized Targeting of an Antipoverty Program. Journal of Public Economics 89, 705-727.
Mansuri, G. and Rao, V., 2004. Community-Based and –Driven Development: A Critical Review. World Bank Research Observer 19 (1), 1-39.
Rosenzweig, M. and Foster, A., 2003. Democratization, Decentralization, and the Distribution of Local Public Goods in a Poor Rural Economy. BREAD Working Paper No. 010.
1 Mansuri and Rao, 2004.
2 Bardhan and Mookherjee, 2000, p. 135.
3 See Rosenzweig and Foster (2003), Galasso and Ravallion (2005), and Bardhan and Mookherjee (2006) for the few examples that exist.
4 These figures only control for inequality and poverty, respectively, and do not include all the variables included in the empirical analysis discussed here.