The Institute for Economic Development at Boston University                                                                    Research Review Spring 2000

A widely anticipated benefit of the decentralization of the provision of public services to local governments is that they enhance accountability of the delivery system. Centralized delivery systems are frequently characterized by substantial corruption and leakages owing to the inability of central governments to effectively monitor the performance of bureaucrats entrusted with local service delivery. Transferring authority over delivery to elected local governments allows dissatisfied citizens to vote against elected officials, thus inducing some accountability. The intuition that local needs are better served by decentralized mechanisms is borne out by cross-country empirical evidence. However, the effects of decentralization may be less benign if local democracy is subverted by powerful elites, local governments are financially constrained, or if local officials lack adequate technical expertise compared to their central counterparts. In this paper, Bardhan and Mookherjee develop an analytical framework that isolates two aspects of the problems associated with decentralization: limited accountability and financing problems faced by local governments. They contrast centralization with three alternative modes of financing revenues by local authorities entrusted with the task of delivery, ranging from full devolution of revenue raising authority, limited autonomy in the form oflevy of user fees, to financing local

In a context where both national and local governments are vulnerable to capture by elites, the effects of decentralization on service volumes, efficiency and equity are shown to depend upon the manner in which expenditures on such programs are financed. In general, decentralization tends to expand the volume of the service delivered. However, such expansion may or may not represent efficiency gains, as it tends to be accompanied by overprovision of the service to local elites. Constraining financial authority of local governments limits such expansion. In order to assess the welfare effects of decentralization, a comparative analysis of the three financing mechanisms is undertaken. Bardhan and Mookherjee find that in general, user-fee-financed decentralization dominates centralization as well as local-tax-financed decentralization, from the standpoint of both efficiency and equity. The intuitive reason is that user-fee-financing combines enhanced accountability with restrictions on the ability of local governments to finance provision to elites by taxes on the less-well-to-do. The welfare comparison between user fees and central grants is more qualified: the former is more sensitive to local needs, while the latter is more effective against overprovision to wealthy users.

The central policy conclusion emerging from the paper is that the choice of financing mode is a key instrument in the design of decentralization initiatives. Vertically unbalanced patterns of decentralization that limit the volume of infrastructure services may actually be in the public interest.