Political Accountability and Public Service Provision in Africa: Evidence from Tanzania and Zambia Barak D. Hoffman [email protected] Clark C. Gibson [email protected] Department of Political Science University of California, San Diego Paper prepared for presentation at the Workshop in Political Theory and Policy Analysis, Indiana University © 2006 Barak D. Hoffman and Clark C. GibsonAbstract Does a government’s source of revenue explain its policies? The predominate view contends that policy variation results from institutional variation. We argue that a government’s source of revenue also strongly affects its expenditures, independent of institutions. Using local government budget data from Tanzania and Zambia, we find that local governments devote a larger share of their budget to public services as local taxes increase. Alternatively, as local governments receive more resources from outside their boundaries - central government transfers and foreign assistance - the share of the budget for employee benefits and administrative costs rises. Because there is no variation in the powers of local governments in Tanzania and Zambia, these results are independent of political institutions. The results suggest that the effectiveness of donor projects to increase local government accountability in part is a consequence of how these programs affect the fiscal link between a local government and its citizens.2I. Introduction Over the past two decades, development policy has changed radically from a narrow concentration on economic issues to a broad concern with political accountability. The 2006 edition of the World Development Report, for example, unequivocally endorses this new policy direction by stating that “politicians have the correct incentives to provide public services only when they have to appeal to the broad mass of citizens.” This new explicitly political approach to development demands that development practitioners understand not only the link between accountability and development but, more importantly, the causes of that accountability. The majority of studies that seek to determine the sources of accountability focus on the role of institutions. According to this line of reasoning, changes in political accountability result from changes in the institutional link between the government and the governed. A smaller group of scholars have posited that the fiscal link between citizens and their government is also a strong determinant of government accountability. Building on a literature which we label fiscal theories of governance, we argue that a government’s revenue source is central to explaining its policy choices. Broadly speaking, fiscal theories of governance contend that a government will defer to its citizens’ policy preferences when it is dependent on its citizens for revenue while a government that is not dependent on its citizens for revenue lacks such incentives (Bates and Lien 1985; Levi 1988; Fearon and Laitin 2004). Empirically however, it is quite difficult to separate the effects of institutions from those of revenue sources, as the two often move together. Thus while fiscal theories of governance have a deep logical3appeal, they have little robust evidence supporting their claims. Ideally, testing fiscal approaches to governance requires testing the theory under the conditions of identical political institutions. In this paper we undertake such a test. Using local level government budget data in Tanzania and Zambia, we find that local governments in Tanzania and Zambia produce more Public Services as a share of total local government expenditure as Local Revenue increases. Alternatively, as central Government Transfers and Foreign Aid increase, public services as a share of the local government budget falls. Not only do our results remain after controlling for socioeconomic factors but far more importantly, the effects we find are independent of political institutions because there is no variation in political institutions at the local level in either country. This study furthers our understanding of the politics of development in several important ways. First, unlike the larger set of studies that link political institutions to policy and development outcomes, we examine the effects of revenue on public expenditure under identical political institutions. The main thrust of the political institutional approach is to tie policy outcomes to political institutions; our approach demonstrates that there is another source of political accountability independent of formal political institutions: the fiscal link between the government and the governed. Understanding these fiscal links, we argue, is central to understanding development in general and the political effect of foreign aid in particular.4Second, our research design allows us to test directly the influence of revenue streams on government choices. Using district level government budget data is a methodological advance over most quantitative studies of revenue’s effect on policy choices that use cross national data and rely on control variables to account for the differences in political institutions. Further, because local level political institutions do not vary within either country but local level political institutions vary widely between the two countries - local governments in Tanzania are dominated by the center while Zambia’s district governments enjoy greater powers - we are also able to explore the strength of revenue effects on policy outcomes under two different sets of institutions. We believe this is the first study to implement such a research design. Third, we examine the entire local government budget. Existing studies that attempt to explain sub-national variation in public policy outcomes tend to examine single sectors. Such research could produce misleading results because demand for different types of public services varies widely and this variation manifests itself in wide differences in expenditures across different types public services. Using the entire local budget provides a comprehensive analysis of how sources of revenue affect policy choices. The results of this study bear directly on current development policy. Through programs of institutional reform, donors are placing considerable effort into increasing the accountability of local governments in these two countries. Our results