1. INTRODUCTION In developing countries, in particular, not only are the governments the largest employers, but also the government budgets constitute the most important resource allocation mechanism. In these countries, on average, 30% of the gross domestic product (GDP) is allocated by the government budgets. The purpose of this article is to investigate the government budgets in relation to a number of political events and institutional factors in Turkey. In this regard, we consider elections and military-backed governments as political events. The institutional factors that are considered include organizational fragmentation of the budgetary institutions and the coalition governments. Examining the effects of the elections on budget deficits might give an indication of the existence of political business cycles in Turkey). (1), (2) The novelty in our approach is that we introduce a new power dispersion index which is suitable for the conditions of Turkey and probably for other developing countries that have fragmented fiscal authorities under the coalition governments. These and similar issues are investigated extensively in developed countries, but less often in developing countries (see Alesina and Perotti 1995, 1996). Therefore, the analysis in this article is expected to be useful for the policymakers in Turkey and other developing countries. In Turkey, since 1983, there have been two separate organizations responsible for the preparation and implementation of the budget, whereas only a single organization was responsible before 1984. These organizations are the Ministry of Finance (MOP) and the Under secretariat of Treasury (UT). UT is responsible for financial aspects of the budget especially including debt management of the budget and some important transfer expenditures such as funds, subsidies, and incentives. The MOF determines the appropriation of current and transfer expenditures, regulates the dates of the expenses, and accrues and collects the revenues. It is known that the State Planning Organization (SPO) is the third organization that is involved in the budget process; but as the SPO mainly prepares the macroeconomic framework of the budget, it may be considered as more an advisory board of the government rather than an administrative board. For this reason, we will consider only the MOF and the UT in our analysis. We claim that the power division between these two organizations exacerbates the political power dispersion of coalitions. To test this claim, we will test the previous power dispersion indices and then introduce an index that takes into account the interaction between these organizations and the number of parties in the coalitions. Analysis will shed light on the necessary fiscal reforms and fiscal policies required to reduce budget deficits. This article is organized as follows: Section II reviews the literature and explains the model used. The historical background of main economic events in Turkey and the data used in the analysis are explained in Section III. Empirical results are provided in Section IV. Policy implications are discussed in Section V. And finally, Section VI gives the conclusions. II. REVIEW OF LITERATURE AND THE MODEL The topic of how political and institutional considerations affect the national fiscal policy formation recently attracted the attention of many researchers. This line of argument starts with the seminal study by Roubini and Sachs (1989a) which is based on a cross-section data of 14 Organization for Economic Co-operation and Development (OECD) countries. They show that the tax smoothing hypothesis cannot fully account for the differing magnitude of the budget deficits because it does not take the various institutional arrangements in the political processes into account. They test a semi-reduced form equation to see effects of the political power dispersion on the budget deficits. This model is consistent with both the tax smoothing hypothesis that is championed by Barro (1979) and the traditional Keynesian model of fiscal deficit discussed by De Haan, Sturm, and Jan Sikken (1999, 166). …