تمرکززدایی در بخش دولتی: جنبه های کمی در کشورهای فدرال و متحد

The purpose of this paper is to investigate empirically the extent and evolution of fiscal (de)centralization in a sample of OECD countries. We ask whether and how some factors derived by the economic theory on this issue are empirically significant either for centralization or decentralization. Then, building a new panel data set, we test the significance of these factors using a random effects model. Our results seem to confirm that land area, population, degree of urbanization and income per capita are negatively correlated with centralization. Moreover, we find that heterogeneity in ethno-linguistic groups is significantly related to the observed decentralization in the patterns of expenditure on education and on religious, cultural and recreational services.

Over the last few decades, the demand for a higher degree of decentralization or even for political independence by some regions or locally based groups has been a hotly debated issue in many countries. A world-wide interest in decentralized forms of government has emerged in “mature” federations (Canada, Australia, United States), in unitary states (for example, Italy, Spain, UK), in developing countries and in some states of Eastern Europe. Furthermore, in the European Union the introduction of the subsidiarity principle in the Maastricht Treaty has led to a renewed emphasis on the virtues of decentralization and a growing pressure to remodel the distribution of power both within the community and the states and within the states. The very recent Treaty of Nice annexes, a declaration on the future of the Union. This declarations calls for a deeper and wider debate about the future of the European Union to be concluded in 2004 when an Intergovernamental Conference will be convened. In such debate the main point on the agenda is how to establish and monitor a more precise delimitation of powers, competences and responsibilities between the European Union and the member states. As a matter of fact, an European Convention is currently working on a preliminary draft of the European Constitutional Treaty and, among other things, it must propose a clear answer to the question of the allocation of powers between the European Union and the member states. In the economic theory, building on Oates’s “decentralization theorem” (1972), many studies in fiscal federalism or local finance literature have indicated some factors that would make either decentralization or centralization desirable. In broad terms, much of this theoretical literature, which has been primarily normative, considers how taxing and expenditure powers can be assigned to various levels of government in order to achieve the most efficient outcome in the whole economic system. In this literature there is a “presumption” in favor of decentralization which results from the assumption that if the central government supplies local public goods it would provide a uniform quantity for all local jurisdictions: the central government is “insufficiently sensitive” to heterogeneity in citizens’ preferences as to their demand for local public goods. Therefore, the most compelling argument in favor of decentralization is that it can better match citizens preferences for public goods. Moreover, decentralization make governments more accountable to citizens and very recently it has also been regarded as a extraordinary institutional framework to enhance economic growth1. This literature also points out the costs of decentralization which can be, for instance, administrative ones, or they may be due to the difficulty of exploiting the economies of scale in local governments. To sum up then, focusing on the trade-off between centralization and decentralization, the literature’s main objective has been to determine the “optimal degree” of decentralization. This means that this literature offers some predictions or testable hypotheses to explain the observed variations in the degree of fiscal decentralization across countries (both federal and unitary) and over time. In other words, according to the theory, we should expect that there are some determinants of the “optimal degree” of decentralization among levels of government. Oates (1972) himself, in the second part of his seminal work, identifies these determinants and makes an attempt to conduct an empirical analysis in order to test them. However, despite these range of benefits and costs in (de)centralization suggested by the theoretical literature, the empirical work and the review of countries’ experience in this area is far from comprehensive. Therefore, our paper is an attempt to fill this gap given the importance of decentralization for increasing efficiency in policy making as well as for designing a better institutional context. Looking at countries experiences by following the economic theory on decentralization can surely help to design a better institutional framework. In this respect the European Union is one of major challenge for policy-making but also countries where recent changes of regime call for a new political system. So, in the paper we ask whether and how some factors derived by the economic theory on this issue are empirically significant either for centralization or decentralization of functions. More specifically, we investigate empirically the extent and evolution of the allocation of expenditure and tax powers in a sample of OECD countries. To introduce the work it is necessary to say that generally a political system includes a national government and at least one level of subnational governments (states, provinces, local governments) and can distribute powers and responsibilities in many ways ranging from system in which the national government possesses most of the decision-making power and the sublevels have a relatively minor role — that is, a centralized system — to system in which the national government is relatively weak and the sublevels are dominant — that is, a decentralized government. In this respect, political studies often present the contrast between unitary and federal governments: unitary governments are simply thought of as centralized governments and federal governments as decentralized. It should be observed that “the contrast” between federal and unitary countries has failed to account for the fact that there is a range of possibilities between centralization and decentralization in performing government activities, both in federal and in unitary countries. In fact, even if the “difference” between the two types of government is that a unitary government presents two levels of government (central and local) and a federal government presents three levels (central, regional and local), in both federal and unitary countries, for each level of government the ability to make decisions is not always defined at the constitutional level. One level may be influenced by another in various ways. Therefore, a federal country may exhibit more centralization than a unitary country. As a matter of fact, it is possible to observe — looking, for example, at a sample of OECD countries — a wide variation in the arrangements of the fiscal structure across countries (both federal and unitary) and over time. This implies that it is possible to address the issue of some determinants of decentralization considering both federal and unitary countries. Thus, in this paper we focus on decentralization in the public sector by considering the fiscal relationships among levels of government in a sample of OECD countries (both federal and unitary) from 1977 to 1994. We have already mentioned that little empirical work has been undertaken on this issue. To our knowledge, this empirical work includes Oates (1972), Pommerehne (1977), Wallis and Oates (1988), Patsouratis (1990), and Panizza (1999). The first two studies deal with a large set of cross country data respectively in 1965 and 1968. Wallis and Oates build a panel data set for the American States during 20th century. Patsouratis (1990) uses a time series analysis in a sample of 11 EU countries. Panizza (1999) considers the degree of centralization mainly in the years 1975, 1980, 1985 by looking at a large group of countries. The key findings of these studies are that the size of the country (both in terms of population and land area) and income per capita are negatively and significantly related to the degree of centralization. Our work build on these previous empirical researches, but our investigation differs to a great extent from them, since we build a new panel data set and consider new variables’s specification accounting for heterogeneity in countries as well as we use a different econometric technique. The study is organized as follows. The next section presents the main concepts relevant to fiscal decentralization and arguments as to centralization versus decentralization. These arguments allow us to state the hypotheses to be tested. Section three shows the patterns of (de)centralization across countries and over time; section four shows data, the model to be tested and empirical findings. Concluding remarks appear in the final section. The appendix describes the data sources and variables included in the work.

The goal of this paper is to identify empirical regularities explaining cross-country differences in the level of fiscal centralization. In spite of strong movements toward decentralization in many countries in recent years, we know relatively little from an empirical point of view about the main theoretical predictions of the fiscal federalism literature. Perhaps such paucity of empirical research is because of data limitations. In our work we make an attempt to partially fill this gap. Our work builds on previous empirical research on this issue. Let us mention these earlier studies to have a better understanding of the extent to which our empirical analysis contributes to the literature as well as to review the research on this subject upto now. The first two studies were Oates (1972) and Pommerehne (1977) which both apply a cross-section analysis. Oates considers fiscal centralization for a sample of 58 countries in 1965 using as data source United Nations Yearbook of National Account Statistics (1968). He proceeds his empirical analysis by estimating different equations, in which in turn more explanatory variables are included such as the population of the country, the geographical size, the income per capita and some measures of diversity. For this last set of explanatory variables, Oates relies on Bank and Textor’s Survey (1963) and simply constructs three dummy variables taking value of 1 if the country is homogeneous considering respectively language, race, religion, and 0 otherwise. He finally adds one more dummy variable taking value of 1 if the previous three dummies are also 1 and 0 otherwise. Summing up, he found that: (i) the degree of fiscal centralization shows a significant inverse relationship to the population of the country; (ii) wealthier countries tend to be more decentralized than poorer countries; (iii) the four dummies described above were not significant. Very similar to Oates’ work is Pommerehne’s empirical analysis (1977). Again, he looks at the degree of fiscal centralization for a sample of 50 countries for the year 1968. Performing a cross section equation, his results are almost the same as Oates’. For example, the dummies accounting for homogeneity are not significant either.29 But, differently from Oates, he finds that the size of a country and the degree of urbanization are significant (i.e., in the equation, these variables present the expected negative sign). Wallis and Oates’ (1988) look at the trends in fiscal centralization during the 20th century in the state and local sector in the United States. They observe that the state-local sector exhibits wide variation in the relative roles of state and local government both over time and across states. In fact, In 1902, local governments accounted for 82% of the tax revenues in the state-local sector; by 1982 this had fallen to 43%. Therefore, using a large panel data set of the US state-local sector reaching back to 1902, they explore econometrically the variation both over time and across states considering two measures of fiscal centralization. The two measures are the state share in the state-local fisc on either an expenditure and revenue basis. To explain the observed wide variation across states, they consider as explanatory variables: the geographic size of states, the population, the degree of urbanization, the income per capita30 and two measures of socio-economic diversity taking account of the proportion of farmers in the population and the proportion of white31. Finally, since over the century the southern states experienced much more centralization than the other states, they introduce a dummy variable to control this “southern” effect32. To test the hypotheses, they make use of a random components approach. The size of the state (measured in term of population), the extent of urbanization and the income variable are significant. In contrast, the “white” and “farmers” variables are not significant. Most notably, the significance of the variables does not change when they also consider as the dependent variable the degree of centralization measured on the revenue basis. Patsouratis (1990) takes a different approach. Through a time series analysis for the period 1960–1986, he focuses on a group of 11 EU countries in order to determine whether the factors which are supported on theoretical grounds are important in explaining the variation in the degree of decentralization among countries. The explanatory variables are: per capita income, population, density of population, a time trend and a political factor variable. The political variable is defined in terms of dummies: each dummy variable denotes the political party which was in power in the period under consideration. Thus, it takes the value 1 for the years in which the party in question was in power and 0 for the years when another political party was in power. The main findings are that per capita income and the time trend are significant (i.e., positively related to fiscal decentralization) for the majority of the EU countries, the population variable is significant only for France, Greece, Ireland, Netherlands, and the UK; the density of population is not significant for the majority of countries. Finally, the political variable is statistically insignificant for all the EU countries except for Netherlands and Spain. More recently, Panizza (1999) considers a large group of developed and developing countries. Data on fiscal centralization (both revenues centralization ratio and expenditure centralization ratio) are obtained from IMF. He considers only the years 1975, 1980, 1985. However in his sample, as he states, for many country the full set of data were not available. The explanatory variables in Panizza’s work are: size, population, income per capita, and data on democracy which are from Gastil (1990) and data on ethnic fractionalism from Atlas Narodov Mira. To estimate determinants of fiscal centralizion he uses OLS estimation and Tobit and semi-parametric estimators to control for outliers in the sample. The findings are that country size, income per capita, ethnic fractionalism and level of democracy are negatively correlated with the degree of fiscal centralization. Concluding, in the light of the theoretical literature and previous empirical works the results of our paper seem quite interesting and encouraging. Our econometric approach builds on Wallis and Oates’s work and we investigate the extent and evolution of fiscal decentralization across some federal and unitary OECD countries. In particular, building a new panel data set and considering a new variables’ specification, our aim has been to test whether certain factors derived from the economic literature could explain statistically the variation in the degree of fiscal decentralization across some OECD countries. Our findings seem to suggest that there are some clear “regularities” in the extent of fiscal decentralization experienced both in federal and unitary countries in the last decades. In fact, land area, population, degree of urbanization and income per capita are negatively correlated with centralization in almost all our regressions. Most important, previous studies attempted to include among the explanatory variables some measures of heterogeneity accounting for differences in preferences. However, these variables in each country were only measured in terms of dummy variables. Clearly, the simple differentiation of the sample between complete homogeneous and heterogenous countries is a very crude measure of preferences for decentralization. On the other hand, Panizza uses data published in the Atlas Narodov Mira. As already mentioned the period covered by these data refers to the years 1960–1965. In this study, we provide instead two more precise continuous measures accounting for preferences in each country. Such measures have not been used before in other studies. Finally we report regressions that use disaggregate measure of expenditure, that is expenditure on education and expenditure on religious and cultural and recreational service. Because of data limitations, we considered only a small group of countries. Most notably, we find that heterogeneity in ethno-linguistic groups is significantly related to the observed decentralization in the patterns of expenditures on education and on religious, cultural and recreational services. Notice that because both types of expenditure imply geographically concentrate benefits our findings lend support to the theory on fiscal federalism as far heterogeneity in preferences for local public goods is concerned. To sum up, decentralization is still an issue worth of investigation as it can be an extraordinarily successful and democratic mode of governance. Moreover in a world of ethnic conflicts and separatist movements, decentralization can be seen as a way to mitigate such phenomena via a more political autonomy.