On the Measurement and Impact of Fiscal Decentralization 1 ON THE MEASUREMENT AND IMPACT OF FISCAL DECENTRALIZATION

The typical post-Bretton-Woods era development approach that put much emphasis on central government-led development efforts has changed dramatically, and local governments have now clearly emerged as players in development policy. Now, the thinking about what is important to achieve development objectives is changing and fiscal decentralization reforms are now being pursued by many countries around the world. Within this context a number of studies attempted to quantify the impact of decentralization by relating some measure of decentralization to the economic outcomes of fiscal stability, economic growth, and public sector size. However, decentralization is surprisingly difficult to measure. In nearly all cases those examining the relationship between decentralization and macroeconomic performance have relied on the Government Finance Statistics of the IMF. However, despite its many merits, GFS falls short of providing a full picture of fiscal decentralization. For some countries, however, data that more accurately capture fiscal responsibilities among different types of governments are available. On the Measurement and Impact of Fiscal Decentralization 2 INTRODUCTION: SCOPE AND PURPOSE For much of the post-Bretton Woods era, the typical development approach emphasized central government plans and programs. The thinking was that if a poor country could come up with a national plan for generating and investing a sufficient amount of funds in a manner consistent with macro-stability, then that country would have met the pre-conditions for development. It would be a state-led (central government) strategy whereby the “flexibility to implement polices devised by technocrats was accorded a pride of place, and accountability through checks and balances was regarded as an encumbrance.” Until perhaps the mid 1990s, this was the main message of not only the two Bretton Woods institutions—International Monetary Fund and World Bank—but also of other multilaterals and many bilaterals. It was not an unreasonable strategy. Bretton Woods reflected a world emerging from the ravages of war, when much of the developing world was gaining its political independence. Development seemed a surmountable and largely technical challenge: good advisors would devise good policies, and technically assisted and institutionally capable governments would implement those policies. There could even be stages, from the first “mission” to an “exit strategy”—words that reveal so well the thinking of the time. There was some progress, especially in infant mortality rates, life expectancy, and adult literacy. There were also many failures. The failures were not just about an inability to demonstrate sustained growth rates. They were also about environmental deterioration, loss of civil liberties, corruption, and a very poor record of delivering “local” public services—clean water, sanitation, education, health, housing, safety nets, and, as some argue, poverty alleviation. 5 These were failures in an era when the scope of central government expanding enormously. 6 Now, the thinking about what is important to achieve development objectives is changing, dramatically so in some countries. Writing in 1994, Dillinger reported (in what has become one of the most quoted World Bank reports) that of the 75 developing countries with populations greater than 5 million, all but 12 claimed to be embarked on some form of transfer of fiscal authority from central to local governments. This transfer of power has been occurring even in “inherently centralized” On the Measurement and Impact of Fiscal Decentralization 3 countries, such as the Kingdoms of Jordan and Morocco (Ebel, Fox and Melhem, 1995; Vaillancourt, 1997; World Bank, 1999), Central and Eastern Europe states that were under the Soviet-type fiscal system (Dunn and Wetzel, 2000; Bird, Ebel and Wallich, 1995), the People’s Republic of China (Wong, 1997), military regimes like Pakistan (Shah, 1996; Pakistan NRB, 2001), countries like Thailand that view decentralization as an efficiency strategy for improving local service delivery in reaction to financial crises (World Bank, 2000); nation-states that are trying to avoid the centrifugal forces of separatism, like Russia (Wallich, 1994; and Martinez-Vazquez and Boex, 2001) and Indonesia (Ahmad and Hofman, 2001; Bird et al., 2001), and Latin America, where participatory budgeting is taking hold (Stein, 1997; Burki, Perry and Dillinger, 1999). The World Bank is very explicit on the importance of all this: the World Development Report on Entering the 21 Century notes that along with globalization (continuing integration of countries worldwide), localization—the desire for self-determination and the devolution of power—is the main force “shaping the world in which development will be defined and implemented” in the first decade of this century. The report argues that these “defining forces of globalization and localization,” which at first glance may seem countervailing, often stem from the same factors and reinforce one another (WDR, 1999/2000). The theme that emerges is that “good governance” matters, where “governance” is about how people determine collectively which services should be delivered by which government, and do so by establishing a set of transparent and competent public institutions they can understand and control. It is a theme that is tied to “getting right” what Bird refers to as the fundamental questions of intergovernmental finance: Who does what? Who levies which taxes (and is there a place for borrowing)? How can the resulting imbalances be resolved? What is the institutional framework to deal with the technical and political problems of decentralization? Within this context a number of studies attempted to quantify the impacts of decentralization by relating some measure of decentralization to the economic outcomes of fiscal stability, economic growth, and public sector size (Davoodi and Zou, 1998; deMello, 2000; Ehdaie, 1994; Fukasaku and deMello, 1998, Oates,1985). Nearly all of these studies draw on Government Finance Statistics (GFS) issued by the On the Measurement and Impact of Fiscal Decentralization 4 International Monetary Fund as the basis for measuring “decentralization.” As emphasized by Bird (2000), however, measurement is surprisingly difficult. And, if one cannot be confident of measuring the independent variable, then one cannot state with much confidence that decentralization is associated with one or more outcomes. The purpose of this paper is to take a critical look at the nature and implications of measuring the fiscal dimension of decentralization. Recognizing that “a curious combination of strong preconceived beliefs and limited empirical evidence” characterizes all too much of the discussion (Litvack et al., 1998; Bird, 2000), we look at two policy issues: (1) the extent to which fiscal decentralization is occurring and (2) the fragility of estimation results depending on how one measures fiscal decentralization (and, therefore, the danger in drawing sweeping conclusions that often have important policy implications). We start with GFS, but supplement this measure with other considerations that recognize more fully subnational autonomy and discretion in expenditure and taxation arrangements. We find substantial differences between GFS indicators and those that capture more accurately fiscal responsibilities among different types of government. We estimate the impact of these various measures of decentralization on economic stability, economic growth, and public sector size. Not surprisingly, we find that the different indicators have markedly different effects on economic performance. THE FRAMEWORK FOR MEASUREMENT The conceptual framework of fiscal decentralization is well established, drawing largely on the contributions by Stigler, Musgrave, Oates, and Buchanan. Here is the core logic: to care about growth and poverty issues, one should be concerned about efficiency–supplying services up to the point at which, at the margin, the welfare benefit to society matches its cost. In the private sector, the market-price system is the mechanism. When the market fails in this objective, there is a case for the public commandeering of resources to supply the activity. Once On the Measurement and Impact of Fiscal Decentralization 5 the public sector intervenes, the efficiency logic is in favor of some form of fiscal decentralization. The argument is that spatial considerations make subnational governments necessary conduits for setting up a system of budgets that best approximates the efficient solution of equating benefits and costs. This leads to the decentralization theorem: The governments closest to the citizens can adjust budgets (costs) to local preferences in a manner that best leads to the delivery of a bundle of public services that is responsive to community preferences. Subnational governments thus become agencies that provide services to identifiable recipients up to the point at which the value placed on the last (marginal) amount of services for which recipients are willing to pay is just equal to the benefit they receive. To implement this, subnational (local) governments must be given the authority to exercise “ownsource” taxation at the margin and be in a financial position to do so. This is the essence of decentralization. How, in practice, does one say that a country is decentralizing? While there is no set of prescribed rules, we draw on Bahl and others to identify 11 characteristics, which range from the requirement for open local elections to the fundamental “essence” question of whether subnational governments have (at least) tax rate-setting authority over locally assigned revenues (Bahl, 1999). A checklist for six transition countries (Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland) serves to explain our selection of countries that we can point to as politically “decentralized” for the purposes of making some statements about whether decentralization matters in terms of its promised benefits. We also have access to new data that goes directly to the point of own-source financial autonomy (OECD, 2001). On the Measurement and Impact of Fiscal Decentralization 6 EMPIRICAL DISCUSSION The literature on the relationship between decentralization and different macro indicators is growing. Most of these studies are crosscountry analyses using the Government Finance Statistics (GFS) of the International Monetary Fund, and all describe the degree of fiscal decentralization as the subnational share of total government spending/revenue or of Gross Domestic Product (GDP). Comparing the degree of fiscal decentralization across countries is a complex task that requires identification of subnational autonomy and discretion on expenditure and revenue arrangements. Although it is widely accepted that subnational share of total government spending/revenue is an imperfect measure of fiscal decentralization and that the need to standardize the fiscal variables in GFS inevitably eliminates details about the design of fiscal systems, many researchers use these measures to represent the degree of fiscal decentralization. What are We Trying to Analyze? Recognizing that GFS has served well as a product of the central government forces of the post-Bretton Woods development model, three major problems emerge when using the data in an empirical study on fiscal decentralization: First, although GFS provides a breakdown of expenditures by function and economic type, it does not identify the degree of local expenditure autonomy. Thus, local expenditures that are mandated by the central government or are spent on behalf of central government appear as subnational expenditure. Second, GFS does not distinguish the sources of tax and non-tax revenues, intergovernmental transfers, and other grants. Hence, there is no information on whether revenues are collected through shared taxes, piggybacked taxes, or locally determined “own-source” revenues. Third, GFS does not disclose what proportion of intergovernmental transfers is conditional as opposed to generalpurpose, and whether transfers are distributed according to an objective criteria or a discretionary measure. We will argue that this On the Measurement and Impact of Fiscal Decentralization 7 distinction between conditional/objective formula grants versus more centrally tied “discretionary”/specific purpose grants can be a useful variable as a country makes the transition from deconcentration to devolution. These aggregation problems limit the use of subnational statistics in the GFS data set. Thus, although GFS has consistent definitions across countries and over time, the subnational expenditure and revenue figures have little relevance in the decentralization context because the data fail to address properly the intergovernmental fiscal structure of countries and ignore the degree of central government control over local tax rates and tax bases. Thus, with GFS, the subnational revenue and expenditure share in total government revenue/spending ends up being an overestimate of fiscal decentralization. This overestimation of the fiscal decentralization indicator can be illustrated by analyzing the revenue structure of subnational governments. Until recently, such a comparison was impossible due to lack of data that were both disaggregated and fit what we identified above as the essence of public sector decentralization—the ability of local governments to set the tax rate at the margin. Such data is available now for a set of EU-accession countries from the Organisation for Economic Cooperation and Development’s survey Fiscal Design Across Levels of Government (OECD, 2001). OECD identifies three sources of subnational revenues—tax revenues, non-tax revenues, and intergovernmental grants—for the Czech Republic, Estonia, Hungary, Latvia, Lithuania, and Poland (Table 1). Tax revenues and intergovernmental grants are further divided into two groups. If subnational governments have total or significant control over a tax as defined by an “own” control over tax rate or a revenue tax base and rate, this is listed as “own tax revenue.” If subnational governments have limited or no control over the rate and base of a tax and the central government determines how to split revenues, it is listed as “revenues from tax sharing.” Non-tax revenues include income from business operations and property, administrative fees and duties, and fines. On the Measurement and Impact of Fiscal Decentralization 8 Table 1: Comparison of GFS Data with Fiscal Design Survey of OECD (1999) Country GFS Expenditure Share 2 Revenue Share 2 Composition of Subnational Revenues