Grant legislation vs. political factors as determinants of soft budget spending behaviors. Comparison between Italian and French regions.

• Author: Josselin, Jean-Michel
• Position: Report

Paper presented at the 2012 PEARLE Seminar, IREF-Condorcet Centre for Political Economy seminar, Public Choice Society meetings in Miami and CREMCNRS seminar. We thank all participants to those conferences and especially Enrico Colombatto, Matz Dahlberg, Emma Galli, Mika Kortelainen, Antonello Maruotti, Gerard Roland, Gilberto Turati, Francisco Veiga and three anonymous referees for helpful comments on previous versions of this paper. Financial support from the IREF is gratefully acknowledged. The usual caveat applies.

1. Introduction

   The aim of this paper is to bring the study of soft budget spending behavior into a comparative framework of analysis. Interpreted in an intergovernmental principal-agent setting, soft budget spending behavior is the result of a dynamic commitment problem, caused by the lack of a "commitment technology" of the central government (the principal) that allows the local governments (the agents) to spend more than their current available resources because they can rationally expect that, in the future, the central government will solve their financial problems by granting them more transfers (Kornai et al., 2003; Qian and Weingast, 1997; Prudhomme, 1995). The empirical literature on the issue has so far tested the implications of the dynamic commitment problem--chiefly the binding force of the rules that discipline the financial relationships between different government tiers and the determinants of these transfer expectations--by means of a variety of empirical strategies and in different samples (Padovano, 2012; Bordignon and Turati, 2009; Pettersson-Lidblom, 2010; Rodden et al. 2003). The common feature to all these studies is the focus on a single institutional setting: they all consider samples drawn from a single country. This approach has the obvious advantage of applying a ceteris paribus condition to the existing institutions, which allows evaluating the credibility of their commitment potential and, by contrast, transfer expectations originating from factors not directly linked to the legislation on the distribution of transfers. It is precisely these "non-institutional" phenomena that, according to theory (Kornai et al., 2003; Rodden et al. 2003), drive the "soft" component of spending behaviors.

   Yet, focusing on just one country makes it impossible to analyze several issues that have never been examined so far in the literature, such as: a) Comparing the binding force of alternative sets of rules that discipline intergovernmental financial relations, to assess which is better able to solve the dynamic commitment problem; b) Comparing the contribution of transfer expectations to the determination of soft budget spending behavior in alternative institutional frameworks; c) Comparing how relevant are political factors, such as alignment effects, "too big to fail" effects, the need to secure local public capital and, more generally, political exchanges and personal contacts between politicians at various government levels, in determining soft budget spending behaviors in different institutional contexts and countries; d) Checking whether and why these political factors play a complementary, or a substituting, role with respect to the standing legislation in the process of grant distribution. More broadly, the adoption of a comparative analysis to the problem of soft budget spending behavior allows verifying the generality of the conclusions reached so far by models estimated on single country samples.

   In this paper, we innovate on the literature by providing the first (to our knowledge) comparative analysis of the role played by alternative financial rules and political factors in solving, or in exacerbating, the dynamic commitment problem that generates soft budget spending behaviors in lower tier governments. To minimize the loss of explanatory advantages due to abandoning the ceteris paribus condition inherent in single country studies, we consider two democracies similar for their cultural legal background, the level of economic development and the type of vertical organization of the state: Italy and France. Both are unitary states with a significant level of decentralization (OECD, 2002; Stegarescu, 2005). Within them, we examine the same intergovernmental relation liable to generate soft budget spending behaviors, namely the distribution of transfers from the central government to the regions. In both countries, regions constitute the government level immediately below the central one: this ensures that grant distribution is both direct and does not get dispersed in a myriad of government units, as regions amount to only 20 in Italy and to 22 in France. As we shall see later on, grant distribution in the two countries differs with respect to one main institutional feature: in France, the legislation that disciplines intergovernmental financial relations is designed to ensure that the distribution of grants to regional and local governments be horizontal, i.e., such that variations in the amount of transfers to one region are matched by proportional variations in funds transferred to all other regions. In Italy, on the other hand, grant distribution is much more redistributive, with the possibility of asymmetries in the (per capita) amounts transferred to each region. The distinction between "special" and "ordinary statute" regions and the presence of re-equilibrating funds aimed at redressing the development gap between the Centre-North regions and those of the Mezzogiorno are examples of the asymmetric nature of grant distribution in the Peninsula.

   The empirical strategy that we adopt in this paper serves two purposes. The first is to disentangle the role played by the legislation disciplining grant distribution on the one hand, and by political factors on the other in determining soft budget spending behavior. The second is to compare evidence related to each of these two dimensions drawn from each country. To perform the first task, we begin by estimating a funding equation, where the amount of per capita transfers, in each region within a country, is a function only of the variables indicated by the legislation regarding grant distribution. Thus, the first round of estimates captures the role played by grant legislation in the allocation of transfers to regions. We then augment this equation by a series of proxies for transfer expectations, drawn from the political economy and public choice literature on the distribution of transfers (Padovano, 2012; Kornai et al., 2003). The explanatory power of these covariates approximates the role played by "politics" in the distribution of grants. The comparison of the respective significance of grant legislation and political variables assesses the relative role played by these two sets of factors in the distribution of transfers to regions in France and Italy. Then, in order to evaluate the impact of those factors on soft budget spending behavior, we estimate an autoregressive empirical model, which is the standard estimating procedure in the analysis of bailout and transfer expectations (Rodden, 2005; Bordignon and Turati, 2009; Padovano, 2011). Specifically, we interpret the fitted values of the most comprehensive model, i.e., the one that includes the covariates related to both grant legislation and political factors, as the rational expectations at time t that regional governments formulate about the transfers they are going to receive at time t+7. This interpretation is legitimate insofar as the fitted values include all the relevant information about the process that generates the distribution of grants available at time t: to ensure that, special attention will be given to consideration of the widest possible set of explanatory variables on the right-hand side of the funding equation. To assess the role of these transfer expectations on regional spending, i.e., to evaluate and compare the dimension of soft budget spending behaviors, we introduce the fitted values into an equation that estimates regional spending. Insofar as transfer expectations affect spending, we obtain an estimate of the role played by transfer expectations on regional expenditures; in other words, we have an assessment of the extent of soft budget spending behavior, "soft" because it is not tightly constrained by the standing legislation about grant distribution. Naturally, to compare the relevance of these phenomena in Italy and France, we apply a similar empirical strategy to the data from both countries. Specifically, two data sets are used for panel data analysis: the one for Italy comprises 19 regions and 2 autonomous provinces studied between 1996 and 2007; for France, 22 regions are considered over the period 1995-2006. The use of basically the same estimating equation for both countries facilitates a comparative interpretation of the estimated coefficients. The panel structure of the datasets allows controlling also for policy changes occurring within a country through time (1).

   To anticipate the results of the estimates, evidence of soft budget spending behaviors is found in both countries. Transfers are in fact partly related to political determinants; their distribution is not uniquely driven by fiscal rules, grant legislation or by automatic formulas. This generates expectations of more transfers in the future, which in turn fuel spending behaviors. In France, political expediencies do not appear to produce significant differences in transfer decisions across regions, yet soft budget spending behaviors are still detected in all regions taken together. In Italy, institutions and political phenomena concur in determining differential treatments in central government's transfer decisions and in regional government transfer expectations. These results are consistent with the two alternative structures of the transfer legislations of the two countries. As far as the role of political factors is concerned, the estimates show...