Essays on electoral accountability and local public finance

This study consists of three essays on electoral accountability and local public finance. ‘A comprehensive test of Yardstick Competition in the Italian Municipalities’ tests the Yardstick Competition hypotheses on a dataset of Italian Municipalities during the period 1995-2004, focusing on the local property tax rate on the main dwelling (ICI abitazione principale). First, a vote popularity equation is estimated with instrumental variables, including an original measure of inter-jurisdictional comparison between the domestic tax rate and the average neighbors’ tax rate. The findings verify popularity concerns of the incumbent, robust to alternative definitions of popularity. Then, we estimate a spatial tax setting equation detecting interactions driven by the positive spatial lag coefficient. Given these evidence, we conclude that Yardstick Competition is present in the dataset. ‘The time dynamics of Yardstick Competition in the Italian Municipalities’ investigates the pattern of strategic interaction from 1995 to 2004. The literature identifies Yardstick Competition with the average interaction in time; we move forward, by estimating the spatial tax setting equation on subsequent time subsamples of the dataset to capture the variation of the interaction due to the introduction of a marginal year. The results show a converging trend towards the lowest level of spatial correlation. This pattern is especially evident in those Regions with a higher density of municipalities, where there are more opportunities to make inter-jurisdictional comparisons. This evidence suggests that the informational spillover generating Yardstick Competition, contrary to the assumption in the literature, changes over time. ‘Do voters learn from past experience? Yardstick Competition and political selection’ extends the model of Yardstick Competition by assuming that the stock of information available to voters accumulates over time. The theoretical results show that when past mimicking is observed there is a range of values of the weight attached to past experience for which the less competent incumbent would not be re-elected. If voters do not observe past mimicking, however, successful mimicking is always feasible. The predictions of the model are tested on an electoral cohort of Italian Municipalities by estimating a probit regression where the dependent variable is the dummy for reelection of the incumbent and the variables of interest are the voters’ beliefs on the tax rate computed through dynamic Bayesian updating. The results, however, fail to verify the predictions of the model as the coefficient associated to the updated belief is never statistically significant. ‘Asymmetric information and Political Budget Cycles: the effect of the local diffusion of newspapers’ examines the expenditure cycles in the Italian Regions from 1984 to 2008 and their nexus with voters’ awareness, proxied with the local diffusion of newspapers. We estimate a dynamic expenditure equation with Least Squares Dummy Variable Corrected (Bruno, 2005) accounting for the small size of the dataset. The results, robust to different specifications of the econometric model, find cycles in total expenditure and in capital expenditure before the electoral and fiscal reforms in the mid-90s. During this period the diffusion of newspapers constrains the electoral expenditure of those same items; the diffusion of generic newspapers is associated to a larger effect than the diffusion of economic newspapers, suggesting stronger popularity concerns towards the newly informed voters. The analyses does not detect neither expenditure cycles nor any effect of the diffusion of newspapers after 1995, indicating a possible shift from a cycle in the size of expenditure to a less visible type of signaling, the cycle in the composition of expenditure.