Fiscal Deficits and Type of Government: A Study of Spanish Local Elections

The literature on public choice has largely argued that when several actors are part of a decision-making process, the results will be biased towards overspending. However, the empirical studies of the effect of minorities and coalition governments on spending have yielded mixed support for this theoretical claim. This chapter argues that the inconclusiveness of the empirical evidence is related to problems of standard regression models to accurately capture unobserved heterogeneity. We use data from Spanish municipalities for the period 2004–2011 to compare the results of four typically used estimation methods: mean comparison, OLS, fixed-effects regression and matching. We argue that out of these models, matching deals better with unobserved heterogeneity and selection bias of the type of government, allowing us to reduce estimating error. The results show that, when we account for these problems in a matching model, minorities run lower surpluses than single party majorities. This result did not arise in simple mean comparisons or OLS models, or even in the fixed-effects specification. These results give support to the law of 1/n (Weingast, Journal of Political Economy 96: 132–163, 1981) and also underscore that in order to identify correctly the impact of government characteristics on policy-making, we need to understand that these are not randomly assigned across our units of observation. This advises the use of more quasi-experimental methods in our empirical research.