Horizontal Versus Vertical Tax Competition: An Empirical Test

In a federation, where different levels of government tax the same base, one can observe two types of externalities: a horizontal externality, working among jurisdictions of the same hierarchical level and leading to tax rates that are suboptimally low; and a vertical externality, working between different hierarchical levels of government and leading to suboptimally high tax rates. By parameterising a model of taxation in a federation that incorporates those two forces (Keen and Kotsogiannis, AER, 2002), we develop an empirically implementable discriminating hypothesis to distinguish horizontal from vertical tax-setting externalities. In essence, the test predicts that increasing jurisdictional fragmentation of a federation lowers the average tax rate of the lower-level jurisdiction if horizontal externalities dominate, but raises their average tax rate if vertical externalities dominate. This test is applied to a new and original data set on Swiss municipal and cantonal taxes, exploiting Switzerland's uniquely decentralised three-tier hierarchical structure of tax-raising powers. The main attraction of this data set is that it allows comparisons across federations (i.e. cantons) as well as within federations (over time), while existing studies of vertical tax externalities had to rely on pure time-series data. Using panel data models with instrumental variables to account for the potential endogeneity of the tax rates applied by higher-level jurisdictions, we estimate the theory-based discriminatory relationship between municipal tax rates and the jurisdictional fragmentation of cantons. This is done separately for corporate and personal tax rates applying to different representative types of tax payers. We find evidence of both vertical and horizontal externalities, depending on the type of tax and tax payer