Generalized Comparative Statics for Political Economy Models

The Median Voter Theorem is an extremely popular result in Political Economy that holds only if the policy space is unidimensional. This assumption restricts its use to a class of very simple problems. In most applications in the literature this implied an oversimplification of the problem studied, which is one of the possible explanations for the lack of empirical support for several predictions derived with this tool. In this paper I show that under suitable restrictions on individual preferences a Median Voter Theorem can be derived even if the policy space is multidimensional and I derive the comparative statics of the resulting model induced by a change in the pivotal voter. I show that this tool can invalidate the predictions of the Meltzer-Richard model of size of goverment and that it can be useful to study other Political Economy problems that cannot be analyzed using the traditional framework, including games in which players have a richer strategy set than the policy vector to be chosen.