INTRODUCTION

According to a widely held view about American politics, the two parties are indistinguishable and both located in the middle of the political spectrum. We emphatically disagree.

The notion of convergence of party platforms emerges from the initial rational choice models of two-party electoral competition in the contributions of Hotelling (1929), Black (1958), and Downs (1957). These seminal works assumed that although voters care about public policy, candidates care only about winning.' In this situation, the candidates propose identical platforms: one should observe complete convergence of policies toward the middle. Which policy is actually chosen by a candidate depends on the specification of the electoral process: the number of issues involved, the dissemination of information within the electorate, the nature of voters' preferences, and the ability of voters to make decisions consistent with their preferences.

The most famous convergence result is the “median voter theorem” (Black (1958)). It implies that, under a certain specification, the candidates converge to the policy most preferred by the median voter. For other specifications, Hinich (1977), Coughlin and Nitzan (1981), Ledyard (1984), and Tovey (1991) analyze cases of convergence not at the median. For our purposes, the important point is not so much “where” the candidates converge (at which particular point of the political “middle”) but the fact that they fully converge.

A different view of party competition holds that political parties, as well as the voters, care about policy outcomes, in addition to winning per se (Wittman (1977, 1983), Calvert (1985), and Roemer (1992)). According to this view, political parties represent the interests and values of different constituencies.