Abstract

This paper provides a unified theoretical and empirical analysis of three long-standing explanations for the consistent loss of support for the president's party in midterm Congressional elections: (1) a presidential penalty, defined as a preference for supporting the opposition during midterm years, (2) a surge and decline in voter turnout, and (3) a reversion to the mean in voter partisanship. To quantify the contribution of each of these factors, an econometric model is developed in which voters jointly choose whether or not to participate and which party to support in both house and presidential elections. Estimated using ANES data from both presidential and midterm years, the model can fully explain the observed midterm gaps, and counterfactual simulations demonstrate that each factor makes a sizable contribution towards the midterm gap, with the presidential penalty playing the largest role.