Learning and Coordination in the Presidential Primary System∗

To analyze the advantages and disadvantages of the U.S. presidential primary system, we develop a model in which candidates with different policy positions and qualities compete for the nomination, and voters are uncertain about the candidates ’ valence. This setup generates two effects, which we call vote-splitting (i.e., several candidates in the same policy position compete for the same voter pool) and voter learning (as the results in earlier elections help voters to update their beliefs on candidate quality). We analyze how different temporal organizations of primaries affect the trade-off between vote-splitting and voter learning: Sequential voting minimizes vote-splitting in late districts, but voters may coordinate on the wrong candidate. Using the parameter estimates obtained from the 2008 Democratic presidential primaries, we conduct policy experiments such as replacing the current system with a simultaneous system or other proposed systems