A dynamic model of primaries

Primary elections are increasingly popular around the world, but historically political parties have typically chosen their candidate selection mechanisms in a decentralized manner. We develop a theory that accounts for variation in the use of primary elections in these settings. In our model, two parties choose candidates for general elections over an infinite horizon. Each party has an elite and a nonelite faction, where the elite faction can choose whether to hold primaries or nominate itself. Primaries produce more electable candidates, but losing a primary also deprives elites of private goods and future elite status. The model predicts that parties adopt primaries under high ideological polarization or when they are electorally disadvantaged. Additionally, we show how rigidities in the ability of winners to change candidate selection mechanisms can increase the universal adoption of primaries in electorally volatile environments.