Primary Elections and the Provision of Public Goods∗

We develop a model of electoral competition in which candidates from two parties compete in a primary and general election. There are three groups of voters, two representing “core” supporters for each party and one “swing” group. In the primary election, each party’s core voters choose a candidate to run in the general election. Candidates within a party share a fixed ideological platform and can promise to distribute a unit of public spending across private goods for the three groups of voters and a public good that distributes benefits evenly across all voters. In a benchmark world without primary elections, candidates promise only the public good when its benefit level is sufficiently high, and only private goods when this benefit is low. Relative to this benchmark, primaries do not always increase the provision of public goods, but do reduce the threshold value of public goods needed for candidates to promise only the public good when the swing group is small. The model also predicts that primary elections shift platforms toward benefits for swing voters as core voters become more ideologically extreme, and that public good provision is non-monotonic in ideological polarization. ∗We thank Deborah Beim and seminar attendees at the University of North Carolina and the 2013 Yale CSAP Conference for useful comments and suggestions.