We examine the interaction between partisan affect and elite polarization in a behavioral voting model. Voting is determined by affect rather than rational choice. Parties are office-motivated; they choose policies to win elections. We show that parties bias their policies toward their partisans if voters exhibit ingroup responsiveness, i.e., they respond more strongly to their own party’s policy deviations than to policy deviations by the other party. Our results suggest that affective polarization is a driver of the growing elite polarization in American politics. Importantly, this observation does not assume any shifts in the voters’ bliss points and is therefore orthogonal to the controversy over whether the American electorate has become more polarized in ideology.

We provide an instrumental theory of extreme campaign platforms. By adopting an extreme platform, a previously mainstream party with a relatively small probability of winning further reduces its chances. On the other hand, the party builds credibility as the one most capable of delivering an alternative to mainstream policies. The party gambles that if down the road voters become dissatisfied with the status quo and seek something different, the party will be there ready with a credible alternative. In essence, the party sacrifices the most immediate election to invest in greater future success. We call this phenomenon tactical extremism. We show under which conditions we expect tactical extremism to arise and we discuss its welfare implications.