ESTIMATING A DYNAMIC GAME OF POLITICAL ADVERTISING

This dissertation studies the effect of political advertising on the outcomes of elections to the U.S. Senate between 2000 and 2018. I first provide descriptive empirical evidence that candidates' advertising choices are driven by their own ideologies and political experience, as well as by their opponents' advertising and their standings in the polls. I then develop and estimate a dynamic game of political advertising with endogenous ad content, namely whether an ad is positive or negative. Parameter estimates show that incumbents have a substantial fundraising advantage over challengers. Positive advertising that similarly benefits challengers and incumbents reinforces this incumbency advantage, while negative advertising that disproportionately benefits challengers works against it. A counterfactual public campaign financing policy that eliminates incumbents' fundraising advantages decreases the incumbent reelection rate by 3.6 percentage points. Hypothetical temporary bans on negative advertising increase the incumbent reelection rate, but decrease ideological extremism among the challengers who replace them.