Information and extremism in elections

We model an election in which parties nominate candidates with observable policy preferences prior to a campaign that produces information about candidate quality, a characteristic independent of policy. Informative campaigns lead to greater differentiation in expected candidate quality, which undermines policy competition. In equilibrium, as campaigns become more informative, candidates become more extreme. We identify conditions under which the costs associated with extremism dominate the benefits of campaign information. Informative political campaigns increase political extremism and can decrease voter welfare. Our results have implications for media coverage, the number of debates, and campaign finance reform.