A model of candidate location with endogenous valence

This work gives a theoretical explanation for the increase in campaign spending and party polarization in U.S. politics. I assume that the effectiveness of persuasive advertising, and of costly valence campaigning in general, positively depends on the share of nonpartisan voters. A decline in the number of partisan voters in a constituency results in greater campaign spending by the candidates. If the voters are risk-averse, the candidates who maximize their expected office rents minus the cost of campaigning will choose divergent policy platforms strategically in order to reduce the costs of subsequent campaign spending. The degree of policy divergence positively depends on the share of nonpartisan voters for a broad class of voter disutility and candidate cost of valence functions.