We consider a model of two-candidate elections with a one-dimensional policy space. Spending on campaign advertisements can directly in ‡uence voters' preferences, and contributors give the money for campaign spending in exchange for promised services if the candidate wins. We …nd that the winner of the election depends crucially on the contributors' beliefs about who is likely to win, and the contribution market tends towards nonsymmetric equilibria in which one of the two candidates has no chance of winning. If the voters are only weakly in ‡uenced by advertising or if permissible campaign spending is small, then the candidates choose policies close to the median voter's ideal point, but the contributors still determine the winner. Uncertainty about the Condorcet-winning point (or its nonexistence) can change these results and generate equilibria in which both candidates have substantial probabilities of winning.
I IntroductionAs candidates compete for votes in an election campaign, they also compete for contributions to …nance their campaigns. To develop a theory of elections, we need to understand the e¤ects of both of these levels of competition, and the interactions between the two levels. When candidates raise funds in return for promises of services after election, a candidate whom contributors expect to lose may be unable to raise funds. Thus the contributors'expectations may become a selfful…lling prophecy, if campaign spending is needed to win. In such situations contributors' expectations can be a decisive exogenous factor in determining the winner of an election. In this paper we show that such equilibria can arise even in electoral situations where voters are only marginally in ‡uenced by campaign spending, if there is a known Condorcet winning position.Furthermore, we demonstrate that less extreme equilibria in which contributors expect two candidates to both have positive probabilities of winning (see for instance Baron 1989a) maybe unstable. In such situations candidates may …nd that their most important activity is the manipulation of contributors'perceptions at the beginning of the campaign.It is di¢ cult to develop an intuitively reasonable and logically complete model of elections that integrates fund-raising competition with vote-getting competition (for general surveys of models of campaign contributions see Morton and Cameron (1991), Austen-Smith (1996) Baron (1989aBaron ( ,b, 1994, Baron and Mo (1993), Hinich andMunger (1989), and Snyder (1991), then the signalling approach begs the question, should the primary message from campaign spending be that the candidate who spends 1 the most has also sold the most promises to special-interest contributors? It is hard to imagine that voters would …nd candidates that provide such favors attractive, although Ashworth (2006), Coate (2004), andPrat (2002) demonstrate that rational votes may have such preference since candidates voters prefer on quality or ideological grounds are forced to obtain contributions as the only way to signal their ...