An Empirical Spatial Model of Congressional Campaigns

Abstract

Testing and estimating formal models of political behavior has not advanced as far as theoretical applications. One of the major literatures in formal theory is the spatial model of electoral competition which has its origins in the work of Black (1948) and Downs (1957). These models are used to make predictions about the policy positions candidates take in order to win elections. A lack of data on these candidate positions, especially challengers who never serve in Congress, has made direct testing of these models on congressional elections difficult.Recently, researchers have begun to incorporate campaign finance into the standard Downsian model. These models of position-induced contributions examine the tradeoff that candidates make between choosing positions favorable to interest group contributors and positions favorable to voters. A major premise of these models is that interest group contributions are based on the policy positions of candidates. This has been borne out empirically in the case of incumbents, but not challengers.To test key hypotheses of these models, we develop a simple spatial model of position-induced campaign contributions where the PAC's decision to contribute or abstain from each race is a function of the policy distance between the PAC and the candidates. We use data from political action committee contributions in order to estimate the locations of incumbents, challengers and PACs. Our model reliably estimates the spatial positions as well as correctly predicts nearly 74 percent of the contribution and abstention decisions of the PACs. Conditional upon making a contribution, we correctly predict the contribution in 94 percent of the cases. These results are strong evidence for position-induced campaign contributions. Furthermore, our estimates of candidate positions allow us to address issues of platform convergence between candidates.