Money and votes: A simultaneous equation model

Conclusions

This paper has specified and estimated a model of the interaction between money and voting in American politics. Convincingly establishing the identification of a system of equations can be difficult even when theory provides a guide. In the study of money in politics, however, no theoretical perspective has been established in the literature. To make the case for identification as unambiguous as possible, this paper has done two things. First, the specification was based, in part, on theory which has formally been developed elsewhere and which satisfactorily describes data at the economic interest group level (Welch, 1980). Second, part of the question of identification revolves around the relationship between party strength and socioeconomic variables. Because there are questions as to the best measurement of party strength, results based on two measures were presented. For a number of reasons, the Presidential vote was the preferred measure.

This study draws four conclusions. First, money influenced voting in U.S. House elections in 1972, although its effect was ‘small’. Second, patterns of total contributions to a U.S. House candidate are consistent with the hypothesis that a majority of dollars come from contributors who prefer to support likely winners. Third, the total contributions made to a candidate are a positive function of the district's median years of schooling completed and the Gini coefficient. Total contributions made to a Democratic candidate are also a positive function of the median income. Finally, had House campaigns been financed in the early 1970s, even to the point of giving each major party candidate $150,000 (in 1978 dollars), few elections would have been changed.

It is also worth mentioning some of the things about the financing of political campaigns that we still do not know. The above conclusions involve only elections to the U.S. House; there are good reasons for suspecting that money is more influential in U.S. Senate elections. OLS statistics support this belief and the greater visibility of those races help justify it. Although money has little effect on House elections, we do not know whether contributions affect legislative voting in the House. Representatives apparently believe that expenditure influences their reelection chances. Given the complexity of the process, it should be no surprise that this belief conflicts with our findings. Acting on this belief, a Representative's vote may very well be influenced by the sources of his contributions, either directly or indirectly through the giving of access to contributors. Finally, we should be reminded that the above model is static while the electoral process is dynamic. The amount contributed to a candidate is affected by his probability of winning in the previous time period (e.g. week, month) and in turn affects that probability in the following time period. As usual, what we know is overshadowed by what we do not know. The above questions are both interesting in their own right and of interest to policy making.