Corporate Political Contributions and Stock Returns

In our paper Corporate Political Contributions and Stock Returns, which was recently accepted for publication in the Journal of Finance, we study whether there is a robust relation between firm contributions and contributing firm returns. Using data from the U.S. Federal Election Commission (FEC), we create a new and comprehensive database of publicly traded firms’ political action committee (PAC) contributions to political campaigns in the U.S. from 1979 to 2004. After merging the FEC contributions data with CRSP/Compustat data, we have approximately 819,000 contributions made by 1,930 firms over the past twenty five years or so – thus, we have a remarkably rich dataset to test for systematic contribution/return effects arising from publicly traded firms’ involvement in the U.S. political process. Our sample captures over 70% of the total dollar volume of all hard-money corporate contributions and represents on average 60% of the market value-weighted capitalization of all publicly traded firms in the U.S.

We develop a simple measure to describe firms’ political contribution practices that takes advantage of the comprehensive nature of the FEC data. We view each firm as supporting a portfolio of candidates and simply sum up, over a rolling multiyear window, the number of candidates that each firm supports. We find that the average firm participating in the political donation process contributes to 73 candidates over any five-year period, 53 of whom go on to win their elections. There is substantial variability across firms in the number of supported candidates, with a standard deviation of approximately 96 candidates.

We perform panel regressions of annual abnormal returns on the lagged number of supported candidates and other control variables. We find that the number of supported candidates has a statistically significant positive relation with future abnormal returns for firms which contribute to political candidates. The relation is evident in univariate regressions of abnormal returns on the number of supported candidates as well as in multivariate regressions after controlling for other established predictors of returns such as book-to-market, firm capitalization, and momentum, measured by lagged 12-month buy-and-hold returns. We find that our results are robust to three alternative contribution definitions: the total strength of the relationships between candidates and the contributing firm (as measured by the length of the firm-candidate relationship), the ability of the candidates to help the firm (as measured by the home state of the firm and the candidate), and the power of the candidates (as measured by a candidate’s committee ranking). We document especially strong effects for a measure related to the ability of the candidate to help the donating firm. Thus, the contribution effect appears to increase for firms that have longer relationships with candidates, support more home candidates, and support more powerful candidates.

We further break the contributions data up into House and Senate categories. We find that there is an incremental House effect after controlling for the Senate effect, although contributions to both branches of government result in positive economic effects for the contributing firms. Our finding of an incremental effect for firms supporting House candidates may be related to the constitutional provision that revenue and appropriations bills must originate in the House. Thus, firms may find that it is more expedient to support House members, where potential firm value increasing actions may be more suitably created. We also split our sample along political party lines. The FEC data show that Republican candidates typically receive higher total dollar contributions than do Democrats and that Republican candidates’ contributions come from a larger number of supporting firms than do Democrat candidates’ contributions. However, despite the fact that Republicans receive more contributions than Democrats, we find an incremental contribution effect for Democrats after controlling for Republican effect, but do not find an incremental Republican effect after controlling for the Democrat effect.

The full paper is available for download here.