Electronic copy available at: http://ssrn.com/abstract=972670 Corporate Political Contributions: Investment or Agency? Rajesh K. Aggarwal Carlson School of Management University of Minnesota 321 19th Avenue South, Room 3-122 Minneapolis, MN 55455 612-625-5679 aggar015@umn.edu Felix Meschke KU School of Business University of Kansas 1300 Sunnyside Avenue Lawrence, KS 66045-7585 347-433-5495 meschke@ku.edu Tracy Wang Carlson School of Management University of Minnesota 321 19th Avenue South, Room 3-122 Minneapolis, MN 55455 612-624-5869 wangx684@umn.edu Draft: November, 2011 We thank Jane Buchan, Craig Brown, Mara Faccio, Eliezer Fich, Murray Frank, Timothy Guinnane, Michael Johannes, Brandon Julio, Jonathan Koppell, Joshua Pollet, Paul Povel, Paolo Volpin, Jeff Zwiebel, and conference and seminar participants at the Western Finance Association Meetings, the European Summer Symposium in Financial Markets in Gerzensee, the Financial Intermediation Research Society Conference, the European Finance Association Conference, the Midwest Finance Association Conference, the NUS Corporate Finance Conference, the Asian Finance Association Annual Conference, the Chinese International Conference in Finance, the Yale Conference on Shifting Capital Markets and Corporate Performance, Claremont-McKenna College, University of Alabama, Indiana University, Georgia State University, University of Cincinnati, the Carlson School of Management, and the University of Minnesota Law School for helpful comments. We thank the Millstein Center at the Yale School of Management for financial support. Mufaddal Baxamusa, Truong Duong, and Yihui Pan provided excellent research assistance.

Electronic copy available at: http://ssrn.com/abstract=972670 1 Corporate Political Contributions: Investment or Agency? Abstract: We examine corporate contributions to political candidates for federal offices in the United States from 1991 to 2004. Firms that donate have operating characteristics consistent with the existence of a free cash flow problem, and donations are negatively correlated with returns. A $10,000 increase in donations is associated with a reduction in annual excess returns of 13.9 basis points. Worse corporate governance is associated with larger donations. Even after controlling for corporate governance, donations are associated with lower returns. Donating firms engage in more acquisitions and their acquisitions have significantly lower cumulative abnormal announcement returns than non-donating firms. We find virtually no support for the hypothesis that donations represent an investment in political capital. Instead, political donations are symptomatic of agency problems within firms. Our results are particularly useful in light of the Citizens United ruling, which is likely to greatly increase the use of corporate funds for political contributions.

Electronic copy available at: http://ssrn.com/abstract=972670 2 Introduction Why do companies make donations to political parties and to candidates for political office? Companies do not have political preferences per se. They may, however, have an economic interest in various legislative actions, regulatory decisions, or other political outcomes. Thus, one reason why companies may donate is to influence the political process in ways that improve firm performance. An alternative reason why companies may donate is that, while companies do not have political preferences, their managers do. In this view, political donations need not be associated with firm performance and, in fact, may damage firm returns as they represent a form of perquisites consumption for the firm���s managers. This form of perquisites consumption is often not transparent or visible to shareholders, and can indicate wider agency problems at the firm. While these two possibilities for why firms donate are not mutually exclusive, they do have very different implications for the firm and its shareholders. Under the first view, donations are an investment in political capital that should, in expectation, generate positive returns for the firm. Under the second view, donations are symptomatic of an agency problem that should lower returns for the firm. We examine which of these views more accurately characterizes political donations in the United States. We use comprehensive data on corporate political donations from the Center for Responsive Politics. Our data span the years 1991 to 2004 and include the four main types of political donations: political action committee (PAC) donations, donations by individuals affiliated with a company, soft money donations, and donations to 527 Committees. We focus specifically on donations made directly from corporate funds���soft money donations and donations to 527 Committees. Of the universe of publicly- traded firms, only 11.27% or 1,381 firms donate directly from corporate funds during our sample period. Our focus on donations made from corporate funds is particularly salient in the wake of the Supreme Court decision in Citizens United v. Federal Election Commission, which essentially eliminates restrictions on the use of corporate funds in elections. We discuss this in greater detail in Section 2. We find that firms that donate directly from corporate funds have operating characteristics consistent with the firms��� facing a free cash flow problem���they are large, slowly-growing firms that have more free cash flow, yet engage in less R&D and investment spending. We also find that donations are negatively correlated with future excess returns. An increase in soft money and 527 Committee donations of $10,000 is associated with a reduction in excess returns of 7.4 basis points in the following year. Similar to Yermack���s (2006) result for CEO personal use of corporate aircraft, this reduction in shareholder value far outstrips the dollar value of the donations. This suggests that political donations may be a useful window into the presence of wider agency problems within firms. To address endogeneity concerns, we instrument for donations and find similar negative associations between donations and returns.

3 Third, we find that better corporate governance (smaller boards, CEOs who are not also chairman of their board, less abnormal CEO compensation, larger block ownership, larger institutional ownership) is associated with smaller donations. However, even after we control for corporate governance, we continue to find a negative association between donations and excess returns. Thus, while better governance may attenuate donations, the negative association between donations and excess returns is not wholly attributable to an omitted governance effect. Fourth, we find that firms that make political donations engage in more acquisitions than do non-donating firms. Further, donating firms��� acquisitions have significantly lower cumulative abnormal announcement returns than do non-donating firms��� acquisitions. We also examine whether donations to the winning party in an election lead to higher event-study returns. We find positive one-day abnormal returns to donating to the winning party in Presidential election years and Congressional election years. This is consistent with the prior literature and has been viewed as supporting the investment hypothesis. For longer-horizon returns, we find no evidence of a positive effect of donating to winners in either Presidential or Congressional elections. Importantly, we generally find that donating to either winners or losers is associated with worse returns than not donating at all. This provides us with a direct test of the two hypotheses, and our findings are consistent with the agency hypothesis but hard to reconcile with the investment hypothesis. We also consider the possibility that politicians may do favors for firms and then firms may donate to politicians. If so, a positive effect of political favors on firm returns may occur prior to the donations. To address this, we examine returns for firms that start donating when there is a shift in political control, since these firms are unlikely to be donating in response to past favors. In this case, we still find a strong negative association between firm returns and starting to donate to a party that wins political control. Thus, our results cannot simply be driven by firms choosing to donate after political favors are bestowed. In addition, we examine a subsample of companies in industries that receive government contracts since for these companies donations are more likely to be an investment in political capital. We do not find evidence of the hypothesized positive association between donations and returns. Overall, we find only limited support for the hypothesis that political donations directly from corporate funds represent an investment in political capital. Taken together, our results suggest that political donations are reflective of an agency problem. Given the magnitude of the destruction of shareholder value that we document, it is more plausibly the case that corporate political donations are symptomatic of wider agency problems in the firm. The existing literature���which we review in the next section���has largely focused on short-term event studies that examine whether corporate political connections or donations are valuable. While event studies can detect changes in firm value attributable to certain events (e.g., elections), they do not measure potential offsetting costs due to potential agency problems from the donations themselves. In contrast, we

4 test two competing hypotheses���donations as reflective of an agency problem between managers and shareholders versus donations as an investment in political capital. In order to do so, our approach differs along three dimensions. First, we use the direct monetary amount of donations, which allows us to test the two competing hypotheses, since donations can either be an investment or a measure of perquisites consumption. Second, our focus on soft money donations and 527 Committee donations, which come directly from corporate resources, eliminates concerns that PAC donations are only an indirect form of corporate donations that cannot be attributed to agency concerns. Third, we study long-horizon returns, which help us to formally distinguish between the agency hypothesis and the investment hypothesis. Our focus on stock returns is appropriate given that political donations over our sample period are not disclosed in a transparent or timely fashion to shareholders, and are therefore likely to be capitalized into stock prices only slowly over time. 1. Literature Review and Experimental Design This paper contributes to a growing literature that looks at the intersection of politics and finance. Several recent papers present event studies examining the effect of the 2000 U.S. Presidential elections on companies aligned with Republicans or Democrats, including Knight (2006), Goldman, Rocholl, and So (2009), Shon (2010), and Jayachandran (2006). In general, these papers find that good (bad) news for Republicans is associated with a positive (negative) stock price reaction for Republican-leaning firms and a negative (positive) stock price reaction for Democratic-leaning firms. These papers necessarily focus on short-run returns, small samples, and isolated events, in contrast to our study. Cooper, Gulen, and Ovtchinnikov (2010) find evidence of a positive effect of donations on firm value and we discuss their results in greater detail in Section 4. By contrast, other studies document the ineffectiveness of campaign donations as a form of gaining influence or buying favorable policies. Ansolabehere, de Figueiredo, and Snyder (2003) show that when one controls for unobserved constituent and legislator effects, there is little relationship between money and legislator votes. Ansolabehere, Snyder, and Ueda (2004) examine the excess returns of firms that give large amounts of soft money and firms that give no soft money, and changes in those excess returns around five key events in the approval of the Bipartisan Campaign Reform Act of 2002. They find no noticeable effect on the valuation of Fortune 500 firms that give large amounts of soft money relative to the firms that give no soft money. We note that much of the prior literature uses short-term event studies to examine the effect of political donations on firm returns. We examine long-term (one-year) stock returns instead of short- horizon stock returns for three reasons. First, a short-run event study around an election does not allow us