discussion

Electronic copy available at: http://ssrn.com/abstract=2128608 Corporate Politics, Governance, and Value Before and After Citizens United First draft: December 16, 2011 Last revised: July 6, 2012 Accepted Journal of Empirical Legal Studies (forthcoming 2012) John C. Coates IV John F. Cogan, Jr. Professor of Law and Economics Harvard Law School Abstract How did corporate politics, governance and value relate to each other in the S&P 500 before and after Citizens United? In regulated and government-dependent industries, politics is nearly universal, and uncorrelated with shareholder power, agency costs, or value. But 11% of CEOs in 2000 who retired by 2011 obtained political positions after retiring, and in most industries, political activity correlates negatively with measures of shareholder power, positively with signs of agency costs, and negatively with shareholder value. The politics-value relationship interacts with capital expenditures, and is stronger in regressions with firm and time fixed effects, which absorb many omitted variables. After the shock of Citizens United, corporate lobbying and PAC activity jumped, in both frequency and amount, and firms politically active in 2008 had lower value in 2010 than other firms, consistent with politics at least partly causing and not merely correlating with lower value. Overall, the results are inconsistent with politics generally serving shareholder interests, and support proposals to require disclosure of political activity to shareholders.

Electronic copy available at: http://ssrn.com/abstract=2128608 2 Corporate Politics, Governance, and Value Before and After Citizens United * In Citizens United, the Supreme Court relaxed constraints on the ability of corporations to spend money on elections. In so doing, it rejected a shareholder-protection rationale for restrictions on spending, in part on the ground that shareholders are generally capable of defending their own interests through “corporate democracy.” 1 Another possible if unstated reason for the Court’s rejection of shareholder protection as a basis for restrictions on corporate political activity (CPA) is that there has been surprisingly little research focused on the relationships among CPA, corporate governance, and corporate value. This article explores those relationships in the S&P 500 before and after Citizens United. The paper finds that before and after Citizens United the data are consistent with companies engaging in political activities that reflect a mix of shareholder-oriented and non-shareholder- oriented goals. In regulated industries (e.g., banking, telecommunications) and in government- dependent industries (e.g., defense), political activity is nearly universal, and does not strongly correlate with measures of shareholder power, managerial agency costs, or value. In these industries, at least, where business strategy or revenues are directly linked to political decisions, it seems hard to imagine that shareholders of any given firm would benefit from unilateral political disarmament.

But the same intuition does not extend to most large public companies. The career paths of S&P 500 CEOs in 2000 points to another possible motivation for CPA: more than one in ten ex-CEOs later obtain political positions, including Cabinet-level appointments. This finding suggests that the extent or nature of the political activity of firms managed by those CEOs – as well as by other CEOs who have not yet left their CEO positions, or who died while CEO, or left under a cloud of scandal – could have at least partly been influenced by personal ambitions. In the majority of industries (e.g., apparel, retail, equipment), political activity is common but varied, and it correlates negatively with measures of shareholder power (shareholder concentration and shareholder rights), positively with signs of managerial agency costs (corporate jet use by CEOs), and negatively with shareholder value (industry-relative Tobin’s q). The negative value-politics relationship is particularly strong for firms making * Coates is the John F. Cogan, Jr. Professor of Law and Economics, Harvard Law School; email:

[email protected]. I extend thanks for comments and discussions on this and related papers to Dick Fallon, Frank Michelman, Rick Pildes, Noah Feldman, Jesse Fried, Heather Gerken, Jeffrey Drope, Andrew Metrick, Josh Fischman, Lucian Bebchuk, Ciara Torres-Spelliscy, Richard Briffault, Darius Palia, Semi Kedia, Mihir Desai, Nell Minow, Barak Orbach, Taylor Lincoln, Charles Fried, participants in workshops at Harvard, Boston College, Columbia, Georgetown, University of Virginia, New York University, Wake Forest, Boston University, the Brennan Center, and the Conference on Empirical Legal Studies, and three anonymous reviewers. Jin-Hyuk Kim graciously provided data on corporate political activity, David Yermack and GMI graciously provided data on corporate jet use by CEOs, and Gail Tan, Ashton Kingsman, Katherine Petti, Alex Trepp, Amanda Vaughn, and Jason George provided excellent research assistance. All errors are mine; all rights are reserved. 1 130 S. Ct. 876, 558 U.S. [ ] (2010) at 46 (quoting Bellotti, 435 U. S. 765, 794 & n. 34). The Court also asserted that the laws at issue in the case were poorly tailored to the goal of shareholder protection. For a critique of the Court’s legal and institutional analysis on that and other points, see Coates 2010a. 3 large capital expenditures, and is stronger in firm fixed-effects regressions than in cross- sectional regressions. The causal relationships between political activity and value likely run in both directions: politics may be one route for a troubled or stumbling firm to pursue to regain profitability, even as politics may distract senior managers and result in business investments that lack focus or are poorly fitted to a firm’s core business strategy. Consistent with this possibility, large capital expenditures have different effects depending on whether the firm making the investments are politically active – politically active firms making expenditures have lower value than other firms; politically inactive firms making expenditures have higher value. As a further test of whether the causal arrow runs at least in part from politics to value, Citizens United is examined as an exogenous shock to aggregate CPA. Although Citizens United changed the law only for “independent expenditures,” registered lobbying and PAC activity by corporations jumped in 2010, in both frequency and amount, particularly at firms that were already politically active in 2008, consistent with the finding that different modes of CPA act as complements – a fact well-established in prior research, as discussed below. Firms that were politically active in 2008 (“treatment” firms) had sharply lower industry-relative value in 2010 than other firms (“control” firms). Because pre-2010 declines in corporate value could not plausibly cause the Supreme Court to rule as it did, the relative decline in value at politically active firms after the decision is most simply explained by politics at least partly causing, and not merely correlating with, lower value.

These results are inconsistent with a simple theory in which CPA can be presumed to serve the interests of shareholders. The results in the politics-value regressions with firm and time fixed effects rule out many potential omitted variables by focusing on the relationship between same- firm changes in politics and changes in value. While unobserved characteristics of firms or managers change over time in tandem with changes in both political activity and value, the combination of the findings – relating to CEO careers, the relationships between firms and shareholder power, firms and signs of managerial agency costs, and value and politics, the industry-based differences in those cross-sectional relationships, and the results before and after Citizens United – are collectively difficult to explain with a model in which managers deploy firm resources solely to pursue firm or shareholder interests, at least in the largest US public companies.

The results have limits. No study of corporate politics can reflect an idealized controlled random double-blind study, in which randomly selected sample of companies were prohibited from engaging in politics, while otherwise identical companies were not. Even if lawmakers were inclined to try to conduct such a study, and even if corporate political resistance did not defeat efforts to do so, the holding of Citizens United makes it legally impossible by labeling CPA as shareholder “speech” protected by the First Amendment. The strength of the politics- value relationship is such that the correlations cannot be interpreted in a naïve way, with (for example) political expenditures being treated as the sole direct cause of lower corporate value. The near-universality of political activity in heavily regulated and government-dependent industries makes it difficult to study politics-value relationships in the very industries where the interests of managers, shareholders, and the polity are strongest. 4 With those caveats, the findings make it more plausible that CPA commonly reflects broader agency problems at large public companies. Together with the likelihood that unobservable political activity is also harmful to shareholder interests (perhaps even more so), the findings provide support for those engaged in efforts to respond to the legal shock to the shareholder- manager relationship at large public companies represented by Citizens United. The cumulative effect of the findings adds support for proposals to require disclosure of such activity to shareholders. 2 If Congress, states, or the SEC adopt rules attempting to give shareholders more information or more authority in the political sphere, the evidence presented here should help demonstrate that such legislation serves as a legitimate and compelling purpose separate from the anti-corruption and other purposes that have traditionally justified campaign finance laws. Contrary to the Supreme Court’s stated assumption, 3 shareholders were not able to protect themselves from misuse of corporate funds for political purposes prior to Citizens United, and the risk of such misuse has increased as a result of the decision. Part I briefly (a) describes the US Supreme Court’s decision in Citizens United, and (b) reviews relevant literatures on (1) corporate governance and its relationship to shareholder value, as measured in the corporate governance literature, and (2) CPA. Part II develops hypotheses to be tested, and describes the data used to test the hypotheses. Part III summarizes data on ex-CEO involvement in politics, corporate governance, CPA, and shareholder wealth. Part IV relates the data on CPA to the data on corporate governance and value. Part V summaries the empirical results and discusses possible interpretations and implications for law and policy. The paper then briefly concludes.

I. Legal Context and Prior Literatures A. Citizens United In Citizens United, the US Supreme Court decided that laws barring corporations (and unions) from making “independent” political expenditures (such as buying television ads supporting a candidate) were unconstitutional under the First Amendment. 4 Those laws banned corporations from actively campaigning in elections on behalf of politicians in the period from World War II through 2010. As a result, they curtailed the amount of money that corporations could spend on election activity, and constrained (though they did not eliminate) the ability of corporations to influence campaigns through contributions. Prior research (summarized below) has established that different kinds of political activity are complements. That is, if two kinds of political activity are employed, both will be more 2 See http://sec.gov/rules/petitions/2011/petn4-637.pdf (August 3, 2011 petition for rulemaking on disclosure of corporate political spending).

3 See text accompanying note 1 supra.

4 Throughout, the word “independent” is in quotes to reflect the fact that the “independence” of such expenditures is difficult to observe and is likely absent in many instances. For example, “Restore our Future” is a nominally “independent” political action committee (PAC) created by three former aides to Mitt Romney, is dedicated to “getting Romney elected president,” and received a $1 million donation from a privately held company (W Spann LLC) formed in March 2011 that promptly dissolved after making its donation. President Barack Obama’s former Deputy Press Secretary formed a PAC, Priorities USA, to back President Obama’s re-election bid. Isikoff 2011. 5 effective and are more frequently used than either on its own. By permitting a new type of corporate political activity – “independent” election expenditures, directly or through intermediaries – the logical implication of Citizens United would be to increase the effectiveness and incidence of other kinds of political activity by corporations. Citizens United generated a great deal of commentary and controversy. To date, however, few have examined the extent to which CPA changed in the 2010 elections, relative to prior periods, nor whether the cross-sectional correlates of that activity changed after the decision. 5 In particular, no study has yet to examine the relationship between corporate value and political activity in 2010, in absolute terms, or relative to prior years.

B. Corporate political activity: channels and regulations Before Citizens United, corporations were barred from donating corporate funds directly to candidates, and they continue to be barred from doing so. 6 Before Citizens United, corporations were permitted to establish political action committees (PACs), and they may still do so. Corporations could not (and still cannot) simply channel corporate funds through the PACs they establish (“connected” PACs) to candidates. 7 Instead, corporate officials must solicit donations to their connected PACs from corporate managers, employees and shareholders. However, corporations could and can continue to pay the fund-raising costs of their PACs, which can amount to a significant share of the nominal budget of the PACs – effectively, they shoulder the substantial fund-raising burden for the political candidates to which the PACs contribute.

Before Citizens United, corporations were also largely unconstrained from lobbying – that is, engaging in efforts to present information and otherwise persuade lawmakers, once elected, to pursue particular policies – and they may still do so. Nevertheless, pre-Citizens United laws limited the ability of corporations to influence the choice of lawmakers by voter. Since lawmaker time and attention is a limited resource, the ban on corporate electoral activities has limited the effectiveness of past lobbying efforts. 8 As discussed below, the complementary relationship between lobbying and election activity is established in the literature on CPA. All studies of CPA are challenged by the fact that only certain kinds of CPA are required to be disclosed, even by public companies. If Exxon hires a registered lobbyist or lobbying firm to act as such, the lobbyist and/or firm must disclose that fact, but nothing requires Exxon to disclose the fact that it may hire a law or public relations firm (not registered as a lobbyist) that 5 One recent study finds (among other things) that in 2010 and 2011 utilities spent the most on politics, that 5% of the firms in the S&P 500 disclosed a relationship with 501(c)(4) firms, 14% disclosed how much of their dues to trade associations are used for political purposes, and 20% disclosed direct political spending. IRRC 2011.

6 A recent court decision that the First Amendment permits corporations to make the same direct contributions as individuals in federal elections, U.S. v. Danielczyk, 788 F. Supp. 2d 472 (E.D. Va. May 26, 2011), has been appealed to the same appeals court that recently rejected a similar attempt to strike down a state law banning corporate contributions, see Preston v. Leake, 2011 U.S. App. LEXIS 22520 (4 th Cir. Nov. 7, 2011). 7 See Federal Election Commission Campaign Guide for Corporations and Labor Organizations (Jan. 2007), available at http://www.fec.gov/pdf/colagui.pdf (last visited Nov. 26, 2011).

8 Lobbying is plagued by asymmetric information, and lobbyists are agents largely undisciplined by conventional market forces, see Heinz et al. 2003, but this feature of lobbying was not directly affected by Citizens United. 6 engages in activities that are essentially political in nature, and would be identified as “lobbying” in ordinary speech. 9 Books, television ads or appearances, op eds, pamphlets, Congressional testimony, efforts to stimulate “grassroots” letter writing campaigns, and public comments on proposed regulations, and all lobbying activities by those whose lobbying activities constitute less than 20 percent of the time engaged in services are all arguably exempt from the legal definition of “lobbying contacts,” depending on the facts. Lobbying disclosure laws are also largely unenforced (Fried 2011).

Even contributions and election expenditures are exempt from disclosure if carefully funneled through “conduits,” i.e., “independent” organizations. While those organizations might be subject to a disclosure requirement if they in turn make contributions to candidates, 10 nothing in the disclosure regime requires a public company that donates money to, for example, a commonly controlled but formally independent non-profit to disclose those donations to the public, or to force the non-profit to disclose to the public the identity of its donors if the non- profit or political committee limits its activities to “independent” election expenditures. Neither FEC nor SEC rules, nor state laws, permit shareholders of public companies to demand such information in any direct way. 11 Top spenders in 2010 included Crossroads GPS, a non- profit organized by Karl Rove that prominently notes on its website:

Any person or entity that contributes more than $5,000 to a 501(c)(4) organization must be disclosed to the Internal Revenue Service on Form 990. However, the IRS does not 9 2 USC § 1602 (definitions of lobbying activities and related terms).

10 A “political committee” – including any corporation or other organization that raises or spends more than $1,000 on, and has a “major purpose” of, influencing federal elections – must register with the Federal Election Commission (FEC), and disclose specified information, including direct contributions to candidates, and committees making direct contributions must disclose the identity of their donors and are subject to limits on both the size of their contributions and the size of donations they receive from others. Federal Election Campaign Act, 2 USC §§ 431-55; SpeechNow.org v. FEC, 599 F.3d 686 (D.C. Cir. Mar. 26, 2010); Buckley v. Valeo, 424 U.S.

612 (1976) (adding “major purpose” qualification to definition of “political committee”). The subset of non-profit organizations permitted under the Internal Revenue Code to engage in political activities (prominently, organizations under IRC §§ 501(c)(4) and 527) are required to disclose large donors to the Internal Revenue Service, but not to the public or the FEC, unless the donations are made specifically “for the purpose of furthering electioneering communications.” 11 CFR § 104.20. Broadcasters must keep records and make available for public inspection the identity of purchasers of election ads, but these records do not include the sources of the purchasers’ funds. 47 USC § 315; 47 CFR § 73.1943.

11 It is possible for a shareholder to propose a bylaw that would require disclosure by a public company of corporate political expenditures, but such a bylaw would face challenges as to its legality under state law, and if the shareholder wanted to try to get other shareholders to vote for the bylaw, the shareholder would have to either incur substantial expenses to solicit proxies, or face legal challenges under the SEC’s Rule 14a-8, which permits shareholders under certain circumstances to include such bylaws in a proxy statement paid for by the company, but only if the bylaw does not concern “ordinary business” of the corporation and is not otherwise in conflict with state corporate law, and the shareholder could expect to face challenges to its ability to include a shareholder resolution relating to disclosure of political activity under Rule 14a-8 unless (as is now typical for active institutional shareholders) the resolution were non-binding. On the legality of shareholder-adopted bylaws, see Coates 2001 and CA, Inc. v. AFSCME Emps. Pension Plan, 953 A.2d 227, 231 (Del. 2008). 7 make these donor disclosures available to the public. Crossroads GPS’s policy is not to provide the names of its donors to the general public. 12 As a result, the sources of half the money spent in the first post-Citizens United election cycle were kept secret. Of $266.4 million spent by outside groups to influence the 2010 elections, $135.6 million was spent by groups that did not reveal any details about their funders. 13 In 2010, such “dark money” was almost double the $68.9 grand total of spending by outside groups in 2006, the previous mid-term election cycle. Six of the top seven spending “independent” groups in the 2010 election cycle kept their donors secret. These included the U.S. Chamber of Commerce, the top spender, at over $31 million. Other top spenders identified themselves only as “Americans for Job Security,” the “American Action Network” or the “American Future Fund.” The control and funding of many organizations active in politics was and remains uncertain. Any research claiming to have assessed the aggregate amount of political activity by businesses or corporations should be viewed skeptically. Nevertheless, for purposes of this article, the effect of there being potentially very large unobservable political activity by corporations is to make it harder for any relationships that might exist among CPA, corporate governance, and value to be detected.

In addition, the fact that political activities act as complements allows us to infer the effects of Citizens United on unobservable political activity by reference to observable political activity. While we cannot know with precision which firms exploited the new “independent” election expenditure channel opened by the Supreme Court, we can estimate the average relationship between the newly permitted activities (mostly through conduits 14) and shareholder value, by reference to which firms were active prior to the decision and could thus be expected to be most likely to exploit the new channel. Empirically, this mode of inference is taken up in Part V below.

C. Corporate political activity: what counts as evidence that it “works”?

Extensive research in management, political science, and economics explore the causes and the narrow consequences of corporate political activity. Many studies have established that different types of CPA are complements. For example, Ansolabehere et al. 2002 and Schuler et al. 2002 find a strong complementarity between lobbying and PAC activity, with over 86% of all contributions coming from firms with both a lobbyist and PAC. Contributions buy 12 https://www.icontribute.us/crossroadsgps (visited October 29, 2011). On Rove’s involvement in Crossroads GPS, and its role in the 2010 elections, see Franz 2010. 13 Public Citizen, Disclosure Eclipse, Nov, 18, 2010. A large amount of expenditures have been made by “super PACs,” new political committees set up in the wake of Citizens United and specifically designed to solicit unlimited sums from unlimited sources (including corporations) and spend the funds on “independent” election expenditures, including television and other media. See Center for Responsive Politics, Outside Spending, available at www.opensecrets.org/outsidespending (last visited Nov. 26, 2011).

14 For evidence that public companies using conduits in this way, see Mcintire and Confessore 2012. 8 access, and lobbying exploits access to affect policy. 15 These findings are confirmed with the data analyzed below.

What does this research say about whether CPA “works” – that is, whether it produces benefits for corporations and their shareholders? Many studies find evidence that interest groups exchange money and/or information for political benefits of various kinds, such as trade barriers, 16 reduced or easier regulatory inspections, 17 and lower tax rates, although these results are sometimes sensitive to specifications and samples, and not all findings are consistent with firms succeeding in influencing policy. 18 Firms withhold contributions from officials who vote against their interests. 19 Researchers focusing on specific issues or industries have found evidence of influence via lobbying or other political activity. 20 Event studies have revealed that US equity markets are affected by the control of Congress (Jayachandran 2006) and policy platforms (Knight 2006). It also seems plausible that some CPA is responsive to political operatives seeking to extract rents from firms – by threatening to impose burdensome regulations or relieve competitors of similar regulations, or provide them with subsidies.

In that sense, it seems clear that CPA “works” – indeed, this is the common intuition behind most efforts to regulate CPA: CPA needs regulation because it affects laws and regulations. However, separate lines of research suggest that CPA does not necessarily “work” for shareholders. A number of studies present evidence that CPA represents a form of managerial “consumption” good – consistent with the possibility that it is pursued at the expense of shareholders (see, for example, Ansolabehere et al. 2003, surveying numerous prior studies).

Brasher and Lowery 2006 find publicly held companies are more likely to engage in lobbying than otherwise similar non-public companies, although they do not develop the potential role of agency costs in explaining their finding, and Kim 2008 21 includes one governance variable in modeling the determinants of CPA, and finds that weak shareholder rights correlate positively with the propensity to lobby and to sponsor a PAC, and with lobbying expenditures. These findings are consistent with the findings reported below.

Which of the two effects is more important, on average, for most public companies? Is the effect of CPA on political outcomes in line with corporate interests, or does CPA align more with the interests of corporate managers than corporate shareholders? Two recent studies have 15 A separate line of research has reached inconsistent conclusions on whether political activity of any kind complements firm business strategy – cf. Baron 1995, finding it does, with Schuler 1996, finding it does not – but prior research is uniform in finding that different modes of politics complement each other. E.g., Wright 1990, Austen-Smith 1995, Gray & Lowery 1996, Lord 2000, Tripathi et al. 2002, Hersch et al. 2008. 16 E.g., Goldberg and Maggi 1999.

17 E.g., Gordon and Hafer 2005.

18 E.g., Richter et al. 2009; but cf. Drope and Hansen 2008; Hansen, Mitchell and Drope 2995 (“Scholars have not been very successful linking the political participation of business to policy benefits”); Lowery and Gray 2004.

19 E.g., Jackson and Engel 2003 (China policy) and Franca 2001 (NAFTA policy).

20 E.g., Schuler 1996 (steel), Kroszner and Stratmann 1998 (financial services), and de Figueiredo and Tiller 2001 (communications). See also Fisch 2005 (case study of FedEx).

21 The author of that study graciously shared his data from 1998-2004, included in the sample tested below. 9 produced contrasting results on this broader question. Cooper et al. 2010 found that companies with PACs that made donations to more candidates in the period 1979 to 2004 had on average higher stock returns than industry peers in the following year. However, companies with PACs that simply made larger donations did not generate excess returns. Cooper et al. control for industry effects in the first-stage of a model (as a predictor of PAC contributions), but they do not do so in their second-stage model (as a predictor of returns), nor do they interact their dependent variables or partition their sample by industry. As a result, one cannot tell if their results are driven by the minority of firms in heavily regulated or government-dependent industries.

In contrast, Aggarwal et al. 2011 found that that companies that made soft money donations to parties or donations to Section 527 committees from 1991 to 2004 (accounting for roughly 11 percent of the universe of U.S. publicly traded firms) tended to be large, slowly growing firms that had more free cash than other firms but spent less on research and development or business investments. They also found that corporate donations were negatively correlated with long-term firm-specific stock market performance. Similarly, in international samples, Faccio 2010 shows that public companies with political connections produce lower accounting returns compared to non-connected firms, and Faccio et al. 2006 find that while politically connected firms are more likely to receive government bailouts, the financial performance of such firms is worse than their non-connected peers at the time of and following the bailout. 22 This article reaches findings that are more compatible with Aggarwal et al. and Faccio – i.e., corporations engage in a mix of shareholder-oriented and non-shareholder-oriented political activity, but the predominant, or average, effect is negatively related to shareholder value.

Finally, a number of studies have reached findings on the firm- and industry-level correlates of CPA (see Hillman et al. 2004 for a survey). 23 Consistent with intuition, ongoing CPA in the US is more common for firms that are larger,24 older, 25 more regulated, and more dependent on government purchases. 26 These correlations are reflected in the research design below.

D. Corporate governance Corporate governance research is vast, multidisciplinary and largely siloed. For surveys, see, for example, Shleifer and Vishny 1997 and Bischoff 2009. Yet few strands of this literature – 22 Aggarwal et al. 2011 also found that better corporate governance – including better board structure, lower CEO compensation, and the presence of large shareholders to monitor corporate behavior – tended to be associated with less political activity. See also Hadani & Schuler 2011, Hadani 2011, and Chaney et al. 2011, which produce findings more compatible with Aggarwal et al. 2011 and this article than with Cooper et al.

23 See also Potters and Sloof 1996, which surveys empirical studies in the public choice, economics, and political economy literatures on political activities of interest groups, including corporations.

24 E.g., Hansen and Mitchell 2000.

25 E.g., Baron 1995, where firm age is interpreted as a proxy for “experience” or “reputation.” 26 E.g., Hart 2001. Firms also match industry-competitors’ political activity. Grier et al. 1994. 10 whether in accounting, law, business, management, or economics – have focused on CPA. 27 Instead, the focus in corporate governance has been agency theory (Jensen and Meckling 1976). Specifically, research has attempted to analyze and test the extent to which and how corporate managers (or dominant shareholders) act in ways that harm or fail to benefit shareholders (or minority shareholders). For example, Berle and Means 1932 posited that shareholder dispersion would increase managerial slack, enabling managers to obtain greater private benefits. Gompers et al. 2003 show that firm-specific shareholder-friendly corporate governance provisions – corporate charters, bylaws, and executive contracts – correlated positively in the 1990s with firm value (as measured by industry-adjusted price/book ratios, often referred to as Tobin’s Q 28).29 Bebchuk et al. 2010 show the correlation between governance provisions and corporate value (measured by industry-adjusted book/price ratios) persisted and even grew through 2008.

30 This article also uses Tobin’s Q as its primary proxy for shareholder value. Related strands of corporate governance research have focused on particular aspects of corporate behavior. Yermack 2006 shows that firms that pay for corporate jets for their CEOs underperform market benchmarks and experience stock price drops upon announcement when jet use is disclosed, and that jet use correlates with personal CEO characteristics, such as long- distance golf club memberships. This article examines whether jet use correlates with CPA.

The empirical study of the causes and effects of corporate governance practices all face design problems (for example, Listokin 2008). It is plausible that corporate governance is set in anticipation of corporate performance, making the direction of causality difficult to establish with certainty. 31 Still, prior studies establish that governance provisions are reliable correlates 27 Bischoff 2009 reviews 141 corporate governance articles published 1997 to 2009 and finds none focused on CPA. A few studies argue that ownership and control structures emerge in response to political pressures, or vice versa, but they rely on country- and not firm-level data. E.g., Roe 1994; Roe 2003; and Morck et al. 2005. 28 This ratio is calculated following Kaplan and Zingales 1997, calculated as [BVA+MVCE-BVCE-DT] / BVA, where BVA is book value of assets, MVCE is current common stock market capitalization – that is, stock prices – BVCE is book value of common equity, and DT is the book value of deferred taxes. 29 Tobin’s idea was to relate an asset’s market value to its replacement value, Tobin and Brainard 1977, but market values of firm assets are not readily observable, and may diverge from book value (as when a firm’s assets include significant intellectual property or other intangibles). When comparing firms in the same industry in the same period, these divergences are unlikely to bias the results, and it has become customary to refer to the ratio described in note 24 as Tobin’s Q and to use it as an indicator of firm value, e.g., Demsetz and Lehn 1985; Morck et al. 1988; McConnell and Servaes 1990; Lang and Stulz 1994; Himmelberg et al. 1999; La Porta et al. 2002; Core et al. 2006; Bebchuk et al. 2009; Bebchuk et al. 2010; Cremers and Ferrell 2011. One prior paper used Tobin’s Q to study politics – Hersch et al. 2008, who find no relationship between Q and political expenditures in the period 1998 to 2000 – but they do not compute each firm’s Q relative to industry peers, which results in pooling firms in industries with very different industries, with very different levels of other intangible assets. 30 Accord Cremers and Ferrell 2011 and Giroud and Mueller 2011.

31 Cremers and Ferrell 2011 argue the direction of causality runs from governance to value because (1) the relationship only appeared after a Delaware Supreme Court allowed boards to resist takeovers, Moran v.

Household 500 A.2d 1346 (Del. 1985), and (2) because cross-sectional variation in that relationship is consistent with theory on how firms impede takeovers. 11 of performance and value, and shift the burden of proof to those who believe such provisions are epiphenomenal. The goal of this article is to do the same with respect to corporate political activity, where currently even basic disclosure rules do not exist.

Among the few corporate legal scholars to address CPA, Brudney 1981 defended restrictions on CPA from a shareholder perspective, noting that early US corporations were limited in their activities by charter restrictions that would effectively have forbidden CPA, and defending a rule requiring a supermajority of (or even unanimity among) shareholders under the First Amendment. Citizens United and potential legislative responses have stimulated a few papers focusing on CPA. Bebchuk et al. 2010 argue that public company shareholders are more vulnerable to managerial agency problems in the CPA context than in other contexts, and argue for new legislative default rules (which shareholders could opt out of) requiring disclosure and prior shareholder approval of CPA. Gilson and Klausner 2010 worry that CPA risks involving public companies in polarizing debates and shareholder votes and argue for shareholder approval requirements so as to minimize the potential costs of such debates. 32 II. Hypotheses and Samples A. Hypotheses As noted above, research on CPA has previously found that firms that are heavily regulated or dependent on government expenditures are more likely to engage in CPA. Even firms that themselves are not heavily regulated or government-dependent, but which operate in industries comprised primarily of firms that have these characteristics, are likely to have business strategies that are interwoven with government affairs. These prior findings lead to the following hypothesis, which can be confirmed in the S&P 500 in the period leading up to and following Citizens United:

Hypothesis 1 (H1): CPA is most common in heavily regulated or government- dependent industries. 32 See also Fisch (2005) (case study of political activity by Fedex); Coates and Lincoln 2011 (study of disclosure policies voluntarily adopted by S&P 500 firms and a positive correlation between those firms and corporate value); IRRC 2011 (same); Regan 1998 (essay on corporate speech and civic virtue); Winkler 2007 (legal analysis of corporations under the First Amendment) and 2004 (history of ban on corporate donations to federal elections). 12 If corporate managers could be trusted to spend corporate money on political activity that would benefit shareholders, then Citizen United’s relaxation in the constraints on corporate political activity might still be of concern to voters generally, because rent seeking beneficial to shareholders might harm taxpayers or consumers, for example. But at least one would not worry about any additional burden of such activity on capital formation or the economic benefits that flow from well-governed public companies. Unfortunately, managers cannot always be wholly trusted with other people’s money, and as reviewed above, corporate governance provisions consistently correlate negatively with shareholder wealth. The literature on corporate governance suggests that agency problems are more acute when shareholders are weak, because in those companies managers can pursue their own interests more freely than in other companies. Managers may pursue CPA that diverges from shareholder interests for a variety of reasons. As elaborated more below, managers may have different political preferences, either based on personal political ideology, or because they have personal political career goals that can be served by using firm wealth in ways that serve those interests but not shareholder interests (e.g., Bebchuk and Jackson 2010). Managers may seek to use politics to create political “shelter” – barriers to entry – that reduce risk but may also reduce risk-adjusted returns (e.g., Faccio et al. 2006). Political advisors and lobbying staff may lead managers to systematically under- or misestimate political risk (e.g., Heinz, et al. 1993). CPA tends to produce opaque strategies, which coupled with the current lack of disclosure regulations (Coates and Lincoln 2011), may impede shareholders’ ability to evaluate managers (e.g., Smith 2000). If, for any combination of those reasons, CPA is harmful to shareholders, then:

Hypothesis 2 (H2): Managers of companies with weak shareholders – those who are more dispersed, are accompanied by fewer block holders, or have fewer rights – will be more likely to engage in CPA.

Prior research suggests that CPA may represent a form of managerial perquisite – a “consumption good” for those who control the CPA – that is, managers. CPA could represent, in this view, pursuit of a pet project that is at best unrelated to shareholder interests, and at worst could actively harm them. Managers might have personal political goals – ideological in nature – that could diverge from the net political interests of shareholders, particularly given that politics can affect a range of issues on which widely dispersed shareholders are unlikely to agree. CEOs inclined to consume perquisites of one kind are more likely to consume perquisites of another kind; alternatively, boards that are more willing to let CEOs consume perquisites of one kind will be more tolerant of consumption of other perquisites. One type of perquisite previously studied – use of a corporate jet for personal travel – is likely to correlate with a CEO’s use of corporate funds to pursue personal goals more generally, including CPA. If CPA is harmful to shareholders, then:

Hypothesis 3 (H3): CEOs who use corporate jets for personal travel – which correlates with harm to shareholders – are more likely to engage in CPA.

13 Most tangibly, corporate managers may have or develop their own personal political ambitions – to run for office or obtain appointed offices such as cabinet posts or ambassadorships. Jon Corzine, ex-CEO of Goldman Sachs, became Senator and then Governor of New Jersey; Dick Cheney, ex-CEO of Halliburton, became Vice President; George H. W. Bush, ex-president and chairman of Zapata Petroleum, became President; and Herman Cain, ex-CEO of Godfather’s Pizza, became chair of the board of the Federal Reserve Bank of Kansas City, then president of the lobbying organization for the restaurant industry, then a Republican candidate for President. Managers’ personal political goals could be furthered if the companies they control engage in political activities, using shareholder funds; or, alternatively, political operatives could reward CEOs for using shareholder funds to support the operatives’ political initiatives, or because they became known to the operatives as a result of their firms’ CPA. This suggests the following hypothesis:

Hypothesis 4 (H4): CEOs who obtain post-CEO political positions are more likely to have engaged in CPA while in office as CEO. The foregoing analysis also suggests that the influence of agency costs on CPA will be most easily observable outside of heavily regulated and government-dependent industries. That is because shareholder-oriented CPA would be common in those industries, reducing the variation across firms in CPA that would allow for differences in the correlation between CPA and shareholder power to be detectable. If CPA serves shareholder interests, as is intuitive in heavily regulated or government-dependent industries, but requires effort or risk-taking by managers, it might even correlate positively with shareholder power and negatively with managerial excess in those industries. That is, the predicted relationships in H2 through H4 might be reversed.

Hypothesis 5 (H5): CPA’s relationships with shareholder power and managerial excess are weakest (or even reversed) in heavily regulated or government-dependent industries, and strongest in other industries.

If the foregoing analysis were correct, one would also expect that CPA would be most likely at firms where CEOs lack strong incentives to maximize shareholder wealth more generally. As a result, CPA should be more common at firms with lower shareholder wealth, when compared to other firms in the same industry. In heavily regulated or government-dependent industries, however, CPA may be so common that the empirical relationship between CPA and value will attenuate (because variation across firms in those industries will be lower, reducing the power of empirical tests of that relationship). In fact, in those industries, where regulation and government are by definition heavily involved with firms’ business strategies, it is possible that CPA might correlate with higher value, if (for example), firms in those industries use CPA to improve regulatory policy (from the perspective of the firms and their shareholders) or defend against politicians’ threats to add regulation that is harmful to firms and their shareholders).

Hypothesis 6 (H6): CPA correlates negatively with industry-relative measures of shareholder value.

14 Hypothesis 7 (H7): CPA’s relationship with corporate value is weakest (or even positive) in heavily regulated or government-dependent industries, and most strongly negative in other industries.

If H6 were true, it would not necessarily mean that CPA itself causes lower shareholder value – it might simply correlate with lower value, because of unobserved firm or manager characteristics, or CPA might be caused by lower value, as managers attempt to lobby their way back to profitability, such as by erecting barriers to competition. How might CPA actually cause harm to shareholder interests? The aggregate amounts that companies spend on CPA are large in absolute terms (see Part V.D.2 below and IRRC 2011), but disclosed CPA expenditures are small relative to these firms’ assets, revenues and even earnings, on average. If CPA were simply a waste of money, but had no effects beyond out-of-pocket costs, it should not significantly affect shareholder interests. However, CPA shaped by managers’ personal interests may produce larger negative effects on firm value through indirect channels. One is strategy. Business schools have long taught that a corporate strategy is best if focused – composed of a small number of elements (e.g., Porter 1980, 1985, 1996), easily communicated, understood and implemented by middle managers and employees. Outside of heavily regulated and government-dependent sectors, CPA may dilute a firm’s strategic focus, and distract and degrade managerial performance, particularly if managers’ personal goals affect CPA. A second indirect channel is large new investments. If manager-influenced CPA affects a firm’s choice of large projects, those projects may be less aligned with shareholder interests than would otherwise be the case. While strategy dilution is difficult to observe across heterogeneous firms, project selection can be partly observed in the form of capital expenditures. Again, one would not expect CPA to reduce the value of capital expenditures to shareholders by as much (or at all) in heavily regulated or government-dependent industries, where CPA can generally be expected to affect project choice. The implication of the foregoing analysis is that firms that are both engaged in CPA and making large capital expenditures are more likely to be making poor project choices, and reducing shareholder value, than firms not so distracted or influenced by CPA. Hypothesis 8 (H8): Outside of heavily regulated and government-dependent industries, capital expenditures correlate positively with industry-relative shareholder value at firms not engaged in CPA, and less positively (or even negatively) with value at firms engaged in CPA.

Finally, what about Citizens United? If prior findings that various forms of CPA are complements are correct and extend to independent expenditures, whether directly or through trade associations or other conduits – that is, if prior findings that firms that engage in one form of CPA (such as lobbying) are more likely to engage in other forms of CPA (such as PAC donations or, when it was legal, making soft dollar donations) – then the decision should have increased CPA overall, and also increased the extent of the activity. These increases should be particularly evident in industries in which CPA is least likely to be of benefit to shareholders. 15 Hypothesis 9 (H9): CPA overall increased after Citizens United, particularly in industries that are not heavily regulated or government-dependent, and the levels of expenditures on CPA increased more at firms that sponsored PACs or paid registered lobbyists in 2008.

In addition to increasing CPA, Citizens United was a largely unexpected and exogenous shock to the restraints on CPA. 33 As a result, it created a natural experiment in which one can better test the effects of CPA on shareholder value. Because CPA of various kinds are complements, the relaxation of constraints on CPA in Citizens United made it more likely that firms already engaged in CPA would continue to do so, at higher levels, in both observable ways (as in H9) and in unobservable ways (through conduits and “independent” expenditures). Firms already engaged in CPA, in effect, can be viewed as “treatment” firms for purposes of the quasi- experiment. The change in the politics-value relationship from before to after the decision (2008 to 2010) at those firms can be compared to the change in that relationship for other firms (which are the “control” firms for purposes of the quasi-experiment), and the difference in differences (before and after Citizens United, treatment vs. control) will represent the estimated impact of politics on value. If managerial agency costs are a principal driver of CPA, on average, that impact is expected to be negative.

Hypothesis 10 (H10): Relative to other firms, firms who either paid registered lobbyists, or sponsored PACs that were active, or both, prior to Citizens United will experience declines in value after Citizens United, relative to firms not so engaged in such observable political activity. B. Samples The foregoing hypotheses are tested on a sample of data on companies in the S&P 500 in the years 1998 to 2004, as well as in 2008 and 2010, the elections immediately before and after Citizens United. For lobbying, the unit of observation is a firm-year; for PAC donations, the unit of observation is a firm in a two-year election cycle (for example, 1997 and 1998, 1999 and 2000, etc.), using firm-year data for the second-year in the cycle. For a subsample consisting of all sample firms in 2000, the post-2000 careers of all CEOs were reviewed to determine whether the CEOs obtained political positions after their tenure as CEO. By construction, both samples consist of large publicly held firms and, as shown below (and in prior research), CPA correlates with firm size. The strength of the relationships reported below 33 The parties to Citizens United (including the Deputy US Solicitor General) initially argued the case before the Supreme Court as a narrow decision applicable to the plaintiff in the case – a small advocacy non-profit specifically formed to engage in political activity. While the case was expected to have implications for campaign finance law more generally, it was not expected to have the sweeping legal effects it did. The Supreme Court chose to ask the parties to return to the Court and reargue the case on the broader grounds that it was ultimately based – that is, that a 50-year-old ban on independent election expenditures by all corporations, for-profit and non-profit alike, was unconstitutional. See http://www.supremecourt.gov/Search.aspx?FileName=/docketfiles/08- 205.htm (last visited Nov. 30, 2011). 16 is likely to fall as one beyond this sample. 34 However, the S&P 500 represents a large fraction of US public company market capitalization, corporate revenues and assets, and economic activity, and is of independent interest, whether the results here can generalize beyond the firms studied.

C. Variables measuring shareholder power Shareholder power is measured on two dimensions: dispersion and rights. First, with respect to dispersion, shareholder power is measured by the logged number (LOG_SHH_NUM) of record stockholders for a given company, as reported by Compustat. 35 The more shareholders, the harder it is for them to overcome collective action problems, and the weaker shareholders are, and the less able they are to protect their own interests against diversion of value by managers (directors and officers). As a second measure of shareholder power due to dispersion, total ownership of block holders (5+% owners) is also measured (BLOCK_OWN), taken from Dlugosz et al. 2004 (for 1998 through 2001) and from the noisier CDA Spectrum database (for later years). The more block holders, the easier it is for such relatively large shareholders to overcome collective action problems, and the more powerful are shareholders as a group. Second, with respect to shareholder rights, two measures commonly used in the corporate governance literature are used. Data on the most widespread measure – the “G-index” from Gompers et al. 2003, based on corporate provisions tracked by IRRC (now RiskMetrics) are taken from IRRC via the WRDS website. 36 IRRC ceased reporting all components of the “G- index” in 2007, so a subset is used, consisting of 13 data items available for all sample years. These items are coded as 0 or 1 and summed (Q_GINDEX – “Q” for quasi). Another measure, the E_INDEX, based on six provisions, is constructed as described in Bebchuk et al. 2009, following Coates 2000. Each measure is constructed such that higher scores indicate fewer shareholder rights. D. Variables evidencing managerial excess 34 Kerr et al. 2011 study a broader sample of companies and find that lobbying is much less common in smaller public firms than in larger public firms. Cf. Drope and Hansen 2006 (finding that the tendency of researchers to study large firms does not bias the picture of CPA overall).

35 The true item of interest is beneficial ownership, but data on the number of beneficial owners for most companies is not available, because the SEC does not require it to be disclosed, even though companies have the data. Nevertheless, interviews with proxy solicitors confirm that the number of record owners is a noisy but correlated proxy for beneficial ownership, and in a separate paper (Coates 2010b), I find that the number of record owners is correlated with a number of merger and acquisition practices (such as contract terms) with which theory suggests ownership dispersion should be correlated. Institutional ownership and board independence, sometimes used as proxies for shareholder power or agency costs, are not used. Institutions have disparate strategies on how to monitor and influence portfolio firms and whether to work together with other institutions in doing so (e.g., Ozer et al. 2010). Listed company boards have been required to be majority independent since 2002 (Coates 2007), eliminating variation in board independence within the samples studied here. 36 http://wrds.wharton.upenn.edu. 17 Data reflecting the possibility of managerial excess consist of two variables: (1) CEOJETPOS, a dummy set to one if the firm reports that the CEO used a corporate jet, and (2) CEOJETVAL, the reported value of that such jet use, both derived for years prior to 2004 from Yermack 2006 and for 2009 from GMI. E. Variables measuring corporate political activity Data on PAC and lobbying activity from the “Open Secrets” website, 37 which has a search engine and summaries of data from the Federal Election Commission (PAC contributions) and U.S. Senate (lobbying) websites. For 1998-2004, these data are derived from Kim 2008 and spot-verified by reference to the Federal Election Commission (FEC) website; for 2008 and 2010, they are derived directly from Open Secrets and spot-verified by reference to the FEC website.

Two variables measure the propensity of firms to engage in CPA: LOBBY_YN is a dummy set to one if the firm participated in lobbying in a given year; CONTRIBUTE_YN is a dummy set to one if the firm’s PAC contributed in the prior two-year election cycle. Two variables measure the extent of participation: LOBBYAMOUNT is the amount in $000s (and LOGLOB is logged amount) of annual lobbying expenditures by the firm (inflation-adjusted), CONTRIBUTEAMOUNT is the amount in $000s (and LOGCONTRIBUTE is the logged amount) of total PAC contributions sponsored by the firm to federal candidates over the prior two-year cycle (inflation-adjusted). Each participation variable is the log of the observed value plus 0.001, to preserve zero observations in the sample. F. Firm and industry variables As in the corporate governance literature, firm value is measured with the log of median- industry-adjusted “Tobin’s Q” see note 26 above, or LOGRELQ. For industry adjustments, Fama’s 48-industry groups are used. LOGRELQ equals the log of the ratio of the firm’s Q and the industry’s median Q. As in prior research, Q is defined as market value of assets over book value of assets, where market value of assets is approximated as the book value of assets plus the market value of common stock less the book value of common stock less the balance sheet value for deferred taxes. The regression analyses below also include proxies customarily included in empirical models of CPA and Tobin’s Q, set out in Appendix A.

III. Summary statistics 37 http://www.opensecrets.org. 18 Table 1 sets forth summary data. Most of the S&P 500 is politically active, with 71% engaged in annual lobbying on average, and 70% sponsoring PACs making donations. For the S&P 500, consistent with prior research, the two types of CPA are complements: the correlation coefficient of lobbying activity and PAC contributions is 0.5, and the correlation among lobbying and contribution amounts is 0.6. S&P 500 firms spent roughly six times more on lobbying than their PACs give in contributions. The distribution of both kinds of CPA is right- skewed (4.2 and 2.9) and kurtotic (29.6 and 23.6), and logged amounts are much closer to a normal distribution (skew of -0.3 and -0.6, kurtosis of 1.7 and 1.6). [Table 1 about here] A third of firms are in heavily regulated industries. Over the whole sample, the average share of revenues derived from government expenditures was 6%. Roughly 4% of sample firms were in industries deriving more than 25% of total revenues from government expenditures. Summary statistics for the EINDEX are comparable to those reported in prior research, as is the shape of the distribution of the Q_GINDEX compared to the full GINDEX in prior research. In this large company sample, the median firm had 16,800 record shareholders, and only a few (between one and 15, depending on the year) had few enough record owners (<300) to be able to “go dark” – that is, deregister with the SEC. In the period through 2003, 23% of companies reported that their CEOs used corporate jets in the period, representing an average of $71,700 worth of value to the CEO for those CEOs who used jets. Reported jet use for personal travel by CEOs was significantly higher in 2009, with 35% of firms reporting their use, at an average cost to the firm of $121,145 for those CEOs who used corporate jets. It should be noted that the GMI data for 2009, which are derived from proxy statements, are not strictly comparable to the Yermack jet data from the pre-2004 period, as the SEC modified its disclosure rules for perquisites in 2006 (SEC 2006), and the differences in expenses may be attributable to changes in disclosure rules, as well as to changes in factor costs (e.g., jet fuel). IV. Data analysis A. Univariate and bivariate analyses The hypotheses developed in Part II are first tested with two sets of simple univariate and bivariate analyses. First, evidence regarding industry effects is presented, to demonstrate that – consistent with past research – CPA in the current sample is strongly correlated, in intuitive ways, with industry groupings that reflect the intuition that CPA may be most shareholder- oriented where government is already crucial to business success – in heavily regulated and government-dependent sectors. Second, relationships among four sets of variables of interest are depicted: CPA, shareholder power, CEO perquisite consumption (in the form of corporate jet use) and CEO career concerns (in the form of post-CEO political appointments). 1. Industry effects and CPA 19 Figure 1 shows that firms in industries that are heavily regulated or dependent on government expenditures are, as predicted, more likely to engage in political activity (with tick-bars showing 95% confidence intervals). “Government dependent” firms are those in industries with a GOV_SHARE of at least 25%. “Heavily regulated” firms are those in industries identified in Appendix A. [Figure 1 about here] Consistent with H1, few firms in those industries do not engage in observable political activity, and this was true before and remains true since Citizens United. By contrast, in other industries, political activity is common, but more varied, and less common than in heavily regulated or government-dependent industries. 2. Shareholder power and CPA Overall, the pairwise Spearman rank correlation coefficient between LOGCSHR and LOBBY_YN is 0.38, and between LOGCSHR and LOGCONTRIBUTE it is 0.43. More shareholders makes shareholder coordination harder, weakens shareholders, strengthens managers – and at firms with more shareholders, political activity is more common, consistent with H2. Between BLOCK_OWN and LOBBY_YN IT IS -0.19 and between BLOCK_OWN and CONTRIBUTE_YN it is -0.17. More block holders, in other words, who face lower coordination costs, make shareholders stronger as a group, and weaken managers – and at firms with more block holders, political activity is less common, consistent with H2.

One might worry that these correlations are driven solely by firm size, which past research has shown is correlated with both CPA and shareholder dispersion. To show the relationship between CPA and shareholder dispersion is not simply an artifact of firm size, Figure 2 graphs the percentage of firms engaged in lobbying, broken down by both a firm’s asset size and its number of record shareholders. Both factors increase CPA: across asset size quartiles, CPA increases, but it also increase within each size quartile as shareholder dispersion increases. Qualitatively similar results hold for PAC contributions.

[Figure 2 about here] Shareholders are weak if dispersed, but they can also be weak if they have few legal rights. Shareholder power on each dimension is distinct – in fact, they are negatively (if weakly) correlated, with a correlation coefficient between LOGCSHR and each of EINDEX and Q_GINDEX of -0.07 (p<.000), and a correlation between BLOCK_OWN and each of EINDEX and Q_GINDEX that is statistically equivalent to zero. This makes examining the relationship between shareholder rights and CPA of interest for two reasons: first, it is of independent interest, given prior research on the relationship between shareholders rights and shareholder value; and second, if the same relationship exists between CPA and shareholder power on both dimensions, it is more likely to be real, and not a spurious relationship driven by some other factor that happens to correlate with either measure of shareholder power on its own.

20 Figure 3a graphs CPA incidence for firms with different E-INDEXes (with tick-bars showing 95% confidence intervals). Figure 3b is the same graph for only firms in heavily regulated industries. Graphs of CPA against the Q_INDEX (not shown) are similar. For the full sample, firms with high E-INDEX scores (weaker shareholder rights) are more likely to engage in CPA than firms with lower scores, and the relationship is nearly monotonic across the E-INDEX, with sharper changes at the ends of the index. Consistent with H2, the negative relationship between shareholder rights and CPA is apparent. The differences are highly statistically significant, with p-values below 0.0001 for both an analysis of variance and ranksum test, which easily reject the null hypothesis that CPA does not vary by EINDEX, but consistent with H2.

[Figures 3a and 3b about here] However, for firms in heavily regulated industries, no relationship between CPA and shareholder rights is apparent, consistent with H5. In those industries, the degree of alignment between manager and shareholder interests caused by strong shareholder rights is unrelated to CPA. 3. CPA and CEO use of corporate jets In both the pre-2004 period, using data from Yermack 2006, and in 2010, using data from GMI, firms whose CEOs use corporate jets are significantly more likely than other firms to engage in political activity, consistent with H3. This relationship is not due to use of jets to engage in trips to engage in lobbying, e.g., to Washington, D.C. – the data only relate to use of a company jet for personal travel. Figure 4a shows the relationship between CPA and one marker of potential CEO excess – the use of a corporate jet by the CEO, at the expense of the firm (with tick-bars showing 95% confidence intervals) – for the period covered by the Yermack 2006 data. Figure 4b is the same graph for only firms in heavily regulated industries, for the same period.

Outside the heavily regulated industries, the difference in lobbying propensity in 2010 is striking: 88% for firms the CEOs of which used corporate jets for personal use in 2009, vs.

66% for other CEOs, a difference is that statistically significant (p<.00001, with a 95% confidence interval for the difference of -32% to -12%). If jet use is a proxy for CEOs who are more apt to take actions that are not in shareholder interests, or for boards willing to allow CEOs to take such actions, the positive correlation between jet use and CPA suggests that CPA may also not be in shareholder interests, just as the negative correlations between shareholder power and CPA does. [Figures 4a and 4b about here] 21 As with shareholder rights, the relationship between CPA and signs of managerial excess is not present in heavily regulated industries. Consistent with H5, lobbying is very common in those industries, whether or not the CEO uses a corporate jet. Even if jets are signs of managerial excess generally, CPA is not intuitively contrary to shareholder interests where regulation is heavy, so one would not expect jet use to correlate with CPA, and it does not. 4. CEOs’ subsequent personal political careers In Part II, it was hypothesized that CEOs and other managers of public companies might have personal interests in directing their companies to engage in political activity, separate and apart from shareholder interests in such activity, and famous examples of ex-CEOs who had gone into politics were noted. Here, more systematic evidence of this potential source of managerial agency problems with respect to CPA is developed. Table 2 presents data on all CEOs (n=438) in the overall sample described in Part III who were in office as CEOs in 2000. [Table 2 about here] As shown in Table 2, most (n=298) of those CEOs had retired by 2011. Of those retired CEOs, over 11% were appointed or nominated to political office between the time of their service as CEOs and 2011. “Office” for this purpose only included positions with authority, and not advisory positions, or such politically influenced recognitions as medals or knighthoods – if that broader mix of political rewards were counted, the number of CEOs receiving post-retirement political rewards roughly doubles in the sample. In the subsample were John W. Snow, ex-CEO of CSX, who became Treasury Secretary; and John E. Bryson, ex-CEO of Edison International, and Carlos Gutierrez, ex-CEO of Kellogg, both of whom became Commerce Secretaries. Also among the group were Carly Fiorina, ex- CEO of Hewlett-Packard, who was nominated as a Republican candidate for US Senate; Charles Price, ex-president and chairman of American Bancorporation, who became U.S. Ambassador to the United Kingdom; and William Donaldson, ex-CEO of Aetna, who became Chair of the SEC. The possibility of CPA being motivated by CEO political ambitions is illustrated by the fact that among the firms in the subsample of CEO careers reviewed, the odds that a CEO obtained post-CEO political employment were significantly higher for CEOs of firms that engaged in lobbying prior the CEO leaving the company (15% vs. 3%, p<0.05), consistent with H4.

These results likely understate the degree to which the prospect of future political careers ignite or shape CPA, for at least three reasons. First, the subsample only includes CEOs, while other, lower-level managers may also expect to obtain private political career benefits if their firms are involved in politics. Government-affairs specialists as well as general counsels and public relations officers can develop personally valuable relationships by directing firm resources in particular political directions. 38 Corporate lobbying is often outsourced to lobbying firms that provide employment and support for former corporate managers entering the political arena. 38 Ten percent of a sample (n=50) of departing general counsels at Fortune 500 companies moved into a government job within a year of their departures. Coates 2011. See also Heinz et al. 2003. 22 Second, political activity can pay off in other ways for corporate managers. Private equity firms employ former politicians and former corporate managers that developed relationships with government officials while serving as managers (e.g., George H.W. Bush, Arthur Levitt). Third, many CEOs and managers may have political interests but never act on them, because they leave office amid scandal, or because they become ill or die before they have the opportunity. In the subsample reviewed, Bruce Karatz, ex-CEO of Kaufman & Broad Home, was convicted of mail fraud, and both Maurice “Hank” Greenberg (AIG’s ex-CEO) and his son Jeffrey Greenberg (ex-CEO of Marsh & McLennan) lost their jobs because of probes by then- New York Attorney General Eliot Spitzer. The evidence presented here, though tentative only, is consistent with H4. In combination with the evidence reviewed above, on the relationship between CPA and shareholder power and managerial agency costs, it seems clear that CPA represents a mix of shareholder-oriented and non-shareholder-oriented activity. In the next part, regression analysis is used to examine how robust the politics-governance relationships is, the extent of the non-shareholder-oriented political activity in the S&P 500, and whether the non-shareholder-oriented political activity has an observable relationship with value. V. Regression analyses In this section, two sets of relationships are modeled with multiple regression analysis: (1) the relationship between measures of CPA and shareholder power, and (2) the relationship between CPA and corporate value, as measured by industry-relative Tobin’s Q, including its direct relationship in cross-sectional regressions, its relationship over time in firm fixed-effects regressions, and its relationship before and after Citizens United. A. Shareholder power and CPA Tables 3 and 4 set forth regression results for CPA in the S&P 500. Each table reports results for logistic models of participation in CPA – that is, whether a firm engages in any lobbying, or sponsors a PAC that made any contributions.

39 Table 3 presents models of lobbying. Table 4 presents models of PAC contributions. In each, shareholder power is proxied by both shareholder dispersion (LOG_SHH_NUM) and shareholder rights (E_INDEX). Qualitatively similar untabulated results are found using other measures of shareholder rights (Q_GINDEX) and shareholder concentration (BLOCK_OWN).

[Tables 3 and 4 about here] In the simple regressions without other explanatory variables (column (1) of Tables 3 and 4), each of the shareholder power variables correlates strongly with the propensity to lobby or 39 Models of the extent of participation – that is, of the amount of lobbying expenditures or PAC contributions – were also estimated, and qualitatively similar results were found, in both OLS and Tobit models. These results are available from the author, but are not tabulated because they are subject to classic selection effects, because no natural subsets of variables can be omitted to achieve identification for use in a Heckman selection model, and because the inverse Mills ratio from first-stage selection models for each type of participation that include all regressors are, not surprisingly, highly correlated (>0.95) with regressors in second-stage spending models.

23 have a PAC donate. More shareholder rights are less likely to engage in CPA, and when shareholders are more dispersed, firms are more likely to engage in CPA. When other explanatory variables are added (column (2) of Tables 3 and 4), the shareholder power variables retain or increase their significance – in both economic and statistical terms. Both LOGCSHR and the E-INDEX are strongly related to lobbying propensity, even after controlling for other factors (such as industry and size) that also correlate with CPA, consistent with H2. Ideally, the robustness of the foregoing results would be tested with firm fixed effects regressions, which would measure the relationship between changes in CPA at one firm as it changes shareholder rights. However, while S&P 500 firms change shareholder rights not infrequently, 40 the changes are minor: the median change in the EINDEX in the sample period is one; and only 5% of the changes are greater than one, amounting to only one percent of the observation years in which the index could change. Ownership dispersion is also stable for a given firm from year to year, with 64% of firms changing by less than 1% on average per year. In a prior paper (Coates 2010c), extreme changes in the G_INDEX were found to correlate in a fixed effects model with some measures of CPA for the period 1998 to 2004, but the correlations were of modest statistical significance, and they do not extend into the 2008 and 2010 period. B. CPA and corporate value Prior research has established that stronger shareholder rights correlate with higher Tobin’s Q. The prior section presented evidence that CPA correlates negatively with shareholder power. It is then natural to ask if these two relationships are connected? That is, does CPA itself correlate negatively with Tobin’s Q, before or after controlling for shareholder power? Table 5, and Figure 5, which is based on Table 5, present evidence that CPA does correlate negatively with corporate value, consistent with H6.

[Table 5 and Figure 5 about here] Panel A of Table 5 shows that LOGRELQ (logged median industry-adjusted Tobin’s Q) is higher for firms that do not engage in lobbying and for firms that do not sponsor PACs making contributions, consistent with H6. These findings hold after including the standard set of explanatory variables used in prior research on Tobin’s Q, as listed in the table. Consistent with the univariate and bivariate analyses above, the relationship between CPA and corporate value is markedly different for firms in heavily regulated industries, consistent with H7. For those firms, the sign on the CPA variables is reversed, and in the models with controls the positive coefficient on the interaction term is larger than that on CPA, suggesting that CPA improves corporate value for such firms. 40 The observed values for the G- and E-INDICES change only every two years, because IRRC publications from which the data on which the indices were released only every other year. In those years, 36.5% of the firms experienced changes in their G-INDEX, consistent with a range of annual change of between 18.3% and 36.5%, and 20.7% experienced changes in the E-INDEX, consistent with a range of annual change of between 10.3% and 20.7%. 24 Panel B of Table 5 presents the results of firm fixed effects regressions, which model the relationship between changes in CPA and changes in Tobin’s Q. These models rule out the possibility that unobserved fixed firm characteristics account for any observed correlation between CPA and corporate value. The results for lobbying are nearly identical to those in Panel A, including both the direction and magnitude of the coefficients for firms in and out of heavily regulated industries. The results on PAC donations, by contrast, differ from those in Panel A. With respect to PAC donations, there may be unobserved firm characteristics that are important to the relationship to value. There are fewer observations for PAC donations, however, since they are only observed in election years, and as a result the fixed effects regressions in Panel B for PAC donations are less precisely specified than those Panel A, and the confidence intervals are wider than in Panel A.

C. Corporate value and the interaction of CPA and capital expenditures Part II of this article hypothesized that one channel for value destruction through CPA was capital expenditures. Panel C of Table 5 presents the results of an OLS regression, similar to that presented in Panel A, that substitutes an interaction term between the CPA variables and CAPEX_ASSETS for the CPA variables on their own. In effect, the models split capital expenditures (scaled by firm assets) into those conducted by politically active firms and other firms. In each regression, heavily regulated firms are excluded, to minimize the need for further interaction terms, which are collinear with the variables of interest. In each case, the results are striking: capital expenditures by politically inactive firms are – consistent with prior research – a positive contributor to shareholder value, but capital expenditures by politically active firms not only produce less value, but substantially erode the generally positive effect of such investments on average industry-relative value, consistent with H8.

D. CPA and corporate value after Citizens United A final question is whether the relationship between CPA and corporate value changed after Citizens United. One cannot model the effects of the decision precisely, since many things changed between 2008 and 2010 in addition to the Citizens United decision. Nevertheless, one can see if there were gross changes between those two elections, whether the value-politics relationships found above persisted, diminished or increased after the case, and, most importantly, whether the value-politics relationship changed in different ways for firms that were politically active prior to the decision than for firms that were not. 25 1. Increases in the propensity to engage in political activity after Citizens United Figure 6 shows that the frequency of lobbying increased from 69% to 74% between 2008 and 2010 among firms in the S&P 500 outside of heavily regulated industries. There was a more modest increase in heavily regulated industries, from 85% to 87%. The frequency of PAC donations also modestly increased, from 82% to 83% in heavily regulated industries, and from 54% to 58% in other industries. These trends are confirmed in an unreported regression of lobbying propensity, similar to those presented in Table 3, in which the odds ratio on 2010 is 1.60, with a 95% confidence interval of 1.01 to 2.52. Citizens United ushered in more CPA, consistent with H9.

2. Increases in cash spent on political activity after Citizens United In real terms, the amount of lobbying expenditures and PAC donations also increased: nominal lobbying expenditures per firm increased by about 10%, and PAC donations by about 15%, both well above the inflation rate, and nearly double the trend from 1998 to 2008. In heavily regulated industries, lobbying expenditures also increased about 10%, while PAC donations declined by about 10%. Lobbying expenditures remained many times larger than PAC donations, and more firms were engaged in both kinds of CPA. As a result, the net change in CPA among S&P 500 firms after Citizens United was strongly positive. The total disclosed cash flows represented by the two types of CPA increased ~15%, by $107 million in nominal dollars, from $689 million in 2008 to $796 million in 2010. Citizens United spurred increases in CPA that were well above inflation and about 60% higher than the 1998 to 2008 trend, consistent with H9.

In addition, data from 2010 confirm the degree to which various political channels are complements. Even though Citizens United only relaxed rules on unobservable “independent” expenditures, firms that were already politically active in 2008 – either through a PAC or by lobbying – increased their observable lobbying expenditures by more than 10% (by $252,000 to $2.4 million in 2010, on average), while firms that were inactive in 2008 only engaged in a modest $30,000 of lobbying expenditures, on average. Since they were inactive in 2008, such firms could not have reduced their expenditures below zero, so the increase in lobbying by such firms is less telling than the fact that it is more than an order of magnitude lower than the increase in previously politically active firms.

3. Difference-in-differences: Citizens United as a quasi-experiment How did the value-politics relationship change in the 2010 elections, relative to the pre- Citizens United period? A final measure of the relationship between corporate politics and corporate value is the change following Citizens United in shareholder value at firms that were politically active before Citizens United, relative to the same change at firms that were not politically active, in each case measuring shareholder value (as above) relative to other firms in the same industry. Table 6 presents the results of a conventional difference-in-difference estimator with the following specification:

26 (1) LOGRELQ it = 0 + 1Xi +2(POSTCITIZENS  CPA i 2008 ) + 3CPA it , where LOGRELQ it is the firm i’s industry-relative Tobin’s Q, a, in year t; i indexes firms; t indexes years; X denotes the same vector of conventional firm and industry characteristics used in Table 5 to control for apolitical factors influencing firm value; POSTCITIZENS is a dummy equal to 1 for observations in year 2010, after Citizens United; CPA i 2008 is a dummy equal to one if firm i engaged in lobbying or made PAC contributions in 2008; and CPA it is a dummy set to one if firm i engaged in lobbying or made PAC contributions in year t. The coefficient of interest is 2, on the interaction term between firms politically active in 2008 and the observation of firm value in 2010, after Citizens United. [Insert Table 6 about here] Because Citizens United was unexpected (see note 31 above), firms were unlikely to have changed their pre-Citizens United political activities in anticipation of the decision. A firm’s post-Citizens United industry-relative value cannot plausibly have caused its pre-Citizens United political activity, which eliminates the possibility of reverse causation that may be present in the models presented in Table 5. The specification in Table 6 also does not have the serial autocorrelation problem identified in Bertrand Duflo and Mullainathan 2004, as there is only one period of post-Citizens United CPA data currently available. In addition, the model includes robust standard errors clustered by firm, and the results are qualitatively similar if one drops all observations prior to 2008, leaving only one pre- and one post-Citizens United observation for each firm (see id., at 252).

The model rules out potential confounding factors by including (for example) industry controls, which absorb any industry-driven changes in shareholder value between 2008 and 2010, such as may have been induced by debates over health care or financial reform. Industry effects are also absorbed because the dependent variable is a firm’s industry-relative Q ratio, which compares the firm’s value to other peer firms in the same industry. The model eliminates the possibility that the results are caused by some unobservable feature of firms that affects politically active and inactive firms in similar ways, or which changed at similar rates across firms over the sample period, or which is industry-specific, or is otherwise absorbed by other controls in the model, such as firm size, leverage, and accounting profitability.

The results of the difference-in-difference model in Table 6 are consistent with the results presented above. Firms that were politically active in 2008 experienced a significant decline in shareholder value in 2010, relative to firms that were politically inactive in 2008. The 95% confidence interval indicates a decline of up to 15% in industry-relative shareholder value, centered at 8%, and for the model with controls, all values in the interval are below zero. 27 The difference-in-difference in value for politically active firms of -0.08 is both economically meaningful and plausible, dramatically reducing the average increase in firm value in 2010 (from the post-crash lows of 2008) of +0.12. In 2010, LOGRELQ (logged median industry- adjusted Tobin’s Q) of sample firms ranged from –0.7 to 1.7, with a mean of 0.14. In sum, firms that were politically active in 2008 experienced a 75% lower increase [(0.12 – 0.08 ) / 0.12] in industry-relative market valuation during the market recovery in 2009 and 2010, as compared to politically inactive firms, consistent with H10, and (as suggested by evidence presented above) such firms being less focused, more apt to waste resources, and more distracted by political activities. VI. Summary and Interpretations In sum, the data are consistent with the hypotheses outlined in Part II. Among S&P 500 firms:

 Corporate political activity is most common in heavily regulated or government- dependent industries.  CPA correlates negatively with two different measures of shareholder power, which are themselves uncorrelated – ownership concentration and greater shareholder rights – and CPA correlates positively with measures of managerial agency costs – greater use by CEOs of corporate jets.

 CPA correlates positively with the significant fraction (11%) of large firm CEOs who gain post-CEO political office.  CPA’s relationships with shareholder power and managerial agency costs are weakest (or even reversed) in heavily regulated or government-dependent industries, and strongest in other industries.

 CPA correlates negatively with measures of corporate value – industry-adjusted Tobin’s Q – and that relationship, too, is weakest (or even positive) in heavily regulated or government-dependent industries, and is stronger in other industries, even after controlling for other factors in various ways, including with firm fixed effects.

 CPA overall increased after Citizens United, particularly in industries that are not heavily regulated or government-dependent, and particularly at firms that were previously politically active, consistent with different political channels serving as complements.

 CPA interacts negatively with capital expenditures, such that capital expenditures correlate with higher shareholder value, on average, at politically inactive firms, but do not do so at politically active firms.

 Firms that were politically active in 2008 experienced an average 8% lower increase in their industry-relative shareholder value from their crisis-era lows when compared to firms that were politically inactive in 2008, consistent with Citizens United inducing an increase in unobservable political activity by previously politically active firms, with a significant attendant drag on shareholder value. In combination, these results are inconsistent with a simple theory in which corporate political activity generally serves the interests of shareholders. They are also inconsistent with theories in which political activity is purely responsive to threats from elected officials and political operatives, since at least simple versions of such theories cannot explain the correlations found 28 here between CPA, shareholder rights, shareholder dispersion, and signs of managerial agency costs. The relationship with shareholder dispersion cannot be explained by noting that more dispersed firms are more salient and large, because the relationship persists after controlling for size and firm fixed effects, which would rule out durable differences in firm salience.

In addition, since managerial control over corporations is justified primarily on the ground that it enhances efficiency by increasing shareholder wealth, these results are also inconsistent with a simple theory in which CPA uniformly improves social welfare. Large amounts of money in politics, from whatever source, can corrupt – with attendant social harms to firms (see Chaney et al. 2010; Yu and Yu 2011; Igan et al. 2009) and polities (see Mo 2001; Podobnik et al. 2008; Apergis et al. 2011; Johnson et al. 2011). Finally, the findings that the politics-value relationship is negated or even reversed in heavily regulated industries suggest that the relevant regulations, far from constraining firms or reducing shareholder value, have been effectively counteracted by regulated firms – consistent with the familiar story of regulatory capture (e.g., Stigler 1971), and at least suggestive that regulations are not as burdensome as often claimed. The evidence presented here does not conclusively prove that CPA reduces shareholder value. It seems likely that politics and shareholder value influence each other, with lower value inducing politically inflected strategic gambles, and political engagements diluting strategic focus and inducing wasteful, politically inflected investments. On the one hand, if the sole explanation for the value-politics relationship were that firms facing difficulties turned to politics as a shareholder-oriented business strategy, one would not also find the correlations between CPA and shareholder power and managerial agency costs. On the other hand, the strength of the politics-value relationship is such that the correlations cannot be interpreted in a naïve way, with (for example) lobbying or PAC expenditures being treated as the sole cause of lower corporate value.

Firms run by poor or self-serving managers might have lower value than other firms even if all corporate political activity were banned. But there are plausible ways in which the availability of corporate political activity could further reduce corporate value, and do so in ways that do not simply reflect the out-of-pocket costs of lobbying or the costs of running a PAC. Corporate politics could fit into a good corporate strategy – and this likely explains why nearly all firms in heavily regulated or government-dependent industries engage in politics, and why those that do have no lower (and possibly higher) value than those that do not. But politics like war is risky, even for large companies (Smith 2000), and hard to predict, even for experts (Tetlock 2005). Significant corporate commitments – such as large capital expenditures – the value of which turn on accurately predicting or influencing political outcomes – will entail significant risks, even in the best of circumstances. For firms without a clear strategy, particularly those with managers that lack a strong shareholder orientation, the costs of politics could extend far beyond direct costs to include opportunity costs of manager time, distraction and confusion for middle managers and employees, the risks of consumer backlash, and the risks that politically contingent operational investments turn sour. Managers may also use CPA to cope with or reduce risk, which may also reduce shareholder value (see Lowery 2007). In addition, firms may use CPA for non-financial reasons not directly related to managerial agency costs, such as mimicry or network effects (e.g., Mizruchi 1992; Darves and 29 Dreiling 2002), or to pursue diffuse benefits such as legitimacy (e.g., Hillman and Hitt 1999), which might accentuate, dampen or be unrelated to the effects of CPA on firm value. Future research could attempt to test these ideas by examining whether the value-politics relationship found here is related to or mediated by other indicators of poor corporate strategy (such as acquisitions the value of which depend on politics), to operational behavior that could be the channel through which political activity produces poor results, to board interlocks or other firm-to-firm networks, or to connections between firms and political officials, as in Faccio 2010.

Even without that research, however, the possibility that political activity often runs counter to shareholder interests – whether as symptom or cause or both – is made more plausible by the finding here that the negative value-politics relationship is strongest among firms in industries where politics has the least obvious potential advantage for shareholders, by the finding that political activity also correlates strongly with other proxies for managerial agency costs, and the finding that politically active firms making large capital expenditures have significantly lower value than inactive firms making similarly sized capital expenditures. Most tellingly, the value of politically active firms diverged downward relative to inactive firms after the unexpected decision in Citizens United, even as political expenditures in already active firms increased. These findings make it plausible that corporate political activity commonly reflects broader agency problems at large public companies, and that the negative value-politics relationship is at least partly caused by politics. The cumulative effect of the findings adds support for proposals to require disclosure of such activity to shareholders. If Congress, states, or the SEC adopt rules attempting to give shareholders more information or more authority in the political sphere, the evidence presented here should help demonstrate that such legislation serves as a legitimate and compelling purpose separate from the anti-corruption and other purposes that have traditionally justified campaign finance laws. Contrary to the Supreme Court’s stated assumption, shareholders were not able to protect themselves from misuse of corporate funds for political purposes prior to Citizens United, and the risk of such misuse has increased as a result of the decision. VII. Conclusion This article has explored the relationships among corporate political activity, corporate governance, and corporate value in large public companies before and after Citizens United. It has found that observable corporate political activity (lobbying and PAC donations) increased sharply after Citizens United, particularly at firms that were already active in politics, despite the fact that the only direct effect of the legal decision was to permit “independent expenditures,” and did not directly change the law governing lobbying or corporate PACs. This finding is consistent with prior research showing that all forms of corporate politics are complements, and is a reminder that legal decisions can have many unintended and unexpected consequences. This article has also found that both before and after Citizens United, corporate politics correlates strongly with both corporate governance and corporate value. Specifically, in industries that are not heavily regulated or government dependent, political activity is 30 associated with weaker shareholder power, greater signs of managerial agency costs, and lower corporate value. The value-politics relationship is strongest for firms making large capital expenditures, suggesting one channel through which politics make lead to value-destroying investments. The precise extent and means by which politics may induce poor performance remains a topic for future research, but at a minimum the findings here reinforce the case that shareholders have a legitimate interest in obtaining better information about corporate politics. Even if political activity were a mere “symptom” of a more serious underlying disease for a given company, and not, as the difference-in-difference results suggest at least a partial cause, shareholders could use that symptom as a guide for where they should invest time and resources in improving corporate governance more generally – but only if disclosure laws are revised to reveal the symptom. Without disclosure reforms, the fact and extent of political activity will remain only partly revealed, with past and prospective investors having to infer the condition of the corporate patient from superficial and often misleading features, such as short- term recent stock-price performance, of the kind that lulled investors into thinking that all was well with Enron and Lehman Brothers until it was too late for them to do anything other than sell into an already plunging market. 41 If, as seems likely, corporate politics outside of heavily regulated or government dependent industries often reflects the personal interests of corporate managers, rather than a shareholder-oriented strategy, then shareholders may do well to try to curb corporate politics lest it disrupt or distract their companies from pursuing legitimate shareholder ends. 41 Compare Chief Justice Roberts, during reargument in Citizens United, “to your [the U.S. Government’s] shareholder protection rationale, isn’t it extraordinarily paternalistic for the government to take the position that shareholders are too stupid to keep track of what their corporations are doing and can’t sell their shares or object in the corporate context if they don’t like it?” Transcript of Oral Argument (Sep. 9, 2009) at 58. 31 Bibliography Ansolabehere, S., Snyder, J., and Tripathi, M. 2002. Are PAC contributions and lobbying linked? New evidence from the 1995 Lobby Disclosure Act. Business and Politics, 4: 131–155.

Ansolabehere, Stephen, John M. de Figueriredo and James M. Snyder, Jr., Why is There so Little Money in U.S. Politics?, 17 J. Econ. Persp. 105-130 (2003) Apergis, Nicholas, Oguzhan C. Dincer and James E. Payne 2011. Live Free or Bribe: On the Causal Dynamics Between Economic Freedom and Corruption in U.S. States, 28 Eur. J. Pol.

Econ. 215-22 (2012).

Austen-Smith, David 1995. "Campaign Contributions and Access." American Political Science Review 89: 566—581.

Baron, D. 1995. Integrated strategy: Market and nonmarket components. California Management Review, 37(2): 47–65.

Bebchuk, Lucian A., Coates IV, J.C., Subramanian, G.. The powerful antitakeover force of staggered boards: theory, evidence and policy. Stanford Law Review 54, 887–951 (2002).

Bebchuk, Lucian A., Alma Cohen, and Allen Ferrell 2009. What Matters in Corporate Governance, 22 Rev. Fin. Stud. 783-827 (2009).

Bebchuk, Lucian A., Alma Cohen and Charles C.Y. Wang, Learning and the Disappearing Association Between Governance and Returns, Working Paper (2010) Bebchuk, Lucian A., Martijn Cremers and Urs Peyer, The CEO Pay Slice, 102 J. Fin. 199-221 (2011).

Bebchuk, Lucian A., and Jesse Fried, Pay without Performance: The Unfulfilled Promise of Executive Compensation (2004).

Bebchuk, Lucian A., and Robert J. Jackson, Jr., Corporate Political Speech: Who Decides?, 124 Harv. L. Rev. 83 (Nov. 2010).

Berle, Adolph A., and Gardiner C. Means. The Modern Corporation and Private Property, (New York: Macmillan Company 1932).

Bertrand, Marianne, Esther Duflo, and Sendhil Mullinaithan, How Much Should We Trust Differences-in-Differences Estimates?, 119 Q.J. Econ. 249-75 (2004).

Bischoff, Steven M., Recent Corporate Research in the United States: The Highlights, Working Paper (2009) Blumentritt, T. 2003. Foreign subsidiaries’ government affairs activities: The influence of 32 managers and resources. Business and Society, 42: 202–233.

Bonardi, Jean-Philippe, Guy Holburn, and Richard Vanden Bergh (2006). "Nonmarket Performance: Evidence from U.S. Electric Utilities." Academy of Management Journal 49:

1209—1228.

Brasher, Holly, and David Lowery 2006. "The Corporate Context of Lobbying Activity." Business and Politics Vol. 8: Iss. 1, Article 1.

Brudney, Victor, Business Corporations and Stockholders’ Rights Under the First Amendment, 91 Y ALE L.J. 235 (1981) Burris, V. 2001. The two faces of capital: Corporations and individual capitalists as political actors. American Sociological Review, 66: 361–381.

Chaney, Paul K., Mara Faccio, and David Parsley, 2011. The Quality of Accounting Information in Politically Connected Firms, 51 J. Acc’t & Econ. 58-76 (2011).

Coates, John C. and Bradley C. Faris 2001. Second-Generation Shareholder Bylaws: Post- Quickturn Alternatives, 56 Bus. Law. 1323 (2001).

Coates, John C. and Reinier Kraakman 2011. Causes and Consequences of CLO Turnover, Working Paper (July 2011).

Coates, John C., 2010a. Citizens United, Corporate Governance, and a New Era of Crisis in the Supreme Court’s Legitimacy, Working Paper (July 2010).

Coates, John C., 2010b. The Powerful and Pervasive Effects of Ownership on M&A, Working Paper (2010) (available at http://ssrn.com/abstract=1544500).

Coates, John C., 2010c. Corporate Governance and Corporate Political Activity: What Effect will Citizens United Have on Shareholder Wealth?, Harvard Law and Economics Discussion Paper No. 684 (available at http://ssrn.com/abstract=1680861).

Coates, John C., 2000. Takeover Defenses in the Shadow of the Pill: A Critique of the Scientific Evidence, 79 Tex. L. Rev. 271-382 (2000).

Coates, John C., and Reinier Kraakman 2010. CEO Tenure, Performance and Turnover in S&P 500 Companies, Working Paper (2010). Coates, John C., and Taylor Lincoln 2011. Fulfilling Kennedy’s Promise: Why the SEC Should Mandate Disclosure of Corporate Political Activity (2011), available at http://ssrn.com/abstract=1923804. Cook, R., and Barry, D. 1995. Shaping the external environment: A study of small firms’ attempts to influence public policy, 34 Bus. and Soc. 317–344 (1995). 33 Cooper, Michael J., Huseyin Gulen and Alexei V. Ovtchinnikov, Corporate Political Contributions and Stock Returns, 65 J. Fin. 687 (2010) Core, John E., Wayne R. Guay, and Tjomme O. Rusticus, “Does Weak Governance Cause Weak Stock Returns: An Examination of Firm Operating Performance and Investors’ Expectations,” Journal of Finance 61: 655-687 (2006).

Darves, D and M. Dreiling 2002. Corporate Political Networks and Trade Policy Formation, 26 Humanity & Soc. 5-27 (2002).

Drope, Jeffrey, and Wendy Hansen 2006. "Does Firm Size Matter? Analysing Business Lobbying in the United States." Business and Politics Vol. 8: Iss. 2, Article 4.

Drope, Jeffrey, and Wendy Hansen 2008. Futility and Free-Riding: Corporate Political Participation and Taxation Rates in the United States, 10 Bus. and Pol. 3:2 (2008). Faleye, Olubunmi, Classified Boards, Firm Value, and Managerial Entrenchment, Journal of Financial Economics, 83, 501-529 (2007) de Figueiredo, John, and Brian Silverman 2006. "Academic Earmarks and the Returns to Lobbying." Journal of Law and Economics 49: 597—625.

de Figueiredo, John, and Emerson Tiller 2001. "The Structure and Conduct of Corporate Lobbying: An Empirical Analysis of Corporate Lobbying at the Federal Communications Commission." Journal of Economics and Management Strategy 10: 91—122.

Denis, David J., Diane K. Denis and Atulya Sarin, Ownership Structures and Top Executive Turnover, 45 J. Fin. Econ. 193-224 (1997).

Dlugosz, Jennifer, Rudgier Fahlenbrach, Paul Gompers and Andrew Metrick 2004. Large Blocks of Stock: Prevalence, Size, and Measurement, Working Paper (July 2004).

Faccio, Mara 2010. Differences between Politically Connected and Non-Connected Firms: A Cross-Country Analysis, 39 Fin. Mgt. 905-927 (2010).

Faccio, Mara, Ronald W. Masulis and John J. McConnell 2006. Political Connections and Corporate Bailouts, 56 J. Fin. 2597-2635 (2006).

Fama, Eugen F. and Kenneth R. French, Industry Costs of Equity, 43 J. Fin. Econ. 153-193 (1997).

Fisch, J., How Do Corporations Play Politics?: The FedEx Story, 58 Vand. L. Rev. 1495 (2005).

Franca, P. 2001. The effects of the North American Free Trade Agreement on corporate and 34 labor PACcontributions. American Politics Research, 29(1): 98–109.

Franz, M.M. The Citizens United Election? Or Same As Ever Was?, The Forum, Vol. 8, Art.

7, available at www.bepress.com/forum/vol8/iss4/art7 Fried, Charles, Lobbying Law in the Spotlight: Challenges and Proposed Amendments, 63 Admin. L. Rev. 419 (2011).

Getz, K. 1996. Politically active foreign-owned firms in the US: Elephants or chickens? in D.

Woodward and D. Nigh (Eds.), Beyond us and them: Foreign ownership and US competitiveness. Columbia, SC: University of South Carolina Press.

Gilson, Ronald and Michael Klausner, That’s My Money You’re Using, Forbes (Mar. 29, 2010).

Goldberg, Pinelopi, and Giovanni Maggi (1999). "Protection for Sale: An Empirical Investigation." American Economic Review 89: 1135—1155. Gordon, Sanford and Catherine Hafer (2005). "Flexing Muscle: Corporate Political Expenditures as Signals to the Bureaucracy." American Political Science Review, Volume 99:

245—261.

Gompers, Paul, Joy Ishii, and Andrew Metrick 2003. "Corporate Governance and Equity Prices." Quarterly Journal of Economics 118: 107—155. Gordon, Sanford and Dimitri Landa 2007. "Consumption or Investment? On Motivations for Political Giving." Journal of Politics 69: 1057—1072.

Gray, Virginia and David Lowery 1996. The Population Ecology of Interest Group Representation: Lobbying Communities in the American States (U. Mich. Press 1996).

Hadani, Michael, SEC Comment Letter dated Oct. 13, 2011 (available at http://www.sec.gov/comments/4-637/4637-8.pdf) Hadani, Michael, and D. Schuler, In Search of El Dorado: The Elusive Financial Returns on Corporate Political Investments, Working Paper (2011).

Hansen, Wendy L., Neil J. Mitchell and Jeffrey M. Drope 2005. The Logic of Private and Collective Action, 49 Am. J. Pol. Sci. 150.

Hansen, Wendy L., and Neil J. Mitchell 2000. Disaggregating and explaining corporate political activity: Domestic and foreign corporations in national politics. American Political Science Review, 94: 891–903.

Hart, D. 2001. Why do some firms give? Why do some firms give a lot?: High-Tech PACs, 1977–1996. Journal of Politics, 63: 1230–1249. 35 Heinz, John P., Edward O. Laumann, Robert L. Nelson and Robert Saintsbury 1993. The Hollow Core (Cambridge: Harvard U. Press).

Hendry John, et al., Owners or Traders? Conceptualizations of Institutional Investors and Their Relationship with Corporate Managers, 59 H UM . REL . 1101, 1102-03 (2006) Hersch, Philip, Jeffry M. Netter and Christopher Pope 2008. Do Campaign Contributions and Lobbying Expenditures by Firms Create “Political” Capital?, 36 Atl. Econ. J. 395-405 (2008).

Hillman, Amy, and Michael Hitt 1999. "Corporate Political Strategy Formulation: A Model of Approach, Participation, and Strategy Decisions." Academy of Management Review 24: 825- 842.

Hillman, Amy, Gerald Keim, and Douglas Schuler 2004. Corporate Political Activity: A Review and Research Agenda, 30 J. Mgt. 837-857 (2004).

Himmelberg, Charles P., R. Glenn Hubbard and Darius Palia 1999. Understanding the Determinants of Managerial Ownership and the Link Between Ownership and Performance, 53 J. Fin. Econ. 353-384 (1999).

Igan, Deniz, Prachi Mishra and Thierry Tressel, A Fistful of Dollars: Lobbying and the Financial Crisis, IMF Working Paper (Dec. 2009), available at http://ssrn.com/abstract=1531520.

Investor Responsibility Research Center (IRRC), Corporate Governance of Political Expenditures: 2011 Benchmark Report on S&P 500 Companies (November 2011).

Isikoff, Michael. Firm Gives $1 Million to Pro-Romney Group, Then Dissolves, NBC News (Aug. 4, 2011), available at http://goo.gl/zz3DT (last visited December 6, 2011).

Jackson, D., and Engel, S. 2003. Don’t bite the PAC that feeds you: Business PAC punishment over the China vote. American Politics Research, 31(2): 138-154.

Johnson, Noel D., Courtney L. LaFountain and Steve Yamarik 2011. Corruption is Bad for Growth (Even in the United States), 147 Pub. Choice 377-93 (2011).

Kim, Jin-Hyuk, Corporate Lobbying Revisited, Business and Politics 10(2):1-23 (2008).

Kroszner, Randall, and Thomas Stratmann 1998. "Interest-Group Competition and the Organization of Congress: Theory and Evidence from Financial Services’ Political Action Committees." American Economic Review 88: 1163-1187.

Kroszner, R., and Stratmann, T. 2000. Does political ambiguity pay? Corporate campaign contributions and the rewards to legislator reputation (NBER Working Paper 7475).

Cambridge, MA: National Bureau of Economic Research. 36 Jensen, Michael and William Meckling, Theory of the Firm: Managerial Behavior, Agency Costs, and Ownership Structure, 3 J. Fin. Econ. 305-360 (1976).

Jensen, Michael and Kevin Murphy, Performance Pay and Top Management Incentives, 98 J.

Pol. Econ. 225-263 (1990).

Kerr, William R., William F. Lincoln and Prachi Mishra, The Dynamics of Firm Lobbying, NBER Working Paper 1577 (2011).

Langbein, Laura (1986). "Money and Access: Some Empirical Evidence." Journal of Politics 48: 1052-1062.

Listokin, Yair, Interpreting Empirical Estimates of the Effect of Corporate Governance, 10 Amer. L. Econ. Rev. 90-109 (2008).

Lord, M. 2000. Corporate Political Strategy & Legislative Decision Making: The Impact of Corporate Legislative Influence Activities, 39 Business & Society 76-93 (2000). Lowery, David and Virginia Gray 2004. A Neopluralist Perspective on Research on Organized Interests, 57 Pol. Res. Q. 163–75 (2004).

Lowery, David 2007. Why Do Organized Interests Lobby? A Multi-Goal, Multi-Context Theory of Lobbying, 39 Polity 29–54 (2007).

Mauro, Paulo, 1995. Corruption and Growth, 110 Q.J. Econ. 681–712 (August 1995). Mcintire, Mike and Nicholas Confessore 2012. Tax-Exempt Groups Shield Political Gifts of Businesses, N.Y. Times at A1 (July 8, 2012).

Mitchell, N., Hansen, W., and Jepsen, E. 1997. The Determinants of Domestic and Foreign Corporate Political Activity. 59 J. Pol. 1096–1113 (1997).

Mizruchi, Mark 1992. The Structure of Corporate Political Action (1992).

Mo, P.H., 2001. Corruption And Economic Growth. 29 J. Comp. Econ. 66–79 (2001).

Morck, Randall, Daniel Wolfenzon and Bernard Yeung, Corporate Governance, Economic Entrenchment, and Growth, 57 J. ECON . LIT. 655 (2005).

North, David S., The Role of Managerial Incentives in Corporate Acquisitions: The 1990s Evidence, 7 J. Corp. Fin. 131-149 (2001).

Ozer, Mine, Ekin Alakent, and Mujtaba Ahsan 2010. Institutional Ownership and Corporate Political Strategies: Does Heterogeneity of Institutional Owners Matter?, 4 Str. Mgt. Rev. 18- 29 (2010). 37 Podobnik, Boris, Jia Shao, Djuro Njavro, Palmen Ch. Ivanov, and H.E. Stanley, Influence of Corruption on Economic Growth Rate and Foreign Investment, 63 Euro. Phys. J. B 547-550 (2008).

Porter, M., Competitive Strategy, Free Press, New York 1980.

Porter, M., Competitive Advantage, Free Press, New York 1985.

Porter, M., What is Strategy?, Harv. Bus. Rev. (Nov. 1996) at 61-78.

Richter, Brian K. and Krislert Samphantharak, Lobbying and Taxes, 53 Am. J. Pol. Sci. 893 (2009) Roe, Mark J., Strong Managers Weak Owners: The Political Roots of American Corporate Finance (1994).

Roe, Mark J., Political Determinants of Corporate Governance: Political Context, Corporate Impact (2003).

Ryan, Lori Verstegen and Marguerite Schneider, The Antecedents of Institutional Investor Activism, 27 A CAD . M GMT . REV . 554, 554 (2002).

Schuler, D. 1996. Corporate political strategy and foreign competition: The case of the steel industry. Academy of Management Journal, 39: 720–737.

Schuler, D. 1999. Corporate political action: Rethinking the economic and organizational influences. Business and Politics, 1(1): 83–97.

Schuler, D., and Rehbein, K. 1997. The filtering role of the firm in corporate political involvement. Business and Society, 36: 116–139.

Schuler, D., Rehbein, K., and Cramer, R. 2002. Pursuing strategic advantage through political means: A multivariate approach. Academy of Management Journal, 45: 659–672.

Securities and Exchange Commission (SEC), Executive Compensation and Related Person Disclosure, Securities Act Release No. 8732A, Exchange Act Release No. 54302A, Investment Company Act Release No. 27444A, 71 Fed. Reg. 53,158 (Sept. 8, 2006) available at http://www.sec.gov/rules/final/2006/33-8732afr.pdf.

Shleifer, Andrei and Robert Vishny, A Survey of Corporate Governance, 52 J. Fin.. 737 (1997).

Smith, Mark 2000. American Business and Political Power (U. Chi. Press).

38 Stigler, George 1971. The Theory of Economic Regulation, 2 Bell J. of Econ. & Mgt. Sci. 3– 21 (1971).

Tetlock, Philip E. 2005. Expert Political Opinion (Princeton: Princeton U. Press).

Tobin, James and W.C. Brainard, Asset Markets and the Cost of Capital, in Economic Progress, Private Values and Public Policy (1977).

Tripathi, Micky, Steven Ansolabehere, and James Snyder (2002). Are PAC Contributions and Lobbying Linked? New Evidence from the 1995 Lobby Disclosure Act, 4 Bus. & Pol. Iss. 2, Article 2 (2002).

Winkler, Adam, Corporate Personhood and the Rights of Corporate Speech, 30 Seattle U.L.

Rev. 863-73 (2007) Winkler, Adam, “Other People’s Money:” Corporations, Agency Costs, and Campaign Finance Law, 92 Georgetown L.J. 8871-940 (2004) Wright, John (1990). "Contributions, Lobbying, and Committee Voting in the U.S. House of Representatives." American Political Science Review 84: 417—438.

Yermack, David. Flights of Fancy: Corporate Jets, CEO Perquisites, and Inferior Shareholder Returns, 103 J. Fin. Econ. 211-42 (2006).

Yoffie, D. R. 1987. Corporate strategies for political action, in A. A. Marcus, A. M. Kaufman, and D. R. Beam (Eds.), Business strategy and public policy: Perspectives from industry and academia: 43-60. New York: Quorum.

Yu, Frank and Xiaoyun Yu 2011. Corporate Lobbying and Fraud Detection, 46 J. Fin’l & Quant. Anal. 1865-1891 (2011). 39 Table 1. Summary Statistics Mean or % positive Median St. dev. Min Max N LOBBY_YN 72.5 -- 44.6 0 1 4316 CONTRIBUTE_YN 68.6 -- 46.4 0 1 2899 LOBBYAMOUNT 1147.5 217.2 2505.3 0 45460.0 4316 CONTRIBUTEAMOUNT 240.9 56.6 485.2 0 4941.0 2899 Q_GINDEX 6.2 6.0 1.9 1 11 4359 EINDEX 2.4 2.0 1.3 0 6 4359 C_BOARD 53.9 -- 49.8 0 1 4359 CSHR 74.9 16.8 222.6 .001 4675.2 3812 INSIDER_OWN 13.1 6.0 15.2 0 1 1863 BLOCK_OWN 16.4 14.4 13.1 0 58.6 1782 CEOJETPOS (1998 to 2003 only) 22.9 0 42.0 0 1 1275 CEOJETVAL (1998 to 2003 only) 16.4 0.0 41.1 0 360.0 1275 CEOAIRCRAFT (2009 only) 34.7 0.0 47.6 0 1 478 CEOAICRAFTEXPENSE (2009 only) 42.1 0.0 101.2 0 1198.1 478 ASSETS 34629.8 8874.5 122657.2 396.0 2264909.0 3924 EMPLOYEES 44.9 19.0 92.8 0.1 2100.0 4272 COMPANYAGE 52.7 36.0 44.6 1 220 3314 REG_FAMA 42.0 -- 49.4 0 1 4363 REG_HEAVY 32.8 -- 46.9 0 1 4363 C4 39.6 35.6 19.5 1.5 98.9 3423 GOV_SHARE 6.4 3.2 9.6 0 91.5 3407 RELQ 1.2 1.0 0.9 0.2 34.0 3978 ROA 3.6 2.6 11.8 -458.3 55.9 4329 ROE 0.2 0.1 3.2 -113.5 141.7 3953 CAPEX_ASSETS 0.05 0.04 0.05 0 0.5 3716 R_AND_D_SALES -0.003 0 0.04 -1.4 0.0003 4344 DEBT 6556.2 1648.3 23802.7 0 448431 3788 LEVERAGE 0.5 0.5 0.3 0 6.2 3905 Note. Table shows summary statistics for sample firms, which include all firms in the S&P 500 for which relevant data are available from the period 1998-2004, 2008, and 2010, pooled unless otherwise noted. Amounts are in $000s or 000s. LOBBYAMOUNT and CONTRIBUTEAMOUNT are inflation adjusted, using 1998 as the base year. Data on political activities are derived from http://www.opensecrets.org, for 1998-2004 as derived from Kim 2008, and directly from the website for 2008 and 2010. Data on corporate governance are derived from IRRC via the Wharton Research Data Services website; data on jet use are from Yermack 2006 and GMI; firm record ownership (CSHR) and financial data are from Compustat; data on industry concentration (C4) and government expenditures (GOV-SHARE) are from the Census Bureau; block holder and insider ownership data are from Dlugosz, et al. 2004.

40 Table 2. Share of CEOs Obtaining or Being Nominated for Office (1) Total # of CEOs in 2000 in dataset 438 (2) CEOs of same company in 2011 88 20% of (1) (3) CEOs of another public company in 2011 31 7% of (1) (4) CEOs died while CEO 13 3% of (1) (5) CEOs who retired before 2011 298 68% of (1) (6) CEOs appointed or nominated for political office 33 11% of (5) (7) CEOs appointed or nominated for office Mean 95% Confidence Interval (a) If the firm lobbied in 1999, the % obtaining office was… 15% 12% 18% (b) If the firm did not lobby in 1999, the % was… 3% 0% 6% Note. Table reflects an analysis by the author of the post-2000 careers of individuals who were CEOs of S&P 500 companies in the year 2000, based on searches of the internet and Lexis/News. “Political office” includes positions with authority, such as cabinet Secretary, Senator, President, SEC Chair, etc., and not advisory positions, or such politically influenced recognitions as medals or knighthoods. In rows 7(a) and 7(b), the sample CEO firms were divided into those that reported lobbying expenditures in the year 2000, and those that did not, derived from http://www.opensecrets.org and Kim 2008. 41 Table 3. Lobbying Participation and Shareholder Power (Logistic) Explanatory variable (1) Each variable on its own (2) Both explanatory variables, plus other variables identified in note Odds ratio 95% confidence interval N Odds ratio 95% confidence interval N % correctly classified E-INDEX 1.15 1.04 1.28 3977 1.26 1.13 1.48 23407 78% LOG_SHH_NUM 1.58 1.43 1.73 3784 1.30 1.07 1.49 The dependent variable is whether a firm reported lobbying expenditures in a given year, derived from http://www.opensecrets.org , and Kim 2008. The E_INDEX measures shareholder rights and is derived from Bebchuk et al. 2009; LOG_SHH_NUM is the logged number of record shareholders of the firm, a proxy for shareholder dispersion, from Compustat. All models reflect robust standard errors, clustered at the firm level. Results reported in column (2) reflect the following additional variables (LOG_ASSETS, LOG_EMPLOY, ROA, ROE, LEVERAGE, REG_HEAVY, and GOV_SHARE25, as well as Fama 12-industry and yearly dummies). See Appendix A for descriptions. Table 4. PAC Donation Participation and Shareholder Power (Logistic) Explanatory variable (1) Each variable on its own (2) Both explanatory variables, plus other variables identified in note Odds ratio 95% confidence interval N Odds ratio 95% confidence interval N % correctly classified E-INDEX 1.04 0.95 1.14 3977 1.27 1.09 1.48 22189 80% LOG_SHH_NUM 1.59 1.43 1.77 3784 1.17 1.01 1.37 The dependent variable is whether a firm sponsored a PAC that made a donation in a given year, derived from http://www.opensecrets.org, and Kim 2008. The E_INDEX measures shareholder rights and is derived from Bebchuk et al. 2009; LOG_SHH_NUM is the logged number of record shareholders of the firm, a proxy for shareholder dispersion, from Compustat. All models reflect robust standard errors, clustered at the firm level. Results reported in column (2) reflect the following additional variables (LOG_ASSETS, LOG_EMPLOY, ROA, ROE, LEVERAGE, REG_HEAVY, and GOV_SHARE25, as well as Fama 12-industry and yearly dummies). See Appendix A for descriptions.

42 Table 5. Political Activity and Corporate Value Panel A. Logged Industry-Adjusted Tobin’s Q (OLS) Explanatory variables (1) Each explanatory variable on its own and interacted with REG_HEAVY (2) With other variables noted below Coef. 95% Confidence Interval N Coef. 95% Confidence Interval N LOBBY_YN -0.18 -0.24 -0.11 3698 -0.07 -0.13 -0.01 3305 LOBBY_YN x REG_HEAVY 0.03 -0.02 0.09 0.11 0.03 0.19 CONTRIBUTE_YN -0.17 -0.23 -0.11 2483 -0.06 -0.11 0.00 2267 CONTRIBUTE_YN x REG_HEAVY 0.02 -0.04 0.07 0.08 0.01 0.16 Ordinary least squares (OLS) models with logged median-industry-adjusted Tobin’s Q as the dependent variable, with robust standard errors, clustered by firm. Results in column (1) reflect explanatory variables set to one for reported lobbying expenditures, sponsored-PAC donations and interactions with a dummy set to one if in an industry in which regulation is particularly comprehensive, from Ken Fama’s website: http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html: alcohol, tobacco, aircraft, drugs, utilities, telecom, transportation, banks, and insurance. Column (2) adds standard controls (LOG_ASSETS, LOG_CO_AGE_MONTHS, DEINC, ROE, ROA, CAPEX_ASSETS, R&D_SALES, LEVERAGE, a dummy for missing R&D data, LOGCSHR, yearly dummies, and Fama 12-industry dummies). See Appendix A.

43 Panel B. Logged Industry-Adjusted Tobin’s Q (Firm Fixed Effects) Explanatory variables (1) Each explanatory variable on its own and interacted with REG_HEAVY, plus firm fixed effects (2) With other variables noted above, plus firm fixed effects Coef. 95% Confidence Interval N obs N firms Coef. 95% Confidence Interval N obs N firms LOBBY_YN -0.11 -0.17 -0.06 3698 -0.07 -0.13 -0.02 3305 LOBBY_YN x REG_HEAVY 0.02 -0.09 0.13 588 0.04 -0.10 0.18 556 CONTRIBUTE_YN -0.01 -0.10 0.08 2483 0.02 -0.06 0.11 2267 CONTRIBUTE_YN x REG_HEAVY -0.01 -0.16 0.14 587 0.02 -0.14 0.19 554 All models are as described for Panel A, with the addition of firm fixed effects in lieu of industry dummies. 44 Panel C. Logged Industry-Adjusted Tobin’s Q (OLS), only firms in non-heavily regulated industries Explanatory variables (1) Each interaction on its own (2) Each interaction with variables noted above Coef. 95% Confidence Interval N Coef. 95% Confidence Interval N CAPEX_ASSETS 2.27 1.34 3.21 3675 1.58 0.78 2.39 3461 LOBBY_YN x CAPEX_ASSETS -2.17 -3.09 -1.24 -0.89 -1.68 -0.11 CAPEX_ASSETS 2.16 1.28 3.03 2469 1.33 0.50 2.166 2346 CONTRIBUTE_YN -2.11 -2.99 -1.23 -0.63 0.00 0.07 x CAPEX_ASSETS All models are as described for Panel A, but the sample is limited to firms outside of heavily regulated industries. 45 Table 6. Difference-in-Differences in Corporate Value Before and After Citizens United Explanatory variables (1) Explanatory variable on its own (2) With other variables noted below Coef. 95% Confidence Interval N Coef. 95% Confidence Interval N POSTCITIZENS x CPA 2008 -0.06 -0.13 0.02 3620 -0.08 -0.15 -0.01 3258 POSTCITIZENS 0.11 0.04 0.18 0.11 0.04 0.19 CPA -0.16 -0.23 -0.10 -0.05 -0.11 0.01 Ordinary least squares (OLS) models with logged median-industry-adjusted Tobin’s Q as the dependent variable, with robust standard errors, clustered by firm. The key variable of interest is POSTCITIZENS x CPA 2008 , a dummy set to one for a firm reporting lobbying expenditures or sponsored a PAC making a donation in 2008, prior to Citizens United, as it was valued in the post-Citizens United period of 2010. Column (2) adds standard controls (LOG_ASSETS, LOG_CO_AGE_MONTHS, DEINC, ROE, ROA, CAPEX_ASSETS, R&D_SALES, LEVERAGE, a dummy for missing R&D data, LOGCSHR, yearly dummies, and Fama 12-industry dummies). See Appendix A for descriptions. 46 Figure 1. Industry and corporate political activity Note: Figure shows, by type of industry, the mean percentage of S&P 500 firms engaging in two types of political activity (reporting lobbying expenditures or sponsoring a PAC that made donations) in the period 1998-2004, 2008 and 2010. “Government-dependent industry” is one deriving more than 25% of its revenues from government expenditures, as reported by the U.S. Census Bureau. “Heavily regulated” industries are those in which regulation is particularly comprehensive, from Ken Fama’s website: http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html: alcohol, tobacco, aircraft, drugs, utilities, telecom, transportation, banks, and insurance. 95% confidence intervals are shown by tick-bars.

50% 60% 70% 80% 90% 100% Heavily Regulated Industry Government Dependent IndustryOther Industry % of firms engaging in political activity % PAC Donations % Lobbying 47 Figure 2. Firm size, shareholder dispersion, and lobbying activity Note. Figure shows the percentage of S&P 500 firms reporting lobbying expenditures, by quartile of assets and by quartile of the number of record common shareholders, as derived from Compustat, in the period 1998-2004, 2008 and 2010.

30% 50% 70% 90% 1 2 3 4 % of firms engaged in lobbying Asset quartile CSHR Quartile 4 CSHR Quartile 3 CSHR Quartile 2 CSHR Quartile 1 48 Figure 3. Corporate political activity, governance, and industry Panel A. Firms not heavily regulated Panel B. Heavily regulated firms Note. Figure shows the mean percentage of S&P 500 firms engaging in two types of political activity (reporting lobbying expenditures or sponsoring a PAC that made donations) in the period 1998-2004, 2008 and 2010, by extent of shareholder rights, measured by the E-INDEX, an index that runs from zero (most shareholder rights) to six (least), as described in Bebchuk et al. 2009, following Coates 2000. On the left are firms that not heavily regulated; on the right are those that are in industries in which regulation is particularly comprehensive, from Ken Fama’s website: http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html: alcohol, tobacco, aircraft, drugs, utilities, telecom, transportation, banks, and insurance. 95% confidence intervals are shown by tick-bars.

49 Figure 4. CPA by Jet Use in Industries That Are Not Heavily Regulated 70% 75% 80% 85% 90%  No  Ye s  CEO Jet Use? Predicted PA C Dona on %  Predicted Lobbying %  Panel A. Firms not heavily regulated Panel B. Heavily regulated firms Note. Figure shows the mean percentage of S&P 500 firms engaging in two types of political activity (reporting lobbying expenditures or sponsoring a PAC that made donations) in the period 1998-2003 and 2010, by divided into firms reporting that their CEOs used the company jet for personal travel, and those that did not. Data from the period prior to 2003 are from Yermack 2006; data on jet use from 2009 from GMI are matched to data on political activity in 2010. On the left are firms that not heavily regulated; on the right are those that are in industries in which regulation is particularly comprehensive, from Ken Fama’s website:

http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html (alcohol, tobacco, aircraft, drugs, utilities, telecom, transportation, banks, and insurance). 95% confidence intervals are shown by tick-bars.

70% 80% 90% 100%  No  Ye s  CEO Jet Use?  Predicted PA C Dona on %  Predicted Lobbying %  50 Figure 5. Corporate political activity and corporate value 0.00 0.05 0.10 0.15 0.20 0.25  No  Ye s  Predicted Logged Industry-Adjusted Tobin’s Q (derived from Table 4, Panel A) PA C Dona ng  Lobbying  Note. Figure shows mean predicted industry-adjusted logged Tobin’s Q – a standard measure of corporate value -- for S&P 500 firms, derived from the model in Table 4, Panel A, by whether the firms engaged in two types of political activity (reporting lobbying expenditures or sponsoring a PAC that made donations) in the 1998-2004, 2006 and 2010. 95% confidence intervals are shown by tick-bars.

51 Figure 6. Change in lobbying activity over time Note. Figure shows percentage of firms in the S&P 500 reporting lobbying expenditures in the period 1998- 2004, 2008 and 2010. The top line shows industries in which regulation is particularly comprehensive, from Ken Fama’s website: http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html (alcohol, tobacco, aircraft, drugs, utilities, telecom, transportation, banks, and insurance). The bottom line shows other firms. Years 2005-2007 and 2009 are omitted due to costs of data gathering. Citizens United was decided in 2009, show the difference between 2008 and 2010 reflects its effects. 52 Appendix A. Control variables for regression models.

The variables included as controls for the regression models of CPA and Tobin’s Q are:

 LOG_ASSETS (logged assets) and LOG_EMPLOY (logged number of employees) serve as proxies for firm size and/or for the size of the pool from which a company may solicit donations to a corporate PAC, from Compustat.

 CAPEX_ASSETS, the ratio of capital expenditures to assets, both from Compustat.

 LOG_CO_AGE_MONTHS, the log of the number of months since the company’s founding, from the Corporate Library (available primarily for years after 2003), with missing years interpolated, as a proxy for firm reputation or credibility.  Return on assets (ROA) and common equity (ROE) are derived from Compustat: ROA equals net income (Compustat item 172) divided by total assets (item 6). ROE equals net income divided by total assets.  DEINC, a dummy set to one if the firm is incorporated in Delaware.

 R_AND_D_SALES, the ratio of research and development expenditures to sales (and a dummy set to one if R&D expenditures are missing from Compustat).

 LEVERAGE, the ratio of debt to assets, where debt is long-term debt (DD1 and DLTT in Compustat).

 FAMA, industry dummies using the Fama-French 1997 mapping of standard industrial classification codes to 48 categories (and another mapping into 12 categories). 42  REG_FAMA, a regulated-industry dummy for firms in regulated industries derived from Fama’s website, 43 and REG_HEAVY, a subset of those industries in which regulation is particularly comprehensive (alcohol, tobacco, aircraft, drugs, utilities, telecom, transportation, banks, and insurance);  GOV_SHARE, the share of a firm’s industry’s revenues derived from government expenditures, as reported periodically by the Census Bureau, as a proxy for potential benefits from lobbying, and GOV_SHARE25, a dummy set to 1 for firms in industries deriving more than 25% of revenues from government expenditures; and 42 http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html.

43 Regulated Fama 48-industries are 4 (alcohol), 5 (tobacco), 13 (drugs), 24 (aircraft), 26 (guns), 27 (gold), 30 (oil), 31 (utilities), 32 (telecom), 40 (transportation), 44 (banks), 45 (insurance) and 47 (finance). Id. 53  YEAR, annual dummies, to control for time trends.