Shareholder Activism and Political Spending

Transcription

1 Shareholder Activism and Political Spending Vishal P. Baloria Boston College Kenneth J. Klassen University of Waterloo and Christine I. Wiedman University of Waterloo July 2014 We thank John Coates, Giri Kanagaretnam, Patricia O Brien, Florin Sabac, Jennifer Wu Tucker, Heidi Welsh, Heather Wier, Bill Zhang, and participants at the 2013 AAA Annual Meeting, the 2013 CAAA Annual Meeting, the 2013 McMaster University Accounting Research Symposium, and the McGill University, University of Alberta, and University of Waterloo workshops for their comments and suggestions. We also thank Khin Phyo Hlaing and Weiming Liu for providing excellent research assistance.

2 Shareholder Activism and Political Spending Abstract: We examine the determinants and consequences of shareholder activism around corporate political spending using a sample of political spending shareholder proposals from 2004 and Proposals calling for enhanced disclosure of spending information are most prevalent and garner the most votes relative to those calling for information on spending philosophy or say on political spending. We find that 20% of the proposals are subsequently adopted, a rate that is high relative to previous studies. Further, we find that while unions and pension funds appear to target firms for private interests, voting patterns and market reaction to implementation decisions reflect a more sophisticated understanding of political spending and an ability to filter out opportunism on the part of activists / management. Our findings contribute to the ongoing debate about the role of disclosure and shareholder voice in an era where firms face fewer restrictions in directing corporate funds towards political spending. JEL Classifications: M41, G34, H32 Keywords: shareholder activism, corporate political spending, disclosure, opportunism, political ideology, agency problems

3 I. INTRODUCTION Corporate political spending has become the subject of considerable scrutiny due to concerns that political spending is symptomatic of agency problems within firms (Aggarwal et al. 2012) and destroys shareholder value (Coates 2012). Concerns have intensified since the landmark Citizens United v. Federal Election Commission ruling in 2010; 1 firms now face fewer restrictions in directing corporate funds towards political spending and investor groups have become more vocal in calling for accountability. Since 2004, activist shareholders have targeted a number of firms with shareholder proposals requesting greater transparency and influence in corporate political spending decisions. More than 500 shareholder proposals related to political spending have been submitted from 2004 to 2012, and, in recent years, proposals on this matter are among the most frequently submitted proposals on any topic. Proponents of this shareholder activism campaign argue that corporate political spending is of increasing economic importance, is associated with a number of risks which threaten shareholder value, and that reform is necessary to ensure political spending is aligned with the interests of the broader shareholder base. Opponents maintain that special interest groups, such as labor union and elected members with oversight responsibilities of public pension funds, are using the campaign to pursue private political agendas. Opponents further assert that greater transparency and shareholder influence impose unnecessary compliance costs on firms and can result in unintended consequences such as revelation of proprietary information. In this paper, we attempt to inform this controversial debate by conducting three sets of empirical analyses. 1 In Citizens United v. Federal Election Commission, the Supreme Court ruled against the ban on corporate political spending in candidate elections. The decision maintains the ban on direct contributions from corporations to candidates, but allows corporations to air advertisements expressly advocating for or against particular candidates. While the ruling lifted the ban on corporate political spending in candidate elections, it also envisioned additional disclosure requirements to hold corporations accountable.

4 Using a comprehensive sample of political spending-related shareholder proposals submitted to S&P 500 firms between 2004 and 2012, we first study the determinants of the decision to target specific firms. In this analysis, we compare firms across a number of dimensions, including political ideology and agency conflicts. Evidence that targeted firms have more conservative ideologies would be consistent with activists targeting firms with political ideologies different from their own, while evidence of agency conflicts within these firms would be consistent with activists being motivated by monitoring incentives. We find that, on average, activists do not target firms with agency problems or conservative ideologies. Instead, activists target decisions appear to be motivated by the usual suspects identified in prior studies targeted firms are large, poorly performing firms, with fewer growth options. We also find that targeted firms have higher levels of observable political spending, though we note that this result does not allow us to discriminate between the two competing motives. When we split observable political spending into a predicted component based on economic determinants and a residual component ( excess spending), we find that both components are positively related to the probability of being targeted, inconsistent with a sophisticated targeting approach. As labor union pension funds and public pension funds (henceforth pension funds ), in particular, have been accused of pursuing private interests through the shareholder activism campaign, we also consider proposals initiated by pension funds separately from those initiated by other activists. We find that, relative to non-targeted firms, firms targeted by pension funds have more conservative political ideologies but similar levels of agency problems, while firms targeted by other activists are characterized by more agency problems, but similar political ideologies. When we directly compare firms targeted by pension funds to those targeted by other activists, we find pension funds target firms have more conservative ideologies and less 2

5 evidence of agency problems. Collectively, this evidence is consistent with pension funds (other activists ) target decisions being motivated, in part, by special interests (monitoring incentives). Second, we conduct an in-depth textual analysis of the 322 political spending-related shareholder proposals that were ultimately voted on. We identify three main types of proposals: (1) spending information proposals, which call for enhanced disclosure of political spending information; (2) spending philosophy proposals, which aim to affect the objective of political spending; and (3) spending voice proposals, which seek to give shareholders a greater voice in political spending decisions. We also capture the arguments made by activists in support of their proposals and those made by management in opposition to the proposals. Activists typically cite business and reputational risks of political spending or the potential for agency conflicts to arise in political spending decisions as a rationale for their proposals. Management support their opposition to proposals by asserting that the proposals are not relevant to their firm, that current mandatory and voluntary disclosures render the proposals unnecessary, or that compliance and/or proprietary costs associated outweigh any benefits associated with the proposals. 2 We then examine the determinants of the voting outcome for this set of proposals to infer the criteria voting shareholders use in evaluating changes to political spending policies. We consider a wide set of potential determinants, including characteristics of the proposal as well as characteristics of the target firm. We find that spending information proposals receive the highest voting support while proposals directed at shaping the objective of political spending or giving shareholder a greater voice consistently receive little support. We also find that voting support is higher for proposals with greater managerial opposition (as measured by the number of specific issues raised by management to support their position). In contrast, we do not find any 2 We do not treat any of these classifications as mutually exclusive; issues raised by activists and management often span multiple categories. 3

6 association between voting support and the extent of activist support for the proposal (as measured by the number of specific issues used by activists to support their position). Analysis of target firm characteristics shows that proposals in firms characterized by agency conflicts receive higher voting support. In sum, shareholders appear to incorporate characteristics of the proposal and the target firm to judiciously support proposals seeking greater disclosure rather than those aiming to impact political spending itself. In our third and final analysis, we examine the consequences of political spending related activism. We consider implementation rates for 541 political spending-related proposals as well as shareholders response to these implementations. We document a 20% implementation rate, which is very high relative to the 5% to 10% rates documented in prior U.S. studies of other types of proposals (Ferri and Sandino 2009; Ertimur et al. 2010; Ertimur et al. 2011). Among firms targeted by proposals that were voted on, we show that the likelihood of implementation increases in the degree of voting support. For the 90 targeted firms that implement the proposal and for which we can identify the precise implementation announcement date, we examine the market reaction to the implementation decision. We find that investors respond negatively, on average, to firms announcing implementation of political spending-related proposals. Further, investor reaction is positively related to the extent of agency problems within the implementing firm and negatively related to the existence of a pension fund proposal sponsor. Taken collectively, the results of our analyses suggest a more nuanced perspective than has been offered by supporters and detractors of the political spending related shareholder activism campaign. While we document some evidence consistent with a subset of activists (pension funds) targeting firms for private interests, shareholders voting patterns and reaction to implementation decisions reflect a sophisticated understanding of corporate political spending 4

7 and an ability to filter out any opportunism on the part of activists and/or management. Our findings speak favorably to the effectiveness of shareholder activism campaigns, as a striking feature of our setting is that almost all of the firms that have made changes to corporate political spending policies have done so due to pressure from shareholder activists. From a policy perspective, our study helps to inform deliberations on regulatory and legislative proposals to grant shareholder greater access to information on corporate political spending. 3 Our study makes several contributions to the literature. First, our research contributes to the literature on shareholder activism. Previous studies document that shareholder proposals are becoming a more effective tool to influence corporate decisions, including governance structure (Bizjak and Marquette 1998; Thomas and Cotter 2007; Guo et al. 2008; Ertimur et al. 2010), financial reporting choices (Ferri and Sandino 2009), and executive pay practices (Cai and Walking 2011; Ertimur et al. 2011; Ferri and Maber 2013). We contribute to this literature by conducting an in-depth textual analysis of shareholder proposals, identifying arguments made by activists and management in support of or against the proposal, and correlating these arguments with voting and implementation outcomes. Our findings are consistent with investors seeing through opportunism on the part of shareholder activists and management in their voting and investment decisions. 4 Our paper also provides evidence on the effectiveness of shareholder activism, as the implementation rates we find are, to the best of our knowledge, the highest ever documented. 3 Recent proposals seek to amend both securities and campaign finance laws to mandate corporate disclosure of unobservable political spending information. A rulemaking petition filed at the Securities and Exchange Commission (SEC) in August 2011 by ten law professors calls on the agency to develop rules to mandate corporate disclosure. To date, over 600,000 comment letters have been received in support of the petition, the highest of any rulemaking petition in the SEC s history. On the legislative front, several attempts to introduce draft legislation in the U.S. House and Senate have been made in recent years with the purpose of promoting corporate disclosure. 4 As noted in Ferri and Oesch (2013), a challenge in identifying the influence of activists (management) on voting shareholders is that activists (managements ) recommendations are typically the same across firms, in favor of (against) shareholder proposals. We overcome this challenge by identifying variation in the strength of support (opposition) by activists (management) through a textual analysis of the issues raised by each party. 5

8 Second, we contribute to the accounting literature that relates shareholder behavior to firms voluntary financial and non-financial disclosures. Previous research has found that investor characteristics impact firms disclosure choices (Bushee et al. 2003; Ajinkya et al. 2005; Karamanou and Vafeas 2005; Ali et al. 2007; Chen et al. 2008) and that increases in disclosure are associated with increases in institutional investor ownership (Healy et al. 1999; Bushee and Noe 2000; Dhaliwal et al. 2011). However, there is scant evidence on the mechanism by which investors preferences translate into changes in firms disclosure policies. We contribute to this literature by identifying shareholder proposals as one method by which investors express their preferences for and obtain desired disclosure outcomes. Our study of disclosure complements Ferri and Sandino (2009) on financial reporting, where the authors documents a positive relation between shareholder activism and the choice to expense employee stock options. Third, we add to the literature examining the influence of pension funds on corporate behavior. A number of prior studies find that public and union pension funds pursue private interests rather than shareholder value (Schwab and Thomas 1998; Woidtke 2002; Cai and Walking 2011; and Agrawal 2012). We contribute to this line of research by identifying a hitherto undocumented motive of pension funds activism decisions targeting firms with opposing political ideologies. We show that investor reactions to implementation decisions are negatively associated with pension fund sponsorship, providing further evidence that investors view pension fund activism as misaligned with the interests of the broader shareholder base. The paper proceeds as follows. Section II discusses the institutional background and related literature. We present our findings on the determinants of the targeting decision in Section III, determinants of voting outcomes in Section IV, and the consequences of political spending-related shareholder activism in Section V. In Section VI, we draw our conclusions. 6

9 II. INSTITUTIONAL BACKGROUND AND RELATED LITERATURE Over the last decade, as corporate political spending in the U.S. has steadily increased, so too has the call for greater transparency and accountability on that spending. Interest has further intensified since the Citizens United v. Federal Election Commission ruling in 2010 that lifted the ban on corporate political spending in candidate elections. Corporate watchdogs, individuals, and investor groups have become more vocal in calling for change. In 2003, the Center for Political Accountability (CPA) launched a disclosure campaign with the express purpose of promoting transparency in corporate political spending. In August 2011, ten law professors, led by Lucian Bebchuk, filed a rulemaking petition at the Securities and Exchange Commission (SEC) calling on the agency to develop rules to mandate corporate political spending disclosure. And in the U.S. House and Senate, draft legislation has been introduced on several occasions to promote corporate disclosure of political spending without success. While the SEC included the issue on its agenda in January 2013, SEC Chairman Mary Jo White suggested later in 2013 that action was not imminent, and chose not to include it in its regulatory agenda for We examine shareholder activism as an alternate mechanism for change in the reporting and governance surrounding corporate political spending. To bring attention to a particular issue, activist shareholders can file a shareholder proposal to be included in a company s Proxy Statement and voted on at the annual shareholder meeting. To be eligible to submit a proposal under SEC Rule 14a-8, a shareholder must, among other things, provide documentation that they have continuously held at least $2,000 in market value (or 1%, of the company s securities) for at least one year as of the date of submission of the proposal. While votes on proposals are nonbinding, they represent a formal communication channel between shareholders and management, 7

10 and proposals receiving higher votes (particularly those receiving majority vote) are more likely to influence management. The academic research on shareholder activism as a mechanism for effecting change in corporate behaviour is mixed. Early studies conclude that there is little evidence that shareholder proposals create value and that they have only a limited impact on firm governance structures (Karpoff 2001). More recently, however, studies document that shareholder proposals are increasingly impacting corporate decisions. Guo et al. (2008) find that firms receiving a shareholder proposal are more likely to de-stagger their boards, and Thomas and Cotter (2007) find that 7% of shareholder proposals filed between 2002 and 2004 on governance issues such as executive compensation and corporate control are adopted by the board. Further, Ertimur et al. (2010) find that 31% of governance proposals submitted between 1997 and 2004 and receiving a majority vote (greater than 50%) are adopted, and Ertimur et al. (2011) find that 5.3% of firms implement proposals relating to CEO compensation. Therefore, shareholders do appear to have a voice through shareholder proposals. Research on shareholder activism around issues relating to disclosure and financial reporting is sparse. Byrd et al. (1998) examine how executive compensation related shareholder activism influences firms voluntary disclosure decisions. They find firms are more likely to voluntarily disclose the comparison group used by the compensation committee to set executive compensation policies when they have received prior shareholder proposals relating to disclosure or capping the level of executive compensation. Ferri and Sandino (2009) find that 7.5% of firms targeted with a proposal to expense employee stock options are implemented, with rates significantly higher for proposals receiving a majority vote. To our knowledge, no studies have considered the efficacy of shareholder activism in the setting of corporate political spending. 8

11 III. SAMPLE SELECTION AND DETERMINANTS OF THE TARGETING DECISION Our sample consists of 541 shareholder proposals related to political spending at firms in the Standard & Poor s (S&P) 500 index over the period obtained from RiskMetrics. Of these 541 proposals, 322 went to vote and 219 did not. 5 Proposals may be withdrawn for two primary reasons: the SEC may grant exclusion of the proposal because of ineligibility (typically if the proposal is vague, false or misleading, the proposal relates to the company s ordinary business operations, the proposal is contrary to applicable law, or on procedural grounds such as the defects in the shareholder s proof of ownership); or, because the company and submitting shareholder have reached an agreement. Because of the latter case, it is important to consider all eligible proposals in assessing the efficacy of shareholder activism. Therefore, our target analysis is based on the full sample of proposals. However, to assess broader investor views of corporate political spending and interpretation of statements made by activists and management, we analyze voting patterns for the proposals that went to vote in Section IV. Activist Motives Firms engage in the political process to derive public and private benefits. Governments can be a source of cost and benefit to the market as a whole or to a particular industry. 6 Therefore, firms engage collectively, through business organizations, coalitions and industry groups, or individually to influence government policy. Further, market or industry effects may filter down more directly to an individual firm, as governments can selectively enforce regulations, exercise powers that may prevent a firm s competitor from entering a market, or create demand for an individual firm s product (Hansen et al. 2005). This gives firms additional 5 The 541 proposals submitted to S&P 500 represent 93% of the 579 proposals (348 went to a vote and 231 did not) submitted to the broader population of S&P 1500 firms. 6 An example of cost is regulation on emission standards; an example of benefit is the protecting of markets through tariffs and import quotas 9

12 incentives to individually lobby and become involved in local, state and federal elections. Firms may also seek other private goods, such as government contracts, through political activity. Shareholder activists target firms with a political spending-related shareholder proposal for a broad set of reasons, including financial and non-financial. Under one view, activists are motivated by monitoring incentives. Activists target firms where there are potential agency conflicts, and spending for political activities may not be aligned with outside shareholders interests. Yu and Yu (2011) report that politically connected firms can evade and delay fraud detection, allowing for greater managerial extraction of shareholder wealth. Activists may also target firms with higher spending, as Aggarwal et al. (2012) report a negative association between political spending and future stock returns. They may also target firms whose political activities expose the firm to business, legal or reputational risks. 7 Activists may also use the disclosure campaign to pursue their own objectives and promote awareness of self-interested initiatives. For example, Wall Street Journal editor Paul Gigot argues, This isn t the effort of public-spirited believers in shareholder rights and transparency. It s part of a multiyear campaign by unions, left-wing activists and their factotums to expose and then vilify companies that disagree with them (Wall Street Journal 2012). Under this view, activists may wish to bring visibility to the broad issue of political spending by targeting large firms, or they may target firms whose political ideology is different from their own. Labor union and public pension funds, in particular, have been accused of pursuing private interests through shareholder proposals. Labor unions are politically active and opponents argue 7 The SEC rulemaking petition and proposed legislation described above are based on the premise that corporate political spending is of increasing economic importance and that disclosure is necessary to ensure spending is aligned with shareholders interests. Similarly, a group of mutual fund and institutional asset managers argue that full disclosure of political spending would allow investors to manage, and to help mitigate, the full range of risks presented by corporate political spending. The risks cited include private rent seeking, conflicts of interest and selfdealing, pay to play political fundraising, and legal risks. 10

13 that union pension fund activism is driven by a desire to adversely impact the spending of companies that disagree with them (Wall Street Journal 2012). As public pension funds are typically managed by elected officials, opponents argue that their activism is motivated by political self-interests (i.e. promoting politically popular causes, including those relating to the environment, corporate political activity, and government use of privatized services) rather than shareholder wealth maximization. In our context of political spending, pursuit of private interests by union and public pension funds could be reflected in those funds targeting firms whose political ideology is not aligned with their own (democratic) ideology. Rather than directly focusing on ideological conflicts, most previous studies have focused more directly on unions role in the collective bargaining process as a primary motive for activism. Agrawal (2012) finds that over a four-year time period, the American Federation of Labor and Congress of Industrial Organizations was more likely to vote against directors at companies that were in the middle of labor disputes with employees. Schwab and Thomas (1998) find that unions use their shareholder power to further organizing and collective-bargaining goals. Woidtke (2002) finds that companies with large ownership by public rather than private pension funds have lower firm valuations. On the other hand, Prevost et al. (2012) do not find that labor unions use shareholder proposals to extract greater gains for their employees as reflected in labor cost and productivity changes, and Ertimur et al. (2011) find no evidence for say-on-pay proposals that unions target firms with a higher percentage of unionized workers, firms involved in labor disputes, or firms renegotiating their collective bargaining agreements. Target Analysis - Research Design To examine the determinants of the targeting decision, we benchmark firms receiving political spending-related shareholder proposals against a control sample comprising all other 11

14 firms. Our sample consists of S&P 500 firms and includes 434 targeted firm-years (541 proposals, less 69 cases of more than one proposal filed within a given year for a given firm, less 38 firms with missing data) and 3,373 non-targeted firm-years over the sample period. We combine these samples and estimate the following pooled firm-year logistic regression with standard errors clustered by firm: Pr (Target i,t ) = Political Spending i,t Political Ideology i,t CEO Duality i,t Entrenchment Index i,t-1 + Control Variables i,t-1 + i,t (1) The dependent variable, Target, equals 1 if the firm receives a political spending-related shareholder proposal in year t, and 0 otherwise. We include several variables to capture activists motives for targeting a firm. First, activists may target firms who spend heavily on political activities. Higher spending can make the firm s political activities more visible. Higher spending may also represent greater risks or greater potential agency conflicts associated with participation in the political process. To capture these different aspects of political participation, we consider lobbying as well as campaign financing, and define observable political spending (Political Spending) as the sum of expenditures on federal lobbying, political action committee (PAC) spending, 527 political committees, state candidates, political parties, and ballot measure committees, scaled by total assets. 8 We obtain this data from the Center for Responsive Politics website. By scaling this measure, we ensure that we are capturing the level of political spending relative to a firm s own size, rather than political spending itself. However, to the extent that activists target those firms 8 Federal lobbying relates to money spent on in-house lobbyists or professional lobby firms in order to influence an elected official. PAC spending related to money spend on election campaign financing. 527 political committees are political groups that advocate for or against a federal candidate through spending on advertising and voter education. In addition to these federal expenditures, firms also expend money on state-level politics including contributions to candidates, political parties, and ballot measure committees. Ballot measure committees are formed to support or oppose the qualification or passage of ballot measures and allow sponsors of a particular initiative to bypass the state legislature and take the initiative directly to the electorate. Political Spending represents a comprehensive measure of a firm s observable political spending footprint. 12

15 that spend the most, our measure may underestimate the overall effect. Also, this variable does not allow us to clearly identify activist motives because political spending could be capturing motives related to agency conflicts or the visibility of political spending for the firm. We also examine firm political ideology to capture the extent to which the political views of the firm may differ from the activist. We measure the firm s political ideology by examining firm-level PAC spending. Political Ideology is defined as the percentage of total PAC spending by the firm for the most recent election cycle ending before year t that is directed to Republican candidates. A significant positive coefficient for 2 would be more consistent with targeting decisions being based on special interests, namely that the activist s motive in targeting a firm relates to differences in political ideology. To capture the degree of potential agency conflict, we focus on measures that we believe best capture the ease with which managers can exercise their own preferences as opposed to those of outside shareholders. We define CEO duality (CEO Duality) as an indicator variable that equals 1 if the firm combines the CEO and Chairman of the Board position, and 0 otherwise. We include CEO Duality as prior research has found that firms that combine CEO and Chairman of the Board positions tend to perform more negatively than firms that separate the two roles (Bebchuk et al. 2009). We also include Bebchuk et al. s (2009) entrenchment index (Entrenchment Index), which is based on six corporate governance provisions: staggered boards, limits to shareholder bylaw amendments, poison pills, golden parachutes, and supermajority requirements for mergers and charter amendments. An indicator variable is created for the existence of each provision, and Entrenchment Index equals the sum of the six variables; the variable thus ranges from zero to six. Bebchuk et al. (2009) find that increases in this index are monotonically associated with economically significant reductions in firm valuation. Positive 13

16 coefficients for 3 and 4 would suggest that activists target firms with potential agency conflicts, and would be consistent with target decisions being made for monitoring purposes. The control variables capture other characteristics found to be associated with the likelihood of being targeted by shareholder proposals. We control for firm size, growth, ownership characteristics, and performance as prior research documents that firms attracting activist efforts around financial issues tend to be large, poorly performing firms, with fewer growth options and low inside and institutional ownership (Karpoff et al. 1996; Thomas and Cotter 2007; Ertimur et al. 2011). Judge et al. (2010) note that motives may differ across the proposal type; they find that prior profitability is lower for firms targeted with financial proposals (including those related to performance, mergers and acquisitions and dividend payout) but higher for firms targeted with social proposals (including those related to corporate social responsibility and environmental issues). Because political spending has both financial and social elements, we make no prediction for the direction of the profitability variable. The final set of controls includes size (Market Capitalization), growth (Book to Market), ownership (% of Inside Ownership, % of Block Ownership), and performance (Return on Assets, Abnormal Returns). All variables are described in Appendix 2. Descriptive statistics are reported in Table 1. Univariate differences reveal targeted firms have higher political spending than non-targeted firms in total and generally across all spending categories. They also have more conservative political ideologies. Firms targeted by pension funds have lower levels of managerial entrenchment as measured by Entrenchment Index but similar levels of entrenchment as measured by CEO Duality. In contrast, firms targeted by other activist investors also have lower levels of entrenchment as measured by Entrenchment Index but significantly higher levels of entrenchment as measured by CEO Duality. For both groups, 14

17 targeted firms are significantly larger than non-targeted firms, as measured by Market Capitalization, and have lower levels of inside and institutional ownership (% of Inside Ownership, % of Block Ownership). However, differences in growth and performance are only evident in the comparisons between firms targeted by pension funds and non-targeted firms. Target Analysis - Results Results from our target analysis are reported in Table 2; Panel A includes the full sample, Panel B considers pension funds and other activists separately relative to non-targeted firms, and in Panel C we compare pension funds to other activists. As reported in Panel A, we find that, overall, activists target firms with higher levels of observable political spending relative to non-targeted firms. In Model (1), the coefficient for our summary measure Political Spending is positive and significant at the 1% level. The coefficients for our agency problem proxies, CEO Duality and Entrenchment Index, and ideology measure, Political Ideology, are not significantly different from zero. As for the control variables, the estimated coefficients for Market Capitalization and Book to Market are both statistically significant in the predicted direction, using a 1% level of statistical significance, suggesting that activists target large firms, with fewer growth options. We also note that the coefficients for Return on Assets and Abnormal Returns are negative and significant at the 1% level. That activists target poorly performing firms suggests that the issue of political spending, although typically classified as a social issue, aligns with financial proposals in terms of target firm profile. The estimated coefficient on % of Block Ownership is negatively significant, at the 5% level, consistent with expectations that ownership characteristics influence target decisions. In Model (2) we break the observable political spending measure into its four components: Lobbying Spending, PAC Spending, 527 Spending and State Spending, and find that the target 15

18 decision is significantly positively associated with PAC Spending and 527 Spending at the 1% level but not significant for the other two measures. 9 In Model (3), we split Political Spending into predicted and residual components, where Predicted Political Spending aims to capture the level of expected spending given its economic determinants, while Residual Political Spending (the difference between Political Spending and Predicted Political Spending) aims to capture the excessive portion of political spending. 10 The results show that the coefficients of both Predicted Political Spending and Residual Political Spending are significantly positive, at the 10% and 5% level, respectively. As union and public pension funds ( pension funds ), in particular, have been accused of pursuing private interests through the shareholder activism campaign, we also consider proposals initiated by these funds separately from those initiated by other activists to identify any differences in targeting criteria. In the first part of Panel B of Table 2, we focus on pension funds. Here we observe that target firms have higher political spending than non-target firms - Political Spending is significant at the 5% level. Model (3) provides some evidence that pension funds follow a sophisticated approach when targeting firms on the basis of political spending as only the coefficient on Residual Political Spending is significantly positive. We also report consistent evidence that, relative to non-targeted firms, firms targeted by pension funds have more conservative political ideologies; the coefficient for Political Ideology is significantly positive at the 10% level in all three models. The evidence on the agency problem variables is 9 In fact, we note that the coefficient on Lobbying Spending is not significant in any of the analyses we employ in the paper. This evidence, in combination with the findings in Table 3 that a relatively small percentage of proposals relate to lobbying, suggests that campaign spending is the primary focus of activists and voting shareholders. 10 Similar to Chaney et al. (2011), we model political spending as a function of the following economic determinants, Book to Market, Headquartered in Washington, Leverage, Market Capitalization, and Operating Cycle, and assign the predicted values from this pooled firm-year OLS regression to Predicted Political Spending. Residual Political Spending is defined as the difference between Political Spending and Predicted Political Spending. Since we require additional data for these measures, the sample sizes for analyses employing these measures are somewhat reduced. Note that because we focus on S&P 500 firms, all large-cap firms, our model likely understates the effects of size in explaining political spending. 16

19 mixed; Model (3) suggests pension funds target firm with less evidence of agency problems but we find no differences in the levels of agency problems in the other two models. In the second part of Panel B, where we focus on other activists, a somewhat different pattern emerges. Although we again observe higher political spending, we find no differences in political ideologies of targeted and non-targeted firms. We also note that targeted firms are characterized by more agency problems; CEO Duality is significantly positive in all three models at the 5% level of significance. Further, when we directly compare firms targeted by pension funds to those targeted by other activists in Panel C, we find evidence that pension funds target firms have less evidence of agency problems. Both CEO Duality and Entrenchment Index are significantly negative in all three models. In Model (2) and (3), we find that Political Ideology is significantly positive at the 10% level, which supports the notion that pension funds target firms have more conservative ideologies. (Activists do not appear to differ in their evaluation of observable political spending, as all of the coefficients on Political Spending are not significantly different from zero.) Collectively, this evidence is consistent with pension funds (other activists ) target decisions being motivated, in part, by special interests (monitoring incentives). III. CHARACTERISTICS OF PROPOSALS AND DETERMINANTS OF VOTING OUTCOMES Proposals Issues, Activist Types and Related Voting As noted previously, 322 of the 541 proposals went to a vote at the annual general meeting. For the 322 that went to vote, we read the proposals from the proxy statements and classify them into four non-mutually exclusive categories: spending information, spending philosophy, spending voice, and other. Spending information proposals call for enhanced disclosure of political spending information, such as campaign contributions, lobbying expenditures, or 17

20 policies and procedures related to these forms of political spending. Spending philosophy proposals aim at affecting the objective of political spending by asking firms to justify their political position (by providing a business rationale for each category of spending) or to affirm their political non-partisanship. Spending voice proposals aim at giving shareholders a say-onspending, such as proposals that seek shareholder approval and advisory votes on spending, or proposals that ask the firm to stop political spending altogether. Proposals in the other category include those that focus on a specific issue, such as global warming-related lobbying, and those that ask for a comprehensive review of all political spending initiatives. We provide examples of proposals in each category in Appendix 1. We also categorize the proposals into six groups based on their proponent: Union Pension Funds, Public Pension Funds, Individuals, Faith-based Funds, Institutional Investors (investment advisors, investment management firms, and mutual funds), and Other Shareholder Groups. Table 3, Panel A presents the frequency of and voting support for political spendingrelated shareholder proposals by proposal type and activist identity. There is an increase in the number of political spending-related proposals and the voting support for those proposals over time, particularly since the landmark Citizens United v. Federal Election Commission ruling in There are 161 proposals in the period (with 18% votes in favor) compared to 161 proposals in the period (with 24% votes in favor), although only four proposals within our sample received a majority vote (untabulated). Proposals requesting spending information are common and enjoy the highest voting support, especially those that ask for disclosure of campaign financing (24% votes in favor during the sample period). Proposals aimed at affecting spending philosophy are less frequent and receive lower voting support in the period (12% votes in favor in period vs. 5% in ). In contrast, 18

21 spending voice proposals exclusively appear in the period (7% in favour), suggesting shareholder activists are seeking an expanded role in the determination of political spendingrelated decisions over time. While the overall average level of support is only 20% during our sample period, support is clearly increasing, and Rummell (2013) argues that 2013 may have signalled a tipping point in support with several proposals garnering a majority vote. 11 Panel A of Table 3 also shows a marked increase in activism by public pension funds and institutional investors, who filed 30% and 28%, respectively, of proposals in the period compared to 17% and 15% in the earlier time period. In contrast, activism by union pension funds and individuals has decreased over time, as these activists submitted only 19% and 16%, respectively, of proposals in the period, compared to 32% and 21% in the earlier time period. The voting support is relatively stable over time for most activists except union pension funds, for which voting support has almost doubled from 14% to 27% over the sample period, and faith-based funds for which voting support increased from 24% to 32%. Proposals filed by individuals receive the lowest support only 10% in favour. Panel A of Table 3 suggest that shareholders are more supportive of proposals that ask for disclosure rather than proposals that aim to affect either the objective or nature of political spending. It further suggests that political spending-related activism has changed over time, with public pension funds and institutional investors accounting for an increasing proportion of activism over time, at the cost of union pension funds and individuals. While concerns have been expressed that political spending related activism has been used to promote the interests of special interests, shareholders appear to have used their voting power to selectively support proposals promoting information disclosure rather than changes in firms spending decisions. 11 Rummel also cites the first lawsuit aimed at compelling companies to disclose more. During 2013, NY State Common Retirement Fund sued Qualcomm when discussions over disclosure of political spending fell apart, with Qualcomm settling and agreeing to adopt a new policy. 19

22 Panel B of Table 3 displays a breakdown of proposal types by the identity of the shareholder activist. Almost all proposals filed by union pension funds and public pension funds ask for enhanced disclosure of political spending information and union pension funds appear further interested in shaping the objective of political spending by requiring firms to justify their spending. Institutional investors file proposals asking for greater information on political spending but also file proposals aimed at increasing shareholder involvement in political spending decisions. Not surprising, individuals have a diverse set of interests and submit proposals across all four categories. Arguments made by Activists and Management In addition to classifying the 322 political spending-related proposals by proposal and activist type, we also capture the arguments being made by activists in support of their proposals and those being made by management in opposition to the proposals. We read the proposals from the proxy statements and classify activist supporting arguments into three key non-mutually exclusive categories: reputational risks, business risks, agency conflicts; all other arguments are accumulated in a catchall category other. An example of an activist citing reputational risks to support its position is included in the 2012 proxy statement of Target, where Green Center Equity Fund notes the following: Corporate political contributions can backfire on a corporation's reputation and bottom line. In 2010, Target Corporation experienced such risks directly after it received unwanted attention, consumer boycotts, and protests for its support of a controversial candidate. In a Harris Poll released in October 2010, a sizable portion (46)% of respondents indicated that if there were option, they would shop elsewhere if they learned that a business they patronized had contributed to a candidate or a cause that they oppose. Activists also cite general business risks as rational for requesting shareholder to vote in favor of proposals. For example, in the 2005 Abbott Laboratories proxy, Mercy Investment Program argues, 20

23 Our company should be using its resources to win in the marketplace through superior products and services to its customers not because it has superior access to political leaders. Political power can change, leaving companies relying on this strategy vulnerable. Finally, activists often urge shareholders to support their proposal because of the potential agency conflicts inherent in political spending decisions. For example, in the 2006 Chevron proxy, the New York City Pension Fund notes that public disclosures of political spending are incomplete. Absent a system of accountability, corporate executives will be free to use the Company s assets for political objectives that are not shared by and may be inimical to the interests of the Company and its shareholders. We also capture arguments made by management in urging shareholder to vote against political spending-related proposals. 12 We read the proposals from the proxy statements and classify management opposing arguments into four key non-mutually exclusive categories: not relevant, sufficient disclosure, compliance costs and proprietary costs. Some firms note that political spending-related proposals are not applicable because the firm either has little or no political spending. For example, in recommending that shareholders vote against a 2004 proposal requesting disclosure of campaign financing, Berkshire Hathaway argued Berkshire Hathaway Inc. made no political contributions during 2003 and did not either recommend or require that any of its 80% or more owned consolidated subsidiaries make any political contributions the Board of Directors believes the proposal serves no useful purpose. Many firms also argue that campaign financing and lobbying laws, as well as voluntary disclosures, already provide shareholders with a sufficient level of disclosure, rendering the proposals unnecessary. As a specific example, Exxon Mobil management advised shareholders to vote against its 2011 proposal because, 12 Prior work suggests management employs a number of other methods to influence voting outcomes, including bundling proposals (Bebchuk and Kamar 2010), opportunistically classifying and timing proposals (Bethel and Gillan 2002; Listokin 2010), soliciting votes from retail investors (Wall Street Journal 2013), and issuing recommendations (Ferri and Oesch 2013). 21

24 The Board believes that because the reporting of corporate political activity is already heavily regulated and the Company already reports political contributions and lobbying expenses, this proposal is unnecessary and would be redundant with information that already is available to the public. The Company discloses political contributions and lobbying expenses in multiple public forums, i.e., our Web site at exxonmobil.com/politicalcontributions, the Company s Corporate Citizenship Report, and Web sites for federal and state regulatory agencies and the U.S. Congress. In firms for which the proposal is potentially relevant and disclosure is lacking, management cites compliance and proprietary costs as a disincentive for adoption. In response to a 2011 proposal asking the firm to detail its lobbying priorities, Walmart management offered the following rationale for recommending a vote against the proposal, Most, if not all, of Walmart s competitors in the retail industry do not issue a report similar to the one the proposal requests. If Walmart were to issue such a report, any strategic advantage, as well as our ability to pursue our public policy initiatives effectively and without counterproductive or untimely publicity regarding our company s efforts, could be negatively affected. Producing a report on such matters would be a complex and expensive undertaking, an expense that, in the opinion of the Board, would far outweigh any benefit to our shareholders. Panels C and D of Table 3 present a breakdown of activist supporting statements and management opposing statements, respectively, by proposal type and activist identity. Panel C demonstrates that activists most often cite agency conflicts (59% of all proposals) and business risks (52% of all proposals) as a rationale for supporting the proposal. Agency conflicts are most often cited for spending information proposals (67%) and for proposals supported by Union Pension Funds (84%). In contrast, Public Pension Funds (83%) and Institutional Investors appear more concerned about business risks (59%). Panel D indicates that firm management most often cite sufficient disclosure (83% of all proposals) as a rationale for opposing the proposal, particularly for spending information proposals (86%). Compliance costs associated with adopting the proposal (60%) is also a popular argument among management, especially for proposals submitted by union pension funds (60%) and public pension funds (71%). Overall, Table 3 Panels C and D document that both activists 22