Changes in Firms’ Political Investment Opportunities, Managerial Accountability, and Reputational Risk

Abstract

We use the U.S. Supreme Court’s decision in Citizens United v. Federal Election Commission to assess the reputational risks created by political investment opportunities that allow managers to spend unlimited and potentially undisclosed firm resources on independent political expenditures. This new opportunity raises important ethical questions, as it is difficult, and perhaps impossible, under current law for shareholders to hold managers accountable for this investment choice and the reputational risks it entails. Using firms’ known political activity as a proxy for managers’ likely future use of independent political expenditures, we examine how market participants reacted to Citizens United, conditional on this prior activity and corporate governance attributes related to the concentration of decision rights in senior management and blockholders. The results of our analyses document that firms with both a high level of known political activity and CEO-chairperson of the board duality experienced negative abnormal returns in reaction to Citizens United. In contrast, firms with concentrated ownership experienced positive abnormal returns; however, as known political activity increased, investors discounted the benefits of concentrated ownership. These findings suggest that investors expect this expansion of firms’ political investment opportunities to amplify principal-agent problems inherent in corporate political activity. Additionally, our findings provide evidence for those deliberating the mandatory disclosure of firms’ investments in politics as a means of increasing managerial accountability to both shareholders and the public.

Introduction

A growing body of theoretical and empirical literature suggests that a firm’s engagement in corporate political activity (CPA) can negatively affect its reputation (den Hond et al. 2014; Torres-Spelliscy 2016; Minfee et al. 2018). As den Hond et al. (2014) argue in their conceptual study of how both CPA and corporate social responsibility affect firm reputation, there are two channels via which CPA can have this negative effect. First, broader society can view firms’ use of CPA as a purely self-interested pursuit rather than as a legitimate attempt to shape public policy, a pattern of behavior that raises clear and substantial ethical questions about corporate power in democracies (Néron 2016). Second, shareholders of firms can have ethical qualms regarding managers’ use of those forms of CPA that are not legally required to be disclosed, as these investments are subject to classic agency problems stemming not only from managers’ use of tangible firm resources in politics but also from the risks created for the intangible asset that is the firm’s reputation (Werner 2017). Under these circumstances, one of the few ways in which shareholders can hold managers to account for the opportunity to engage in such forms of covert CPA is to exercise their rights to “exit” the firm by selling their shares.

We empirically explore the latter of den Hond et al.’s (2014) two potential channels of reputational risk in CPA by analyzing the financial market’s reaction to the 2010 U.S. Supreme Court ruling in Citizens United v. Federal Election Commission (FEC), as the case expanded managers’ political investment opportunities without creating a disclosure mechanism that would allow shareholders to hold managers accountable for their spending and the reputational risks it entails. Managers are responsible for determining the nature of their firms’ political activities and the amount of resources invested in them. Building on prior research (Semin and Manstead 1983; Tertlock 1992), we define managerial accountability as the implicit or explicit expectation that managers will be asked to justify their investment decisions to shareholders. Historically managers’ investments in electoral activity such as direct contributions to candidates have been restricted or significantly regulated, affording shareholders and the broader public the opportunity to hold managers accountable for these investments, as well as limiting their downside reputational risk. However, Citizens United significantly expanded managers’ political investment opportunities at all levels of government in the U.S. by granting managers the right to allocate firms’ resources directly to independent political expenditures (IPEs) that are used to influence elections by promoting the election or defeat of specific candidates.

For two reasons that we detail further in the next section, managers’ investments in IPEs post-Citizens United are distinct from other forms of current and historic CPA. First, there are no legal limits on the amount of firm resources that managers can invest in IPEs. Second, and more importantly, there is no mandatory disclosure of the objective or the amount spent on IPEs when a firm contributes to a non-profit organization that then engages in electoral activity. Consistent with the predictions of den Hond et al. (2014), we argue that these two factors increase CPA-related reputation risks, as the IPEs allowed by Citizens United have led to questions among the broader public (Stohr 2015) and activists (Salant 2013) regarding the ethicality of this form of CPA, as well as among investors regarding not whether or not it creates specific ethical considerations for them due to agency problems (Croce 2018).

Employing a similar methodological approach to Karpoff (2012), Karpoff and Lott (1993), and Sampath et al. (2018), we conduct an event study of the potential changes in electoral activity ushered in by Citizens United to examine whether or not these differences resulted in negative shareholder reactions in expectation that the firms most likely to engage in new, less accountable forms of CPA would face greater reputational risks. In doing so, we are also able to test the investment theory of CPA versus the agency theory of CPA by examining whether the concentration of decision rights within the firm—as proxied by CEO duality (i.e., the CEO holding that position while simultaneously serving as chairperson of the board) and blockholdings—affected shareholder reactions to the decision. We posit that concentration of decision rights in senior management or a subset of powerful owners (Deloitte 2011) will increase agency costs and thus perceived reputational risks by decreasing the accountability of key decision makers to the average shareholder. As prior literature suggests that firm governance can mitigate agency costs (Gompers et al. 2003) and enhance managerial accountability more broadly (Brennan and Solomon 2008), we focus on whether these two governance attributes affect the market’s reaction to Citizens United.

Ultimately, we document that market participants anticipated that firms’ Citizens United-related reputational risks and thus, shareholders’ future ability to hold management accountable for these new forms of CPA would be highly contingent on firms’ existing governance structures, as well as their preexisting investments in CPA. Specifically, consistent with extant research (Werner 2011; Burns and Jindra 2014), our results indicate no significant market reaction, on average, to Citizens United. However, we find that firms with high levels of known political activity and CEO duality experienced consistent negative abnormal returns around the four Citizens United events we examine. Cumulatively, the results indicate that politically active firms with CEO duality, on average, experienced a − 0.2% to − 0.9% decline in share value as a result of Citizens United. Further, although the market reacted more positively to Citizens United as firms’ blockholdings increased, these positive reactions were discounted as known political activity also increased. Taken together, these findings suggest that market participants anticipated that previously politically active firms with concentrated decision rights in the form of CEO duality and blockholdings posed greater agency and thus, reputational risks for the average shareholder.

Our paper makes several contributions. First, our study adds to the literature examining the relationships between managers’ investment decisions, agency costs, and accountability. For example, McNichols and Stubben (2008) and Shroff et al. (2014) suggest that information asymmetries inherent in managers’ investment decisions can contribute to greater agency costs and lessen shareholders’ ability to hold managers accountable, and Jensen (1993) and Shivdasani and Yermack (1999), as well as others, discuss how governance mitigates these agency costs. Within the realm of CPA, Hadani (2012), Coates (2012), Sobel and Graefe-Anderson (2014), and Hadani et al. (2015) all provide associational evidence regarding potential agency problems and firm engagement in and returns to investment in CPA, as conceptualized using tactics legally available to firms pre-Citizens United. We contribute to this literature not only by utilizing a research design that lends itself to causal identification but also by providing evidence that in the specific context of IPEs, governance structures moderate—in more subtle ways that previously documented—perceived risks created by CPA.

Second, our research also contributes to the academic and policy debates over the mandatory disclosure of firms’ CPA as a mechanism for enhancing managerial accountability and decreasing reputational risk (see, e.g., Bebchuk and Jackson 2010; Stoll 2015). Although Shleifer and Vishny (1997) and Lambert (2001) suggest that enhanced transparency via disclosure results in improvements in managerial decision making and reduction of managerial consumption of firm resources, others, including Durnev and Mangen (2009) and Zingales (2009), argue that mandatory disclosure of an activity can affect the decision to engage in the activity, potentially to the detriment of shareholders. This debate occurs in the realm of CPA as well: one side advocates for the disclosure of CPA so that investors are fully informed of the decisions made by management in this realm (Bebchuk and Jackson 2010), as well as the brand and reputational risks these investments create (Torres-Spelliscy 2016); whereas the other side, including prominent associations such as the U.S. Chamber of Commerce, argues that requiring firms to disclose their IPE investments leaves firms open to harassment, boycotts, and other reputational threats that can lead to inefficient changes in product market competition (Salant 2013). By examining firm-specific effects around the events in Citizens United, we contribute by evaluating, whether investors, at the time, viewed the decision as increasing agency and reputational risk due to the lack of mandatory disclosure. Thus, the findings of our study provide evidence to inform those policymakers, scholars, and shareholder activists deliberating the mandatory disclosure of firms’ political spending as a mechanism to hold managers accountable for their firms’ CPA.

Third, although, as is suggested by the first two contributions, we focus on testing whether or not CPA affects firm reputation due to the ethical questions raised by the lack of mandatory CPA disclosure and thus, managerial accountability, our findings also speak to the broader ethical debate regarding corporate political power in society. If CPA is indeed largely oriented toward self-interested rent seeking (Néron 2016), then our findings suggest that these rents may accrue not to politically active corporations per se (and thus, their shareholders broadly) but instead to the even narrower set of actors within firms who hold the greatest decision-making rights. This conclusion can help inform and tailor public policies aiming to regulate the role of corporate money in politics.

The remainder of the paper begins by providing background on the regulation of firm engagement in electoral activities and how Citizens United changed these regulations. Based upon these changes in institutional arrangements, we motivate three hypotheses, two of which focus on the concentration of decision-rights within the firm and the interaction of that concentration with known political activity. We then introduce the empirical context—i.e., the history of the Citizens United case—that we use to test our hypotheses. Next, we present the results from our cross-sectional analyses of investor reactions to the events in the case, as well as those from various sensitivity analyses we conduct. The paper concludes by discussing the study’s limitations, implications—focusing on the specific accountability and reputational issues that investments in IPEs raise—and avenues for future research.

Institutional Background and Theory

The regulation of firm involvement in electoral politics at the national level in the U.S. dates to 1907. In order to contextualize the ways in which Citizens United altered firms’ political investment opportunities, we provide a brief background on these institutional rules in this section. After documenting some high-level information on other means by which firms are (or were) able to engage in electoral politics, we specifically discuss the regulation of IPEs pre- and post- the events in Citizens United. This discussion focuses solely on electoral activity and does not address lobbying, as it is subject to a separate and less complicated set of regulations.

To begin, direct contributions funded by firms’ resources to federal candidates or political parties to support their political campaigns are prohibited. Direct contributions from firms to candidates have been banned since the Tillman Act of 1907, and although direct, unlimited contributions from firms to political parties (also known as ‘soft money’) were allowed beginning in 1977 due to a FEC advisory opinion, they became unlawful in 2002 as a result of the Bipartisan Campaign Finance Reform Act of 2002 (BCRA, also known as McCain-Feingold). When these ‘soft money’ contributions were allowed, firms were required to disclose them on a quarterly basis to the FEC, and there was significant legal ambiguity as to whether the political parties could use these funds to engage in express advocacy—that is, advocate for the election or defeat of a clearly identified candidate by using such words as “vote for,” “defeat,” or “support” in broadcast advertisements, promotional materials, billboards, internet sites, etc. (Werner 2017). Neither of these prohibitions were challenged in Citizens United, and thus, they went unaffected by the case.

At the time of the events in Citizens United, the primary vehicle through which firms engaged in electoral activity was the corporate political action committee (PAC). Corporate PACs are separate and segregated funds from the corporation, and firms form them to make contributions to political candidates, political parties, and other PACs (e.g., trade association PACs). The Federal Election Campaign Act (FECA) of 1971, its 1974 (post-Watergate) Amendments, and BCRA govern PACs and set their contribution limits. Firms can set up a PAC and pay its administrative costs, but firms are not allowed to contribute resources to the PAC that are later contributed to candidates. Rather, firm-linked PACs can raise funds only from a restricted class of individuals connected to the firm such as executives, managerial staff, and their families, provided they are U.S. citizens or permanent residents. Foreign firms may establish PACs, provided these PACs fundraise only from the members of the firms’ restricted class that are U.S. citizens or permanent residents. Otherwise, as foreign firms are considered foreign nationals, they are strictly prohibited from engaging in campaign finance.

The funds raised by firms’ PACs are often referred to as ‘hard money.’ PACs can contribute up to $5000 per candidate per election and face no aggregate limit on their giving. The names and occupations of all individuals who contribute more than $200 in a year to a PAC (along with their total contribution amount, which is subject to an annual cap of $5000) must be disclosed to the FEC, and PACs must also report all contributions they make. The first row of Table 1 summarizes these regulations, and as its last column notes, none of these rules changed as a result of Citizens United, as they were not at issue in the case.

Table 1 Changes in the legal regulation of firms’ electoral engagement around Citizens United v. FEC (2010)

A second form of electoral activity that was legal prior to and went unchanged by Citizens United is issue advocacy. Operating either directly or through a 527 organization (named after the section of the tax code that governs it), firms can spend unlimited funds from corporate resources to advocate for a public policy position, provided they do not expressly advocate for/against a candidate using the language identified earlier. Were firms to engage in issue advocacy directly, they would have to report these expenditures quarterly to the FEC and identify themselves as the sponsor of any issue advertisements; were firms to use a 527 to engage in issue advocacy, the 527 would be responsible for reporting the firm’s contribution to the Internal Revenue Service (IRS), which would then make these data publicly available.

As a result of the disclosure requirements for PACs and issue advocacy, as Table 1 highlights, market participants were at the time of Citizens United, and are still today, able to determine the firms that form PACs and engage in issue advocacy. In addition, market participants can track to whom the firm’s PAC is contributing and on what political issues it is advocating using its own resources. As disclosure is a necessary condition for accountability, the regulations on PACs and issue advocacy at a minimum put shareholders in a position to monitor the behavior of those with political investment decision rights in the firm.

Lost in the debate and controversy surrounding Citizens United is the fact that IPEs—which can be used for express advocacy, subject to the constraint that the actor making the IPEs does not coordinate with the candidate or party on whose behalf they are advocating—were legal and occurring prior to the decision. That being said, as we discuss below and is detailed in Table 1’s final section, the case and its implementation by the FEC, lower federal courts, and state governments fundamentally altered virtually all aspects of IPE regulation.

Prior to Citizens United, firms could make IPEs but only by using their corporate PACs (i.e., with so-called ‘hard money’ that originates from individuals in the firm’s restricted class; not from the firm’s resources). Firms’ PACs could pay for these IPEs directly, or they could make a PAC contribution to the PAC of a 501(c)4 social welfare organization or the PAC of a 501(c)6 trade association. With the exception of 501(c)3 charities that are strictly prohibited from engaging in electoral politics, 501(c)s are non-profit corporations that are allowed to engage in some electoral activities as long as political activities are not their “primary purpose.” Examples of 501(c)(4) groups include Crossroads GPS run by Republican strategist Karl Rove and Americans for Prosperity affiliated with David and Charles Koch. Examples of 501(c)(6) groups include the U.S. Chamber of Commerce, the National Restaurant Association, or the Edison Electric Institute. If firms chose to invest in IPEs directly pre-Citizens United, these investments would have been disclosed by their own PAC and would have been limited by the amount in the firm’s PAC’s coffers. If firms chose to invest in IPEs indirectly pre-Citizens United, these investments would have been capped at the $5000 PAC-to-PAC contribution limit and disclosed both by the firm’s PAC and the receiving non-profit corporation’s PAC. In either case, shareholders had the ability to monitor any funds flowing from the corporation for the purposes of IPEs pre-Citizens United. More generally, as the fifth column of Table 1 makes clear, all direct and indirect firm investment in electoral activity (as opposed to lobbying via non-profit organizations) was publicly disclosed prior to the case (see, Rosenberg 2010).

Post-Citizens United, the regulation of both existing forms of IPE investment changed, and a new investment category was created. As Werner (2017) discusses, the two significant changes to IPEs ushered in by Citizens United were (1) firms’ ability to use their own resources, without legal limit, to engage in express advocacy (potentially making IPEs far more potent than the old ‘soft money,’ which suffered from legal ambiguities about its uses, and freeing firms from the constraint of raising large amounts of ‘hard money’ via their PACs) and (2) firms’ ability to evade disclosure to their shareholders and the broader public, provided they engage in IPEs via a 501(c)4 or 501(c)6 and do not specifically earmark their contribution to the 501(c) organization as being for IPE/electioneering purposes. This tactic is commonly referred to as ‘dark money,’ since neither the contributing nor the receiving entity has to disclose it.¹ The first two rows of the last section of Table 1 highlight these changes due to Citizens United; the final row of the section and table introduces ‘Super PACs’ or IPE-only committees, which like 501(c)s can receive unlimited contributions for the purposes of making IPEs, but which unlike 501(c)s have to disclose their contributors, as Super PACs exist solely for electioneering purposes and can be tied to specific candidates or parties even though they are technically independent of those actors.

The above changes, particularly with regard to IPEs made via 501(c) organizations, are the shifts in the legal environment that we posit shareholders will respond to around the events in the case, as we argue that shareholders’ reactions capture whether they anticipate those with investment decision rights within the firm will exploit the newly available political investment opportunities and whether those actions will exacerbate CPA-related reputational risk. The next section articulates these expectations more formally.

Hypotheses

As mentioned in the introduction, following Karpoff (2012), Karpoff and Lott (1993), and Sampath et al. (2018), we employ a financial market event study to assess investor reactions to the events in Citizens United, which we view as a proxy for investors’ beliefs regarding the new reputational risks created by the decision and its implications for managerial accountability with regard to firm investment in these new forms of CPA. In particular, our approach is similar to Karpoff’s (2012), who views investors as key stakeholders to observe when exploring whether a firm has suffered a reputational loss in the wake of an adverse event. In the context of an event study examining firms’ dissemination of misleading financial reports, Karpoff (2012) argues that reputational loss occurs because investors expect, after the revelation of misconduct, firms to incur greater operating costs associated with implementing effective monitoring and control policies, face higher cost of capital due to elevated operating and information risk, and have altered cash flows from operations when other stakeholders (e.g., customers) change their engagement with the firm. As we discuss further below in motivating the agency theory of CPA, we believe that both the first and third argument made by Karpoff can inform investors’ anticipatory reactions to the events in Citizens United.

Our theorizing begins with the well-documented finding that a firm’s future CPA is positively related to its prior CPA (Hart 2001; Hadani and Schuler 2013). In other words, once a manager elects to invest in CPA, he or she stays committed to continuing investing in CPA. The first explanation for this long-term commitment to invest considers managers’ political spending to be a non-product market, strategic investment undertaken by the firm in its political and regulatory environment for the purpose of increasing firm value (Baron 1999; Bonardi et al. 2005). Under the strategic investment perspective, it is assumed that managers will engage in CPA when doing so leads to a more favorable business environment, new revenue opportunities, and/or lower costs (Buchanan and Tullock 1962; Hillman et al. 2004). As evidence for this investment theory of CPA, in the pre-Citizens United period, Cooper et al. (2010) find that the number of Congressional candidates that a firm’s PAC contributes to is positively correlated with its financial market performance, and Richter et al. (2009) find that as firms increase their lobbying, their effective tax rates decrease.

This perspective predicts that post-Citizens United managers will commit firm resources to IPEs until the marginal benefit derived from this investment strategy equals the marginal cost, thereby increasing firm value. Werner (2017) shows that this was indeed the case for a small number of firms that were accidentally disclosed as having made contributions for IPE purposes to the Republican Governors Association in 2014. However, he cautions that his findings might not extend to IPEs done through less reputable, non-political party organizations, and he also finds that the benefits of even party-linked IPEs decrease if firms’ shareholders have previously expressed their displeasure with the tactic via a proxy resolution, suggesting that agency and reputational risks color shareholders’ reactions to this form of CPA. Nevertheless, this perspective would predict a positive market reaction, on average, to Citizens United for those firms with known political activity, as shareholders anticipate that this set of firms will benefit more significantly from the case’s relaxation of constraints on CPA.

The second explanation for managers’ commitment to invest in CPA is that CPA is a mechanism whereby managers use firm resources for self-serving purposes (Bebchuk and Jackson 2010; den Hond et al. 2014). Recent empirical research argues that firms’ CPA is not value enhancing for shareholders but rather a manifestation of the principal-agent problem in which CPA is a vehicle for managers to extract rents from shareholders (see, e.g., Aggarwal et al. 2012; Hadani and Schuler 2013; Werner and Coleman 2015). For example, Coates (2012) finds that CPA (as measured by engaging in lobbying or having a PAC) positively correlates with other indicators of agency problems, such as executive jet use, and Sobel and Graefe-Anderson (2014) find positive correlations between CPA and executive compensation but no relationship between CPA and firm performance. Post-Citizens United, the lack of mandatory disclosure of the objective of and total resources spent on IPEs could heighten such fears among shareholders, as non-disclosure more generally limits managerial accountability, which in turn leads to a heightened risk of moral hazard, greater agency costs for shareholders, and a reduction of firm value (see, e.g., Healy and Palepu 2001; Jensen and Meckling 1976).

As discussed in the introduction, within the specific context of IPEs, managers are potentially risking not only firm tangible resources in the form of cash to engage in CPA for self-serving purposes but also strategic intangible assets, including their firms’ brands and broader reputations (Torres-Spelliscy 2016). These assets are potentially in danger when firms engage in IPEs, as the broader public’s dislike for Citizens United has not waned with time and as activists have been quick to protest and even boycott firms when their use of IPEs are disclosed. The use of IPEs has many of the characteristics of discreditable action likely to damage the organization’s reputation (Reuber and Fischer 2010). That is, and in-line with Karpoff’s (2012) arguments above, a firm’s investment in IPEs is under the firm’s control, poses a threat to the interests of the actors monitoring the firm, and is deviant in the eyes of society (especially if run through a third-party that does not need to disclose its contributors). If the IPE investment is ever disclosed, it can be traced to the firm and will likely receive a great deal of media coverage, as was the case with Target Corporation’s contribution to a Republican Super PAC in the summer of 2010 (Torres-Spelliscy 2016). Similar to other legal but discreditable actions (e.g., aggressive tax strategies; see, Hardeck and Hertl 2014), it is unlikely that the legality of the firm’s investment in IPE will spare it from reputational damage if it comes to light. Further, as McDonnell and Werner (2016) document, such negative reputational shocks can introduce significant political and economic costs for firms, including decreased access to policymakers, as well as reductions in revenue from government contracts. Thus, the agency theory perspective on CPA would expect a negative market reaction, on average, to Citizens United for firms with known political activity, indicating that shareholders anticipate that their inability to hold management accountable for their potential use of IPEs could impose additional agency costs.

Operating under the assumption that shareholders view known political activity as a strong predictor of future CPA (including IPEs), these opposing theories lead us to the following null hypothesis:

H1

Financial market participants will not react differentially to the deregulation of IPEs by Citizens United based upon firms’ known political activity.

When the Supreme Court announced its decision in Citizens United, considerable heterogeneity was present in firms’ known political activity and agency risk that, when modeling an average reaction to the decision, could amplify or potentially mask shareholders’ long-run expectations for Citizens United’s effect on firm reputation and value. That is, the marginal shareholders most likely to move firms’ share prices in reaction to the specific events in Citizens United, when motivated by fears of agency problems, unaccountability, and reputational risk, could react in a more conditional manner by focusing on particular accountability mechanisms and the interaction of those mechanisms with firms’ known political activity.

We expect shareholders’ reaction to Citizens United to vary across firms conditional on two distinct governance attributes related to the concentration of investment decision rights: CEO duality and ownership concentration. We focus on these two governance attributes in our analysis because they capture firm-specific demand for accountability in decision making by the minority shareholders most likely to trade based upon the information revealed by Citizens United. In contrast, other attributes of governance, such as the presence of institutional investors or the managerial entrenchment index, may not have clear implications for accountability because of the lack of consistent incentives among institutions to monitor managers (Johnson and Greenings 1999; Hadani 2012) and the multiple, non-decision making-related dimensions of managerial power they capture (Bebchuk et al. 2009), respectively.

Beginning with CEO duality, the work of Cohen et al. (2008) suggests looking at top management’s power to determine whether shareholders anticipate that firms will earn returns on CPA. We use CEO duality as a proxy for top management power, as when the CEO is also chairperson of the board there is a significant concentration of decision rights within the firm (see, e.g., Imhoff 2003; Larcker et al. 2011). Some argue that CEO duality can lead to greater agency risk because the firm lacks the checks and balances necessary to prevent one leadership position from dominating over the other, and prior research provides evidence suggesting that CEO duality has negative economic consequences for firms (Shivdasani and Yermack 1999). Specifically, with regard to CPA, Hadani et al. (2015) document that among firms engaged in those forms of disclosed CPA that were legal pre-Citizens United, CEO duality was negatively associated with firm market value.

Following from this research, we posit that the concentration of power via CEO duality will be priced negatively if shareholders expect Citizens United to result in lower accountability of the top management team to shareholders and potentially greater agency risk via the firm’s IPEs. The latter effect could occur through wastage of the funds spent on the IPEs, through the economic consequences produced as a result of IPEs’ effect on public policy (e.g., the creation of management-friendly corporate law), or through potential damage to the firm’s brands and reputation if it is ever disclosed as making use of IPEs.

Given our arguments above in support of H1 and the effect of known political activity, we also posit that known political activity will interact with CEO duality. Assuming agency problems are at the heart of firms’ CPA and that they are exacerbated when the CEO is empowered with strong decision rights, we expect the market’s reaction to Citizens United to be more negative for firms with concentrated power in the form of CEO duality and higher known political activity.

Together, this set of arguments leads us to the following hypotheses:

H2a

CEO duality will negatively affect the market’s reaction to the deregulation of independent political expenditures as a result of Citizens United.

H2b

The interaction of CEO duality and known political activity will negatively affect the market’s reaction to the deregulation of independent political expenditures as a result of Citizens United.

The second governance attribute that we posit affects the market’s reaction to Citizens United is concentrated ownership via blockholdings. The separation of ownership and control in firms leads to information asymmetry problems between shareholders and managers (Jensen and Meckling 1976). However, when a firm’s shares are concentrated in the form of blockholdings, the blockholders have a financial interest to scrutinize firm managers and can place more demands on them via their voting rights (Jensen 1993; Shleifer and Vishny 1997). Although blockholders tend to leave management in charge of day-to-day operations, they take a more active role in the monitoring of managers’ more significant investment decisions. Moreover, blockholders typically have direct access to management, and the concentration of ownership in the hands of blockholders allows them to influence proxy voting that affects management’s decisions (Bhojraj and Sengupta 2003), as well as the public policy positions advocated for on behalf of the firm (Gourevitch and Shinn 2005).

The work of Hadani (2012) documents that the likelihood of firms engaging in CPA declines as the single largest institutional investor in a firm holds more shares. This finding suggests that a significant shareholder or set of shareholders will affect firms’ IPE either by assisting managers in allocating resources to IPE more efficiently or by restricting managers from investing in self-serving IPE. Either of these expectations leads to a positive market reaction to Citizens United for firms with concentrated ownership via blockholdings.

However, in those instances of firms with large blockholdings and greater known political activity, it could be the case that the blockholders have historically used their power and influence vis-à-vis management not to hold the firm accountable but to direct investment toward CPA that is in their own interest instead of the firm’s. Although this pattern of behavior is not a classic-principal-agent problem, it could be viewed as ethically problematic by non-blockholding shareholders, as they would find themselves incapable of holding either management or large shareholders accountable for their investment decisions. As a result, the non-blockholding shareholders who are more likely to trade around Citizens United may react negatively toward, and thus discount the benefits of blockholdings in, those firms with high levels of known political activity as the case unfolded.

These differing expectations lead to the following interrelated hypotheses:

H3a

Blockholdings will positively affect the market’s reaction to the deregulation of independent political expenditures as a result of Citizens United.

H3b

The interaction of blockholdings and known political activity will negatively affect the market’s reaction to the deregulation of independent political expenditures as a result of Citizens United.

Research Design

Citizens United Event History

Citizens United v. FEC began as a case that challenged the timing and disclosure requirements on political advertising. Citizens United, a 501(c) corporation, wanted to air advertisements for its documentary Hillary: The Movie during the 2008 Democratic presidential primaries. Citizens United used contributions from firms to finance the advertisements and the production of the documentary, and the documentary was sharply critical of Hillary Clinton. However, the advertisements for the documentary did not expressly advocate voting against her. Citizens United, fearing that the documentary’s advertisements might nonetheless be considered electioneering activity and thus subject to regulation, preemptively sought an injunction against the FEC under the belief that BCRA’s rules on non-express advocacy violated the First Amendment.

The Supreme Court heard oral arguments on this narrow question on March 24, 2009. On June 29, 2009, in a surprise move (see, Smith 2010), the Court ordered a second oral argument instead of issuing an opinion in the case. Importantly, the Court broadened the scope of the case by asking the parties involved whether or not existing restrictions on corporate IPEs in federal, state, and local elections had continuing validity. Given the surprise nature of this order and the way in which it broadened the case to include challenges to the prohibitions on corporate IPEs in federal and state laws, it is the first Citizens United event that we analyze.

Citizens United was reargued on September 9, 2009, and this is the second Citizens United event that we examine. The second oral argument is important to analyze because the content of the argument informed market participants that agency risk and accountability problems exist with regard to IPE. For example, Associate Justice Elena Kagan stated that “when corporations use other people’s money to electioneer, that is a harm… to the shareholders themselves.” Moreover, the tenor of the majority of the justices’ questions strongly suggested that the Court’s decision would lead to a change in regulation of IPEs at all levels of government.

The third event that we analyze is the issuance of the Court’s opinion that took place on January 21, 2010. The decision itself revealed that the majority of the Court supported Citizens United and, more importantly, the extensive deregulation of firms’ IPEs. The decision was well covered in the mainstream and financial press.

Following Citizens United and a subsequent ruling in Speechnow.org v. FEC by the D.C. Circuit Court of Appeals, the FEC issued a pair of advisory opinions on July 22, 2010, which created ‘Super PACs’ as an additional campaign finance investment option. As noted above, Super PACs are FEC-registered IPE-only committees, and if a firm contributes directly to a Super PAC, then this contribution will be disclosed via the Super PAC’s filings with the FEC. However, Super PACs are also allowed to receive contributions from 501(c) corporations, meaning firms can avoid disclosure of their IPE investments by tunneling contributions to a Super PAC via a 501(c) corporation. Thus, this event further enhanced firms’ ability to invest in IPEs in an undisclosed fashion by allowing them to channel their IPEs through multiple third parties. The issuance of the FEC advisory opinions is the final event we examine.

Examining financial market reactions to these four events allows us to investigate whether shareholders anticipated the changes in IPE investment created by Citizens United would have consequences for firms’ reputations and whether these reactions were moderated by investors’ ability to hold managers accountable via corporate governance. In our main analyses and our sensitivity tests, we focus on the cumulative effect of the four key events in Citizens United, as any legal ambiguities surrounding firms’ new campaign finance investment opportunities dissipated following the issuance of the FEC’s advisory opinions in July 2010 (Briffault 2012).

Sample

Panel A of Table 2 summarizes the construction of our sample. We begin with all unique firms with daily return data in CRSP during 2009 and 2010 (n = 6485). We first eliminate U.S. foreign issuers as the prohibition on their investment in U.S. campaigns beyond having a PAC went unaffected by the deregulation effects of Citizens United (n = 891). These firms are important to our study, however, as we utilize the relation between domestic and foreign issuers’ returns prior to the first Citizens United event date to estimate domestic firms’ abnormal returns (see, methodology discussed below). Next, we exclude 1816 firms operating in the financial industry, identified as firms in industry 11 using the 12-industry Fama–French classification system. We remove these firms due to a significant confounding event: On the morning that the Supreme Court announced its Citizens United ruling, President Obama publicly endorsed the ‘Volcker Rule’ that would prevent financial firms from engaging in proprietary trading. We also delete 189 firms that have share prices of less than one U.S. dollar at the beginning of our analysis period to avoid the small price effect (Blume and Stambough 1983). Firms missing data from Compustat (n = 609) or Corporate Library (n = 984), which is the source of blockholdings, or missing CEO duality data (n = 78) are also excluded from our sample. This leaves us with a sample of 1918 unique firms.

Table 2 Sample

In Panel B, we report the distribution of the 1918 sample firms across the Fama–French 12 industry categorization, as well as the distribution of all U.S. domestic issuers in CRSP. Even after our sample restrictions, the two distributions remain highly correlated (r = 0.974). Our sample’s industry distribution is also highly correlated with that of Larcker et al. (2011) (r = 0.977), which examines the effects of corporate governance regulation on firm value and also excludes financial firms using the same industry classification. These correlations suggest that our sample has an industry distribution reflective of the whole market.

Methodology

Following Karpoff (2012), we use firms’ cumulative abnormal returns, measured over the Citizens United events discussed above, as the estimate of firms’ expected reputational losses, and test our hypotheses estimating the following regression equation:

$$CAR_{i} = \alpha_{0} + \beta_{ 1} KNOW\_POL\_ACTIVITY_{i} + \beta_{ 2} CEO\_DUALITY_{i} + \beta_{ 3} CEO\_DUALITY_{i} *KNOW\_POL\_ACTIVITY_{i} + \, \beta_{ 4} BLOCKHOLDINGS_{i} + \beta_{ 5} BLOCKHOLDINGS_{i} *KNOW\_POL\_ACTIVITY_{i} + \, \theta {\mathbf{CONTROLS}} + \psi_{j} + \, e_{i}$$

(1)

CAR_i is firm i’s cumulative abnormal return. Since Citizens United can have investment consequences for every U.S. domestic issuer but has no effect on U.S. foreign issuers (as they were prohibited from investing in IPEs both pre and post the ruling), we use U.S. foreign issuers’ returns prior to the first event date to calculate CAR_i.² Specifically, we begin calculating CAR_i by using return data from CRSP to create a daily value-weighted return for an index composed of all foreign issuers for an estimation window consisting of all trading days from January 2, 2008 to June 28, 2009, as June 29, 2009 is the first Citizens United event that we analyze.³ Using these return data, we model the relationship between the value-weighted returns for sample firm i (US_RET_i) and the all- foreign issuer index (AF_RET_t) using this regression:

$$US\_RET_{i} = \, b_{0} + b_{ 1} AF\_RET_{t} + e_{t}$$

(2)

Then, using the parameter estimates from Eq. (2), we derive expected returns and use the prediction errors (e.g., AB_RET_i= US_RET_i−(β₁ _RET_t)) as sample firm i’s abnormal return for the day. Finally, we sum firm i’s abnormal returns over the relevant event window. The first three events (the order for a second oral argument, the second oral argument, and the decision itself) all occurred at the Supreme Court, which is an institution that is highly secretive and from which there is no information leakage prior to announcements. For example, the decisions that the Court announces on any given day that it is sitting are not known until the moment they are announced. Thus, for the first three events, we employ [0,1] event windows. For the last event, the advisory opinions issued by the FEC, we employ a slightly longer event window of [− 1,1] to account for potential information leakage about the FEC opinion. Although we believe that these event windows best capture the institutional context in which our Citizens United events unfolded, in Table 7 in Appendix B we present sensitivity tests that demonstrate that our results are consistent across various event windows.

KNOWN_POL_ACTIVITY is our first variable of interest and allows us to test H1. Following Hadani and Schuler (2013), we measure it as the log plus one of the sum of each firm’s lobbying expenditures and PAC contributions (i.e., the known political activity associated with the firm).⁴ The other two variables of interest are CEO_DUALITY, which is a binary variable coded 1 if the CEO is also chairperson of the board and allows us to test H2a, and BLOCKHOLDINGS, which is the percentage of shares held by 5% or greater blockholders and allows us to test H3a. The interactions of these two variables with KNOWN_POL_ACTIVITY are used to test H2b and H3b, respectively. To ensure all three variables were measured prior to our first event, we record their values as of the end of 2008; we provide further technical information on all employed variables in Table 6 in Appendix A.

CONTROLS represents a vector of control variables. First, given the well-established finding the regulated firms are more sensitive to politics (Grier et al. 1994), we use Coates’ (2012) mapping of regulated industries to control for whether or not the firm operates in a non-financial regulated industry (NF_REGULATED) coded one if the firm operates in the (Fama–French 48) industries of (4) alcohol, (5) tobacco, (13) drugs, (24) aircraft, (26) guns, (27) gold, (30) oil, (31) utilities, (32) telecom, and (40) transportation, and zero otherwise.

Second, we include several variables from Compustat to capture firm performance and operations that are influenced by firms’ investment strategies. These include earnings per share (EPS), return on assets (ROA), total liabilities/total assets (LEVERAGE), market-to-book ratio (MTB), and one-year change in sales (ΔSALES); all of these variables are measured as of the end of 2008. Third, we control for systemic risk (IRISK, calculated using a market model based upon firms’ monthly returns from 2008) and firm size (MVE, defined as the log of a firm’s market value of equity as of December 31, 2008). Finally, to control for the possibility that industry-specific events or trends might confound our analysis, we include industry fixed effects (ψ_j) based upon 2-digit SIC membership. We report the descriptive statistics and correlation matrix for our independent variables in Panels A and B of Table 3, respectively.

Table 3 Descriptive statistics

The descriptive statistics indicate that 38% of our sample firms have concentrated power in the form of CEO duality. Based on KNOWN_POL_ACTIVITY, over 36% (700 of 1918, not tabled) of our sample firms have invested resources in PAC and/or lobbying efforts prior to the Citizens United ruling. One-fourth of our sample firms operate in non-financial regulated industries, and sample firms hold, on average, systematic risk of 1.33. Turning to performance metrics, the average sample firm reports a positive but small return on assets (mean ROA = 0.04), and a sales growth rate of 20%.

The correlation matrix provides some initial insights into the importance of CEO duality and concentrated ownership for firms’ potential investment in IPE conditional on known political activity. CEO_DUALITY is positively correlated with KNOWN_POL_ACTIVITY, whereas BLOCKHOLDINGS is negatively correlated with KNOWN_POL_ACTIVITY, hinting that these two governance attributes have conflicting influences on firms’ decisions to invest in CPA. The significant positive correlation between KNOWN_POL_ACTIVITY and NF_REGULATED is consistent with prior research that notes that regulated firms invest in political activities at higher rates (Masters and Keim 1985), as are the positive correlations between KNOWN_POL_ ACTIVITY and firm performance (EPS and ROA) and size (MVE). The remaining significant correlations in Panel B also appear to be in-line with prior research, and to assuage potential concerns regarding multicollinerarity, we note here that none of the variables in any of our estimated models had a variance inflation factor greater than ten, which is the conventionally employed VIF cut-off for multicollinearity (Hair et al. 1995).

Finally, we note that in estimating Eq. 1, we account statistically for the fact that our events affect all firms at the same time, as this can bias asymptotic standard errors due to contemporaneous correlation across returns (see, Sefcik and Thompson 1986; Bernard 1987). If unaddressed, such correlations may lead to other regularities producing associations between cross-sectional firm characteristics and abnormal returns. To address this possibility, in addition to the p-values calculated from our asymptotic standard errors, we also present p-values associated with bootstrapped standard errors calculated via Monte Carlo simulations using non-event day returns (Zhang 2007; Larcker et al. 2011) when reporting the significance of the results discussed in the next section.⁵

Results

Main Findings

We present the results of our main tests in Table 4. We focus the discussion of our findings on the first set of columns in Table 4 (“All Events”), which capture Citizens United’s cumulative effect on CPA-related reputational risk, given legal scholars’ beliefs that the case’s total impact could be understood only after the FEC acted in July 2010 to further shield corporations from disclosure of their political activity and thus prevent managers from being held accountable.

Table 4 Abnormal returns of U.S. domestic issuers in response to Citizens United v. Federal Election Commission

In this model, as in all of the individual events in Citizens United modeled in the remaining columns of Table 4, we find no evidence to reject the null hypothesis articulated in H1. That is, shareholders did not differentially react to firms around the events in Citizens United based upon firms’ known political activity. None of the coefficient estimates for KNOW_POL_ACTIVITY in Table 4 are significant, and the sign of the coefficients also varies by event. These non-findings are consistent with a prior event study of Citizens United that did not find an effect for firms’ known political activity (Werner 2011), and we interpret them as suggestive evidence for the theoretical arguments that motivate our hypotheses 2 and 3: that is, given the potential agency issues and accountability problems created by Citizens United, the marginal investor trading around events in the case likely had a more nuanced or conditional view of the decision’s distributional effects on reputational risk based upon both firms’ known political activity and their existing governance attributes.

Turning to H2a and H2b that posit that concentrated decision rights in the hands of the CEO will lead to a negative reaction by investors to the events in the Citizens United case, our findings are mixed. We find no evidence for H2a, as the coefficient for CEO_DUALITY is statistically insignificant. However, with regard to H2b, firms with CEO duality experience larger negative CARs as their known political activity increases. More formally, the CEO_DUALITY*KNOWN_POL_ACTIVITY interaction is statistically significant using either set of p-values and is negatively signed.

To ease interpretation of this interactive effect—i.e., to consider jointly the effect of all three variables—following Brambor et al. (2006), Fig. 1 plots predicted firm CARs on the y-axis while varying KNOWN_POL_ACTIVITY on the x-axis. We plot separate lines for each value of CEO_DUALITY. The plot reveals that our model predicts higher CARs for firms with no KNOWN_POL_ACTIVITY and CEO_DUALITY, but the plot also shows that as KNOWN_POL_ACTIVITY increases for firms with CEO_DUALITY, these firms’ predicted CARs quickly underperform those of firms that are similarly politically active but without CEO_DUALITY.

Fig. 1

Predicted cumulative abnormal returns (CAR) from the “All Events” model of Table 4, varying CEO duality, known political activity, and their interaction and setting all other variables at their means

In more concrete terms, firms with CEO_DUALITY that engage in the mean level of known political activity experience CARs due to Citizens United that are − 0.1% lower than normal. For these firms, a one standard deviation increase in KNOWN_POL_ACTIVITY is associated with a CAR across the case’s events approximately − 2% lower than the average firm. In dollar terms, these effects are roughly equivalent to effects of − $1.8 million and − $38 million on market capitalization, respectively.

These results suggest that, in-line with our H2b, the market views the deregulation of IPE to be particularly problematic for firms that engage in high levels of known political activity and have concentrated decision rights, as proxied by CEO_DUALITY. The lack of a significant finding on the main effect of KNOWN_POL_ACTIVITY bolsters our claim that shareholder reactions to firms’ expanded investment opportunity set as a result of Citizens United are conditional on the concentration of decision rights in the CEO and thus, the decreased ability of shareholders to hold him or her accountable for CPA-related reputational risk.

In contrast to the finding on CEO duality’s direct effect, the result of our test of H3a that posits that concentrated ownership has a positive effect on the market’s reaction to the deregulation of IPE under Citizens United is significant and positively signed. Concentrated ownership, as measured by BLOCKHOLDINGS, has an average effect of producing returns that are + 2.2% above normal around Citizens United events (approximately + $41.8 million in market capitalization). Further, a one standard deviation increase in BLOCKHOLDINGS increases returns by + 0.57% above normal (approximately + $11 million in market capitalization). Finding a significant positive coefficient on BLOCKHOLDINGS is consistent with the market anticipating better monitoring of firms’ investments in IPEs by blockholders leading to greater managerial accountability in the form of a more efficient (or less reputationally risky) allocation of firm resources to this type of CPA relative to firms with more diversified ownership.

However, as was predicted by H3b, the positive effect of concentrated ownership on shareholder value is discounted as a firm’s level of known political activity increases. That is, the coefficient for the interaction between KNOWN_POL_ACTIVITY and BLOCKHOLDINGS is statistically significant and negative. And, for each standard deviation increase in KNOWN_POL_ACTIVITY, holding BLOCKHOLDINGS and all other variables constant at their means, firms have a predicted CAR that is − 0.1% or approximately − $2 million (in terms of market capitalization) lower. To illustrate this interactive effect more dynamically, in Fig. 2, we plot KNOWN_POL_ACTIVITY on the x-axis with predicted CAR on the y-axis and draw best-fit lines for four set values of BLOCKHOLDINGS (0.25, 0.50, 0.75, 1.00), all of which fall in the observed range of the variable.

Fig. 2

Predicted cumulative abnormal returns (CAR) from the “All Events” model of Table 4, varying blockholdings, known political activity, and their interaction and setting all other variables at their means

As Fig. 2 reveals, the main effect of BLOCKHOLDINGS (captured by the distance between the four lines) is positive for most values of KNOWN_POL_ACTIVITY. However, as KNOWN_POL_ACTIVITY increases, the predicted anticipated benefits of decision rights being concentrated in a small group of owners dissipates and ultimately turns negative as the log of KNOWN_POL_ACTIVITY surpasses five. Relating this threshold to the observed range of KNOWN_POL_ACTIVITY, these predictions suggest that shareholders anticipated that decision rights being concentrated in a small group of owners would be problematic for those firms that were in the upper quartile of political activity at the time of the events in Citizens United. This set of findings for BLOCKHOLDINGS and its interaction with KNOWN_POL_ACTIVITY provide strong support for H3a and H3b.

The estimated parameters on the control variables are either insignificant or in-line with expectations. Of the significant control coefficients, NF_REGULATED and IRISK are negatively signed, and ROA and ΔSALES are positively signed. Additionally, the interaction between NF_REGULATED and KNOWN_POL_ACTIVITY is also negative, suggesting that politically active regulated firms may be especially vulnerable to the reputational risks involved in IPE.⁶ Collectively, the results of our analysis examining cumulative shareholder reactions to Citizens United suggest that the agency threats and accountability issues created by unlimited, undisclosed IPE are heterogeneous, with governance and known political activity jointly affecting how shareholders perceived the reputational risks stemming from the deregulation of CPA induced by Citizens United.

In the remaining columns of Table 4, we re-estimate Eq. 1 for each of the four individual Citizens United events. Since all four events produced outcomes that signal an increased ability of corporations to invest in unlimited, undisclosed IPE, we expect the individual events to produce results similar to those in our cumulative analysis. The results for all four events for our key variables, as well as for our control variables, are consistent with our cumulative findings: the abnormal returns around each event are significant and negative for firms that have CEO_DUALITY and high levels of KNOWN_POL_ACTIVITY, and the abnormal returns around each event are significant and positive for firms that have higher levels of BLOCKHOLDINGS; although, again, this positive effect is discounted as the amount of KNOWN_POL_ACTIVITY increases. The consistency of these results provides strong evidence in favor of H2b, H3a, and H3b and suggests that the agency and accountability threats of firms’ undisclosed investment in IPEs are detrimental to firm value and signal greater reputational risk.

Sensitivity Tests

We hypothesize that CEO duality and concentrated ownership are key governance attributes for our research question because more powerful CEOs and more significant shareholders have greater decision rights over their firms’ investments. However, as prior research notes, firms have numerous governance attributes, which have heterogeneous associations with firm outcomes and share value (see, e.g., Larcker et al. 2007). To see if our findings are sensitive to other features of governance, which may enhance managerial accountability on dimensions unrelated to or beyond investment decision rights, we add four additional measures of corporate governance to Eq. 1, including other measures that capture ownership composition and potential agency issues.

The four additional governance measures that we include are the log of the number of common shareholders (SHAREHOLDERS), which measures the severity of the collective action problem shareholders might experience when attempting to constrain management’s CPA (Coates 2012); a binary capturing whether (1) or not (0) a majority of the firm’s shares are held by institutional investors (INSTIT_MAJORITY), which tests whether there is a threshold effect for majority institutional ownership that insulates shareholder value from the heterogeneous threats created by undisclosed IPE (Hadani 2012); the percentage of shares held by insiders (INSIDER_OWN), which controls for the degree to which management’s interests align with shareholders’ and might mitigate the agency-related effects captured by CEO_DUALITY; and the entrenchment index (E_INDEX) of Bebchuk et al. (2009), which counts how many of six governance provisions (staggered board, limits to amend bylaws, limits to amend charters, supermajority, golden parachutes, and poison pill) are adopted by a firm. The data on the number of shareholders came from Compustat and on the remaining governance provisions from RiskMetrics. All variables were measured as of the end of 2008.

Due to data availability, when we control for these additional governance attributes, our sample is reduced from 1918 to 1040 firms. This reduction is largely driven by the data needed to calculate the E_INDEX, which primarily covers the S&P 1500. Panel A of Table 5 documents this reduction. In Panel B, we report the distribution of our governance subsample across the Fama–French 12-industry categorization, and we find that our governance subsample’s cross-industry distribution correlates highly with our full sample (r = 0.957) and the market as a whole, after excluding financial firms (r = 0.879). In Panel C, we present the descriptive statistics for our governance subsample. In comparison with our full sample, the firms in our governance subsample have higher levels of known political activity, are more likely to have CEO duality, have higher levels of performance (as measured by EPS), and are larger (as measured by MVE). The difference in size is also principally responsible for the difference in known political activity, as firm size is a key determinant of engagement in politics (Sadrieh and Madan 2005).

Table 5 Sensitivity analysis considering other governance attributes

In Panel D, we report the results of two OLS specifications. In the first specification, we add these five additional measures to our main regression Eq. (1). In the second specification, we include these variables and an interaction between each of them and KNOWN_POL_ACTIVITY. Additionally, in the second specification we also include an interaction between NF_REGULATED and KNOWN_POL_ACTIVITY. For simplicity of presentation and due to the way in which of each of the four Citizens United events build upon one another legally, in our sensitivity tests, we present only the cumulative abnormal returns across all four Citizens United events; unreported results for each event are consistent with the below findings.

The results across both specifications in Panel D mirror those of our main analysis, including for our control variables. The interaction between CEO_DUALITY and KNOWN_POL_ACTIVITY is significant and negative, and BLOCKHOLDINGS has a positive and significant effect that is mitigated by the negative interaction between it and KNOWN_POL_ACTIVITY. In contrast, none of the additional governance attributes nor their interactions with KNOWN_POL_ACTIVITY are statistically significant. We interpret these results as supportive of the notion that around the events in Citizens United what shareholders are responding to is not simple engagement in political activity nor ‘bad’ or ‘weak’ governance or even the interaction of these two variables; rather, shareholders are reacting negatively toward the subset of politically active firms that have concentrated decision rights in a small set of actors, principally the CEO and major investors. In demonstrating that only those governance attributes related to decision rights have such an effect, we posit that these results reflect that shareholders, within the context of CPA, have an expectation that managers should have to justify their political investment decisions and that their firms will suffer reputational penalties when they are unable to do so.

In Appendices B and C, we present two additional sensitivity tests. In Table 7 in Appendix B, we again replicate both our main analysis and the second specification of our governance sensitivity test, but now we vary the length of the event windows over which we calculate the cumulative abnormal returns over all four Citizens United events. In Panel A of Table 7 in Appendix B, we assign all four events a 3-day event window [− 1,1]; in Panel B, we assign all four events a 5-day event window [− 1,3]; and in Panel C, we assign all four events a 7-day event window [− 1,5]. The justification for the first window extension is that by elongating the front-end of the window, we capture any potential (however unlikely) information leakage for all of the events; the remaining two extensions to the back-end of the windows are tested primarily to examine the sensitivity of our findings, but they also allow for the possibility that market participants need additional time to determine which firms engage in CPA and potentially suffer from agency or accountability problems. The results presented in Panels A–C of Table 7 in Appendix B demonstrate that our findings are minimally sensitive to the windows over which we calculate our abnormal returns: our main findings regarding heterogeneous agency and accountability threats consistently hold when examining the bootstrapped p-values, and only our finding for the interaction between CEO_DUALITY and KNOWN_POL_ACTIVITY fails to retain its statistical significance using an asymptotic test (at p < 0.10) and only when we use the 7-day event window.

In Table 8 in Appendix C, we replicate both our main analysis and the second specification of our governance sensitivity test using different market models to generate our CARs over all four Citizens United events. In Panel A of Table 8 in Appendix C, we use a market model that employs the value-weighted return of a market index composed of Canadian issuers that are traded on U.S. exchanges. We choose to isolate Canada since its national economy is so closely linked to the U.S. economy and since it is home to more foreign issuers than any other nation (just under 25% of all foreign issuers traded on U.S. exchanges are Canadian) (Zhang 2007). In Panel B, we use a market model that employs Bermudian issuers to construct a market index. We employ Bermuda as a second market model check since, during the estimation window in 2008 and 2009 (prior to the first event) over which we estimate our baseline market models, Bermudian returns were the most highly correlated with U.S. returns (r = 0.915). In both panels our key findings hold in terms of both their statistical and substantive significance, demonstrating that our findings of heterogeneous agency and accountability threats are not sensitive to the choice of market model used to generate our CARs.⁷

Discussion and Conclusion

Our study investigates the equity market’s reaction to the Supreme Court case of Citizens United v. FEC to explore the reputational consequences and ethical considerations introduced by a new form of CPA that grants shareholders little to no ability to hold managers accountable for their investment decisions. In doing so, we test and find evidence for a channel of reputational risk created by the agency issues endemic to this form of CPA that den Hond et al. (2014) develop in their conceptual model linking CPA to reputation. Specifically, we find that the share prices of firms that had both CEO duality and high levels of known political activity were negatively and significantly affected by the case, and in contrast, we also find that the share prices of firms with more concentrated ownership increased in reaction to Citizens United but that the increase in value for firms with concentrated ownership was discounted when such firms had higher known political activity. Collectively, these findings suggest that the expansion of managers’ political investment opportunities induces heterogeneous agency threats across firms based upon the concentration of decision rights within firms.

As a result of these findings, we argue that shareholders anticipate that their ability to hold management accountable for political activity—and subsequent reputational threats that may develop from it—as being highly contingent on firms’ existing governance structures. That is, we believe that the negative impacts on firm reputation that we document in our short-run event study of Citizens United capture shareholders exiting those firms in which agency risk vis-à-vis CPA is greatest. Stated differently, because of the lack of on-going disclosure of these new forms of CPA, the events surrounding the decision itself provided the shareholders of firms with concentrated decision rights the only information they would ever receive in order to hold managers and dominant shareholders accountable for their IPE investment decisions. The average effects in our model suggest that firms paid a reputational penalty for such investors’ discomfort with this on-going lack of accountability.

Beyond documenting this link between firm CPA and reputation, our study makes four additional contributions to the literatures on CPA, governance, and accountability. First, this study is among the first to provide empirical evidence for the agency theory of CPA in the electoral context. Since most prior studies on CPA and agency have explored lobbying or PAC contributions, they have been unable to explicitly examine how the use of firms’ own monies in the electoral process might affect firm performance. In the IPE context, we find mixed evidence for the agency theory of CPA, and we find no evidence for the investment theory of CPA. Further we advance this debate and theorizing in the CPA subfield broadly by focusing on the impact of CPA on firm reputation, a dependent variable that has only recently received attention in the CPA literature but one that represents a firm’s most important intangible asset.

Second, by employing an event study research design, we view our study as providing causal evidence to complement the associational studies we cite earlier that show similar governance-related qualifications on the relationship between CPA and firm performance. That is, by exploiting the exogenous shocks created by events in the Citizens United ruling we make a methodological contribution to this literature by avoiding many of the sources of endogeneity that have plagued previous work.

Third, our results suggest that researchers in all fields should take greater care to consider corporate governance not as a general construct but to unpack mechanisms such as decision rights that, depending upon the context, have more power in explaining how market actors react. For example, scholars of corporate social responsibility (CSR) might find examining decision rights a fruitful way to disentangle the ambiguities in the relationship between CSR and financial performance. And, within the CPA subfield, scholars of lobbying, which is subject to many of the same agency issues as IPEs and also represents a political investment that is an order of magnitude greater than firms’ electoral activity, should give greater consideration to the role of governance as a means for shareholders to hold both managers and external, contract lobbyists accountable for the reputational risks that lobbying can introduce or heighten.

Finally, our findings provide empirical evidence useful in the ethical debate over whether more disclosure of CPA is needed in order to hold managers accountable (Stoll 2015). A broader movement pushing for greater corporate accountability is growing and to date has focused on how stakeholders may be better off pursuing corporate accountability over CSR (see, e.g., Utting 2008). The arguments made in support of this movement in the social realm could very easily carry over to the political realm (see, e.g., Lyon et al. 2018), and actors such as the Center for Political Accountability are pursuing such an agenda via public pressure and proxy resolutions. Much of the practical and legal debate around CPA disclosure relates to its reputational and competitive consequences and whether they may be at odds in the CPA context. We view this as an open question ripe for further engagement by scholars. Further a resolution of this discussion in favor of full transparency when it comes to CPA would also inform the bigger ethical question raised in our introduction of whether it is appropriate for firms to engage in CPA at all, as it is only when we have a full picture of their activity that we can attempt to answer this question.

As with all papers, our study has its limitations. First, as we are analyzing shareholder reactions to a court case deregulating CPA activity, we are measuring only anticipatory responses in terms of reputational effects to the potential use of IPEs and not their actual use. However, since few firms voluntarily disclose their investments in IPE and there have been few leaks of such covert investments to analyze (see, e.g., Werner 2017), it remains difficult for scholars to make broad empirical inferences with regard to the actual use of IPEs. Second, although the event study research design aids us in making causal claims, we can speak only to short-term reactions using it, and although our analyses and robustness checks give us confidence in the direction and significance of our effects, there is nevertheless the possibility that their magnitudes are overstated due to market over-reactions in the short-run (McWilliams and Siegel 1997). We acknowledge this but are again limited by the lack of disclosure in studying the long-term consequences of firms’ investment in IPE. Finally, although we believe it is unlikely that firms’ known political activity and governance were jointly determined in a process that was coincident or endogenous to Citizens United, firms do select into both CPA and governance arrangements, and thus, our findings should be read with this caution in mind.

In both its written opinion and its questions during oral argument, the five-justice majority in Citizens United dismissed the shareholder protection argument made by the U.S. government under an implicit assumption that firms will act ethically to disclose IPEs to investors and that investors could exit by selling their shares when they disagreed with managers’ actual investments in IPEs. Yet, more than 9 years after the ruling it remains difficult, if not impossible, for both shareholders and the public to determine what IPE investments managers are making, as few firms disclose these investments voluntarily. And, neither Congress nor the relevant regulatory bodies appear likely to mandate disclosure in the immediate future. In fact, under President Trump, the U.S. Treasury Department is even eliminating the requirement that 501(c)4 s and 501(c)6 s confidentially report their contributors to the IRS (Rubin 2018). Beyond raising the obvious questions related to managerial accountability and reputation, this state of affairs makes the broader ethical questions of the legitimacy and appropriateness of firms’ engaging in the political and policymaking processes to any degree that much more important for business ethicists and CPA and corporate governance scholars to continue to explore.

Appendices

Appendix A

See Table 6.

Table 6 Variable definitions

Appendix B

See Table 7.

Table 7 Expanded event windows

Appendix C

See Table 8.

Table 8 Alternative estimates of abnormal returns