Board design and governance failures at peer firms

Research Summary: Our study introduces board committees as a crucial determinant of board actions. We examine how directors who structurally link different board committees — referred to as multi-committee directors (MCDs) — explain why some board actions are merely symbolic while others are more substantive. As a baseline, we argue that boards in general respond to financial restatements at peer firms by symbolically appointing new directors who are relatively inexperi-enced and unlikely to have a substantive impact. In contrast, boards with audit – nomination MCDs are more likely to take the substantive action of appointing new directors with the prior experience necessary to reduce the risk of their own future financial restatement. We combine qualitative interviews and a causal identification strategy using an original dataset cover-ing Russell 3000 firms from 2001 to 2014. Managerial Summary: Committees play a central role in how boards carry out their governance responsibilities. This study shows that assigning directors to multiple board committees can improve governance performance under certain conditions. Specifically, when external events focus attention on the board's monitoring role, the presence of multi-committee directors (MCDs) that link the audit and nomination committees is associated with the subsequent appointment of more experienced directors and a reduction in the likelihood of future financial restatements and lawsuits. Because these specific board con-figurations can have substantive implications for governance effectiveness, we recommend that board leaders be deliberate in how they allocate board committee assignments and treat board committee structure as an intentional organizational design choice that can meaningfully improve governance outcomes.

enced and unlikely to have a substantive impact. In contrast, boards with audit-nomination MCDs are more likely to take the substantive action of appointing new directors with the prior experience necessary to reduce the risk of their own future financial restatement. We combine qualitative interviews and a causal identification strategy using an original dataset covering Russell 3000 firms from 2001 to 2014. Managerial Summary: Committees play a central role in how boards carry out their governance responsibilities. This study shows that assigning directors to multiple board committees can improve governance performance under certain conditions. Specifically, when external events focus attention on the board's monitoring role, the presence of multi-committee directors (MCDs) that link the audit and nomination committees is associated with the subsequent appointment of more experienced directors and a reduction in the likelihood of future financial restatements and lawsuits. Because these specific board configurations can have substantive implications for governance effectiveness, we recommend that board leaders be deliberate in how they allocate board committee assignments and treat board committee structure as an intentional organizational design choice that can meaningfully improve governance outcomes.

K E Y W O R D S
board committees, board monitoring, corporate governance, new director nomination, peer financial restatements

| INTRODUCTION
A common criticism levied at boards centers on their inability to effectively fulfill their monitoring responsibilities. Shareholders increasingly expect more from the boards representing them. In response to these calls for greater accountability, some boards have taken action by adopting practices like increasing communication with shareholders (Brandes, Goranova, & Hall, 2008) and putting directors up for annual reelection (Ertimur, Ferri, & Stubben, 2010). These actions are especially common following precipitating events such as negative press (Cho & Hambrick, 2006), shareholder proxy fights (Campbell, Campbell, Sirmon, Bierman, & Tuggle, 2012), SEC rule changes (Campbell et al., 2012), social movement pressure (Tihanyi, Graffin, & George, 2014), and peer firms' misconduct (Greve, Palmer, & Pozner, 2010).
At the same time, skepticism lingers around whether actions taken by boards-especially following negative events-lead to substantive improvements in monitoring or whether they serve merely as symbolic actions (Hambrick, Misangyi, & Park, 2015). Research on impression management suggests that some board actions reflect only superficial adjustments that lack any substantive change. Although a merely symbolic act would signal compliance, it would not meaningfully and durably improve monitoring performance. Extant research uncovers avenues through which boards take such action to influence shareholders strategically, including the timing of material announcements (Graffin, Carpenter, & Boivie, 2011), corporate partner selection (Graffin, Haleblian, & Kiley, 2016), and CEO compensation . These symbolic actions do not necessarily improve-and may even impair-the performance of the firm, especially if they substitute for more substantive action (Davidson, Jiraporn, Kim, & Nemec, 2004). Thus, it remains imperative to understand the conditions under which boards can act more substantively.
In this study, we explore how the structure of board committees enables boards to take substantive-rather than merely symbolic-action to improve monitoring and meaningfully improve the future performance of the firm. Going beyond the traditional conception of the board as a unitary entity, a growing body of research argues that board committees represent the locus of decision-making for the board (Brandes, Dharwadkar, & Suh, 2016;Kolev, Wangrow, Barker, & Schepker, 2019;Spira & Bender, 2004). However, despite the significant role of board committees, limited research addresses the potential implications of board committee structure. We focus on one particular structural attribute of board committees: having a director who simultaneously sits on multiple committees. We refer to these directors as multi-committee directors (MCDs) and contend that they play an integral role in shaping committee decisions (Chen & Wu, 2016). Prior work on MCDs finds that they channel information between otherwise siloed committees (Brandes et al., 2016). We argue that a board structured specifically with MCDs linking the audit and nomination committees (i.e., audit-nomination MCDs) may facilitate decisions that integrate different committee priorities, which results in a higher likelihood of choosing more substantive over merely symbolic actions.
Through a program of 16 interviews with Fortune 500 directors in 2019, a large-scale online survey of over 5,000 board directors from 2015 to 2016, and a panel dataset of 1,243 firms from 2001 to 2014, we document how board committee structure affects whether a board takes more substantive action following a financial restatement at a peer firm, which we refer to as a "peer restatement event." We focus on the context of firms facing peer restatement events because they serve as specific but widespread occurrences that are likely to prompt boards to engage in symbolic impression management tactics. These events at peer firms do not implicate the focal firm in any wrongdoing but nevertheless serve as a salient reminder of the negative consequences of poor board monitoring. As such, financial restatements at peer firms serve as a plausibly exogeneous shock that allows us to cleanly identify the role of board committee structure in determining board actions. In our baseline analysis, we find that, overall, boards respond to peer restatement events by appointing new directors. However, because these new directors are relatively less likely to possess prior board or audit committee experience, we argue that this type of board action is more symbolic when compared to the more substantive act of recruiting an experienced director who could meaningfully improve the board's monitoring ability (Naumovska, Wernicke, & Zajac, 2019). In our main analysis, we find that, in contrast, boards that have audit-nomination MCDs are more likely to take substantive board actions in response to a peer restatement event by appointing new directors who possess prior board and audit committee experience. Consistent with our hypotheses, we find that these boards are then less likely to file their own financial restatements in the future, suggesting that boards with auditnomination MCDs do indeed improve their monitoring performance.
Overall, our study sheds light on an important and long-lingering question in corporate governance research: when and under what conditions do boards become more effective in their monitoring responsibilities? Building on prior work that uses governance failures to explain when boards take action, we focus on the structural features of the board to explain what type of actions are taken. Our findings thus highlight the role of board committees in enhancing board monitoring and their importance in governance more broadly. We also contribute to existing work on impression management by highlighting when and why certain boards default to merely symbolic actions while others engage in more substantive actions. Finally, by examining the consequences of certain structural attributes within the board, we join the broader narrative of organizational design by laying the groundwork for a new subfield-board design-that identifies how certain structures contribute to a board's overall performance.

| THEORETICAL AND CONTEXTUAL BACKGROUND
Our theoretical background and hypothesis development combine insights from prior research on corporate governance with qualitative data generated from a program of 16 semistructured interviews with current and former board directors of Fortune 500 firms in the United States and a large-scale online survey of over 5,000 board directors conducted from 2015 to 2016. 1 We include excerpts to provide greater color, clarity, and context to the conceptual narrative.

| Board monitoring responsibility, responses, and limitations
Boards have a governance responsibility to monitor their respective firms by preventing managers from engaging in misconduct (whether intended or accidental). Agency theory suggests that a board's primary function is to curb managerial activities that may hurt shareholder interests (Fama & Jensen, 1983). The Sarbanes-Oxley Act (SOX) of 2002 further sharpened this responsibility, and boards now command an increasingly active role in protecting shareholder interests and overseeing regulatory compliance (Cheng, Groysberg, Healy, & Vijayaraghavan, 2021;Ghosh, Marra, & Moon, 2010). As one director whose board career started in the late 1990s recounted, "So much has changed from the so-called 'good ol' days' to the post-Enron period. I think the responsibility is greatly heightened now, and it's just very critical that companies have great boards who really intently and wisely and studiously and meticulously look after the interest of shareholders" (President and CEO of private investment firm, interview, 2019).
However, an extensive body of research casts substantial doubt as to whether boards can meet these expectations. Even when directors seek to diligently monitor managers, they may face cognitive limitations that confine what they attend to and prioritize (Ocasio, 1997). Furthermore, directors often sit on multiple boards and frequently find themselves dealing with multiple competing demands (Field, Lowry, & Mkrtchyan, 2013). Even the meeting context can influence discussions (Tuggle, Schnatterly, & Johnson, 2010). One director shared that "there's a lot that you miss in the day-to-day goings-on at a big, public company" (Partner of private investment firm, interview, 2019).
Additionally, even when directors clearly acknowledge their monitoring duties, they may not always prioritize monitoring. As one director concluded, sometimes directors fall short of "what we know we should do until we are forced to do something" (President and CEO of professional service firm, interview, 2019). Prior research shows that it often takes external pressure-such as negative media coverage (Bednar, Boivie, & Prince, 2012), attention from analysts (Harrison, Boivie, Sharp, & Gentry, 2018), or pressure from shareholders and stakeholders (Campbell et al., 2012;Coombs & Gilley, 2005;Kesner & Johnson, 1990)-to motivate an otherwise inert board to take action.
Notably, responses to external events raise a different set of concerns. Specifically, when boards respond to outside scrutiny, they do not always engage in substantive actions. Research on impression management-defined as any action carried out with the intent of influencing an audience's perception of the organization (Elsbach, Sutton, & Principe, 1998;Graffin et al., 2011)-predicts that although boards are likely to engage in symbolic, visible changes as a way to influence external evaluators, these choices are unlikely to include actions that alter the status quo. Visible changes to board composition can also serve as an important impression management tactic (Westphal & Graebner, 2010). As such, although impression management tactics may differ between boards, the underlying concern remains the same: when and under what conditions do boards undertake substantive actions versus more symbolic actions?

| Board committees and their role in board responses
To shed light on this question, we peer into the black box of board decision-making to investigate how board structure can facilitate more substantive-rather than merely symbolicactions. In particular, we take a nuanced view of board structure that highlights the key role of board committees in shaping board decision-making.
As the primary intra-board subunit, board committees are responsible for many of the key decisions traditionally attributed to boards of directors. The passage of SOX prompted stronger requirements over the activities of the three committees tasked with fulfilling many of the board's core monitoring responsibilities: the nomination of new directors for the nomination committee; CEO and senior executive compensation for the compensation committee; and financial reporting for the audit committee (Conyon & Peck, 1998;Kesner, 1988). Because overseeing all of these activities would be difficult for directors to simultaneously take on as a whole, these tasks are delegated to board committees, which allow for increased efficiency through specialized decision-making (De Kluyver, 2009).
The early literature on board committees argues that much, if not most, of the actual board decision-making process takes place at the committee level (Kesner, 1988;Klein, 1998). In our interviews, directors overwhelmingly agreed that this still holds true. One director shared his belief that "a really good board has much, much of the governance work taken by the committees" (Former CEO and chair of global technology firm, interview, 2019), while another stressed that the committee is "where everybody has to participate" (Former CEO and chair of global power firm, interview, 2019). One director elaborated: It is absolutely the case that the work gets done in the committees. Especially as the expectations have moved up, it's essential that a significant amount of workmaybe even the vast majority of work-get done during committee meetings because there simply isn't enough time during a board meeting to cover everything. (Former CEO and chair of global software firm, interview, 2019) Given the central role of committees in board functioning, a body of extant work examines the consequences of committee decision-making. Specifically, scholars seek to understand how certain features of board committees may affect firm-level outcomes such as performance (Chan & Li, 2008;Nguyen & Nielsen, 2010), firm value (Aggarwal, Erel, Ferreira, & Matos, 2011;Davidson, Xie, & Xu, 2004), and financial misconduct (Badolato, Donelson, & Ege, 2014;Faleye, Hoitash, & Hoitash, 2011); another stream examines individual-level outcomes such as executive compensation (Conyon & He, 2004;Cyert, Kang, & Kumar, 2002) and director nomination (Cohen, Frazzini, & Malloy, 2012;. Much of the research on board committees adopts an agency theory perspective and focuses on the association of board committees with traditional monitoring objectives. Notably, an underexplored element of board committees relates to their specific decisions and whether these decisions amount to symbolic or substantive changes to the board. In particular, the nomination committee's mandate of recruiting new directors can result in changes that could either be more substantive or merely symbolic. On one hand, this committee can take substantive action by recruiting directors who bolster the board's monitoring capabilities. Recruiting a new director involves balancing a multitude of factors, such as board independence (Aguilera & Jackson, 2003), diversity (Terjesen, Aguilera, & Lorenz, 2015), firm strategy (Lungeanu & Zajac, 2016), and individual director expertise (e.g., cybersecurity) (Cheng & Groysberg, 2017). By nominating directors with certain types of characteristics, the nomination committee can substantively enhance the human capital of the board (Hillman & Dalziel, 2003). Importantly, new director nominations may have long-term effects on the firm (Garg & Furr, 2017) since directors often serve for multiple terms: the average tenure of an S&P 500 outside director is 8 years (Spencer Stuart, 2019). As a result, director nominations can shift the capabilities of the board and the oversight of the firm for years to come.
At the same time, new director nominations can serve as a symbolic way for a board to manage external impressions, given that the nomination committee has broad flexibility in whom it chooses to nominate. Prior work examining director nominations finds that shareholders respond positively to new director appointments (Certo, 2003;Fich, 2005) because they may reflect a change in strategic direction (Chen, Crossland, & Huang, 2016;Lungeanu & Zajac, 2019). Other studies find that certain board appointments are meant to be highly symbolic while lacking more substantive effects (Westphal & Graebner, 2010;: for example, appointing a woman with no prior board experience to the board may increase board diversity, but the lack of necessary mentoring may nevertheless lead to a higher likelihood of her leaving (McDonald & Westphal, 2013). Given the potential long-term implications of new director appointments, it is imperative to understand the factors contributing to a nomination committee's recruitment decisions ).

| Siloed committee structures and the role of multi-committee directors in facilitating different committee responses
Much of the extant research on board committees examines their activities in isolation. This is perhaps unsurprising given that board committees typically have siloed structures that help directors specialize in specific tasks (Klein, 2002). As a result, board committee responsibilities often segregate information to different committees: an outcome that may come at the expense of the entire board having access to all the available information (Reeb & Upadhyay, 2010). Notably, prior research details how committees are often isolated from one another (Brandes et al., 2016). One director noted that "there's certainly a tendency to form silos" (Former CEO and chair of global software firm, interview, 2019). When asked to elaborate why these silos hinder communication between committees, he shared: In most cases, it's not that people don't want to communicate, it's just they don't think about it as being essential to doing their job. Like I'm doing my job on comp and uh, gee, I didn't realize that governance needed to know this information, or vice versa. So it's more: out of sight, out of mind.
At the same time, the siloed structure of board committees does not preclude committees from ever interacting with each other, and the ability to foster inter-committee interactions could have important implications. We identify multi-committee directors (MCDs) as a key structural feature present in some-though not all-boards, and we suggest that they play a pivotal role in shaping committee decision-making. Specifically, the structural position of MCDs allows them to be aware of the priorities and strategic considerations of multiple committees (Brandes et al., 2016) and to facilitate "cross-pollination" across committees (Former CEO and chair of global technology firm, interview, 2019) (e.g., Aggarwal, Hsu, & Wu, 2020). One director noted that being an MCD "was a good opportunity to provide communication both between and amongst committees" (Former CEO and chair of global software firm, interview, 2019). Another director echoed his sentiments: Sitting on the same board on multiple committees is as a good thing because it allows me to deal with issues in a more nuanced way. Having that benefit of that interaction makes for richer content and better discussions. (President and CEO of professional service firm, interview, 2019) Nevertheless, research examining interactions between different committees is relatively scarce, and the efficacy of MCDs remains unclear. Although one study purports that auditcompensation MCDs (directors who serve on both the audit and compensation committees) facilitate knowledge transfer between the two committees (Brandes et al., 2016), another study finds that multiple committee memberships come at the cost of director busyness and actually result in poorer committee decision-making (Liao & Hsu, 2013). Furthermore, it is unclear when MCDs engage in sharing information between committees. As research on interlocking directors highlights, simply having a director connecting two different boards does not guarantee that she will transmit experience or knowledge of organizational practices between them (Shropshire, 2010). As such, more needs to be done to ascertain when and why having MCDs can be beneficial for the board and the overall firm.
One apt way to answer these questions is to examine whether the presence of MCDs results in a board action that is merely symbolic or more substantive. 2 In the following section, we develop hypotheses around when and under what conditions a board will choose a merely symbolic versus a more substantive response depending on its board committee configuration.

| HYPOTHESIS DEVELOPMENT
In the prior section, we discussed how heightened expectations for boards and board committee structure have important implications for board decision-making. We shift now to examine board responses to a particular shock: a peer firm issuing a financial restatement, which we refer to as a "peer restatement event." First, we establish a baseline of how boards in general respond to peer restatement events. From there, we narrow our focus to boards with a specific structure-namely the presence of an MCD-to distinguish their response from the baseline response of boards in general. In doing so, we demonstrate that although boards in general may default to a merely symbolic response, boards with MCDs engage in more substantive actions.

| Peer restatement events and the likelihood of nominating new directors
External shocks can force directors to refocus on their monitoring capabilities. By increasing the uncertainty in which a board operates, certain environmental shocks-particularly those 2 It is important to note that symbolic actions and substantive actions do not occupy a single sliding scale where high symbolism means low substance, and vice versa. Symbolic and substantive need not be a binary choice, and a course of actions can be both symbolic and substantive. We seek to highlight instances when a move can be merely symbolic compared to one that is symbolic and substantive. that reflect governance failures-can bring the short-term attention of directors back to monitoring (Boivie, Bednar, Aguilera, & Andrus, 2016). The severity of such events often focuses the attention of boards on the immediate situation, prompting a board to exercise more control (Mizruchi, 1983). Among the spectrum of different types of governance failures, we focus on financial restatements at peer firms and investigate how firms that share an interlocked director with a restating firm respond to this shock. Importantly, "restating firm" refers to firms issuing the restatement while "focal firm" refers to firms affected by a restatement at a peer firm.
Financial restatements reflect a failure of board monitoring and can negatively affect the restating firm's performance due to concerns that its managers and directors either engaged in misconduct or failed to prevent it (Greve et al., 2010;Palmrose, Richardson, & Scholz, 2004). One director described restatements as atypical, though "if it were to happen, it would be a major event" (Chair of private equity firm, interview, 2019). Importantly, the firm that files the financial restatement is not the only affected entity: a financial restatement can precipitate changes in firms associated with the restating firm (Johnson, Joshi, & Hogan, 2020;Nalick, Kuban, Hill, & Ridge, 2020). Building on prior research that focuses on board reactions to peer restatement events (Cowen & Marcel, 2011;Srinivasan, 2005), we suggest that financial restatements sharpen the saliency of restatement risk for firms linked to the restating firm. Once a peer restatement event "opened his eyes" to the possibility of a similar event occurring at his own firm, a director mentioned that he would "ask to go back and challenge [the managers]: how are we exposed to the potential problem they have that might result in our having to make a restatement?" (President and CEO of professional service firm, interview, 2019).
Despite capturing the attention of board directors, peer restatement events may nevertheless be insufficient to prompt the board to respond in a more substantive way. Drawing on the impression management literature, firms may be motivated to engage in strategic actions that influence an external audience's perception of them in a merely symbolic manner (Elsbach et al., 1998;Graffin et al., 2011). Because financial misconduct raises serious concerns about a firm (Palmrose et al., 2004), a restating firm may respond by deploying certain changes, such as establishing ties with non-profits to gain social approval (Lungeanu, Paruchuri, & Tsai, 2018). These responses attempt to assuage external stakeholder concerns by signaling that the restating firm is distancing itself from malfeasance without necessarily generating meaningful change.
Importantly, despite no wrongdoing on their part, firms linked to the restating firm via an interlocking director (i.e., the focal firm) may similarly engage in impression management given that they often suffer from negative reputational spillover (Jonsson, Greve, & Fujiwara-Greve, 2009). Indeed, prior work finds that directors associated with a restating firm not only experience higher rates of dismissal from the firm that filed the restatement, but also higher rates of dismissal from other board directorships (Cowen & Marcel, 2011;Fich & Shivdasani, 2007;Srinivasan, 2005). By removing the director linking them to the restating firm, the focal firm may avoid becoming further "stigmatized by association" (Devers, Dewett, Mishina, & Belsito, 2009;Neuberg, Smith, Hoffman, & Russell, 1994;Pozner, 2008), which acts as an impression management tactic meant to influence the perception of external shareholders by distancing themselves from accusations of financial misconduct.
Notably, in addition to distancing themselves from negative events, firms can engage in impression management tactics meant to signal positive changes. Specifically, we expect that a peer restatement will prompt boards to appoint new directors. Prior work finds that increasing board independence heightens the congruence between "visible attributes of the board and widely shared investor beliefs about good governance" (Westphal & Graebner, 2010, p. 18). As such, appointing a new independent director can function as an impression management tactic to engender confidence in the board's independence and monitoring effectiveness (Elsbach & Elofson, 2000). Appointing new directors may allow the focal firm to replace the interlocking director who was on the board of the firm that filed a restatement, and thus further distance themselves from the stigma of being associated with the financial misconduct of another firm. Additionally, as part of ex post settling up, if a peer restatement event causes other directors to depart due to concerns of further stigmatization (Pozner, 2008), appointing new directors to replace departing directors can function as an impression management tactic given that neglecting to fill board vacancies could be perceived poorly by external stakeholders (Larmou & Vafeas, 2010). The nomination of new directors may thus evoke change and greater accountability, regardless of whether overall board size changes. We therefore expect the following baseline effect of peer restatements on director nominations: Baseline Hypothesis A: After experiencing a peer restatement event, boards are more likely to appoint new directors compared to boards that did not experience a peer restatement event.
At the same time, the act of appointing a new director can be merely symbolic without being substantive, especially if the director lacks the experience to affect change. Specifically, although experienced directors confer numerous advantages to their boards (Duchin, Matsusaka, & Ozbas, 2010), there are significant costs associated with recruiting them. The limited pool of experienced board directors may impede a board's efforts to appoint a director with prior board experience (Fernandez-Mateo & Fernandez, 2016). Additionally, boards are increasingly reluctant to allow their directors to join multiple other boards due to concerns around "overboarding" (Boivie, Graffin, & Pollock, 2012;Khanna, Jones, & Boivie, 2014). These limitations on the number of board seats that directors can hold further constrains the supply of experienced directors. Moreover, talented directors can be quite selective when deciding whether or not to join a board (Brands & Fernandez-Mateo, 2017). As one director imparted, "Over time, I've become more picky." (Former CEO and chair of global software firm, interview, 2019). For firms that have been affected by a peer restatement, the stigma-by-association could make it especially difficult to attract discerning directors to join the board (Pozner, 2008).
Similarly, boards seeking to recruit a director with prior audit committee experience may face challenges in finding directors with this type of experience. The pool of directors with audit-related expertise is even more limited; serving on the audit committee typically requires a higher degree of time and commitment compared to the other committees, and thus circumscribes the number of directors who opt to serve on this committee. When asked about serving on the audit committee, one director expressed that "it gives you great insight to the company… but it's very time-consuming" (President and CEO of professional service firm, interview, 2019). Furthermore, director recruitment challenges are compounded by listing requirements that mandate that companies have at least one member with "accounting or related financial management expertise" on their audit committee (New York Stock Exchange, 2020). Given the limited number of candidates with this type of expertise, the demand for directors with audit committee experience often outpaces the supply.
For boards seeking to influence external perceptions following a peer restatement event, the extra effort required to recruit a director with prior board or audit committee experience may be too challenging considering the pressure for a rapid response, especially if the act of appointing any new director-regardless of prior experience-is often sufficient in shielding the board from scrutiny. Put differently, although boards may be motivated to respond to a peer restatement by nominating a new director, they may opt to take a largely symbolic approach that does not require difficult, substantive actions. This could be driven by a deliberate choice to pursue the faster solution of simply recruiting any new director, or a forced choice due to the heightened difficulties of attracting experienced directors. In either case, the decision to appoint a new director nevertheless functions as an impression management tactic given that not adding a new director-which could potentially mean leaving the board short-staffed-would risk further damaging external perceptions (Larmou & Vafeas, 2010). Therefore, we expect that even if the likelihood of appointing a new director increases following a peer restatement (Baseline Hypothesis A), boards will be less likely to take the substantive action of appointing directors with either prior board experience or prior audit committee experience in response to the event.
Baseline Hypothesis B: After experiencing a peer restatement event, boards are less likely to appoint directors with prior board experience compared to boards that did not experience a peer restatement event.
Baseline Hypothesis C: After experiencing a peer restatement event, boards are less likely to appoint directors with prior audit committee experience compared to boards that did not experience a peer restatement event.

| Audit-nomination MCDs and the likelihood of nominating experienced directors
While our theoretical discussion thus far suggests that nomination committees will generally respond to a peer restatement with symbolic director appointments, we now turn to the question of whether having an MCD may result in more substantive actions. Notably, the siloed nature of board committees can impede a board's ability to respond to a peer restatement event because the shock is not equally salient for different board committees. When asked which committees respond to external events such as peer restatements, one director offered this assessment: "It depends on the subject and the risk factor that this particular issue raises. A financial restatement would end up in the audit committee" (Partner of private investment firm, interview, 2019). The reason that peer restatements are especially salient for audit committees is that audit committee members possess a specific skill set that allows them to assess the quality of financial statements and thereby identify weaknesses in other firms' filings. As one survey respondent noted, the audit committee is responsible for "look[ing] for emerging sources of risk." As a result, audit committee members are more likely to recognize the severity of a peer restatement event (Cowen & Marcel, 2011).
However, even when prompted by a peer restatement, the audit committee has limited ways in which it can unilaterally take action to shore up the monitoring capabilities of the board as a whole. Besides firing its auditor if the auditor served a restating firm (Chen & Zhou, 2007), there are few other recourses available to the audit committee to improve the board's monitoring capabilities. A director shared that although the audit committee takes actions to ensure that the board is "meeting all of the controls and standards-which are very importanttypically [the audit committee] is not the one that drives a board's response" (President and CEO of professional service firm, interview, 2019).
In contrast, the nomination committee can play a critical role in augmenting the monitoring capabilities of the overall board. Due to its purview over new director nominations, the nomination committee can recruit board candidates who have the experience and expertise to fulfill the board's monitoring responsibilities. However, nomination committee members expressly noted that the audit and nomination committees typically have limited interactions with each other: "Basically the audit committee's work plan and the nomination committee's work plan are in totally different parts of the world. So the notion that they would be chattering a lot back and forth seems like a waste of time" (President and CEO of private investment firm, interview, 2019).
In contrast, we suggest that the presence of an audit-nomination MCD who connects the audit committee to the nomination committee may increase the likelihood that the latter will be responsive to audit-related concerns. Given that directors on the audit committee are most likely to be aware of a peer restatement event and more cognizant of the negative implications it may have for the focal firm, they are more likely to encourage the adoption of stringent measures to protect the focal firm from their own risk of issuing a restatement in the future. Importantly, audit-nomination MCDs can relay these concerns to the nomination committee and prompt them to "search for candidates competent to address the weaknesses" (Survey respondent). One former audit-nomination MCD recounted how he would react to a peer restatement event: It would prompt you to ask the question: "Look at what happened to Joe across town," so to speak. "Are we vulnerable to the same weakness here at our company that would cause a similar, horrible result? What are we doing to protect ourselves against this issue?" When a problem happens at a peer company, it almost always in the very next meeting causes you to ask, "What are we doing to avert those sets of risks?" (Chair of private equity firm, interview, 2019) One effective way to reduce risk would be to appoint more experienced directors to the board. Because directors with prior board experience have been "vetted" by other firms, the focal firm is more likely to trust that these directors can effectively fulfill their responsibilities. Importantly, their experience on other boards could give them insight as to best practices at other firms (Shropshire, 2010). As such, we expect that firms with audit-nomination MCDs will be more likely to appoint directors with prior board experience because directors with this background will be more likely to provide the oversight necessary to prevent future restatements from occurring. Thus, our theory suggests that having an MCD who structurally links the audit and nomination committees can facilitate the recruitment of more experienced directors following a significant governance failure by a peer firm. Formally, we hypothesize that: Hypothesis 1. After experiencing a peer restatement event, boards with audit-nomination MCDs are more likely to appoint directors with prior board experience compared to boards without audit-nomination MCDs.
We expect that audit-nomination MCDs will have a similar effect on the likelihood of appointing new directors with a specific type of prior board experience: experience on the audit committee. Because peer restatements reflect a weakness in the restating firm's auditing capabilities, the most direct way to protect against a similar failure would be to bolster the focal firm's audit capabilities. As one survey respondent noted, audit committee members undergo an "extensive onboarding process, including meetings with key leaders including external auditors, internal audit, general counsel, compliance, finance and accounting." Furthermore, a new director with audit committee experience will have had experience with another firm's financial reporting, and thus have a point of comparison to judge whether certain procedures at the focal firm may constitute "a systemic failure" (Survey respondent). New directors with audit committee experience are thus uniquely suited to improve a board's monitoring capabilities.
However, as previously discussed, recruiting a director with this specific expertise can be quite challenging. Furthermore, given that the nomination committee is charged with identifying directors based on the board's needs, they could be concerned that an emphasis on one particular skill may leave the overall board unbalanced in terms of expertise (Kilduff, Angelmar, & Mehra, 2000). Nevertheless, following a peer restatement event, the usual tradeoffs to appointing directors with audit committee experience may be more acceptable, particularly when there is an auditnomination MCD present to raise awareness of how critical audit expertise would be to protecting the focal firm. As such, we expect that boards with audit-nomination MCDs can translate the audit committee's awareness of a restatement risk into actions by the nomination committee to appoint directors with audit committee experience. We therefore hypothesize: Hypothesis 2. After experiencing a peer restatement event, boards with audit-nomination MCDs are more likely to appoint directors with prior audit committee experience compared to boards without audit-nomination MCDs.

| Consequences for future financial restatements
Thus far, we have highlighted the role of MCDs in the appointment of new directors with prior board or audit committee experience following a peer restatement event. We now switch to examine firm-level consequences of MCDs. Past research suggests that having experienced directors leads to a lower likelihood of future restatements (Arthaud-Day, Certo, Dalton, & Dalton, 2006). When asked how directors could protect shareholders, one director opined that being a good director requires "the ability and the talent to be appropriately inquisitive [and] to ask the right questions" (President and CEO of private investment firm, interview, 2019). Experienced directors in general and those with audit experience specifically are more likely to exhibit these qualities (Lungeanu & Zajac, 2019), which would be essential to enhancing a board's ability to avoid future governance failures. If our predictions are correct in terms of when and why certain boards are more likely to appoint directors with either prior board experience or audit committee experience, then we would expect the same conditions to lead to a lower likelihood of the focal firm issuing a financial restatement in the future.
Hypothesis 3. After experiencing a peer restatement event, boards with audit-nomination MCDs are less likely to issue a financial restatement in the future compared to boards without audit-nomination MCDs.

| Data and sample
We construct a comprehensive panel dataset of large, publicly traded firms in the U.S. to test our hypotheses. Detailed annual data on board committee composition and director characteristics come from Equilar, which provides granular information on intra-board structure and committee composition for Russell 3000 companies. We use these data to construct measures of the number of MCDs, the prior experience of newly appointed directors, and several control variables. Additionally, we gather information on financial restatements from Audit Analytics, which compiles details on financial restatements filed by publicly traded firms in the United States. We derive the remaining control variables from the Compustat and Execucomp databases. 3 Our dataset of Russell 3000 companies is significantly more expansive than the samples analyzed in many prior corporate governance studies, which are typically limited in scope to the S&P 1500 due to the constraints of more commonly used datasets. Furthermore, our panel data structure allows for the inclusion of firm and year fixed effects in our multivariate analyses. As a result, we have robust controls for unobservable firm and year characteristics that cannot be included in studies that use only cross-sectional data.
For our baseline analyses (testing Baseline Hypotheses A, B, C), we examine the full set of Russell 3000 companies. We define our treatment sample as firms affected by an adverse financial restatement at a peer firm, where peer firms are identified based on the presence of a shared director who is a member of both boards. (Again, we refer to the firms affected by a financial restatement at a peer firm as "focal" firms in our analysis.) 4 We chose to define peer firms as those connected through director interlocks due to the tendency of reputational penalties to spread through shared directors (Cai, Dhaliwal, Kim, & Pan, 2014;Chiu, Teoh, & Tian, 2012;Kang, 2008). Our control sample consists of firms not affected by an adverse financial restatement at a peer firm, and we exclude firms that filed their own adverse financial restatement. We specifically focus on adverse financial restatements and eliminate restatements that resulted in an improvement in financial position or those that did not affect financial position. 5 Adverse restatements are more likely to indicate the presence of overly aggressive accounting, operational and/or managerial problems, downward changes to future performance expectations (Callen, Livnat, & Segal, 2006;Srinivasan, 2005), and a corresponding governance failure. Therefore, these types of restatements more likely spur changes in board-level policies due to the severity of the misconduct perpetuated at the restating firm (Arthaud-Day et al., 2006). In summary, we design the sample of firms around those that encounter peer restatements that directly reflect a failure in board oversight. The sample of analysis consists of 16,279 firm × year observations (covering 2,646 unique firms), over the period 2001-2014.
In our main analyses (testing Hypotheses 1, 2, 3), we analyze the subset of firms affected by a peer restatement event and examine new director nominations during the 5 years preceding and the 5 years following the peer restatement event. These limitations result in a final sample 3 As a robustness check, we run our main analyses on the subset of S&P 1500 firms with additional controls that were not available for the full Russell 3000 data sample (CEO tenure and an indicator for founder CEOs). Our results are similar across data samples. 4 To ensure that interlocked directors were present at the board during the period for which financial performance was restated, we limit our sample of restatements to those where the end of the restatement period occurred within a year of the filing date of the restatement. 5 We follow the Audit Analytics restatement classification of "adverse." These restatements classified as adverse reveal one or more among a broad set of underlying issues, commonly including debt and equity securities, cash flow classification, revenue recognition, taxes, compensation, receivables, payables, expense recording, EPS, etc. of analysis consisting of 6,302 firm × year observations (covering 832 unique firms) over the period 2001-2014. 6

| Dependent variables
New director nominations. To examine changes to patterns of new director nominations, we construct three measures. # New directors counts the number of new directors who were appointed to the board in a given year. Our next two measures are based on newly appointed directors' prior experience. # New directors w/ board exp. counts the number of new directors who were appointed to the board in a given year who possess prior board experience at a publicly traded firm. This measure takes a value of 0 if the board did not appoint any new directors or if the board appointed new directors who did not possess prior board experience. We construct a similar measure, # New directors w/ audit exp., for newly appointed directors who possess prior experience on an audit committee on the board of a public firm.
Number of financial restatements. To test for changes in patterns of financial restatements, # Restatements counts the total number of adverse financial restatements filed within each firm in our panel dataset on an annual basis, using the same Audit Analytics data used to identify restatements at peer firms. 7 To account for the time that elapses between the issuance of inaccurate financials and the subsequent filing of a restatement, we use a leading count of financial restatements by matching each firm-year observation to financial restatements (if any) filed in the subsequent year.

| Independent variables
Post-restatement period. We construct Post-restatement period based on the filing date of adverse financial restatements. We identify all boards that had an independent director who was also a director at a restating firm in the year that the restatement was filed. We categorize this set of interlocked firms as being treated by a "peer restatement event" in a given year. Postrestatement period is a dummy variable that takes a value of 1 during the 5 years following the year of the restatement and takes a value of 0 during the 5 years prior to the filing (we excluded the year of the restatement filing from our analysis). To ensure the balance of our panel, we limit our analysis to the 5 years preceding each peer restatement event and the 5 years following each peer restatement event.
Audit-nomination multi-committee directors. To create our variable # Audit-nomination MCDs, we identify, by year, each director who is a concurrent member of the board's audit committee and nomination committee. We aggregate this director-year level measure up to the firm-year level to create a total count of audit-nomination MCDs on a yearly basis for each firm.
Control variables. In each of our empirical specifications, we control for several measures that could influence new director nomination patterns. We include measures of financial performance (the natural log of annual firm revenues, Annual firm revenues [ln]; average ROA for 6 For Hypotheses 1 and 2, we excluded the firms that filed their own restatements. We relax this restriction in our analysis of future restatements for Hypothesis 3. 7 Due to the relative infrequency of financial restatements, this variable often takes a value of 0. the four-digit SIC code in which a firm operates, Industry-average ROA); the sizes of the audit, nomination, and compensation committees (Audit committee size, etc.); CEO characteristics (an indicator CEO duality for whether the CEO simultaneously holds the CEO and board chairman roles); as well as year and firm fixed effects. In our analysis of future restatements filed by the focal firm, we also control for a lagged cumulative count of financial restatements to date (Lagged restatement stock).

| Descriptive statistics
Summary statistics and univariate correlations between the key variables used in our main analysis are presented in Table 1. Although adverse financial restatements are relatively rare, being linked to a peer firm that filed an adverse restatement is quite common. Fifty-five percent of the firm-year observations in the sample for our main analyses occur following a peer restatement event (63% of firms in our full Russell 3000 sample experienced a peer restatement at some point during our study period). MCDs are highly prevalent among public firms. The average board had 1.65 MCDs linking the audit and nomination committees (in any given year, approximately 80-86% of boards had at least one audit-nomination MCD). The average board appointed 0.37 new directors with prior board experience and 0.31 new directors with audit committee experience each year. Firms in our sample are large, with average annual revenues over $5 billion. The average audit committee had 3.52 members, the average nomination committee had 3.45 members, and the average compensation committee had 3.39 members.
The absolute correlation between our two independent variables of interest (postrestatement indicator and number of audit-nomination MCDs) is 0.02, suggesting that collinearity is not a major concern in our empirical model. Correlations between the post-restatement indicator and control variables are also weak, suggesting that peer restatement events do not disproportionately affect firms of certain sizes or certain CEO types.

| Statistical methods
We focus on firms that were affected by a financial restatement at a peer firm; the effect of the peer restatement can be viewed as a plausibly exogenous shock to the firm that is not correlated with the quality of the board's governance capabilities. 8 Because we use a panel dataset that contains repeated observations from the same firm over time and because our main dependent variables (# New directors; # Restatements) are non-negative integer counts, we employ a conditional fixed effects negative binomial model (Hausman, Hall, & Griliches, 1984).
We employ a difference-in-differences approach with firm fixed effects to control for unobserved time-invariant firm-specific characteristics. We use year fixed effects to control for year-to-year changes that affect our entire sample, allowing us to identify within-firm, across-time changes to new director nominations and subsequent financial restatements. This allows us to control for many unobserved firm-level characteristics that cannot be directly 8 We exclude firms that filed their own financial restatement to preserve the identification strategy (which relies on an exogeneous shock) and to remove boards that may have taken actions as a direct response to their own governance failure.  accounted for using archival data, and thus holds constant a greater number of firm-level differences than typically employed in much prior research. We estimate bootstrapped standard errors to account for potential serial correlation between repeated observations from the same firm. 9 For our baseline analyses (testing Baseline Hypotheses A, B, C), we compare changes to new director nominations in firms that experienced a peer restatement event to firms that did not experience a peer restatement. For our main analysis (testing Hypotheses 1-3), we focus on the subset of firms that were treated by a peer restatement. We make this additional limitation to our data sample to eliminate any potential selection effects that are correlated with the likelihood that a firm is linked to a peer restatement through an interlocking director. The timing of the treatment (being affected by a peer restatement) is staggered and takes place across different years in our sample. The control group in our model is the set of boards that have no MCDs, and the treatment group is the set of boards that have one or more MCDs. An important assumption underlying difference-in-differences analyses is that there are common pre-treatment trends in the dependent variables between the treatment and control groups. To confirm that this assumption holds, we conduct several tests, which we report following the main results. We did not find any evidence of a violation of this assumption.
Our primary empirical specification for our analysis of new director nominations (Hypothesis 1) takes the following form for firm i and year t: #New directors w=board experience =β 0 +β 1 post −restatement period it + β 2 #Auditnomination MCDs it where X 0 it is a vector of control variables, γ t are year fixed effects, and η i are firm fixed effects. We compare pre-versus post-restatement numbers of newly appointed directors with prior board experience within boards with audit-nomination MCDs relative to pre-versus postrestatement numbers of newly appointed directors with prior board experience within boards without MCDs. To test Hypothesis 2, we use the same methodology but focus on a different dependent variable: the number of newly appointed directors with prior audit committee experience. Finally, to test Hypothesis 3, we use the same specification as in Hypotheses 1 and 2, but we instead focus on the number of financial restatements as the dependent variable. By comparing within-firm changes over time following a plausibly exogeneous shock, we can identify a causal effect and advance prior work that has focused on cross-sectional variation across boards. 9 We use the xtnbreg, fe command in Stata to run our conditional fixed effects negative binomial model. We estimate bootstrapped standard errors to account for potential serial correlation because this particular estimator does not allow for the calculation of robust standard errors. Due to the dispersion of our data (variance greater than the mean), a negative binomial model is more appropriate than a Poisson model, which assumes that the variance equals the mean; however, as an additional check, we run our main analyses using a conditional fixed effects Poisson model (xtpoisson, fe in Stata), which allows for the calculation of robust standard errors. Our findings are similar using this alternative estimator.

| Nominating new directors
Baseline Hypothesis A predicts that firms that experience a peer restatement will nominate more new directors to their boards. Consistent with this hypothesis, we find that in the postrestatement period, boards increased the average number of new directors appointed each year, relative to boards that were not affected by a peer restatement (Model 1 of Table 2: b = 0.127, pvalue = .016). Baseline Hypotheses B and C predict that boards will nominate fewer directors with prior board experience and fewer with prior audit committee experience, respectively. As shown in Models 2 and 3 of Table 2, we find evidence consistent with these hypotheses (Model 2 of Table 2: b = −0.199, p-value = .001; Model 3 of Table 2: b = −0.251, p-value = .001). Relative to the pre-peer-restatement period and to boards that were not affected by a peer restatement, boards were less likely to appoint directors with prior board or prior audit committee experience following a peer restatement.
Hypothesis 1 predicts that among firms that experience a peer restatement event, boards with audit-nomination MCDs will be more likely to appoint new directors with prior board T A B L E 2 New director nominations following a peer restatement event: Baseline analysis experience following the restatement. We test this hypothesis in Model 1 of Table 3. Consistent with Hypothesis 1, we find that pre-versus post-restatement, the number of new directors appointed per year with prior board experience increases when there are audit-nomination MCDs present following a peer restatement event (Model 1 of Table 3; Post-rest. × # audit-nom. MCDs; b = 0.076, p-value = .082). In terms of incidence rate, this corresponds to a 1.079-times greater rate of new director nominations with prior board experience. Hypothesis 2 predicts that boards with audit-nomination MCDs will be more likely to appoint new directors with prior audit committee experience. We find that the number of new directors with prior audit committee experience increases (Model 2 of Table 3; Post-rest. × # audit-nom. MCDs; b = 0.082, pvalue = .090) when there are audit-nomination MCDs present following a peer restatement

| Consequences for future focal firm restatements
Finally, to test Hypothesis 3, we examine whether boards with audit-nomination MCDs are less likely to file their own financial restatements after experiencing a peer restatement event. As shown in Model 3 of Table 3, boards with audit-nomination MCDs that experience a peer restatement are less likely to file their own financial restatement in the subsequent years (Postrest. × # audit-nom. MCDs; b = −0.117, p-value = .088). This corresponds to a decrease in the rate of financial statements by a factor of 0.890. We therefore find support for Hypothesis 3.

| Mediation model
We argue that by appointing new directors with prior board experience, boards can improve their monitoring capabilities and reduce the likelihood of future risks, such as financial restatements. To test this directly, we run a mediation analysis to see whether the appointment of new directors with prior experience is a pathway through which the likelihood of future restatements is reduced. While this analysis requires a different data structure from the main analysis, we find confirmatory evidence that the effect of MCDs on future restatements is partially mediated by the appointment of new directors with prior board experience and prior audit committee experience (Online Appendix A.2). Taken together, these findings lend further support to the conceptual assumptions underlying our theoretical arguments.

| Alternative specifications and other contexts
To corroborate our findings, we re-run our main analyses using an ordinary least squares estimator with firm and year fixed effects, and we find that our main findings are unchanged. Additionally, we examine whether firms endogenously add MCDs to their respective boards in response to peer restatement events, and we do not find any evidence that firms do so.
We run a falsification test where we test whether the presence of audit-compensation MCDs is associated with changes in patterns of new director nominations. We do not find systematic evidence of a relationship; however, we do find that the presence of audit-compensation MCDs following a peer restatement event is associated with a decrease in the proportion that equity 10 We find a negative relationship between the number of audit-nomination MCDs and the likelihood of appointing new directors with either prior board experience (Model 1) or prior audit committee experience (Model 2). This relationship arises mechanistically because the addition of an MCD can alleviate the need to appoint a new director to fill a committee vacancy. On average, boards with audit-nomination MCDs tend to appoint fewer new directors on an annual basis. 11 In testing the parallel pretreatment trends between the treatment and control groups (Mora & Reggio, 2012Rowley, Shipilov, & Greve, 2017), we fail to reject the test for equivalence across the different pretreatment specifications (p-values range from .31 to .63 depending on the pre-treatment period used). compensation comprises of total CEO compensation, thus tempering the likelihood of financial misconduct (Harris & Bromiley, 2007). Together, these findings suggest that board actions correspond to the specific committees that MCDs link. To test whether the presence of auditnomination MCDs is associated with other outcomes, we analyze the likelihood of subsequent lawsuits filed against the focal firm and find a negative relationship, suggesting that board committee structure could have implications for minimizing internal governance failures (e.g., financial restatements) and avoiding external issues (e.g., lawsuits). We provide details on these supplemental analyses in Online Appendix A.3-A.7.

| DISCUSSION AND CONCLUSION
Despite heightened expectations for boards in the post-SOX era, a growing body of evidence highlights how difficult it is for boards to fulfill their monitoring responsibilities (Boivie et al., 2016). As a result, many board actions are criticized for being impression management tactics meant to assuage concerns while eschewing meaningful change (Westphal & Graebner, 2010). In contrast, we offer a more contingent perspective by examining when and under what conditions boards engage in merely symbolic versus more substantive actions. By focusing our analysis at the board committee level, we find that when an external event motivates a board to take action, certain internal structures allow the board to move beyond more expedient symbolic responses and instead engage in more substantive changes. We highlight one key feature of board structure-the existence of MCDs that link board committees together-and find evidence for why their presence, when combined with a motivating shock, can result in substantive changes to committee decisions and, ultimately, a measurable impact on firm outcomes.
To isolate the structural mechanism by which certain boards choose more substantive over merely symbolic change, we rely on an established type of governance failure-a financial restatement filed by a peer firm-which has been shown in prior studies to be a significant shock that compels a variety of board responses. Expanding upon the extant impression management literature that primarily analyzes director dismissals, we focus on its converse: new director nominations. We hypothesize and find causal evidence that, while boards were generally more likely to appoint new directors following a peer restatement event, boards with auditnomination MCDs were more likely to take substantive action by appointing new directors with the necessary experience (i.e., prior board or audit committee experience) to prevent a similar governance failure from occurring in the future. Importantly, these board actions have material consequences; we find that these types of new director nominations are associated with a lower likelihood of future financial restatements and lawsuits at the focal firm.
Notably, in our extensive follow-up analyses, we sought to rigorously test our claim that structure determines action. Given that much of the research on board committees studies committees in isolation, we deliberately chose to examine committees in concert with each other and to test different MCD combinations to assess whether certain board responses were indeed contingent on specific structural configurations. Ultimately, the data passed our comprehensive examination: it is not the case that just any combination of committees will result in our hypothesized changes. Rather, these changes are contingent on which two committees are currently linked by the MCD. Furthermore, our mediation analysis suggests that these structures ultimately contributed to better firm outcomes in terms of lowered future restatement risk.
Ultimately, we chose to examine board responses to adverse financial restatements at peer firms because they represent a clear, plausibly exogeneous shock that would allow for a clean empirical test of board actions. The focus of our study sets important boundary conditions for board actions. In general, we expect that less severe external events would decrease the likelihood of a board response. Although we cannot determine the exact inflection point that triggers a response, future work can map the spectrum of shocks that compel changes to boards.

| Contributions
We seek to make three primary contributions through this study. First, we illustrate the importance of studying corporate governance at the level of board committees, which represents a fundamental shift from the vast majority of extant literature that focuses on the entire board or on individual directors as the unit of analysis. Our empirical results thus highlight the role that board committees-and MCDs as a structural feature of board committees-play in enhancing board monitoring capabilities and broader firm outcomes. These findings are bolstered by insights from our interviews and survey responses that emphasize how most board activity takes place within committees, rather than at the board level. Taken together, this suggests that focusing on board committees may speak more directly to the experiences of board directors and may also provide a valuable lens through which to study governance as a phenomenon.
Second, we contribute to the literature on impression management in corporate governance by advancing a contingent perspective on the role of board committees. We address a longstanding debate surrounding the practical relevance of the board: whether boards act in a merely symbolic or more substantive way (Westphal & Zajac, 1998). By analyzing new director nominations following an exogeneous shock, we can the identify how certain structural features of the board correspond to substantive actions that are associated with a reduction in future risks to the firm. Our findings thus help advance the impression management and corporate governance literatures by providing a test of the conditions under which boards are more likely to take more substantive-rather than merely symbolic-action.
Last, we seek to contribute to the novel field of board design. A subfield of organizational design, board design explores core features of board structure and the ways in which these structures align with the organization's goals. There are numerous tradeoffs that boards need to make when deciding on how to staff committees. Our study illustrates how the decision to have MCDs that link separate committees can facilitate substantive board actions. More broadly, this study suggests that decisions around board structure and design can have important implications for board functions and governance outcomes.

| Policy implications
Our findings have practical implications for how boards approach decisions around the structure and design of their boards. When determining board committee memberships, board leaders contend with managing board size, balancing needs for certain types of expertise, and ensuring that directors are not "overboarded," especially in light of increasing restrictions on the number of boards and number of audit committees that board members are allowed to join (Spencer Stuart, 2019). Although there are established board design best practices such as having a lead independent director, separating the CEO and board chair positions, and limiting the number of inside directors, for many boards, having MCDs (assigning individual directors to multiple committees) is not a deliberate governance decision. Rather, this arrangement often occurs more by happenstance. In lieu of this approach, we suggest that determining when and how to structure MCD director roles can-and should-be an intentional governance decision. Indeed, given demands for board accountability, several prominent boards have already recognized the strategic benefits of having MCDs that link certain committees. 12 While our main analyses focus on the effect of audit-nomination MCDs on director nominations and future restatements, our robustness checks highlight how MCDs that link other committees (e.g., audit-compensation MCDs) can have effects on other types of committee decisions (e.g., CEO compensation).
Recent studies suggest that while regulations have increased CEO accountability in the post-SOX era, additional pressures for firms to undertake substantive corrective action in the event of a governance failure come from the effect of greater media scrutiny of misconduct (Pozner, Mohliver, & Moore, 2019). This not only sharpens the risks to boards that directly engage in misconduct; risks of stigma-by-association also rise for boards connected to the misbehaving firm. Put simply, we find that appointing new directors can function as an impression management tactic, and that only on certain boards do these nominations function in a more substantive rather than merely symbolic way. Ultimately, increased demands for board accountability make it imperative that board leaders think strategically about how to design board structures, and deliberately consider whether valuable synergies can be created by having directors serve on multiple committees.
Our findings also have implications for how the nomination committee can be more strongly prioritized in board activities. A sobering insight we gleaned from one of our interviews was that nomination committee members typically receive less compensation for their committee activities than members of the audit and compensation committees. While we do not suggest that the pay differential constitutes a material concern for directors on the whole given that directors in the United States are generally compensated quite handsomely (Ryan & Wiggins, 2004), this differential could have the unintended consequence of dissuading talented directors from joining the nomination committee or from investing substantial time and energy in the committee. Especially considering our finding that the nomination committee decisions can shape long-term board composition and firm-level outcomes, a greater emphasis on the value of this committee's activities could better position boards for success. In the words of one director, the members of the nomination committee are "the ones that are driving the future" (President and CEO of professional service firm, interview, 2019).
Notably, although having more directors with audit-related experience may help decrease the likelihood of the focal firm issuing a restatement, it could also potentially increase the risk of the board becoming co-opted by directors who are predisposed to audit-related issues. Put differently, while it may be beneficial in some cases to add directors with prior audit committee experience, it is not always beneficial. Kolev et al. (2019Kolev et al. ( , p. 1177) acknowledged how board turnover could "alter committee dynamics and functioning." In our setting, if departing directors who lack audit experience are replaced with directors who are attuned to audit-related issues, the previous mix of director perspectives could become skewed in favor 12 Take for example, this statement from the 2012 Goldman Sachs proxy statement: "The overlap in membership between our Compensation Committee and our Risk Committee provides our Compensation Committee with a comprehensive picture of our firm's risk management process, which informs and assists the Compensation Committee in its review of our compensation program" (Goldman Sachs, 2012). of the audit committee. Given that individuals who are drawn to accounting careers may be more "prevention-focused" compared to others, an emphasis on risk avoidance may be helpful in some contexts, but not necessarily beneficial in others. For example, an overly riskconscious board could prove to be detrimental in advising on strategic issues for a firm in an innovation-focused, high-growth industry (Cheng & Groysberg, 2018;Wu, Gonzalez, & Wang, 2018), even if this could potentially minimize the risk of future financial restatements. Additionally, an overt focus on recruiting directors with prior board experience could also come at the expense of board diversity initiatives, given that recruiting from the pool of existing directors reinforces current board demographics and can perpetuate the underrepresentation of minority groups (Cheng, Groysberg, & Healy, 2020). Thus, extra care must be paid so that boards do not react to external shocks by overcorrecting. While monitoring is a core responsibility of the board, the board's role is multi-faceted . As such, rather than maximizing across only one governance dimension, boards need to balance across multiple dimensions.

| Limitations and future research
Many opportunities exist to advance our understanding of MCDs. Our study focused on the presence or absence of MCDs rather than the specific characteristics of MCDs. In doing so, our theory and data prohibit us from directly commenting on what features make an individual audit-nomination MCD more or less effective in their particular position. Given that the extant research on MCDs is largely inconclusive with respect to their overall efficacy (Brandes et al., 2016;Liao & Hsu, 2013), understanding variance among MCDs offers a compelling extension of our current study. In addition, our contingent approach identifies a set of conditions under which MCDs contribute positively to board decisions. However, our findings cannot rule out that there might be conditions under which MCDs may be detrimental to board performance. If, for example, MCDs are burdened by competing committee responsibilities, they may be less effective in fulfilling certain duties. Further studies in this vein would help to broaden our understanding of the effects of MCDs.
Future studies can also build on our understanding of board committees. For example, studies on group-level dynamics within committees could shed light on decision-making and director turnover and retention. Future research on specific board policies-such as the roles of committee chairs, the frequency and timing of committee meetings, and involvement of the CEO in committee activities-could advance our understanding of how committees function. Finally, especially in light of our findings on the appointment of new directors with audit committee experience, future work on the relative prevalence of different types of director expertise on boards could help us better understand how boards strike a balance between monitoring and advising activities.
In summary, while our study enhances our understanding of the role of MCDs and the implications of board committee structure, there remains more to explore. We advocate for additional studies on board design that can provide better predictive and prescriptive findings to help guide the choices of key governance leaders and aid them in making more informed decisions about how directors are appointed and how responsibilities are assigned. We hope that our work will motivate additional research in this area and establish board design as an important research stream within the broader corporate governance arena.