Traveling Blockholder Governance: Evidence from Voluntary Adoption of Clawback Provision

Transcription

1 Traveling Blockholder Governance: Evidence from Voluntary Adoption of Clawback Provision Abstract We find that firms decision to adopt clawback provisions is affected by clawback adoption behavior by firms that share common institutional blockholders, and the effects are driven by activist shareholders. Our results imply that governance travels from one firm to another through common institutional activist blockholders, and such travelling effect is independent of other channels of contagion effects bases on industry affiliation, location, and board interlock. This travelling effect is positive and causal. Further analyses show that the travelling effect is more pronounced for firms with lower board independence and past restatements, consistent with the theoretical prediction that indirect external governance substitute rather than complement internal governance. Moreover, we find that the travelling effect is stronger when product market competition is higher, suggesting that these two external governance mechanisms complement each other. Our findings suggest that institutional activist blockholders can effectively govern firms they own through a less costly indirect governance mechanism. We re-visit consequences of clawback adoption using the travelling effect as an instrument variable for clawback adoption, and find that earning quality, firm valuation, and R&D increase after clawback adoption. Keywords: CEO Compensation; Clawback; Activism; Institutional Investors

2 1. Introduction Clawback provision (compensation recovery provision) enables firms to recover incentive compensation paid to top executives based on misstated financial reports. It was initially introduced by Sarbanes-Oxley Act of 2002 and attracted considerable attention following the financial crisis of 2007/2008 when academics, policymakers, and shareholder activists called into question the structure of compensation contracts for top executives. The clawback provision aims at assuring the integrity of financial reporting by adjusting excess executive pay ex post. The proponents of clawback provision suggest that clawback discourages managerial manipulation of executive pay by reporting earnings with an upward bias, and leads to positive stock-valuation consequences, whereas the opponents argue that such a provision can add noise to the underlying performance measure, reducing managerial efforts and firm value. Through a series of debate among regulators and professionals, on July 1, 2015, SEC proposed Rule 10D-1 to implement Section 954 of the Dodd-Frank Act which will require listed companies to mandatorily adopt clawback provision. 1 The ultimate form and effective date of the rule remain uncertain. While the benefits of mandatory adoption remain to be seen, the percentage of firms in the Russell 3000 index that adopt clawback provisions has increased dramatically from 17% in 2009 to 53% in 2014, implying investors preference for such a provision. Prior studies have found mixed evidence regarding the effectiveness of internal governance (e.g., impact of boards: Addy, Chu, and Yoder, 2014; Huang, Lim, and Ng, 2015) in promoting voluntary clawback adoption. These mixed findings call for alternative explanations for the recent dramatic increase in percentage of firms voluntarily adopt clawback. As a main concern that clawback provision is reducing executives incentives to work hard, investigation of firms behave voluntarily can bring 1 1

3 insights into the potential benefits of the clawback provision. In particular, it is likely that managers will work hard and at the same time accept the disciplinary role of clawback provision if an external, instead of internal, governance factor induces the voluntariness. In this paper, we study a novel external governance mechanism (from common blockholder) on firms voluntary adoption of clawback provision. Conventional wisdom suggests that common ownership weakens governance by spreading an investor too thin. In contrast, Edmans, Levit, and Reilly (2016) model governance of multiple firms and suggest that managers of firms governed by shared ownership have greater incentive to work hard than firms governed by separated ownership where one investor owns one firm. We apply the theory by Edmans, Levit, and Reilly (2016) to investigate whether common ownership has an impact on the compensation-related governance mechanism across multiple firms. Since clawback provision was controversial but gain popularity in recent years, we investigate the power of common ownership in this area. In particular, we investigate whether firms mimic clawback adoptions by other firms that have common active institutional blockholders with them. Clifford (2009) finds that among a representative sample of U.S. public firms, 96% of them have blockholders. We focus on institutional blockholders which, unlike inside blockholders, are likely to exert external influence over the firm. Following Aggarwal, Erel, Ferreira, and Matos (2011), we use traveling to describe the spreading of good governance mechanisms. However, instead of traveling of governance from country to country, in our paper, we study traveling of governance among firms sharing common institutional activist blockholders. As we are investigating the effectiveness of governance by blockholders, we investigate whether common activist blockholders that engage in at least one investor activism activity have a stronger effect on the traveling of governance. 2

4 The empirical evidence has important implication for institutional monitoring effort. That is, if governance can travel through common ownership, then institutions can focus on serving as an activist for one firm they own and expect the effect will travel to some other firms that they own. The cost of monitoring will be much lower for the society as a whole. Our findings may also bring insights into why studies find economic benefits when firms adopt the clawback provision voluntarily. 2 A major concern of clawback provision is that it may have a negative impact on manager s incentive to work hard. If managers sense the pressure of shared ownership and intend to work hard, then adoption of clawback provision should meet less resistance from the managers; in turn, a less adverse impact on managers incentives to work hard. It is important to note that traveling of governance through common blockholder is distinct from spillover effect of governance through common industry affiliation and common location. The travelling of governance in this paper complements but also differs from board interlock effect of clawback (Addy, Chu, and Yoder, 2014). These two papers differ dramatically about the mechanism of governance. We explicitly compare these visible sources of spillover effects in our analyses and find that traveling of governance through common blockholder dominates other contagion effects. Meanwhile, we find that focal firm s clawback is responding to its peers clawback rather than other aspects of peers governance. A natural question is how the traveling governance interacts with existing governance mechanisms that are related to clawback adoption. To address this question, we conduct our analyses conditional on quality of internal governance. In our analyses, 2 A few studies have documented the economic impacts of voluntary clawback provision: Chan, Chen, Chen, and Yu (2012) document that clawback provisions help to improve accounting quality by reducing restatements, increasing earnings response coefficients, and reducing audit risk. Babenko, Bennett, Bizjak, and Coles (2015) find that executives respond to clawbacks by lowering firm risk, holding more cash, and reducing R&D. Chen, Greene, and Owers (2015) find that a clawback provision effectively lengthens the horizon of incentives and curbs misreporting but it can also reduce managerial effort and firm value. 3

5 we use two proxies for the quality of internal governance, which are (1) independence of board that is meant to perform crucial function of monitoring top management and (2) historical restatements that are often treated as triggering requirement of clawback. Ferreira, Ferreira, and Raposo (2011) find that board independence substitute the informativeness of a firm s stock price (a form of external governance). However, Cohn and Rajan (2013) argue that internal and external governance can be either substitutes or complements depending on the strength of external governance. In our setting, peer effect induced by common activist blockholders can be viewed as a form of indirect external governance as implied by Cohn and Rajan. Whether this indirect external governance should work as a substitute or complement for the internal governance is therefore an empirical question. Our empirical findings will shed light on this debate. Next, we examine the association between the traveling blockholder governance and another type of external governance, i.e., product market competition. Admati and Pfleiderer (2009) argue that impact of the threat of exit on manager decreases with investor s exit cost. Blockholders incur lower opportunity cost in exiting firms operating in a more competitive industry, because investors can easily maintain their exposure to the same industry after exiting these firms by purchasing other similar firms. Therefore, the threat of activist investors is more impactful for firms in more competitive industry. Thus, we predict a positive association between product market competition and the prevalence of the traveling blockholder governance. In our paper, we use voluntary clawback adoption data from firms in Russell 3000 index from 2009 to 2014, and identify activist blockholders using activism records obtained from 13D filings and institutional investor holding data from Thomson Reuters. We refer to institutions that have engaged in at least one activism as activists and refer to institutions that hold at least 5% of shares outstanding as blockholders. We define firms with common blockholder ownership as all firms that share at least one common 4

6 institutional activist blockholder in the prior year. We refer to firms sharing common activist blockholders as activist-portfolio members. Our findings can be summarized as following: First, we find that firms likelihood of clawback adoption is positively and significantly related to their activist-portfolio members clawback adoption decisions. For instance, a one-standard-deviation increase in the fraction of a firm s portfolio members adopting clawback corresponds to an increase in firm s likelihood to adopt clawback provisions in the next year by 12.6 percentage points. Meanwhile, we find that traveling of governance through common blockholder dominates other spillover effects through industry, location and board interlock. Furthermore, we find that activist-blockholders, which are presumably experienced in governing firms, play a more crucial role than other types of blockholders in propagating clawback adoption across firms in their portfolios. We also find that traveling of clawback can t be completely explained by traveling of board governance, implying that there is clawback-specific factors that facilitate the spillover. Second, to show that the adoption is due to traveling governance instead of omitted variables driving both, we conduct several additional analyses. A potential issue that will go against our above interpretation of empirical results is that blockholders may invest in firms with similar clawback adoption status. We mitigate the potential issue in several ways. We start with a change analysis. Consistent with the notion that clawback adoption is a reaction to traveling governance, we find that the likelihood of clawback adoption increases with the change in fraction of members adopting clawback in the prior year. Next, we examine a firm s adoption decision after its incumbent member adopts clawback for reasons exogenous to institutional holdings. We identify the clawback adoption by member firms that join the Troubled Asset Relief Program (TARP) in 2008 as exogenous to institutional holdings as these firms are required to 5

7 implement mandatory clawback provisions by Secretary of the Treasury. Consistent with our conjecture that member effect influences firm clawback adoption, we find that a firm s likelihood to adopt clawback is higher when it has members that joins TARP and adopts clawback mandatorily. Third, we test whether the traveling governance substitute or complement internal governance. We find a substitution relation between traveling governance and internal board governance in governing executive compensation policies. Using board independence as a proxy for quality of internal governance, we conduct subsample analyses for high and low board independence firms. We find a positive and significant association between the fraction of portfolio members that are clawback adopters and the likelihood of clawback adoption for low board independence firms, and an insignificant association for high board independence firms. We also find consistent results using firms historical restatements as proxy for quality of internal governance. In particular, we find that the traveling governance is stronger for firms that have restated earnings in the past. Fourth, we test whether product market competition magnify the effect of traveling governance or not. We show that traveling governance and product market competition are complements. Using Herfindahl Index of sales among firms in the same industry as the proxy for competition, we conduct subsample analyses for high/low competition group. We find that traveling governance significantly influences clawback adoption decision for firms in high competition group whereas its impact is insignificant for firms with low product market competition. For example, when we measure market competition using Fama-French 48 industry classification, the marginal effect of fraction of adopter members is 0.29 (0.15) for firms with high(low) product market competition. This finding is robust to alternative industry classifications, such as Fama-French 12 industries, four-digit SIC code, and NAICS code. 6

8 To further shed lights on the driving force of traveling governance, we conduct a few robustness tests. First, we take into account the cross-sectional relative importance of a firm s portfolio members to the firm. We calculate a weighted fraction of member clawback adoption using common activist shareholder ownership between two members, number of years two member firms are connected, number of common activist shareholders between two members, and member firm size as weight. We find a stronger association between a firm s clawback adoption decision and value-weighted fraction of clawback adoption than our baseline results based on equal-weight average. This evidence suggests that a firm is influenced more by members that have closer connections than other members that have looser connections, and by members that are larger. Next, we include firm fixed effects to control for biases caused by unobservable, firm-specific, and time-invariant omitted variables that are correlated with member effects, and we find results robust and consistent. Our results are generally consistent with the notion that traveling governance influences clawback adoption decision, and the traveling governance substitutes internal governance and complements external governance in terms of product market competition. We also extend our discussion by revisiting consequence of clawback adoption using traveling of governance as an instrumental variable for firm s voluntary clawback adoption. A key problem in prior literature to interpret the effect of clawback lies in the difficulty to infer causality mentioned by Denis (2012). We contribute to this strand of literature by advocating the usage of traveling of governance as an instrumental variable for voluntary clawback adoption. Using the instrumented clawback adoption, we confirm findings in previous literature (e.g., Chan, Chen, Chen, and Yu (2012)) that clawback adoption increases earnings quality, but in contrast to existing findings (e.g., Babenko, Bennett, Bizjak, and Coles (2015)), we find that Tobin s Q and R&D expense increases after clawback adoption. 7

9 Our study makes the following contribution. First, we contribute to corporate governance literature by documenting the effect of common activist ownership on governance. We identify member firms as firms that share common activist blockholders, and explore the implication of connected stocks in corporate governance literature 3. We are the first to provide empirical evidence that governance travels through linkage by common activist blockholders. Specifically, we find that a firm s likelihood to adopt clawbck increases with the fraction of firms in the common activist investors portfolio that have adopted clawback. Our paper provides empirical evidence supporting Edmans, Levit, and Reilly (2016) s argument about the role of common ownership in governing multiple firms. Our paper differs from existing connected stocks or peer effect studies (e.g. Leary and Roberts, 2014; Massa, Reham, and Vermaelen, 2007; Popadak; 2014; Kedia, Koh, and Rajgopal, 2015) by advocating that the governing role of activist blockholder can spread across connected firms. Second, we contribute to literature on clawback provision, a widely-adopted voluntary executive compensation policy to discipline management reporting behavior. The clawback provision has been debated among the regulators and professionals over more than a decade and it is still not mandatorily required. Understanding the determinants of clawback provision will shed lights on the desirability of such a provision from both the management and shareholder s sides. If an external force (with low cost) can successfully govern firms to adopt the clawback provision, voluntary adoption may be the best choice as not all firms need such a provision. Third, our paper also talks to recent discussion about the optimal governance in the presence of activist investor and how multiple governance mechanisms interact with 3 Anton and Polk (2014) propose the concept of connected stocks referring to stocks connected through their common active mutual fund owners and provide evidence that shared ownership forecasts cross-sectional variation in return correlation. This pioneering work motivate a few follow-on papers that examine the role of shared ownership in explaining other asset pricing patterns, such as liquidity co-movement driven by common investors (Koch, Ruenzi, and Starks, 2016). 8

10 each other. Cohn and Rajan (2013) argue that internal and external governance could be complements or substitutes. Our empirical evidence shows that the effect of external governance in the form of traveling governance is stronger with lower board independence, supporting the traditional view that external governance substitutes internal governance (e.g., Ferreira, Ferreira, and Raposo, 2011). Meanwhile, our finding that product market competition complements the traveling governance supports Admati and Pfleiderer (2009) argument that governance by shareholder is more effective with lower exit cost. The rest of the paper is organized as follows. Section 2 provides background information and develops our main hypotheses. Section 3 describes our sample and research design. Results are presented in Section 4. Section 5 presents our analyses to address endogeneity. Section 6 presents the relation between traveling governance effects with other governance mechanisms. We conduct a few robustness checks in section 7, re-visit consequences of clawback adoption in Section 8, and conclude in Section Background and hypothesis development 2.1 Background on clawback provision Clawback provision (i.e. compensation recovery provision) allows firms to recoup compensation from executives in the event of accounting restatements. It was first introduced by Section 304 of the Sarbanes-Oxley Act in 2002 (SOX 304). However, SOX 304 has only been successfully implemented in a few cases (Chan, Chen, Chen, and Yu, 2012). In the 2010 Dodd-Frank Wall Street Reform and Consumer Protection Act Section 954 (hereafter, DFA 954), which is on the recovery of erroneously awarded compensation, was included to facilitate the enforcement of clawbacks. DFA954 9

11 mandates all U.S. public companies to include a clawback provision that is triggered by any material accounting restatement. On July 1, 2015, the SEC proposed Rule 10D-1 implementing Section 954 of the Dodd-Frank Act. These rules facilitate national securities exchanges and associations (including NYSE and NASDAQ) to establish listing standards that require public companies to adopt, implement and disclose compensation clawback policies providing for recovery of excess incentive-based compensation from current and former executive officers. Although mandatory clawback has yet to be enforced, a number of firms have voluntarily adopted clawback provisions and this voluntary adoption of clawback policies attracts great attentions from researchers. One stream of literature explores the consequences of voluntary clawback adoption and documents that clawback provisions improve financial reporting quality and firm valuation (e.g., Chan, Chen, Chen, and Yu, 2012; Dehaan, Hodge, and Shevlin, 2013; Iskandar-Datta and Jia, 2013). Chan, Chen, Chen, and Yu (2012) document that firm-initiated clawback provisions help to improve accounting quality by reducing restatements, increasing earnings response coefficients, and reducing audit risk. Dehaan, Hodge, and Shevlin (2013) provide further supporting evidence that after adopting clawback provisions, firms have lower analyst forecast dispersion, higher pay-for-performance sensitivity, and higher CEO compensation. Iskandar-Datta and Jia (2013) document positive market reactions to voluntary clawback adoption news, especially for firms with past restatements. However, mixed evidence is provided on the effectiveness of clawback provision in reducing financial restatements. For instance, Babenko, Bennett, Bizjak, and Coles (2015) find no evidence on the reduction of financial restatements after the adoption of clawback provisions. Given the importance of clawback provisions in mitigating agency problems in CEO compensation, researchers start to investigate the reason behind firms voluntary adoption of clawbacks. Babenko, Bennett, Bizjak, and Coles (2015) find that firms are 10

12 more likely to adopt clawback provisions when probability of misconducts by CEO is higher or when corporate governance is stronger. Addy, Chu, and Yoder (2014) find that probability of clawback decreases in management entrenchment and increases in board interlocks by directors on the compensation committee with other clawback firms. Huang, Lim and Ng (2015) demonstrate that the likelihood of clawback adoption is negatively associated with board co-option, i.e. the fraction of board appointed the CEO assumed office. The above studies mainly focus on explaining the phenomenon of voluntary clawback adoption from the perspective of internal governance, knowledge on external driving factors behind clawback adoption decision is still limited. 2.2 Hypothesis development Shareholder activism improves targets cost efficiency or input factor, leads to capital redeployment, and modifies capital structure to reduce agency costs. We define investors as activist blockholders when they have engaged in activism in the past and hold a large block in a firm. When firms in the investment portfolio adopt clawback provisions, activist blockholders can update their knowledge on this policy change and obtain information on the cost and benefits of clawback provisions. Activist blockholders may express their preference for clawback provisions and facilitate the spillover of the information to other firms in the investment portfolio through communication with management. Therefore, we propose that common activist blockholders facilitate the spread of knowledge on clawback adoption among firms in the same investment portfolio. H1. Firms likelihood of clawback adoption increases in the incidence of clawback adoption by its member firms (fraction of members in a blockholder s portfolio that adopt clawback provisions) Influential institutional investors can exert governance effect via either persuasion or threating to exit. Some investors, such as activist blockholders, are likely to use both. Therefore, activist blockholders is more influential on firm management than 11

13 blockholders that have not engaged in activism activity. The latter is likely to influence firm management relying merely on threat to exit. Investors past activism records can increase their bargaining power with firm management team in negotiating clawback adoption policies, and thus facilitate faster propagation of clawback adoption across firms in their portfolios. H2. Traveling governance effect on clawback adoption is stronger for members connected by common activist blockholders than members connected by nonactivist blockholders. We regard the member effect arising from common activist blockholders as a form of external governance that traveling across the portfolio members. Prior research suggests that internal governance and external governance can be either substitutes or complements. For example, Cohn and Rajan (2013) argue that board governance and activism governance are complements if boards can learn from activist investors. Giroud and Muller (2011) find that G-indexes is effective only for firms in non-competitive industries, indicating that internal and external governance mechanisms are substitutes. On one hand, external governance can act as a substitute for internal board governance as weak board is ineffective in protecting against management misconduct. On the other hand, as suggested by Babenko et al. (2013) that firms with strong corporate governance are more likely to adopt clawback, external governance may act as a complement to internal governance. Therefore, it is an empirical question whether the traveling governance effect induced by common activist blockholders is a substitute or complement to internal governance mechanism. Using board independence as the proxy for internal governance, our third hypothesis is presented in an alternative format. H3a. Traveling governance effect on clawback adoption is stronger for firms with weak board independence. H3b. Traveling governance effect on clawback adoption is stronger for firms with strong board independence. 12

14 Furthermore, we study the interaction between two sources of external governance, i.e. our traveling governance and market competition. The threat of activist investor, which is the underlying driver of our traveling governance, becomes more creditable when the investor has more bargaining power over the firms. For firms in more competitive industries, their investors have more choices when they threat to replace the firm with another similar firm in the same industry. Therefore, industry competition should be positively related to the prevalence of our traveling governance. H4. The positive effect of traveling governance on clawback adoption is stronger for firms in more competitive industries. 3. Research design 3.1 Sample and data We obtain the clawback dataset from GMI. This dataset covers firm adoption of clawback provisions during the period from 2008 to Since our member variables are lagged by one year, we skip the 2008 data in our regression analyses. Our main sample starts from 2009 and ends in We extract stock price and return information from CRSP, financial accounting information from Compustat. We get restatement data from AuditAnalytics, financial analyst information from I/B/E/S, executive information from ExecuComp, board information from GMI and BoardEx, and institutional investor information from Thomson Reuters. We exclude TARP recipients from our analysis because we try to focus on voluntary clawback adoption while these firms are required by Secretary of the Treasury to implement mandatory clawback policies. 3.2 Members identified by common activist blockholders 13

15 We identify shareholder activism activities by checking 13D filings from SEC EDGAR, and we manually match institutions from Thomson Reuters to filer of 13D filings. We only retain institutions that have engaged in at least one activism prior to the year we analyse and hold at least 5% of shares outstanding in the firms. We refer institutions that satisfy these two requirements as institutional activist blockholders. For each firm i in our sample in year t, its members refer to all firms that share at least one common institutional activist blockholder with firm i in year t-1. In other words, if two firms are commonly held by the same activist investor in year t-1, and the investor s holding in each firm exceeds 5% of shares outstanding, then the two firms are treated as members to each other in year t. We refer to members sharing common activist blockholders as activist-portfolio members. 3.3 Regression specifications To examine the impact of common activist shareholders on member firms clawback adoption, we estimate the following logit regression models. Clawbacki,t = a0 + b1dpclawback i,t-1 + b2mtb i,t-1 + b3lev i,t-1 + b4roa i,t-1 + b5size i,t-1 + b6total Accruals i,t-1 + b7restatement i,t-1 + b8 Institutional Ownership i,t-1 + b9ceo Compensation i,t-1 + b10board Size i,t-1 + b11independence i,t-1 + ei,t-1 (1) Clawbacki,t = a0 + b1pclawback i,t-1 + b2mtb i,t-1 + b2mtb i,t-1 + b3lev i,t-1 + b4roa i,t-1 + b5size i,t-1 + b6total Accruals i,t-1 + b7restatement i,t-1 + b8 Institutional Ownership i,t-1 + b9ceo Compensation i,t-1 + b10board Size i,t-1 + b11independence i,t-1 + ei,t-1 (2) The dependent variable, Clawback, is an indicator variable that equals one if a firm has clawback provision in year t, and zero otherwise. The independent variable in Eq. (1), Dpclawback, is an indicator variable that equals one if a firm has member firms that adopt clawback provisions, and zero otherwise. The independent variable in Eq. (2), Pclawback, is the fraction of peer firms that have clawback provisions. If common 14

16 activist blockholders have an influence on firms clawback provision adoption, we should observe b1 > 0 for both Eq. (1) and Eq. (2). We control for several firm characteristic variables following prior studies (e.g., Addy et al. 2014; Chan et al. 2012; etc.). MTB is market-to-book ratio; LEV is leverage ratio; ROA captures a firm s accounting profitability; Size is natural logarithm of total assets; Total Accruals is the difference between income and cash flow from operation. We include Restatement, which is an indicator variable that equals one if a firm has earnings restated in the past three years, and zero otherwise, to control for the possibility that historical accounting actions would have an impact on a firm s decision to adopt clawback provisions. We include institutional ownership (Institutional ownership) to control for shareholders external monitoring effect on a firm s clawback adoption decision. We also control for several CEO and board characteristics. We include CEO compensation (CEO Compensation), total number of directors on the board (Board Size), and percentage of independent directors on the board (Independence) in our analyses. 4. Results 4.1 Summary statistics Table 1, Panel A provides the distribution of firms that have clawback provisions in our sample. Over our sample period from 2009 to 2014, we find that the number and the percentage of firms that adopt clawback provisions are increasing, which is consistent with prior research (e.g. Chan et al., 2012; Huang et al., 2015). Panel B presents the descriptive statistics of key variables used in our analyses. Clawback has a mean value of 0.329, which suggests that on average 32.9% of firms in 15

17 our sample have clawback provisions. The mean of Dpclawback is 0.962, indicating that 96.2% of firms in our sample have at least one peer firm that adopt clawback provisions. Pclawback has a mean value of 0.256, and this suggests that firms included in our sample have 25.6% of peer firms that adopt clawback provisions. As for control variables, the mean of Restatement is 0.344, suggesting that 34.4% of firms in our sample have restated earnings in the past 3 years. On average, firms have 62.5% of institutional ownership. The mean value of Independence indicates that firms on average have 70% of independent directors on the board. In Panel C of Table 1, we report the descriptive statistics of key variables for firms that adopt clawback provisions and firms that do not have clawback provisions. Our sample has 3,461 clawback observations and 7,054 non-clawback observations. We find that the percentage of peer firms that have clawback provisions is significantly higher for clawback adopters than for non-adopters. Clawback adopters has 30.4% of peer firms with clawback provisions, while non-adopters have 23.2% of peer firms with clawback provisions, and the difference is statistically significant. With respect to fundamental firm characteristics, compared with non-adopters, clawback adopters are larger (Size) and have significantly higher leverage (LEV), better earnings profitability (ROA), and higher total accruals (Total Accruals). Clawback adopters have similar market-to-book ratio (MTB) and similar likelihood of financial misstatement (Restatement) to nonadopters. The institutional ownership (Institutional Ownership) for clawback adopters is significantly higher than that for non-adopters. As for CEO and board characteristics, CEOs in firms that have clawback provisions receive significantly higher compensation (CEO Compensation). Clawback adopters are likely to have a larger board (Board Size), and more independent directors on the board (Independence). [Insert Table 1 here] 4.2 Traveling governance effect on clawback provision adoption 16

18 Table 2 provides regression results for our main hypotheses. Column (1) presents the results based on Eq. (1) and results of Eq. (2) are presented in Column (2). The coefficient of Dpclawback is positive and significant (0.741, z = 3.77), which indicates that the likelihood that member firms have clawback provisions has a positive impact on the likelihood that a firm adopt clawback provisions. The coefficient of Pclawback is significantly positive (1.318, z = 0.268), suggesting that the percentage of member firms that are clawback adopters is positively associated with the likelihood that a firm adopt clawback provisions. Taken together, these findings imply the existence of traveling governance effect on clawback provisions among firms that are connected via common activist blockholders. With regard to control variables, we find a higher likelihood of clawback provisions adoption for larger firms (Size), more profitable firms (ROA), and firms with lower accruals (Total Accruals). The positive and significant coefficient of Institutional Ownership suggests that clawback adopters have a higher level of institutional investor participation. In addition, firms that have higher CEO compensation (CEO Compensation), larger board size (Board Size), and more independent directors (Independence) on the board have a higher likelihood of clawback provisions adoption. These findings are consistent with prior studies (e.g. Abby et al., 2014; Huang et al., 2015). [Insert Table 2 here] 4.3 Traveling governance effect Controlling for other channels of adoption Besides peer pressure from members that share common activist blockholder, prior research suggests that governance practices can also travel through other channels (Chiu, Teoh, and Tian, 2013; Kedia, Koh, Rajgopal, 2015; etc). In this section, we aim at differentiating the common activist blockholder effect we document from contagion effects from other networks. We investigate three channels of contagion, which are (1) 17

19 industry peer relationship, (2) board interlocks, and (3) geographic peer relationship. Industry peers are defined as firms that share the same 2-digit SIC industry code. Members connected by board interlocks are firms that have common directors on board. Geographic peers are defined as firms that are located in the same state. To differentiate the impact of common activist blockholders from the impact of other networks on clawback adoption, we control for the clawback adoption by industry peers, board interlock peers, and geographic peers in our main regression. Results are reported in Table 3. In Panel A, we present results after controlling for industry peer effect of clawback adoption (Dpclawback_Ind and Pclawback_Ind). We find that the coefficients of Dpclawback and Pclawback are both positive and significant (0.742, z = 3.77; 1.318, z = 3.50 in the first two columns), which are similar to our main findings in Table 2. The coefficients of Dpclawback_Ind are positive but insignificant, and the coefficients of Pclawback_Ind are positive and significant. This suggests that industry peers also have an impact on a firm s clawback adoption. Results in Panel A shows that peer pressure from members with the same activist is different from peer pressure from industry competitors. Panel B reports results with contagion effect of board interlock controlled. The coefficients of Dpclawback and Pclawback remain positive and significant, after controlling for contagion effect of board interlock (Dpclawback_Board and Pclawback_Board). This suggests that the peer pressure we document is different from that from board interlocks. We find that the coefficients of Dpclawback_Board, Pclawback_Board are both positive and significant, indicating that firms that are connected by common directors exhibit similar clawback adoption policies. This finding is consistent with prior research on corporate governance s role in influencing clawback provision adoption decision. 18

20 In Panel C, we control for the impact of geographic neighbors. We find that our results remain robust after controlling for the clawback adoption by firms located in the same state (Dpclawback_State and Pclawback_State). Taken together, the results in Table 3 further support our view that the peer pressure we document is different from peer pressure through other channels of connections. [Insert Table 3 here] 4.4 Traveling governance effect Activist Blockholder vs Other Blockholder In this section, we investigate the role of activist versus non-activist blockholders in facilitating the traveling governance. To accomplish this, we extend our traveling governance analysis to another set of members that share non-activist blockholders. In light of influential role of dedicated institutional investors in governance issues documented in prior literature, we restrict our non-activist blockholders to dedicated institutional investors based on Bushee s classification. We first identify non-activist-connected members, members that share at least one common non-activist blockholders. Then, we construct two new variables for each firm based on its non-activist-connected members. Pclawback_Non-Activist is the fraction of non-activist-connected member firms that have clawback provisions. Dpclawback_Non- Activist is a dummy variable that equals to one if the firm has at least one non-activistportfolio member that adopts clawback provisions, otherwise zero. As in table 2, we include the variables constructed based on activist-portfolio members, and other control variables. In table 4, we present results of using two types of members: activist-portfolio members and non-activist-portfolio members. The dependent variable is Clawback, a dummy variable that equals one if a firm has clawback provision, otherwise zero. We find that the traveling governance effect only shows up for activist-connected members, 19

21 but not for non-activist-connected members. The coefficients of Pclawback_Non-Activist and Dpclawback_Non-Activist are positive but not significant. The marginal effect of activist-connected members is (0.128) in the first (second) regression specification, while the marginal effect of non-activist-connected members is (0.019) in the first (second) regression specification. [Insert Table 4 here] 4.5 Traveling governance effect What do Firms React to? Our previous analysis is mute on what do firms react to. In particular, since clawback adoption is closely related to board governance, the proxy we use for peer pressure (i.e., average of peers clawback adoption) may also capture variations in peers board governance. Therefore, it is unclear what do the focal firms react to, either peers governance in general or peers clawback adoption. For example, a firm may adopt clawback simply because its peer firms in general have high board independence, which also leads to a large fraction of peers adopting clawback. In this section, we analyze the source of peer pressure with a two-stage approach: we first decompose peer pressure with a regression model into a predicted component (predicted by peers board governance and other control variables) and a residual component; in the second stage, we estimate the effect of residual peer pressure on focal firm s clawback adoption. Since the residual of peer pressure is orthogonal to peers board governance by construction, we can use it to test whether focal firm is reacting to peer firms clawback-specific variations or board-governance-related variation in the second stage. In the first stage, we regress peer pressure on peer members board governance structure and other control variables. The reason why we include peer members board 20

22 governance in our regression to decompose the peer pressure is the innate relation between clawback and board: boards are responsible to implement clawback. We use the fraction of peer members that have adopted clawback (Pclawback) as our proxy for peer pressure, and run the following estimation model in the first-stage. Pclawbacki,t = a0 + b1pboardsize i,t-1 + b2pindependence i,t-1 + b3mtb i,t-1 + b4lev i,t-1 + b5roa i,t-1 + b6size i,t-1 + b7total Accruals i,t-1 + b8restatement i,t-1 + b9institutional Ownership i,t-1 + b10ceo Compensation i,t-1 + b11board Size i,t-1 + b12independence i,t-1 + ei,t-1 (3) We estimate Eq. (3) in the first stage and obtain the predicted value and residual value of Pclawback (Pclawback_Predicted and Pclawback_Residual). The control variables we use include average of peer members board size (PBoardSize i,t-1), average of peer members board independence (PIndependence i,t-1), and other characteristics of the focal firm. The predictive component of the dependent variable from Eq. (3), Pclawback_Predicted, represents the part of peer pressure driven by observable fundamentals of the firm and its peer members board governance. The residual component, Pclawback_Residual, represents the peer pressure driven by other factors unobservable to researchers (but observable to firm mangers), e.g., private communications/negotiations between the firm and institutional blockholders that are the common investors of the firm and its peer members. In the second stage, we run a logit regression and regress a firm s clawback adoption indicator on Pclawback_Predicted and Pclawback_Residual to examine their impact on a firm s clawback adoption. Clawbacki,t = a0 + b1 Pclawback_Predicted i,t-1 + b2 Pclawback_Residual i,t-1 + b3mtb i,t-1 + b4lev i,t-1 + b5roa i,t-1 + b6size i,t-1 + b7total Accruals i,t-1 + b8restatement i,t-1 + b9institutional Ownership i,t-1 + b10ceo Compensation i,t-1+ b11board Size i,t-1 + b12independence i,t-1 + ei,t-1 (4) Results are presented in Table 5. In column 1, we present the first stage results. The instrumental variables (PBoardSize i,t-1 and PIndependence i,t-1) are highly correlated 21

23 with Pclawback. In the columns 2 and 3, we present the second stage results. The predicted value of Pclawback (Pclawback_Predicted) is positively related to a firm clawback decision. The relation is both statistically and economically significant. The residual Pclawback (Pclawback_Residual) is positively associated with a firm s clawback adoption and the relation is significant. The marginal effect results in column 3 suggest that 10% increase in unexpected peer members probability to adopt clawback will increase this firm s likelihood to adopt clawback by 1.1% (=10%*0.11). This result suggests that focal firm s clawback is responding to peer firms clawback unexplained by peer firms board governance. The residual of peer firms clawback adoption is due to unobservable factors (such as private communications and negotiations between firms and institutional blockholders) which also affect focal firm s clawback. Column 3 also shows that the marginal effect of predicted Pclawback is 0.9, implying that 10% increase in expected peer members probability to adopt clawback will increase this firm s likelihood to adopt clawback by 9% (=10%*0.9). This result suggests that the positive impact of peer pressure clawback adoption works partially through focal firm responding to peers board governance and focal firms other observable characterisctics. [Insert Table 5 here] 5. Endogeneity of the travelling governance The peer pressure is endogenous because of the reflection problem mentioned by Manski (1993). In this section, we adopt two approaches to mitigate this problem. 5.1 Traveling governance effect Regression with changes in adopting members 22

24 To further support our argument of members influence on firms likelihood of clawback provision adoption, we test the incremental impact of change in clawback adoption by member firms on firms likelihood of clawback provisions adoption by estimating the following logit regression models. Clawbacki,t = a0 + b1 Dpclawback i,t-1 + b2 Dpclawback i,t-1 + b3mtb i,t-1 + b4lev i,t-1 + b5roa i,t-1 + b6size i,t-1 + b7total Accruals i,t-1 + b8restatement i,t-1 + b9institutional Ownership i,t-1 + b10ceo Compensation i,t-1 + b11board Size i,t-1 + b12independence i,t-1 + ei,t-1 (5) Clawbacki,t = a0 + b1 Pclawback i,t-1 + b2 Pclawback i,t-1 + b3mtb i,t-1 + b4lev i,t-1 + b5roa i,t-1 + b6size i,t-1 + b7total Accruals i,t-1 + b8restatement i,t-1 + b9institutional Ownership i,t-1 + b10ceo Compensation i,t-1 + b11board Size i,t-1 + b12independence i,t-1 + ei,t-1 (6) Clawbacki,t = a0 + b1 Pclawback_num i,t-1 + b2 Pclawback_num i,t-1 + b3mtb i,t-1 + b4lev i,t-1 + b5roa i,t-1 + b6size i,t-1 + b7total Accruals i,t-1 + b8restatement i,t-1 + b9institutional Ownership i,t-1 + b10ceo Compensation i,t-1 + b11board Size i,t-1 + b12independence i,t-1 + ei,t-1 (7) Dpclawback is the change in member firms likelihood of clawback adoption, which takes the value of -1 (stop clawback adoption), 0 (no change), or 1 (begin clawback adoption). Pclawback is the change in fraction of member firms that have clawback provisions. Pclawback_num is the change in number of member firms that have clawback provisions. Results are presented in Table 6. We find that the coefficient of Dpclawback is positive and significant, suggesting that the decision of clawback adoption by member firms have a positive influence on firms likelihood of adopting clawback provisions. The significantly positive coefficients of Pclawback and Pclawback_num support our argument that firms likelihood of clawback provisions adoption is affected by their member firms behaviour of clawback adoption. [Insert Table 6 here] 5.2 Traveling governance effect Members with mandatory clawback provisions 23

25 In this section, we investigate a firm s reaction to its members mandatory adoption of clawback. Firms that join the Troubled Asset Relief Program (TARP) in 2008 are required to implement mandatory clawback provisions by Secretary of the Treasury. These TARP participants clawback adoption is exogenous to other firms decision because the adoption is subject to regulation over a subset of firms that obtain inject from government in 2008 crisis. We use these mandatory adoptions as a shock to other Non-TARP firms voluntary clawback adoption. We examine how a Non-TARP firm s member connection to those TARP participants influences its own adoption decision. To accomplish above, for each Non- TARP firm in our sample, we participate all of its activist-connected members into TARP members and non-tarp members. (Non-)TARP members refer to members that (do not) join the Troubled Asset Relief Program (TARP). We construct two variables: Dpclawback_TARP which equals to one if a firm shares a common activist blockholder with at least one TARP member that have adopted clawback; pclawback_tarp, the number of TARP members with clawback scaled by number of all TARP members. Although our focus of interest is a firm s connection with its TARP members, we also control for two similar variables constructed based on other of its members. Our regression sample is restricted to Non-TARP firms, although their members could be either TARP or Non-TARP. Results are reported in Table 7. We find that connection with TARP members increases firms likelihood of clawback provision adoption. Meanwhile, we find similar results for Non-TARP members. This suggests that firms decision to adopt clawback provision is affected by their member firms behaviour, no matter whether member firms action is mandatory or voluntary. In unreported table, we obtain similar results using number of TARP members and number of TARP members with clawback as alternative measures. [Insert Table 7 here] 24

26 6. Relation with other governance mechanisms In this section, we explore relation between the traveling governance effect and other governance mechanism. We proxy incumbent internal governance with board independence and historical restatement, and we use product market competition as a proxy for external governance. We then conduct traveling governance effects analysis in sub-samples formed based on these governance mechanisms. 6.1 Traveling governance effect Relation with board independence To test the conditional impact of board independence on the traveling governance effect on clawback adoption, we partition our sample into low and high board independence groups based upon median of board independence (Independence), and compare the marginal effect of Dpclawback (Pclawback) between low and high independence groups. Table 8 presents the results of the conditional impact of board independence on the traveling governance effect we document. In Column (1) and (2), we find that the marginal effect of Dpclawback is higher for low independence group than for high independence group. As for the Pclawback, the coefficient of Pclawback is positive and significant under low board independence group, whereas the coefficient of Pclawback is positive and insignificant under high board independence group. Overall, results suggest that when board independence is lower, firms decision to adopt clawback provision is more likely to be affected by their common activist blockholder members, and this implies that for firms with weak corporate governance, external activist shareholders may have an enhanced role in monitoring. [Insert Table 8 here] 25