Corporate Governance and Financial Peer Effects

Transcription

1 Corporate Governance and Financial Peer Effects Douglas (DJ) Fairhurst * Yoonsoo Nam August 21, 2017 Abstract Growing evidence suggests that managers select financial policies partially by mimicking the financial policies of peer firms. This paper documents that the use of these peer effects in capital structure choice is more pronounced for firms operating in a weak external corporate governance environment. This finding suggests that the practice of peer mimicking is suboptimal from the perspective of shareholders. Further, cross-sectional tests suggest this finding is best explained by a quiet life hypothesis. In sum, the findings suggest that managers may be able to avoid both the cost and effort required to optimize financial policies and the scrutiny of financial market participants. Keywords: Peer Effects, Corporate Governance, Capital Structure JEL Classifications: G32, G34 * Douglas Fairhurst and Yoonsoo Nam are from the Department of Finance and Management Science, Washington State University, Pullman, WA, Douglas can be reached at dj.fairhurst@wsu.edu and Yoonsoo can be reached at yoonsoo.nam@wsu.edu. We thank Jonathan Lee and Daniel Greene for helpful comments. We are grateful to Robert Bird and John Knopf for sharing with us their state-level data on the strength of enforceability of non-competition agreements.

2 I. Introduction The financial policies of peer firms play a significant role in the selection of a firm s own financial policies. For instance, Leary and Roberts (2014) document the impact of peer firms on capital structure choice. Additionally, peer firms influence payout policy decisions (Popadak (2017)), the choice to enter the public equity markets (Kaustia and Knüpfer (2012)), and the choice a firm faces whether to split its stock (Kaustia and Rantala (2015)). However, it is unclear whether following the financial policy choices of peer firms is optimal from the perspective of shareholders. For instance, firms may act in shareholders interests by learning optimal financial policy choice through observing the financial policies of more successful peer firms. However, firms that are unable to determine their own optimal financial policy may also lack the sophistication to infer optimal financial policies by observing more successful peers. In this case, firms may incorrectly establish sub-optimal financial policies based on the observation of the policies of peer firms. However, it is empirically challenging to determine the optimality of basing financial policies on peer firms. In this study, we consider the impact of firms external corporate governance environments on the propensity to make financial policy decisions based on peer firms. Managers of firms with strong corporate governance are more likely to make decisions, including financial policy decisions which are optimal from the perspective of shareholders. If the use of peer firms financial policies in the selection of a firm s own policies is valueincreasing, we expect the presence of these peer effects to be most apparent in firms with strong corporate governance. On the other hand, managers in poorly governed firms may make financial policy choices which are detrimental to shareholders. For example, poorly governed firms tend to have lower cash holdings as managers spend the cash on self-serving projects (Harford, 1

3 Mansi, and Maxwell (2012)). Managers of firms operating under weak corporate governance may simply pursue the quiet life (Bertrand and Mullainathan (2003)) in regards to setting financial policies. Specifically, managers may maximize their own utility by minimizing the cost and effort required to select financial policies. However, shirking the responsibility to optimize financial policies creates scrutiny from financial market participants (e.g., shareholders, lenders, etc.). By mimicking peers, managers may be able to avoid the scrutiny of financial market participants. For instance, the models of Scharfstein and Stein (1990) predict that it is potentially costly to financial managers reputations to deviate from the herd. In sum, peer mimicking may avoid both the costs of optimizing policies and the scrutiny of financial market participants. Under this argument, corporate peer effects should be more pronounced for firms operating under weak governance. We use the empirical models of Leary and Roberts documenting peer influence in the selection of corporate capital structure to discriminate between these two competing hypotheses. These models test for the impact of leverage changes in peer firms resulting from idiosyncratic stock returns on an individual firm s capital structure. The use of peer firms idiosyncratic returns ensures that the effect is driven by firm-specific fluctuations in peer firm capital structure and not simply omitted observable characteristics that may impact capital structure choice at the industry level. The models developed by Leary and Roberts also provide a setting where the financial policy decision is faced by a broad cross-section of firms. As such, the methodology allows for tests of a large, pooled sample of firm-years. We first update the tests in Leary and Roberts and find that the results are robust to a more recent time period. Using these models, we then split the sample of firms into those with a weak/strong corporate governance environment to determine whether the use of peer effects to determine capital structure policy varies based on the quality of corporate governance. We initially test this prediction using the concentration of institutional 2

4 ownership as the proxy for corporate governance as institutional owners with high ownership concentration have greater incentives to monitor the firm (see, e.g., Hartzell and Starks (2003)). We find evidence of peer effects in capital structure choice only for firms with a low concentration of institutional ownership or, in other words, relatively weak corporate governance. This finding is robust to the use of either market or book leverage as the measure of capital structure and to the use of ordinary least squares (OLS) or two-stage least squares (2SLS) to the estimation procedure. The finding that peer effects are more driven by firms with relatively weak corporate governance provides preliminary evidence that mimicking peer firms financial policies is not in the best interest of shareholders. One potential concern with this finding is that the concentration of institutional ownership is an endogenous measure of governance. To address this concern, we use the takeover index developed in Cain, McKeon, and Solomon (2017) which measures the strength of the external takeover market as an additional measure of corporate governance. This measure has several advantages. For instance, the use of the external takeover market as an important part of the corporate governance environment is consistent with numerous other studies. 2 Also, the takeover index allows for the inclusion of a long sample period and broad cross-section of firms which ensures that results aren t driven by unique macroeconomic environments or a misrepresentative sub-sample of firms. Last, and most importantly, the index is based on the legal environment surrounding the firm, macro-economic conditions, and an exogenous firm-level characteristic. As such this measure likely captures an exogenous dimension of corporate governance. Using the takeover index, we continue to find that peer effects are more pronounced for firms with weak corporate governance (a low takeover index value). This finding holds 2 For example, see Bertrand and Mullainathan (1999, 2003) and Gompers, Ishii, and Metrick (2003). 3

5 regardless of the measure of leverage (market or book) or the estimation procedure (OLS or 2SLS). In other words, firms with poor corporate governance are more likely to follow idiosyncratic changes to their peers capital structure when setting their own level of leverage. This evidence is consistent with managers in poorly governed firms enjoying the quiet life by mimicking their peers financial policies and suggests that basing financial policy choices on peers is not optimal from the perspective of shareholders. We next look at a variety of cross-sectional tests to provide support for the argument that this finding is driven by managers seeking to pursue the quiet life in selecting financial policies when corporate governance is weak. If managers are truly seeking their own interests, then the results should be most pronounced when the cost to mimic and scrutiny from market participants is high. We focus on cross-sectional variation within firms with a relatively weak corporate governance as these firms are more likely to display peer effects. However, we also tabulate the results for firms with high governance also to ensure that the cross-section variation is unique to firms with weak governance. We utilize the takeover index as our measure of corporate governance for the remaining tests based on the advantages of this measure discussed above. We first consider variation in incentives for managers to maximize the value of their human capital. These incentives may at least partially offset the problems of weak external corporate governance as managers concerns about their own value may drive them to act in the best interest of shareholders regardless of the external corporate governance environment. As such, if the increased use of peer effects by firms with low threat of takeover is actually a manifestation of managers seeking the quiet life, then the effect should be most pronounced when managers have relatively low incentives to maximize the value of their human capital. 4

6 We test this prediction through the use of the enforceability of non-competition agreements. Stronger non-competition agreement enforcement increases executive stability but also reduces incentives for managers to maximize human capital (Garmaise (2011)). We measure the enforceability of non-competition agreements based on an index developed in Garmaise (2011) and extended to earlier time periods in Bird and Knopf (2015). We find capital structure peer effects only in the sub-sample of weak governance firms that also have high non-competition enforceability. This finding supports the quiet life hypothesis as managers seeking their own best interest will only optimize financial policies when they value the need to maximize their own human capital. We next consider the reputation of the manager. This concept is similar to but distinct from the human capital channel discussed above. The previous test captures cases where managers have high incentives to increase future human capital values. Here we consider how incentives change when the manager has a high level of reputation based on experience. If the use of peer effects in firms with poor external governance is evidence of managers pursuing the quiet life, then the effect should be most pronounced when managers are concerned about their reputation. This argument relies on models that predict that financial managers herd to protect their reputation despite social inefficiencies (Scharfstein and Stein (1990)). As such, we predict that peer effects in weakly governed firms will be driven by managers with high reputation as these managers have incentives to protect this reputation. We test this prediction using a measure of managerial ability developed in Demerjian, Lev, and McVay (2012). Importantly, consistent with this measure of ability also capturing managers reputations, Demerjian, Lev, and McVay document that the announcement returns for firms with departing CEOs are more negative if the CEO has a higher value for managerial ability. Consistent with the quiet life hypothesis, we find that peer mimicking in 5

7 firms operating in weak governance environments is only found for managers with relatively high reputation as proxied by managerial ability. We next consider the impact of litigation risk. To test this prediction, we exploit the staggered adoption of universal demand (UD) laws (see Lin, Liu, and Manso (2017)). In brief, these laws make it more costly for shareholders to successfully file lawsuits against firms, leading to a reduction in litigation risk. This potentially leads to two distinct effects. First, making shareholder litigation more costly through the adoption of UD laws may weaken governance by removing one governance mechanism (namely, shareholder lawsuits). Alternatively, increasing the cost of shareholder litigation in this way may protect managers from litigious shareholders who pressure managers to make sub-optimal decisions. Consistent with this pressure channel, Lin, Liu, and Manso find that introducing these laws increases long-term investment through innovation. In our setting, this pressure channel suggests that managers will make decisions to avoid litigation that might tarnish the manager s reputation. This channel predicts that firms with weak governance are more likely to base financial policies on peer firms policies in the absence of UD laws as the potential for shareholder litigation is high. We find that firms with weak external corporate governance are more likely to base their financial policies on peer firms in the absence of these laws. In other words, when shareholders are more able to place pressure on firms, poorly governed firms are more likely to mimic their peers in capital structure choice. Finally, we predict that under the quiet life hypothesis, managers are most likely to mimic peer firms when the cost to optimize financial policies is high. To proxy for this, we use the industry volatility of cash flows following Bates, Kahle, and Stulz (2009). Greater volatility of cash flows creates uncertainty regarding profitability, the ability to meet financial obligations, and other determinants of optimal financial policies making it more difficult to identify optimal financial policies. Consistent with the quiet life hypothesis, we 6

8 find that the use of peer effects in capital structure choice by firms with weak governance is more pronounced for firms in industries with relatively high cash flow volatility. In sum, our cross-sectional tests support the argument that greater use of peer effects by firms in weak governance environments is evidence of managers seeking the quiet life by avoiding both the cost of optimizing financial policies and the scrutiny that comes from the financial markets. Our paper contributes to the literature of corporate peer effects. Several papers document the presence of peer effects in the selection of various corporate policies. However, these papers often leave open the question of the optimality of the influence of peer firms from shareholders perspectives. We contribute to this literature by providing evidence that peer effects are more pronounced for firms operating under weak corporate governance. Further, this relation varies in ways consistent with the explanation that managers pursue the quiet life by avoiding the costs to optimize financial policies. Also, managers appear to avoid the scrutiny of financial market participants by mimicking the financial policies of peer firms. Another implication of our findings is that, to the extent that the practice of mimicking peer firms financial policies is sub-optimal from a shareholder perspective, it can be discouraged by strengthening the corporate governance environment that the firm operates in. II. Related Literature and Empirical Predictions A. Related Literature Economic agents partially make decisions by observing peers decisions in various settings. For instance, subjects who were asked to fill letters in envelopes displayed higher levels and lower volatility of production when working in pairs than when working alone (Falk and Ichino (2006)). In some cases, peers push economic agents to make decisions that appear to increase social welfare. For example, pineapple farmers in Ghana alter their inputs 7

9 to align with those of farmers who were surprisingly successful in previous periods (Conley and Udry (2010)). Alternatively, basing decisions on the observed actions of peers may also lead to sub-optimal decisions at times. For instance, the probability that a high school student smokes increases significantly when moving from a school where no students smoke to a school where smoking is relatively common (Powell, Tauras, and Ross (2005)). Traditionally, theoretical and empirical models of corporate financial policy choice suggest that decisions regarding these policies are only made based on firm characteristics. For instance, managers set leverage policies based on firm characteristics such as the firm s tax rate, profitability, and the proportion of assets that are fixed. 3 Recent work provides evidence that financial managers are also impacted by peer firms. For instance, survey evidence from CFOs recognizes the importance of peer firms on financing decisions (Graham and Harvey (2001)). Further, most firms report that they utilize peer firms in setting executive pay (Bizjak, Lemmon, and Naveen (2008)). Leary and Roberts (2014) provide evidence of firms leverage ratios being impacted by idiosyncratic shocks to the capital structure of their peer firms. Popadak (2017) provides additional evidence that managers base payout decisions on similar decisions of peer firms. Finally, two papers show that managers decisions in the financial markets are impacted by firms. Specifically, firms are more likely to enter the stock market if local peer firms have successfully entered the stock market (Kaustia and Knüpfer (2012)). Further, firms are also more likely to split their stock following peer firms which have done the same (Kaustia and Rantala (2015)). However, it is unclear whether basing financial policy decisions on the decisions of peer firms is optimal. The findings in Leary and Roberts are suggestive that basing capital structure decisions on peer firms is evidence of learning as less successful firms mimic their 3 For a few examples, see DeAngelo and Masulis (1980), Titman and Wessels (1988), Rajan and Zingales (1995), and Shyam-Sunder and Myers (1999). 8

10 successful peers. However, the authors conclude, an open empirical question is whether this mimicking behavior is optimal (see p. 173). While Leary and Roberts provide evidence of learning behavior, other findings suggest that little to no benefit is derived from the mimicking firm (e.g., Kaustia and Rantala (2015)). B. Empirical Predictions In sum, the optimality of basing financial decisions on the observed decisions of peer firms is ambiguous. On one hand, some managers may be able to observe the financing decisions of peer firms and glean information regarding their own target financial policies. If this is less costly than expending the resources to determine optimal policies in other ways, then managers may act in the best interest of shareholders mimicking their peers. As such, in this scenario, then mimicking behavior should be pronounced in firms where the interests of managers and shareholders are tightly aligned. Empirically, if basing financial policy decisions on the decisions of peer firms is optimal, we predict that firms in a stronger shareholder governance environment will be more likely to display mimicking behavior. On the other hand, there may be sub-optimal reasons to mimic peer firms. For instance, determining optimal financial policies requires effort on the part of managers. Selfinterested managers may prefer to enjoy the quiet life by shirking this responsibility. Yet, market participants (e.g., shareholders, analysts, creditors, etc.) may punish firms which are too far from optimal financial policies. One approach to avoiding both the effort required to optimize financial policies and the scrutiny of financial market participants is to set financial policies partially based on the observed policies of peer firms. This choice reduces the cost to select financial policies and also protects managers by reducing outside concerns over financial policies that are not aligned with industry standards. In this case, mimicking peer 9

11 firms is sub-optimal. Empirically, this quiet life hypothesis predicts that firms operating in a weak governance environment will be more likely to mimic in this case. III. Empirical Approach and Sample Description This section describes the empirical approach taken to test the predictions outlined above. Specifically, the approaches to measuring both peer effects in financial policy choice and the proxy for the corporate governance environment are described. Also, we describe the sample used throughout the paper. A. Empirical Approach To test the paper s empirical predictions, we use a methodology which allows us to observe peer mimicking behavior in the selection of capital structure policies. Further, we measure the firm s corporate governance environment multiple ways. Our approach to each is described below. 1. Measuring Peer Mimicking To measure peer mimicking behavior, we follow the methodology established in Leary and Roberts (2014). Specifically, we regress measures of firm i's leverage on characteristics of that firm which have previously been shown to impact capital structure choice, peer firm averages (excluding firm i) of these same characteristics, and industry and year fixed effects. These control variables hold constant known determinants of capital structure choice and further control for the possibility that peer firms determinants provide additional information regarding the firm s determinants through industry characteristics. To identify exogenous peer mimicking, we continue to follow Leary and Roberts by including the average of idiosyncratic shocks to the equity of peer firms. The use of idiosyncratic shocks to equity values is beneficial for two reasons. First, an increase 10

12 (decrease) in the equity of a firm mechanically decreases (increases) that firm s market value of leverage which would induce a mimicking firm to make a corresponding change. Second, because the fluctuation to the peer firms equity values are idiosyncratic, their impact on a firm s capital should only be evident if that firm is mimicking. In other words, this methodology avoids potential problems with simply using the average of leverage for peer firms (e.g., feedback problems). Idiosyncratic returns are measured as realized returns minus expected returns and are calculated each year on a rolling basis using historical monthly returns. Expected returns are calculated using up to 60 months of returns and requiring a minimum of 24 months with the estimation ending in the month prior to the beginning of the fiscal year. Expected returns are based on the market factor to account for systematic movements in equity prices. Expected returns also include an industry factor to ensure that the returns do not include information regarding industry returns that may induce leverage changes. Idiosyncratic returns are then cumulated over the fiscal year. For a more detailed description, see Leary and Roberts. Table I documents the results of the estimation procedure by tabulating the average factor loadings and adjusted R-squared values for the calculation of expected returns. A few observations are noteworthy. Specifically, our average estimated factor loadings and adjusted R-squared values are very similar to Leary and Roberts (despite an extension of the sample period). Also, the factor loadings on the excess market return and the excess industry return nearly sum to one consistent with asset pricing models. In summary, extending the sample period from Leary and Roberts has little impact on the estimation of the equity shock variable which is key to testing for peer effects. 11

13 2. Measuring Corporate Governance We use two measures for corporate governance. As our preliminary measure, we follow Hartzell and Starks (2003) and use the proportion of the institutional ownership accounted for by the top five institutional investors in the firm as a measure of corporate governance. When concentration is high, institutions have a greater incentive to monitor firms (e.g., Shleifer and Vishny (1986)). While this measure is frequently used to proxy for the governance environment, one potential concern is that this measure is endogenous. For example, Chung and Zhang (2011) provide evidence that institutional investors gravitate to stocks in a good governance environment. To address this concern of reverse causality, we use the takeover index developed in Cain, McKeon, and Solomon (2017) as our second and primary measure of corporate governance. This index is based on three factors. The first is legal determinants of hostile takeovers based on 17 takeover laws. The use of legal determinants is advantageous because it focuses on time-varying factors which impact the propensity for hostile takeovers. As such, it is not subject to the criticism garnered by individual laws that the law does not actually impact hostile takeover activity. Further, the index uses aggregate capital liquidity to ensure that bidders have access to capital to execute hostile takeovers. Last, the index takes into account firm age. The values of the index indicate the probability of being subject to a hostile takeover. As higher values of the index are correlated with a higher propensity to hostile takeover, this suggests that higher values of the index imply stronger external corporate governance. Consistent with this conjecture, Cain, McKeon, and Solomon show that this index is positively correlated with firm value. The takeover index is advantageous in answering the research question for several reasons. Specifically, the external takeover market is an important mechanism to align the 12

14 interests of managers and shareholders. In fact, in his review of corporate governance, Gillan (2006) says, In many regards the market for corporate control is the ultimate corporate governance mechanism. Further, the takeover index captures this important governance mechanism and is not subject to criticisms of other indexes meant to capture the threat of takeover. For instance, the index is available for a large cross-section of firms and for a significant time series. This feature of the data allows us to replicate the work of Leary and Roberts and provide evidence on how their findings are impacted by the governance environment faced by the firm. Also, this index measures multiple legal dimensions (as opposed to a single dimension), a factor capturing the capital market, and a firm-level characteristic. Finally, the components of the index are exogenous to managerial choice. As such, we can rule out reverse causality that may stem from managers choosing higher takeover protection as a result of their choice to mimic peer firms. Thus, the use of this index strengthens the ability to rule out endogeneity claims regarding the impact of corporate governance on the tendency to mimic peer firms. B. Sample Description Our primary data are created from the intersection of three sources. First, we draw our sample from the Compustat database for the period 1965 to Utility firms (SIC codes ), financial firms (SIC codes ), and quasi-public firms (SIC codes greater than or equal to 9000) are excluded. We merge these firm-years with data from the Center for Research in Security Prices (CRSP). To remain in the sample, each firm-year must have valid data for all analysis variables (see Appendix A). As the third source of data, we require all firm-years to have data for the takeover index. We note that this index is available through As such it determines the end of our sample period. Our sample results in 104,960 firmyears. Additional analyses use institutional holding (13f) data from the Thomson Reuters 13

15 Ownership Database, managerial characteristics, and changes to the legal environment the firm faces. However, we only remove firm-years from each analysis if the required corresponding data are not available. Table II tabulates the summary statistics for the sample. These statistics are tabulated for both firm-specific factors and peer firm averages. We also tabulate the measures of corporate governance for the full sample as well as the top and bottom tercile of each measure. 4 A few observations are notable. First, the magnitudes of the average values of firmspecific and peer firm variables are consistent with past literature. Second, the difference in the governance environment for the weak governance and strong governance firms are economically meaningful. Specifically, the top five institutions own 88.9% (0.0%) of a firm s shares for firms in the strong (weak) governance sample on average. Also, the average value of the takeover index between these two sub-samples is economically significant. Firms in the low takeover index group face a 6.6% probability of hostile takeover attempt. This probability is more than three-times larger for firms in the high takeover group (25.7%). These means suggest a meaningful difference in the corporate governance environment between the two groups. IV. Empirical Results A. The Impact of Corporate Governance on Capital Structure Peer Effects We first extend the work of Leary and Roberts (2014). Their analysis utilizes data through Using the same data sample and restrictions, we extend the sample through In untabulated results, we run regressions which mirror those in Leary and Roberts. Our regressions include controls for firm-specific factors which have been shown to affect 4 Note that we select the top and bottom tercile of each measure for our cross-sectional tests to be consistent with the cross-sectional tests in Leary and Roberts (2014). 14

16 capital structure choice (e.g., market-to-book ratio, EBITDA and Net PPE). We also include peer firm averages of these factors to control for the possibility that observing peer firm averages provides information about these firm-specific factors. Finally, we include industry fixed effects, year fixed effects, and cluster standard errors at the firm level. We find that the documented pattern of choosing leverage partially based on idiosyncratic returns to peer firms is robust to this extended time period. Further, this finding holds whether the dependent variable is market leverage or book leverage. Also, the result is robust to using the peer idiosyncratic return variable as the variable of interest in an ordinary least square (OLS) model or if it is used as an instrument for peer firms leverage in the first stage of a two-stage least squares (2SLS) model. Table III first addresses the research question of whether the use of peer firms financial policies by managers to set their own policies varies based on the corporate governance environment. Using market leverage as the dependent variable, the sub-sample in column (1) includes only firm-years with a value for institutional ownership concentration in the bottom tercile for a given year. In other words, column (1) includes firm-years for which institutional owners have a lower incentive to align the interests between shareholders and managers. In this case, consistent with the presence of peer effects, we find a significantly negative coefficient on the equity shock variable when considering market leverage. In other words, an idiosyncratic increase (decrease) in the equity value of peer firms leads to a mechanical decrease (increase) in the peer firms leverage values. The negative coefficient on the equity shock variable implies that, on average, firms in this group mimic their peers. This finding is robust to the use of book leverage, as tabulated in column (2). In contrast, the coefficients on the equity shock variable are insignificant for the firms with institutional ownership concentration in the top tercile, regardless of whether market 15

17 or book leverage is used (see columns (3) and (4)). 5 Taken together, these findings provide preliminary evidence consistent with the quiet life hypothesis which suggests that managers mimic other firms to avoid the cost of optimizing their leverage ratio. We next perform the same tests but instead use the takeover index as our proxy for corporate governance. We find evidence of peer effects for both low and high takeover index when market leverage is used as the dependent variable (see columns (5) and (7)). However, the magnitude of the coefficient is larger for the low takeover index firms. Further, when using book leverage as the dependent variable, we find evidence of peer effects for the low takeover index firms (column (6)) but not the high takeover index firms (column (8)). The results using takeover index as our measure of governance are consistent with the results using institutional ownership concentration as the measure of governance. Specifically, the observed pattern of peer mimicking in capital structure choice as documented in Leary and Roberts is more pronounced for firms operating under weak governance. To further rule out concerns of endogeneity, we next readdress the Table III regressions in a 2SLS framework. Our variable of interest in this framework is the peer average of the dependent variable (either market or book leverage). If firms based capital structure decisions on peer firms, then their leverage ratio values should be positively correlated to the peer average leverage ratio. However, this positive correlation may simply reflect firms simultaneously observing information relevant to the industry that is not captured completely by our set of control variables. To account for this endogeneity, we use the average peer firm equity shock variable as an instrument for the peer firm average leverage ratio in the first stage. Because the equity shock to peers is idiosyncratic (to both market and industry equity movements), it should only impact the firm s leverage ratio 5 This result is robust to the inclusion of the proportion of all shares owned by institutions as a control variable. 16

18 through the mechanical change in the peer firms leverage ratios. As such, this variable likely meets the exclusion restriction. Table IV documents the results of the two-stage least squares analysis. The pattern documented in Table III is similar to the pattern documented in Table IV. Specifically, regardless of the dependent variable or the measure of corporate governance, the positive correlation between a firm s leverage and the average leverage of its peer firms (after accounting for endogeneity) is more pronounced for firms operating in a weak governance environment. Further, the peer firm average equity shock has a negative and significant impact on the average peer firm leverage ratio in all models suggesting that this instrument is not a weak instrument. In sum, our findings based on OLS models are also supported by models utilizing a 2SLS approach. For the remainder of the paper, we show all results using OLS and a 2SLS approach for both market and book leverage. B. Cross-Sectional Variation in the Relation between Corporate Governance and Peer Effects To this point, we have documented that firms are more likely to base their capital structure decisions on the decisions of peer firms if they face a lesser threat of the loss of corporate control from the takeover market. This finding is consistent with the quiet life hypothesis. However, there may be other explanations for such a finding. We next consider whether this finding varies cross-sectionally in ways that suggest that the result is a manifestation of poor corporate governance driving managers to avoid the cost of optimizing financial policies. Specifically, we consider whether the observed peer effects for firms with low threat of takeover are more pronounced for 1) managers with reduced incentives to maximize their human capital, 2) managers with high reputation, 3) firms facing pressure due to high litigation risk, and 4) firms which face costly optimization of financial policies. 17

19 1. Peer Effects, Low Threat of Takeover, and Managerial Human Capital We first consider variation in incentives for managers to maximize the value of their human capital. The incentive to maximize the value of human capital may offset the problems of weak external corporate governance as managers seek to act in the best interest of shareholders regardless of the external corporate governance environment in order to maintain their marketability. As such, if the increased use of peer effects by firms with low threat of takeover is actually a manifestation of managers seeking the quiet life, then the effect should be most pronounced when managers have relatively low incentives to maximize the value of their human capital. We test this prediction through the use of the enforceability of non-competition agreements. Greater non-competition increases executive stability but also reduces incentives for managers to maximize human capital (Garmaise (2011)). We measure the enforceability of non-competition agreements based on an index developed in Garmaise (2011) and extended to earlier time periods in Bird and Knopf (2015). This index is based on the answer to 12 questions regarding the enforceability of non-competition agreements for a particular state. If the response to a particular question exceeds a given threshold, then a value of 1 is assigned to that question (and zero otherwise). The resulting index ranges from 0 to 12 with 12 being the highest enforceability of non-competition agreements. The index values vary both across states and over time. The index values are assigned to firms based on the location of their headquarters. In Panel A of Table V, we test the prediction that peer effects in weakly governed firms should be most pronounced when non-competition enforcement is high. To do so, we split firms in the bottom tercile of the takeover index into terciles of the non-competition enforceability index. We consider firms located in states with relatively low non-competition enforcement in columns (1)-(4). Regardless of the form of the dependent variable and whether 18

20 we use OLS or a 2SLS approach, we find no evidence of peer mimicking for this sub-sample of firms. In contrast, we find evidence of peer mimicking in each of the four models considering firms with low values for takeover index but relatively high non-competition enforceability in columns (5)-(8). The results suggest that managers with little incentive to maximize their own human capital are more likely to mimic peer firms when corporate governance is weak. In Panel B of Table V, we again split firms into the top and bottom tercile of noncompetition enforceability, but this time we do so for firms in the top tercile for the takeover index. This test can be viewed as a falsification test. As there is no systematic evidence of peer mimicking in these firms, we would also expect no cross-sectional variation. Consistent with this argument, we find no evidence of peer mimicking for any of the models in Panel B of Table V. 2. Peer Effects, Low Threat of Takeover, and Managerial Reputation We next focus on variation in managerial reputation. We note that this concept is related to but distinct from the incentive to maximize human capital as discussed in the previous section. Here we focus on the reputation previously accumulated by the manager. Under the quiet life hypothesis, the tendency of managers of poorly governed firms to base financial policies on peer firms policies should be most pronounced for managers with a highquality reputation. This argument is similar to one made in the model in Scharfstein and Stein (1990). In this model, it is costly for financial managers to deviate from the herd. Specifically, an incorrect decision that is different from other financial managers would damage an individual s reputation. However, an incorrect decision that is similar to that of other financial managers masks a manager s reputation from penalty. In our setting, a manager can avoid both the cost of choosing optimal financial policies and any reputational 19

21 impact by simply mimicking peers financial policies. As such, we predict that peer mimicking in weak governance firms will be most pronounced for managers with valuable reputations. Of course, managerial reputation is not an easily observable trait. We utilize the measure of managerial ability as developed in Demerjian, Lev, and McVay (2012). The intuition of their measure is relatively straight-forward. A firm s efficiency captures both firm-specific and manager-specific drivers of efficiency. The authors estimate firm efficiency and then regress this measure on firm-specific drivers of efficiency (e.g., firm size, market share, firm age, etc.). The residual, or unexplained portion of efficiency is attributed to managerial efficiency. A number of tests support the validity of this measure. For instance, the ability of managers that switch firms is better explained by managerial fixed effects than by firm fixed effects suggesting that the use of the residual term captures managerial ability and not firm ability. In addition, managers with high ability demonstrate significantly higher operating performance after being hired with the firm. Finally, the market response to an executive departure is significantly more negative for managers with high ability. This final observation is essential as it suggests that managerial ability captures an important dimension of managerial reputation. We predict that the use of peer effects by firms operating in a weak governance environment is most pronounced for firms managed by individuals with relatively high ability. We test this prediction in Panel A of Table VI. We find no evidence of peer mimicking in firms with poor corporate governance when the manager has relatively low ability as tabulated in columns (1)-(4). This is true regardless of whether market or book leverage is used as the dependent variable or whether the model is estimated using OLS or a 2SLS approach. Alternatively, regardless of the dependent variable or the estimation procedure, we find evidence of peer mimicking for firms with weak governance when managerial ability is relatively high as tabulated in columns (5)-(8). In Panel B, we confirm that there is no 20

22 cross-sectional variation in peer mimicking based on managerial ability when external governance is relatively strong. Collectively, these findings provide added support for the argument that managers of poorly governed firms pursue the quiet life by mimicking peer firms financial policies, especially when the potential cost of damage to the manager s reputation is high. 3. Peer Effects, Low Threat of Takeover, and Shareholder Litigation While managers may prefer to minimize effort in optimizing financial policies, this may increase scrutiny from financial market participants. One form of scrutiny managers might faces when shirking responsibilities to prudently manage financial policies is shareholder litigation. To capture shareholder litigation risk, we exploit the staggered adoption of universal demand (UD) laws across 23 states from 1989 through UD laws are described in detail in Lin, Liu, and Manso (2017). In sum, these laws make it more costly for shareholders to file derivative lawsuits against firms. A derivative lawsuit is a lawsuit brought against a corporate entity by an individual shareholder but on behalf of all shareholders for wrongdoing that is harmful to the entire corporate entity. Importantly, these derivative lawsuits are applicable to the mismanagement of firms. For example, a derivative lawsuit was brought against City Holding Co. because it was alleged that the executive officers and various directors did not properly manage the company, resulting in substantial charges against earnings. 6 These lawsuits typically proceed in two steps. First, the plaintiffs must request that the board takes actions to address the demands made by the plaintiffs (i.e., the demand requirement ). The board has the option to reject the demand, ignore the demand, or consider 6 See 21

23 the demand. In most cases the board rejects the demand and the court sides with the board based on the business judgement rule. However, shareholders can avoid this through the futility of the demand if they can show that the board will not present a fair decision regarding the demand. As such, plaintiff shareholders typically plead this futility exception to avoid being rejected by the board and increase the probability that the demand is reviewed in court. UD laws require that firms make the demand requirement in every derivative lawsuit for firms incorporated in the state that implements the law. In other words, UD laws remove the possibility of the futility exception and significantly reduce the likelihood of a successful derivative lawsuit. As a result, the implementation of UD laws reduces risk of shareholder litigation for firms incorporated in states that adopt these laws. Importantly, there are at least two potential ways in which a reduction in litigation risk will impact peer mimicking by firms in a weak governance environment. First, reduced litigation risk through the adoption of UD laws may amplify weak corporate governance because shareholders are less able to discipline managers not acting in their best interest. Alternatively, shareholder litigation may magnify a manager s reputational concerns leading managers to avoid any action that increase the probability of litigation. Because of this pressure from shareholders, managers may focus on their reputation when litigation risk is high, leading them to mimic peer firms and avoid reputational penalties. Lin, Liu, and Manso (2017) find that, in the setting of corporate innovation, UD laws leads to greater innovation consistent with this pressure hypothesis. In our setting, managers would be less likely to mimic peers following the adoption of UD laws as these laws provide protection for shareholder litigation. In other words, managers in firms with weak governance would be more likely to mimic if litigation risk is high. 22

24 We present the tests of this prediction in Table VII. The sample in Panel A only includes firms with relative low hostile takeover probability based on the takeover index. Note that this sample only includes firm-years from 1986 through 2008 (three years before the implementation of the first UD law and three years after the implementation of the last UD law). We match the adoption of UD laws to the firm s state of incorporation. Consistent with Lin, Liu, and Manso, we find that the pressure channel dominates following the adoption of UD laws. Specifically, we find evidence of peer mimicking for firms with weak governance that are incorporated in states that have not adopted UD laws. This is true for both market and book leverage as well as OLS and 2SLS models (see columns (1)-(4)). Further, we find no evidence of peer mimicking in three of the four models which include firms that are incorporated in states that have adopted universal demand laws. The only exception is the OLS model that uses book leverage as the dependent variable (column (6)). However, the magnitude of the peer effects in this model is lower than the magnitude for the corresponding model including firms incorporated in states that have not adopted UD laws (column (2)). We continue to find no evidence of peer mimicking for firms with relatively strong corporate governance regardless of whether the firms operate in a state that has adopted UD laws as tabulated in Panel B. In sum, peer effects by firms with weak corporate governance are most pronounced in firms that face high litigation risk. This finding supports the quiet life hypothesis and suggests that managers in weakly governed firms mimic peers to avoid reputational costs stemming from shareholder litigation. 4. Peer Effects, Low Threat of Takeover, and the Cost of Optimizing Financial Policies For our final cross-sectional test, we consider the cost of optimizing financial policies. Under the quiet life hypothesis, managers seek to minimize effort and/or costs. As such, 23

25 managers would be most likely to avoid a particular action if that action were costly. We use industry cash flow volatility as a proxy for the cost to optimize financial policies. Particularly, volatile cash flows make it difficult to forecast cash flows that may be used to service debt payments, make dividend payments, etc. which are common determinants of financial policy choice. As such, it becomes difficult to select an optimal leverage ratio. Under the quiet life hypothesis, we predict that peer mimicking by firms with relatively weak corporate governance will be more pronounced for firms with volatile cash flows. We test this prediction in Table VIII. We measure industry cash flow volatility following Bates, Kahle, and Stulz (2009). Industry cash flow volatility is calculated as the average of standard deviation of cash flow to assets for the previous 10 years with minimum of 3 years each year across each three-digit SIC code. 7 Consistent with the prediction above, we find evidence of peer effects in the selection of firms leverage ratios for firms with relatively weak corporate governance for all models that include firms with industry cash flow volatility in the top tercile (see columns (5)-(8) of Panel A). We find no evidence in any of the models of peer firm mimicking for firms with weak governance and industry cash flow volatility in the bottom tercile in one of four models (see column (1)-(4) of Panel A). Once again, Panel B documents no evidence of peer mimicking in any model where the sub-sample includes firms with relatively strong external corporate governance. These findings suggest that the propensity for firms to mimic peers when corporate governance is relatively weak is more pronounced for firms when the cost to optimize financial policies is high, consistent with the quiet life hypothesis. 7 While we use three-digit SIC codes to measure industry cash flow volatility, our results are robust to using two-digit SIC codes as in Bates, Kahle, and Stulz (2009). 24