FACULTY OF ECONOMICS AND BUSINESS
CEO/CFO inside debt and risk-taking behavior
Master thesis
MSc Accountancy & Control, variant Accountancy Student name: Cindy Lepei Lie
St. Nr.: 10183256
Supervisor: Mario Schabus Date: 22th of June, 2015 Word count: 18165
Statement of Originality
This document is written by student [Lepei Lie] who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
The role of equity-based compensation is widely examined in prior literature and it is believed that this compensation component aligns an executive’s interests with those of stockholders. However, less is known about debt-based compensation in the form of deferred compensation and pension benefits, also referred to as inside debt. A growing body of research documents that inside debt represents a substantial component of an executive’s compensation package. In addition, theory posits that inside debt holdings help align executives’ incentives with those of the firm’s debtholders by altering the executives’ risk preferences. Because empirical evidence is limited in this aspect, I investigate whether inside debt held by CEOs and CFOs is associated with less risk-taking behavior, using different measures for risk-taking. The results of my multivariate analysis reveal that inside debt held by the CFO, but not CEO, is
negatively associated with risk-taking behavior. Additionally, the robustness analysis which makes a distinction between firms with low and high default risk, shows some stronger results, indicating that CEO inside debt is negatively associated with risk-taking behavior. However, due to insignificant results in some specifications, I cannot conclude whether this reduced risk-taking is more pronounced for firms with higher default risk. The results from the multivariate analysis regarding CFO inside debt are robust to this additional test. The additional test also shows that the negative relation between CFO inside debt and risk-taking is more pronounced in firms with high default risk, consistent with my predictions.
Keywords: inside debt, pension benefits, deferred compensation, executive compensation, risk-taking behavior, chief executive officer (CEO) / chief financial officer (CFO) incentives
Acknowledgements
The goal of writing this master thesis is to obtain the Master’s degree in Accountancy & Control at the University of Amsterdam. I would like to thank my supervisor, Mario Schabus, for his continuous support, suggestions and useful feedback throughout the thesis process. His encouragement has contributed a great deal to the completion of this thesis. I would also like to thank my fellow students for their support and friendship during these years. My special thanks go to my parents for their moral and financial support throughout my years of education.
Cindy L. Lie
Table of Contents
1 Introduction ... 6
2 Literature review ... 8
2.1 Inside debt ... 8
2.2 Effects of the use of inside debt in executive compensation ... 9
2.3 CEO risk-taking behavior ... 12
2.4 CFO risk-taking behavior ... 14
3 Methodology ... 18
3.1 Sample ... 18
3.2 Variable measurement ... 19
3.2.1 CEO/CFO inside debt holdings ... 19
3.2.2 CEO risk-taking behavior ... 21
3.2.3 CFO risk-taking behavior ... 22
3.2.4 Control variables ... 23
3.3 Empirical model ... 26
4 Results ... 28
4.1 Descriptive statistics ... 28
4.2 Multivariate analysis ... 34
4.2.1 CEO inside debt and capital expenditures ... 34
4.2.2 CEO inside debt and R&D ... 36
4.2.3 CEO inside debt and diversification ... 37
4.2.4 CFO inside debt and discretionary accruals ... 39
4.2.5 CFO inside debt and leverage... 40
4.3 Robustness analysis ... 42
5 Conclusion ... 48
References ... 50
Appendices ... 53
Appendix A: Sample reconciliation and distribution – CEO sample ... 53
Appendix B: Sample reconciliation and distribution – CFO sample 1 ... 54
Appendix C: Sample reconciliation and distribution – CFO sample 2 ... 55
Appendix D: Variable description Model 1 ... 56
Appendix E: Variable description Model 2 ... 57
6 | P a g e
1 Introduction
Since Jensen and Meckling’s study (1976) about managerial behavior, agency costs and ownership structure, a large number of academic studies have emerged on executive compensation. These studies focus primarily on equity-based compensation that are paid in the form of stock options or restricted stock awards (Sundaram and Yermack, 2007).
However, recent empirical studies show that the practice of compensating executives such as chief executive officers (CEOs) and chief financial officers (CFOs) with inside debt is prevalent and often substantial (Sundaram and Yermack, 2007). Inside debt refers to the portion of executive compensation with claims that are equal to those of debt securities held by outside creditors.1 The most common forms of inside debt are defined benefit pensions and deferred compensation. The inclusion of inside debt in executive compensation packages may solve the shareholder-bondholder conflict due to differential risk-taking incentives of
debtholders and shareholders. Specifically, inside debt can possibly mitigate risk-shifting of managers by aligning incentives between shareholders and debtholders (Jensen and Meckling, 1976).
Research about inside debt in the literature has been absent due to limited disclosure requirements. In 2006, the U.S. Securities and Exchange Commission (SEC) disclosure reform required public firms to disclose detailed information about the computation and value of executive pension benefits and deferred compensation. Therefore, this disclosure reform has increased the transparency of defined pensions and deferred compensation. This rule has encouraged a number of recent studies that mainly examines the implications of CEO inside debt holdings. Wei and Yermack (2011) find that more than 80 percent of CEOs in their sample hold inside debt. Further, they find that investors respond significantly when a firm reports that the CEO has large inside debt holdings in relative to the firm’s equity investment. Under these circumstances, equity prices are inclined to fall, while bond prices are inclined to rise.
According to prior literature, larger CEO inside debt holdings result in lower promised yields and fewer restrictive covenants in loan contracts (Chava et al., 2010; Anantharaman et al., 2013). Besides, CEO inside debt holdings are also negatively related to credit default swap spreads (Bolton et al., 2010). Finally, larger CEO inside debt holdings are associated with higher cash balances but lower marginal value of cash (Liu et al., 2014). Prior literature
7 | P a g e
that examines CFO (and CEO) inside debt finds that CFOs with higher inside debt invest less in innovation and that the relation between inside debt and tax avoidance is significant for the CFO, but not for the CEO (Kubick et al., 2014).
Several studies provide empirical evidence that with the use of equity-based
compensation, managers engage more in risk-taking behavior (Guay, 1999; Coles et al., 2006; Low, 2009). This is consistent with the notion that CEOs with large equity holdings are incentivized to engage in risk-taking behavior in order to increase their own wealth since the value of the stock options and common stock they hold, is sensitive to the volatility of the stock returns (Guay, 1999). In contrast to equity-based compensation, inside debt negatively affects risk-taking because managers with inside debt holdings are subject to the same default risks and insolvency treatment as debtholders. Hence, large inside debt holdings might
incentivize managers to take less risk than shareholders may want and thus tend to align the interests of the manager with those of the debtholders of the firm (Edmans and Liu, 2011).
Empirical evidence about inside debt and its effect on the risk-taking behavior of executives is limited (Cassell et al., 2012). Therefore, this study investigates the impact of CEO/CFO inside debt on risk-taking behavior. This study makes several contributions to the existing literature. First, this study contributes to the literature that examines the effects of the different executive compensation components. This study is motivated by the demand of Edmans and Liu (2011) for more research about inside debt due to its emerging widespread presence. Recently, studies on inside debt are emerging in the academic literature (Jensen and Meckling, 1976; Sundaram and Yermack, 2007; Edmans and Liu, 2011; Cassell et al., 2012; Anantharaman et al., 2013). To date, only Cassell et al. (2012) have examined the effect of CEO inside debt on risk-taking behavior. The research that I conduct is closely related to Cassell et al.’s (2012) study, since I examine the same relation. However, this thesis extends the aforementioned study by looking at CFO inside debt as well. Specifically, I will test whether CEO inside debt is associated with risky operational/strategic decisions and CFO inside debt to risky finance/accounting decisions. Prior research on executive compensation has primarily focused on corporate (finance) decisions that are made by the CEO. However, recent studies provide evidence suggesting that the CFO has considerable influence over finance/accounting decisions whereby sophisticated financial knowledge is required (Chava and Purnanandam, 2010; Jiang et al., 2010; Kim et al., 2011; Graham and Puri, 2014).
Second, this study contributes to the literature which examines executive
compensation incentives as a determinant of managerial risk-taking behavior (Jensen and Meckling, 1976; Guay, 1999; Coles et al., 2006; Low, 2009). This study can be of interest for
8 | P a g e
shareholders and debtholders who want to know to what extent their incentives are aligned with those of the CEO/CFO. It can be relevant for policy makers who are interested in the effects of executive compensation contracts on managerial risk-taking behavior. Finally, this study can be of interest for the board’s compensation committee when considering the benefits of inside debt in CEO/CFO compensation packages.
The remainder of the thesis is organized as follows: section 2 discusses prior literature related to inside debt and risk-taking behavior of executives. Following this, my hypotheses are developed. In section 3 I explain my sample selection and further, I introduce my empirical models and the variables. Section 4 provides main results of the multivariate and robustness analysis. Finally, section 5 concludes.
2 Literature review
2.1 Inside debt
The use of a compensation plan has three main purposes (Larcker and Tayan, 2011). First, it should attract people with skills, experience and behavioral characteristics that are required to succeed in the function. Second, the level of compensation should be sufficient to retain those people in the firm. People tend to leave and work at another firm that offers a higher
compensation for their skills. Third, it should give the ‘right’ incentives so they are inclined to act in a way that is beneficial and/or appropriate to the firm. This encompasses stimulating behaviors that are in accordance with the firm’s strategy and risk profile and discouraging opportunistic behavior (Larcker and Tayan, 2011).
The executive compensation package generally consists of annual salary, annual bonus, stock option plans and restricted stock awards, defined benefit pensions and deferred compensation (Larcker and Tayan, 2011). The equity-based compensation component has been widely examined in prior literature, however less is known about the incentives of debt-based compensation such as the aforementioned defined benefit pensions and deferred
compensation. These compensation components are commonly defined as ‘inside debt’ in the academic literature (Jensen and Meckling, 1976; Sundaram and Yermeck, 2007). Inside debt holdings of the CEO represent obligations for the firm to pay its CEO fixed amounts at or after retirement as long as the firm is in a solvent condition. Because CEO inside debt
9 | P a g e
under these plans. This is why these plans are considered to be ‘debt-like’ (Sundaram and Yermack, 2007).
According to Anantharaman et al. (2013), in order for pensions and other deferred compensation plans to act as inside debt, their payoffs should be debt-based. That is, payoffs in solvent condition must be fixed and proportional to the firm’s value during bankruptcy. Secondly, to determine whether these plans are able to improve the incentives between CEOs and outside debtholders, depend on whether plan balances are superior to the claims of outside debtholders (Anantharaman et al., 2013). Anantharaman et al. (2013) examine each observed form of debt-like compensation in the United States. First, they state that executive pension plans usually include tax qualified plans that include all employees (the so-called rank-and-file or RAF plans), and supplemental executive retirement plans (SERPs) that are only meant for top executives, like CEOs. In contrast to SERPs, RAF pension plans should be funded and secured in accordance with ERISA of 1974.2 Secondly, other deferred plans involve CEOs voluntarily deferring current compensation that are withdrawn later, based on a predefined timetable. Like SERPS, ODC plans are not required to be funded and secured, but they do differ in terms of withdrawal flexibility and payoff method (Wei and Yermack, 2011).
2.2 Effects of the use of inside debt in executive compensation
According to the agency theory, a firm is the nexus of a set of contracting relationships among different parties such as managers, directors, shareholders, lenders, customers, employees, and suppliers. An agency relationship arises when one or more individuals (the principal) engage one or more individuals (the agent) to perform certain tasks on their behalf and further, delegate decision-making authority to the agent (Jensen and Meckling, 1976). Agency theory argues that individual agents (the managers) choose actions that maximize their own self-interest and the principals (the shareholders) bear the wealth consequences of those actions. This conflict of interests (or the agency problem) between managers and shareholders can result in agency costs which are borne by the shareholders. Agency costs consist of monitoring costs by the principal, bonding costs by the agent and residual loss of the principal (Jensen and Meckling, 1976).
2 The Employee Retirement Income Security Act of 1974 (ERISA) is aimed at protecting the interests of employee benefit plan participants (Anantharaman et al.,2013).
10 | P a g e
The shareholder-manager agency conflict can be attenuated in two main ways. First, managerial actions could be monitored by the shareholders themselves or these control rights are delegated to the board of directors (Denis et al., 1999). Second, by compensating
managers with equity, incentives are provided to them to take actions that benefit
shareholders. That is, equity-based compensation can encourage managers to increase stock prices in order to benefit from selling shares, which leads to a better alignment of interests between managers and shareholders (Jensen and Meckling, 1976).
When the interests of shareholders and managers are assumed to be aligned with each other, a new type of agency conflicts arises – those beween shareholders and creditors. This type of agency conflict is a result of the asymmetric payoff structure of shareholders and creditors, with shareholders having residual claims on the firm’s assets and creditors having fixed claims. Creditors care about the actions by managers that increase shareholders’ wealth while reducing the creditors’ wealth (Amstrong et al., 2010). Because the CEO/CFO is assumed to act on behalf of the shareholders, he/she tends to engage in actions that transfer wealth from debtholders to shareholders and increase the firm’s default risk, resulting in agency cost of debt for the firm. Such actions include dividend payout, claim dilution, asset substitution and underinvestment (Jensen and Smith, 1985).
Awarding CEOs with debt-based compensation represents a possible means to reduce agency costs of debt in leveraged firms (Jensen and Meckling, 1976). Respectively, Jensen and Meckling (1976) propose an optimal compensation scheme that awards the CEO with debt and equity in equal proportions. However, Edmans and Liu (2011) state that a
compensation scheme does not involve pure equity compensation nor debt and equity in equal proportions. According to them, equal proportions of debt and equity is rather inefficient. In general, equity compensation in relative to debt, leads to a greater effort, so in this case it is desired that the CEO’s equity stake is greater than his debt holding. Edmans and Liu (2011) call this ‘equity bias’. Although this could lead to more risk-taking behavior by the CEO, a greater effort will compensate this. However, if effort leads to an increase in liquidation value or bankruptcy is probable, a ‘debt bias’, whereby debt holding exceeds equity stake, is
desirable since it can improve effort and discourage risk shifting as well. Thus, Edmans and Liu (2011) advocate the use of inside debt as a superior to the agency cost of debt rather than bonuses and salaries proposed by prior studies, since its payoff (inside debt) not only depends on the risk of bankruptcy but also on the liquidation value of the firm at the occurrence of bankruptcy. Therefore, inside debt holdings could enhance the alignment of CEO incentives with those of debt holders by encouraging CEOs to engage less in risk-taking behavior.
11 | P a g e
Due to the limited disclosure requirements, research about inside debt has been absent until 2006. In 2006, a disclosure reform by the SEC required public firms to disclose detailed information about the computation and value of executive pension benefits and deferred ocmpensation. The new disclosure rule has facilitated a number of studies that investigates the implication of CEO inside debt. Wei and Yermack (2011) find that more than 80 percent of the CEOs in their sample have inside debt which amount to about circa 10 million dollar. In their study they examine stockholder and bondholder reactions to firms’ initial disclosures of CEOs’ inside debt holdings following the SEC disclosure reform. They find that when CEOs hold large pension and deferred compensation claims in relative to the firm’s equity
investment, equity prices are inclined to fall, while bond prices are inclined to rise. Wei and Yermack (2011) also find that the volatility of both securities decreases for firms whose CEOs hold large inside debt. Further, they find similar changes in the credit default swap spreads and exchange traded options. In overall, the results show a reduction in firm’s default risk, a shift of value from equity toward debt, and an overall devastation of firm’s value when CEOs have large inside debt holdings.
Several studies find a negative relation between CEO inside debt and the use of restrictive covenants in debt contracts (Anantharaman et al., 2010; Chava et al. 2010; Chen et al., 2010). Anantharaman et al. (2010) find that as CEO inside debt increases, private lenders require a lower promised yield and include fewer covenants in loan contracts. According to them, private lenders perceive inside debt as aligning CEO incentives closer to their own. Anantharaman et al. (2010) call this the ‘incentive-alignment effect’ that is driven mainly by pension benefits accrued under SERPs, in contrast to pension benefits under RAF plans or balances in other deferred compensation plans. Furthermore, promised yields are lower when the CEO can withdraw his/her compensation claim, only after expectation that outside debt is going to be settled. This finding suggests that lenders perceive the seniority of debt-like compensation also of importance, in addition to its magnitude.
Chava et al. (2010) document a negative relationship between CEO inside debt, in particular pensions, and the use of all major types of covenants (i.e. restrictions on
investments, dividend payouts, subsequent financing), except for the event-specific
restrictions such as financial distress and takeover threats. These findings suggest that large CEO pensions, relatively to their current compensation, can reduce managerial agency risk for bondholders due to the full pension payout’s sensitivity to default or bankruptcy. Chen et al. (2010) also find a negative association between CEO inside debt and the restrictiveness of debt covenants. This is consistent with the theoretical assumption that a better alignment of
12 | P a g e
the incentives between CEO and debtholder may lead to a lower cost of debt or fewer restrictive covenants. Besides, Chen et al. (2010) find that CEO inside debt is negatively associated with accounting conservatism, using various measures. This indicates that as debt-based compensation mitigates the debt-equity conflict, lenders demand less of accounting conservatism.
Lastly, Liu et al. (2014) examine the effect of CEO on firm cash holdings and the value of cash. They first examine the determinants of CEO debt compensation incentives. They find that CEO inside debt is higher in firms with poor corporate governance, which indicates that CEO inside debt results in conflicts between stockholders and bondholders and thereby affecting firm cash balances. Further, they find that higher CEO inside debt is
associated with higher firm cash balances, which is congruent with the view that CEOs are incentivized to lower risk-taking as their incentives become more aligned with those of the bondholders. This result also supports the risk aversion hypothesis which argues that CEOs with higher inside debt carry more cash to protect the value of their compensation. Finally, they find a negative relation between CEO inside debt and the marginal value of cash to shareholders.
2.3 CEO risk-taking behavior
In prior literature, executive compensation has mainly been investigated from the shareholder’s viewpoint. Agency theory assumes that, with the design of executive
compensation contracts, incentives are provided for the agent (CEO) to take actions that are in the best interests of the principal (shareholders), assuming that there is no certainty and no perfect monitoring (Jensen and Meckling, 1976).
The separation of ownership and control in large firms often causes agency problems due to differences in the incentives of principals (shareholders and debtholders) and agents (CEOs). Agency problems between CEOs and shareholders arise because CEOs carry the total cost of their efforts to generate returns for the shareholders, but they only retain a portion of the generated returns. Furthermore, because CEO wealth is usually less diversified than that of shareholders, CEOs are less likely to make risky choices relative to shareholders (Jensen and Meckling, 1976).
Several studies have examined executive compensation contract designs that mitigate these agency conflict costs. In general, empirical evidence suggests that equity-based
13 | P a g e
the shareholders (Jensen and Meckling, 1976; Guay, 1999; Coles et al., 2006; Low, 2009). Although both CEO and shareholders are risk averse, only shareholders can become risk neutral since they can eliminate the idiosyncratic risk from their portfolio by using
diversification. In contrast, CEOs cannot diversify this risk because with the use of equity-based compensation (stock and options), their wealth is tied to the value of the firm, or stock price, performance (delta) (Jensen and Meckling, 1976; Guay, 1999). A higher delta could encourage CEOs to work harder because they share gains and losses with shareholders. Therefore, in this case, delta is seen as a mechanism to align the interests of CEOs to those of shareholders. However, higher delta could expose CEOs to more firm risk. In this respect, CEOs tend to reject positive net present value (NPV) projects that are accompanied with increased firm risk (Smith and Stulz, 1985).
Guay (1999) states that the convexity of the relation between CEO wealth and stock price, in addition to the slope, should be managed to encourage CEOs to make optimal
investments and financing decisions. The convexity of this relation is the sensitivity of CEOs’ wealth to the volatility of stock return or vega. According to Guay (1999), convexity in a compensation plan creates a positive relation between a CEO’s wealth and firm risk.
Therefore, convex payoffs such as stock options and common stock can encourage risk-averse CEOs to make valuable risky investments that they may otherwise reject. Guay (1999) finds that stock options, but not common stockholdings, are related with a higher sensitivity of the CEO’s wealth to equity risk. This finding is consistent with CEOs having incentives to make risky investments in case when the possible loss of investment in risk-increasing projects is highest. Further, the stock return volatility is positively associated with the convexity
provided to CEOs, which indicates that convex incentive plans affect investing and financing decisions (Guay, 1999).
Coles et al. (2006) examine the relation between vega and investment policy, debt policy, and firm risk. In their study they find that, after controlling for delta, a higher vega gives the CEO incentives to engage in riskier activities such as more investment in research and development (R&D), less investment in property, plant, and equipment (PPE), more firm focus and higher leverage. This result is consistent with their hypothesis that higher sensitivity to stock price volatility in the CEO compensation package incentivize CEOs to invest in riskier assets and apply more aggressive debt policies. In contrast to the findings on vega, higher delta incentivize CEOs to decrease risk-taking, such as decreased R&D expenditures, increased capital expenditures and decreased leverage.
mid-14 | P a g e
1990s to examine the impact of equity-based compensation on managers’ risk-taking behavior. She finds that managerial risk aversion is a worrisome problem that causes managers to reduce firm risk at the expense of shareholder wealth. This firm risk reduction (about 6 percent) as a reaction to the regime shift, is especially found among firms that provide low equity-based incentives in their executive compensation packages, in particular firms with low CEO sensitivity to stock return volatility (low vega). This result is consistent with the standpoint that increased vega encourages managers to engage more in risk-taking behavior.
In the existing literature there is little evidence regarding the relationship between CEO inside debt and risk-taking behavior (Cassell et al., 2012). While prior studies suggest that equity-based compensation increases the CEO’s risk-taking behavior, I predict that higher CEO inside debt holdings will be associated with lower risk-taking behavior, since they improve the alignment of incentives between debtholders and the CEO. Thus, the following hypothesis tests how CEO inside debt holdings is related to risk-taking behavior, and based on prior literature a negative relation is expected:
Hypothesis 1: CEO inside debt is negatively associated with risk-taking behavior
2.4 CFO risk-taking behavior
Prior literature on executive compensation has primarily focused on corporate (finance) decisions that are made by the CEO. However, recent studies provide evidence suggesting that the CFO, compensated with equity, has considerable influence over finance and accounting decisions whereby sophisticated financial knowledge is required (Chava and Purnanandam, 2010; Jiang et al., 2010; Kim et al., 2011; Graham and Puri, 2014).
Chava and Purnanandam (2010) show that both CEOs and CFOs have significant influence over the financial policies of their firm. That is, CEOs and CFOs adopt certain financial policies based on risk-taking incentives (vega and delta) created by their stock holdings and options in the firm. In particular, Chava and Purnanandam (2010) find that CEOs with higher delta are associated with lower leverage but in case of higher vega, they take higher leverage. They do not find any association between CFOs’ incentives and leverage decision. Besides, CEOs with higher delta have more cash holdings in their firms, whereas this association reverses for CEOs with higher vega. Controlling for CEOs’
risk-15 | P a g e
taking incentives, they find that CFOs’ incentives do not have much influence on the firms’ cash balances.
Next, Chava and Purnanandam (2010) examine two other forms of corporate financial decision-making, namely the debt maturity structure3 and accrual management. In contrast to the findings with regard to leverage and cash policies, they do not find that CEOs’ incentives play an important role in the determination of the debt maturity structure. In contrast, CFOs have significant influence in this decision-making. CFOs with higher delta are associated with lower short-term debt, whereas CFOs with higher vega are associated with higher short-term debt. Chava and Purnanandam (2010) state that short-term debt causes firms to refinance their debt and consequently, expose them to interest rate risks. Therefore, short-term debt is
considered as a riskier financing decision compared to the long-term debt. Finally, they find that CFOs with higher delta engage more in income smoothing through accounting accruals, whereas CFOs with higher vega engage less in accrual management.
Jiang et al. (2010) show a positive relationship between CEO and CFO equity incentives and accrual-based earnings management. Similar to Chava and Purnanandam’s (2010) findings, their evidence suggests that CFOs’ incentives have more influence on accrual-based management than CEOs’ incentives. According to Jiang et al. (2010) this is because CFOs’ primary responsibility is preparing the financial statements, so the effect of their equity incentives on financial reporting is stronger than the effect of CEOs’ equity incentives. In addition, they state that only CFOs are skillfull enough to use accrual management as a discretionary tool, while CEOs have more oversight tasks. Chava and Purnanandam (2010) conclude that CEOs’ incentives have more influence on general corporate decisions such as capital structure and cash policies. However, decisions which require more financial expertise, such as the debt maturity structure and accruals
management, are delegated to the CFO.
Kim et al. (2011) find that the CFOs’ incentives generated from their stock options are significantly and positively associatied to future stock price crash risk. However, they find weak evidence that CEOs’ incentives lead to crash risk. This weak effect vanishes with the inclusion of CFO option incentives. Furthermore, they find no evidence that CEO/CFO stock incentives are associated with crash risk. Their results are consistent with Benmelech et al.’s (2010) expectation that managers’ equity incentives encourage them to hide bad news and increase stock price crash risk. While Benmelech et al. (2010) only focus on CEO incentives,
3 Following Barclay and Smith (1995), Chava and Purnanandam (2010) make a distinction between short-term debt and long-term debt. Debt is reclassified as short-term when its maturity is less than three years.
16 | P a g e
Kim et al. (2011) find that CFO incentives play a more important role than CEO incentives in determining future crash risk. This is consistent with Chava and Purnanandam (2010) and Jiang et al.’s (2010) results, who demonstrate that CFO incentives have more influence in cases where financial knowledge is required. Besides, they find that only option incentives are related to crash risk, not stock holdings. According to them, option holdings provide more powerful incentives to induce managerial bad news hoarding behavior because losses from such holdings are limited in case of a stock price crash. In addition, Kim et al. (2011) provide evidence that the positive relation between CFO option incentives and crash risk is more pronounced for firms with more ex ante incentives to hide risk taking. With financial leverage level as a proxy for the ex ante incentive to hide risk taking, they show that the positive
association is only present in high leveraged firms. They consider this finding as evidence that CFO option incentives are more associated with future crash risk when they have more
incentives to hide risk taking. In addition, they note that CFOs may have more skills to hide risk taking through accounting techniques.
Graham et al. (2014) examine how CEOs and CFOs share the decision-making process with regard to five corporate decisions: mergers and acquisitions (M&A), capital structure, payout policy, capital allocation and capital investment. They find evidence suggesting that the degree of delegation is not uniform for a given firm. In particular, delegation varies by the informational input demands by the CEOs. That is, for those decisions CEOs have the greatest informational advantage, in this case M&A decision-making, they are least likely to delegate this decision. On the contrary, they find CEOs delegate the most investment decisions like capital allocation and corporate investment, for which they need the most informational input. Besides, Graham et al. (2014) find that CEOs are more likely to delegate the capital structure and capital allocation decisions when they are distracted by recent multiple M&As. CEOs are less likely to delegate decisions as job tenure increases and when they have a financial background. In addition, evidence suggests that CFOs relatively have great influence over capital structure (leverage) decisions (relative to CFO influence over other decisions).
While prior literature suggests that equity-based compensation increases the CEO’s risk-taking behavior, recent studies provide evidence that CFO equity incentives also induce risk-taking behavior like accrual-based earnings management (Chava and Purnanandam, 2010; Jiang et al., 2010) and hiding bad news and increasing stock price crash risk, especially in high leveraged firms (Kim et al., 2011). In particular, they find in some cases that CFO incentives are even stronger than CEO incentives. Studies also show that CEOs are more
17 | P a g e
likely to delegate finance decisions such as debt maturity choice to the CFO when the CFO have more financial expertise (Chava and Purnanandam, 2010; Graham et al., 2014). According to prior studies, inside debt can possibly mitigate risk-shifting of managers by aligning incentives between shareholders and debtholders (Jensen and Meckling, 1976; Wei and Yermack, 2011). Sundaram and Yermack (2007) argue that managers with large inside debt are expected to pursue firm risk reducing strategies like investing less in risky projects, unlevering the capital structure, reducing dividend payouts, or increasing the average maturity of debt. The question is whether this effect applies to CFOs. Consistent with the first
hypothesis that CEO inside debt holdings decrease risk-taking behavior, the second hypothesis should show that CFO inside debt holdings induce CFO to engage less in risk-taking behavior as well. Hence, the second hypothesis can be formulated as followed:
18 | P a g e
3 Methodology
3.1 Sample
The empirical approach of my study will be archival research. Data on CEO/CFO inside debt will be retrieved from Compustat Executive Compensation (ExecuComp) database.
ExecuComp database offers data on CEO’s salary, bonus, stock holdings, stock options, defined benefit pensions and other components of total CEO pay. Only the executives that are identified as CEO/CFOs (ExecuComp item CEOANN=CEO & CFOANN=CFO) are used for the tests. Because the U.S. SEC’s disclosure requirements for defined pension benefits and deferred compensation became effective in 2006, I obtain data on CEO/CFO inside debt holdings starting from year 2006 to 2014 from ExecuComp database. Annual firm data are retrieved from the Compustat Fundamentals Annual database, which includes firms in the Standard and Poor’s 1500 index.
The initial CEO sample from ExecuComp consists of 16,293 firm-year observations. Because CEO inside debt is my variable of interest, I delete all observations that have missing values for inside debt. The remaining data are merged with annual firm data and historical segment data from Compustat, reducing the CEO sample to 7,324 firm years. After dropping missing values for equity compensation and for the calculation of the control and dependent variables,4 the CEO sample size is reduced to 2,803 firm-year observations, representing 542 unique firms from 2006-2014. Panels A-C of Table 1 (see Appendix A) summarize the sample reconciliation and distribution by year and industry.
To analyze CFO inside debt and risk-taking behavior (proxied by discretionary accruals and leverage) I use two separate samples. This is due to the different control variables I use for both tests. Although some control variables are similar, two separate samples will eliminate the likelihood that one control variable with a large number of missing values also restricts the sample for the other test in which the variable is not used. The final samples, after dropping missing values, consist of 5,014 and 2,9365 firm-year observations, respectively (see Appendix B and C).
4
Compustat Fundamentals Annual database is used to calculate my control and dependent variables.
5 The final CFO sample to test leverage has nearly the same number of observations as the CEO sample since I use the same control variables.
19 | P a g e
3.2 Variable measurement
3.2.1 CEO/CFO inside debt holdings
The independent variable of interest is CEO/CFO inside debt. According to Jensen and Meckling (1976), CEO inside debt holdings could induce CEOs to manage the firm in such a way that the agency cost of debt is reduced. They suggest, together with Edmans and Liu (2011), when the CEO’s debt-to-equity ratio reflects that of the firm (ratio is one), CEO incentives are in line with those of debt and equity holders. The higher the CEO’s debt-to-equity ratio relative to the firm’s debt-to-debt-to-equity, the more CEO incentives are aligned with debtholders than with equity holders. Thus, the CEO has more incentives to reduce firm risk (Edmans and Liu, 2011).
The first measure for the strength of CEO/CFO incentives, is constructed as CEO/CFO debt-to-equity ratio divided by firm’s debt-to-equity ratio (DER1):
CEO (CFO) to firm debt/equity ratio (DER1) =(𝐹𝐼𝑅𝑀_𝐷𝐸𝐵𝑇/𝐹𝐼𝑅𝑀_𝐸𝑄𝑈𝐼𝑇𝑌𝐶𝐸𝑂_𝐼𝐷𝐻/𝐶𝐸𝑂_𝐸𝐻 ) (1)
CEO/CFO inside debt holding (CEO_IDH) is calculated as the sum of the present value of pension benefits accumulated (CS item PENSION_VALUE_TOT) and the balance of other deferred compensation (CS item DEFER_BALANCE_TOT) as reported in Execucomp. The CEO/CFO equity holding (CEO_EH) is the sum of the fair value of both stock holdings6 and stock options by multiplying the number of shares (CS item SHROWN_EXCL_OPTS) by the stock price (PRCC_F) at the fiscal year-end. To calculate the value of the stock options, the Black-Scholes (1973) model is used. The firm’s debt-to-equity ratio
(FIRM_DEBT/FIRM_EQUITY) is calculated as total debt (CS items DLC + DLTT) scaled by the market value of equity (CS items CSHO * PRCC_F) (Edmans and Liu, 2011).
Following Sundaram and Yermack (2007), the second measure of CEO/CFO inside debt is an indicator variable (DER2) which equals one if CEO (CFO) to firm debt/equity ratio is greater than one, and zero otherwise. The expectation is when the CEO/CFO’s debt/equity ratio exceeds the firm’s debt/equity ratio, the CEO/CFO act more conservatively.
For the third measure, I follow Wei and Yermack (2011) who state that a limitation of the first measure (1) is that it only considers levels of the value of debt and equity and not
20 | P a g e
changes. Hence, Wei and Yermack (2011) propose the ‘CEO’s relative incentive ratio’ which estimates the marginal change in CEO inside debt over the marginal change in his inside equity claims, given a one-dollar increase in the overall firm value, scaled by the ratio of the marginal change in the firm’s debt over the marginal change in the firm’s equity, given the same one-dollar change in the overall firm value:
CEO (CFO) Relative incentive ratio (DER3) = (∆𝐹𝐼𝑅𝑀_𝐷𝐸𝐵𝑇/∆𝐹𝐼𝑅𝑀_𝐸𝑄𝑈𝐼𝑇𝑌 ∆𝐶𝐸𝑂_𝐼𝐷𝐻/∆𝐶𝐸𝑂_𝐸𝐻 ) (2)
Like Wei and Yermack (2011) and Cassell et al. (2012), I assume that:
∆CEO_IDH/∆FIRM_DEBT ≈ CEO_IDH/FIRM_DEBT (3)
Wei and Yermack (2011) provide two arguments for this simplifying assumption. First, since most firms in the S&P 1500 index are not financially distressed, each of ∆CEO_IDH and ∆FIRM_DEBT will be small and hard to estimate. Second, there is insufficient information in Compustat about the firm’s debt maturity structure, since U.S. firms are not mandated to disclose information about the maturity of debt with a remaining life exceeding five years. ∆CEO_EH, which is regarded as the CEO/CFO’s ‘total delta’,7
is defined as:
∆CEO_EH = S + ∑𝑖𝑁i (∆Ni) (4)
where S is the CEO/CFO’s total share delta (number of shares multiplied by an assumed delta of one) and ∑𝑖𝑁i (∆Ni) is the CEO/CFO’s total option delta, where Nis the number of options held by the CEO/CFO and ∆N is the option delta.8 To calculate the ∆𝐹𝐼𝑅𝑀_𝐸𝑄𝑈𝐼𝑇𝑌, I follow the same approach as for ∆CEO_EH, where S is the number of common shares outstanding (CS item CSHO) multiplied by an assumed delta of one; N is the number of options outstanding at end of year (CS item OPTOSEY) and ∆N is firm’s total option delta.
7 Wei and Yermack (2011) call the equation (4) the manager’s ‘total delta’ since it captures the overall slope of the manager’s equity holdings for a one-dollar change in the stock price.
8
Option delta = e–dTN (Z) * (price/100). The options are divided into tranches (e.g. unexercisable and
exerciable), indicated by the subscript i in equation (4) that have different exercise prices and times-to-maturity (Core & Guay, 2002).
21 | P a g e
3.2.2 CEO risk-taking behavior
Consistent with Coles et al. (2006), I use three measures for CEO risk-taking behavior, or the riskiness of firm investment and financial policies. The first measure is R&D expenditures and is labelled as RND, calculated as research and development expenditures scaled by total sales. R&D expenditures are considered to be high risk investments compared to other investment decisions like capital expenditures on property, plant and equipment (PPE) given the high level of uncertainty regarding their future economic benefits (Kothari et al. 2002). Thus, higher values of RND indicate more risk-taking behavior of the CEO.
The second measure is CAPEX, defined as net capital expenditures (capital
expenditures minus sales of PPE) scaled by total assets. According to Coles et al. (2006), if the CEO is incentivized to engage in risk-taking behavior, one way to do this is by
reallocating investment dollars from capital expenditures towards R&D. Thus, lower values of CAPEX, indicate lower investments in capital expenditures, so higher risk-taking CEO behavior.
Finally, CEOs can change the riskiness of their firms’ operations by altering the level of diversification. Also, managerial risk aversion is considered to be a motive for
diversification (May, 1995). A firm that is not diversified may collapse if the market for its product or service falls apart. In contrast, a diversified firm has a lower risk to go bankrupt as other operations of the firm are not affected by such event. Diversification (HHI) is measured using the sales-based Herfindahl index, calculated as the sum of the square of business
segment (CS item BUSSEG) sales divided by the square of firms’ total sales. Specifically, the Herfindahl index shows the extent of concentration of sales across the business segments. A high Herfindahl index (up to a value of one) represents a high firm focus (low diversification), while a low Herfindahl index represents high level of diversification.
22 | P a g e
3.2.3 CFO risk-taking behavior
Prior literature suggest that CFOs have considerable influence over accounting decisions such as accrual-based earnings management (Chava and Purnanandam, 2010; Jiang et al.,2010). Another stream of research documents that accrual-based earnings management is negatively associated with future firm performance, and consequently lower firm value. This may increase the likelihood of bankruptcy and reduce the firm’s liquidation value in case of bankruptcy (Bartov et al., 2002; Marquardt and Wiedman, 2004). Therefore, I choose accrual-based earnings management as a measure for CFO risk-taking behavior. Specifically, higher levels of discretionary accruals indicate more CFO risk-taking behavior. I use the modified cross-sectional Jones Model (1991) to calculate discretionary accruals (Dechow et al., 1995): TACCi,t
Ai,t-1 = α1 1 Ai,t-1 + α2
∆𝑅𝐸𝑉i,t - ∆RECi,t Ai,t-1 + α2
PPEi,t Ai,t-1 + ɛit where
TAi,t = Total accruals in year t (Income before extraordinary items (CS item IBC) – Cash flow from operations (CS item OANCF))
Ai,t-1 = Total assets in year t-1
∆𝑅𝐸𝑉i,t = Change in revenues (CS items SALEi,t – SALEi,t-1)
∆RECi,t = Change in accounts receivables (CS items RECTi,t – RECTi,t-1) PPEi,t = Gross property, plant and equipment in year t (CS item PPEGT) α1, α2, α3 = Coefficients in the model
ɛit = Firm-specific discretionary portion (DA) of TA
In addition, prior literature shows that CFOs also have considerable influence over finance decisions such as capital structure decisions (Chava and Purnanandam, 2010; Graham et al., 2014). Besides, higher leverage increases the firm’s risk (Coles et al., 2006). In this respect, the second measure for CFO risk-taking behavior is the level of debt in the firm’s capital structure or total book leverage (LEV), which is defined as the ratio of total debt (CS items DLC + DLTT) to total assets (CS item AT).
23 | P a g e
3.2.4 Control variables
3.2.4.1 CEO sample
Prior empirical research on CEO compensation has provided evidence that larger equity-based compensation, in the form of stock and option holdings, could encourage CEOs to engage in risk-taking behavior (Guay, 1999; Coles et al., 2006). However, due to their linear payoff structure, equity-based compensation could cause CEOs to become more risk-averse. Therefore, I control for the change in CEO wealth for a one percent change in stock price (DELTA) and for a 0.01 change in stock return volatility (VEGA), as high vega (delta) may encourage CEOs to make more (less) risky investment choices (Core and Guay, 2002; Coles et al., 2006).
For the calculation of the option values the one-year approximation method is used, which is based on the Black-Scholes (1973) model, modified by Merton (1973) to account for dividend payouts.
Option value = [Se –dT N(Z) – Xe –rT N(Z - σ T(1/2))] where
Z = [ln(S/X) + T(r – d + σ2/2)] / σ T (1/2)
N = cumulative probability function for the normal distribution S = price of the underlying stock (CS item PRCC_F)
X = exercise price of the option (CS item OPTPRCGR)
σ = expected stock-return volatility over the life of het option (CS item OPTVOL) r = natural logarithm of risk-free interest rate (CS item OPTRFR)
T = time to maturity of the option in years (CS item OPTLIFE)
d = natural logarithm of expected dividend yield over the life of the option (CS item OPTDR) I calculate the vega and delta as follows:
CEO delta = [∂(option value) / ∂(price)] * (price / 100) = e–dTN (Z) * (price/100) CEO vega = [∂(option value) / ∂(stock volatility)] * 0.01 = e–dTN’ (Z)ST(1/2) * (0.01), where N’ = normal density function.
The six inputs of the Black-Scholes Model that are required to estimate the value of current options grants, are available in Compustat. Only with regard to the time to maturity, I make
24 | P a g e
assumptions for current grants as well for unexercised unexercisable and unexercised
exercisable options. Following Core and Guay (2002), I assign a time to maturity of ten, nine and six for current grants, unexercisable and exercisable options, respectively.9
In addition, I estimate the exercise price for previously granted options (i.e. unexercised exercisable and unexercisable options) using the realizable values (excess of stock price over exercise price). Because unexercisable options include current grants, the number and realizable value of current grants must be deducted from the number and
realizable value of unexercisable options. Dividing the total value of unexercisable (excluding current grants) and exercisable options by the number of unexercisable and exercisable
options, shows whether each option tranche is ‘in the money’. Substracting this from the stock price generates the average exercise price of unexercisable and exercisable options.10 To calculate the value of CFOs’ option holdings, the procedure is similar as explained above.
The control variables that I further use as determinants of CEO risk-taking behavior are based on prior literature. First, I include the natural logarithm of CEO cash compensation (TOT_CUR_COMP) to control for the CEO’s outside wealth. The higher level of cash
compensation that the CEO can invest outside the firm, the more he/she is diversified and the lower the risk-aversion effect is (Guay, 1999).
Second, I include the natural logarithm of total assets to control for firm size (SIZE), since larger firms are less likely to adopt risky policy’s and/or investments (Pastor and
Veronesi, 2003; Coles et al., 2006). In particular, Low (2009) finds that doubling the firm size reduces total risk by about six percent in his sample.
Third, I use the market-to-book ratio (MTB) and sales growth (SALESGROWTH) as proxies for investment and growth opportunities (Coles et al., 2006). I include cash surplus (CASH_SURPLUS), defined as the available cash (net cash flow from operations less depreciation and amortization expenses plus R&D expenses) for financing new projects and market leverage (MKT_LEV), defined as the ratio of the firm’s total debt to equity, to account for previous finance decisions. Finally, I include industry (two-digit SIC codes) and year
9 OPTLIFE provides expected times to maturity, but since those are likely to be lower than the stated times to maturity, assumptions are made. Following Core & Guay (2002), unexercisable options have times to maturity one year less than new grants, based on the assumption that option vesting periods are approximately two years; exercisable options have an average times to maturity of three years less than unexercisable options, based on the expectation that some older grants have already been exercised. Specifically, the times to maturity of these options are set at nine and six years, respectively, since most options are granted with ten years to maturity. 10 The following formulas are used to calculate the average exercise price of (1) unexercised exercisable options = S-(OPT_UNEX_EXER_EST_VAL/OPT_UNEX_EXER_NUM) and (2) unexercised unexercisable options = S – (OPT_UNEX_UNEXER_EST_VAL/((S-X)*OPTION_AWARDS_NUM)) / (OPT_UNEX_EXER_NUM – OPTION_AWARDS_NUM)
25 | P a g e
fixed effects, and winsorize all control variables at the 1 percent and 99 percent levels to mitigate the potential impact of outliers.
3.2.4.2 CFO sample
For the test that examines the relationship between CFO inside debt holdings and risk-taking behavior (proxied by LEV), I follow a similar procedure as explained above. Thus, to
calculate the value of CFO’s option holdings, I also use the Black-Scholes Model. Besides, I use the same set of control variables as aforementioned for the CEO test. However, I
eliminate control variable MKT_LEV (debt/equity ratio) since in this case my dependent variable is the firm’s financing decision (LEV).
For the test between CFO inside debt holdings and accrual-based earnings management I use another set of control variables. First, I control for LOSS, an indicator variable that equals one if a firm reports losses, and zero otherwise. According to DeAngelo et al. (1994), troubled firms may be inclined to use income-decreasing accruals. Second, I include the natural logarithm of total assets to control for firm size (SIZE). According to the political cost hypothesis larger firms are more sensitive to political scrutiny when they report high profits. In particular, larger firms are more likely to manage their earnings downwards (Watts and Zimmerman, 1990). Next, I include an indicator variable (BIGN) that equals one if a firm is audited by a big-4 audit firm, and zero otherwise.11 Big-4 client firms are assumed to have higher financial reporting quality and are less likely to engage in earnings management. In addition, I control for the firm’s return on assets (ROA), sales growth (GROWTH) and leverage (LEV) (Cheng and Warfield, 2005).
Finally, I control for the CFO’s equity incentives using Bergstresser and Philippon’s (2006) incentive ratio. They provide evidence that in firms where the executive’s total compensation is closely tied to the value of stock and option holdings, more accrual-based earnings management is used. The incentive ratio is calculated as followed:
INCENTIVE_RATIOi,t = ONEPCTi,t / (ONEPCTi,t + SALARYi,t + BONUSi,t) 12 where
ONEPCTi,t = 0.01 * PRICEi,t * (SHARESi,t + OPTIONSi,t)13
11 CS item AU = 1 if AUCD (auditor code) is 04 (EY), 05 (Deloitte), 06 (KPMG) or 07 (PWC), and 0 otherwise. 12 Salary + bonus are directly obtained from Compustat’s Execucomp.
13
Number of shares (CS item SHROWN_EXCL_OPTS) and stock price (CS item PRCC_F) are directly obtained from Execucomp.
26 | P a g e
ONEPCTi,t is the dollar change in the value of the CFO’s stock and option holdings that comes from a 0.01 increase in the firm stock price (0.01 * share price * option delta * number of options). Following Core and Guay (2002), the Black-Scholes Model is used to calculate the option delta for current grants, unexercised exercisable options and unexercised unexercisable options. For a detailed procedure, see section 3.2.4.1.
For both tests I include industry (two-digit SIC codes) and year fixed effects, and winsorize all control variables at the 1 percent and 99 percent levels to mitigate the potential impact of outliers.
3.3 Empirical model
To test hypothesis one, I develop a model which examines the effect of CEO inside debt holdings on the risk-taking behavior of the CEO. The regression to test for the first hypothesis can be specified as followed (i represents firm and t represents time):
Model 1:
CEO Risk-taking behaviori,t = αi,t + β1*CEO Inside debti,t + β2*CEO_VEGADELTAi,t + β3*SIZE i,t + β4*MTB i,t + β5*SALESGROWTH i,t + β6*CASH_SURPLUSi,t +
β7*TOT_CUR_COMPi,t + β8*MKT_LEVi,t + Industry fixed effectsi,t + Year fixed effectsi,t + ɛi,t
The variable of interest is CEO inside debt. Consistent with the first hypothesis that CEOs with larger inside debt holdings engage less in risk-taking behavior, I expect a positive relationship between CEO inside debt and CAPEX and a negative relationship between CEO inside debt and RND and HHI. Specifically, in the first regression that tests the effect of CEO inside debt on RND, a negative β1 coefficient is expected. The second regression tests for the relationship between CEO inside debt and CAPEX, whereby a positive sign for β1 is expected. The last regression that is performed for the model, tests the relationship between CEO inside and firm diversification. For this test, a negative sign for β1 is expected.
To test hypothesis two, I develop two models that examine the effect of CFO inside debt holdings on the risk-taking behavior of the CFO. The first model tests the relationship between CFO inside debt holdings and accrual-based earnings management, which can be specified as followed (i represents firm and t represents time):
27 | P a g e
Model 2:
CFO Risk-taking behaviori,t = αi,t + β1*CFO Inside debti,t + β2*INCENTIVE_RATIOi,t + β3*LOSSi,t + β4*SIZEi,t + β5*BIGNi,t + β6*SALESGROWTHi,t + β7*LEVi,t + β8*ROAi,t + Industry fixed effectsi,t + Year fixed effectsi,t + ɛi,t
Consistent with my second hypothesis, I expect a negative coefficient on CFO Inside debt, indicating that larger inside debt holdings lead to less risk-taking proxied by the level of discretionary accruals. The next regression that tests the relationship between CFO inside debt holdings and the extent of firm’s debt can be specified as followed:
Model 3:
CFO Risk-taking behaviori,t = αi,t + β1*CFO Inside debti,t + β2*VEGAi,t + β3*DELTAi,t + β4*SIZE i,t + β5*MTBi,t + β6*SALESGROWTHi,t + β7*CASH_SURPLUSi,t +
β8*TOT_CUR_COMPi,t + Industry fixed effectsi,t + Year fixed effectsi,t + ɛi,t
The last test should demonstrate that inside debt is negatively related to the firm’s leverage (β1<0). Because leverage is my dependent variable and this is also included in the denominator of my variable of interest (inside debt ratio), it is likely that any association is caused by a mechanical relationship between the two variables. Therefore, any reported association between CFO inside debt holdings and the firm’s leverage should be interpreted with caution. For detailed variable definitions of Models 1-3, see Appendix D, E and F.
28 | P a g e
4 Results
4.1 Descriptive statistics
Panel A of Table 4 gives an overview of the descriptive statistics of the CEO risk-taking behavior measures, CEO inside debt and control variables for the CEO sample. The means (medians) of HHI and RND are positive and have values of 0.631 (0.278) and 0.043 (0.072), respectively. These values are in line with Cassell et al.’s results (2012).
With regard to my primary variable of interest, CEOs have average inside debt holdings of 8.9 million dollar. This indicates that CEOs hold substantial amount of inside debt which is consistent with prior literature (Wei and Yermack, 2011; Cassell et al., 2012).
Further, the absolute mean (median) value of CEO debt/equity ratio is 0.227 (0.122), which is similar to the reported value 0.22 (0.15) by Wei and Yermack et al. (2011). The mean (median) of 0.227 (0.122) suggests that the average CEO has more equity holdings than inside debt holdings. Besides, the average CEO_DER1 ratio is 0.88 which is consistent with reported means in prior literature (Cassell et al., 2012; Kubick et al., 2014; Liu et al. 2014). Jensen and Meckling (1976) state that the CEO_DER1 should equal one to create balance between the incentives of equity holders and debtholders. However, Edmans and Liu (2011) argue that the optimal CEO debt/equity ratio should be smaller than the firm’s debt/equity ratio, because equity compensation should be relatively more to increase the CEO’s effort. Consistent with this statement I find that only 29 percent of all firm-year observations have a CEO_DER1 above one.
29 | P a g e
Table 4: Descriptive statistics
Panel A
Variables N Mean St. Dev. 25% Median 75%
CAPEX 2803 .0407561 .032258 .0193604 .0318461 .0510698 HHI 2803 .6314996 .2783103 .3718298 .5561001 1 RND 2803 .0430827 .0722542 .0031897 .0191742 .0493074 CEO_DER1 2803 .8801747 1.130855 .1800533 .5439702 1.159791 CEO_DER2 2803 .2932572 .4553366 0 0 1 CEO_DER3 2803 2.682998 1.946369 1.133126 2.470052 3.836478
CEO inside debt (tsd) 2803 8936.441 17004.28 786.238 3300.927 9965.959 CEO debt/equity ratio 2803 .2268653 .3252296 .038575 .1221126 .29838
VEGA (tsd) 2803 275.6327 338.7857 43.85099 147.7814 367.6951 DELTA (tsd) 2803 368.682 474.4577 60.1958 196.2204 474.3774 Current compensation (tsd) 2803 1114.458 850.6553 750 950 1173.077 CEO_VEGADELTA 2893 .4303349 .9214666 .0260636 .1051471 .3449392 SIZE 2803 8.189623 1.410288 7.119716 8.09013 9.10731 MTB 2803 2.897398 5.751207 1.609472 2.356018 3.60211 SALESGROWTH 2803 .0654352 .18245 -.0099071 .0633034 .1307935 CASH_SURPLUS 2803 .0922281 .0859002 .0461471 .0836499 .1301655 TOT_CUR_COMP 2803 6.87274 .4773276 6.620.073 6.856462 7.067386 MKT_LEV 2803 .4267137 1.057611 .097394 .1986313 .3973357
Panel B of Table 4 provides descriptive statistics for the variables used in my analysis of the relationship between CFO inside debt and the level of discretionary accruals (DA). The mean of DA is negative which is an indication of downward (income decreasing) earnings
management (Cohen et al., 2008). The absolute mean value of CFO inside debt is 2.14 million dollar, suggesting that inside debt holdings are substantial. CFO debt/equity ratio is 42
percent on average, which indicates that the average CFO has more equity holdings than inside debt holdings. The mean of CFO_DER2 indicates that 27 percent of all firm-year observations has a CFO debt/equity ratio that exceeds the firm’s debt/equity (CFO_DER1>1), which is consistent with prior literature (Kubick et al., 2014). Further, the mean of
INCENTIVE_RATIO is 0.097, which is comparable to the one reported by Kim et al. (2011). The firms in the sample are quite diverse in terms of (mean and median of Total assets are 14442 and 4023 million dollar , respectively). About 98 percent of all firms are audited by the big-four auditors and only 14 percent of the firms has net losses.
30 | P a g e
Panel B
Variables N Mean St. Dev. 25% Median 75%
DA 5014 -.1155216 .123845 -.1674653 -.0955866 -.0466817
CFO_DER1 5014 .8051512 .9928414 .1601574 .4809259 1.06841
CFO_DER2 5014 .27144 .4447469 0 0 1
CFO_DER3 5014 1.735442 1.506998 .5777175 1.389031 2.48375
CFO inside debt (tsd) 5014 2140.246 3734.808 205.259 719.299 2354.995 CFO debt/equity ratio 5014 .4229414 1.625725 .0484962 .1501428 .3907681 Total assets (mln) 5014 14442.33 41839.22 1603.949 4023.219 12000.14 INCENTIVE_RATIO 5014 .0971474 .0973573 .0285032 .0659247 .1312977 SALESGROWTH 5014 .0528698 .1689018 -.0237433 .0502104 .121033 SIZE 5014 8.412185 1.426739 7.380224 8.299837 9.392673 BIGN 5014 .9790586 .1432024 1 1 1 LEV 5014 .2624729 .1622067 .1496885 .2460452 .3516156 ROA 5014 .0548678 .0721483 .0253834 .0562892 .0922796 LOSS 5014 .1372158 .3441093 0 0 0
Panel C of Table 4 provides descriptive statistics for the variables used in my analysis of the relationship between CFO inside debt and the firm’s leverage. The mean of LEV is 0.25 which is lower than the 0.62 reported by Cassell et al. (2012) but similar with Coles et al. (2006). The mean of DELTA is higher than VEGA (88>75), what is in line with results of prior literature (Coles et al., 2006; Cassell et al., 2012; Kubick et al., 2014; Srivastav et al., 2014). The higher value of delta means that the CFO’s sensitivity to one percent change in stock price, is stronger than a 0.01 change in stock-return volatility.
Panel C
Variables N Mean St. Dev. 25% Median 75%
LEV 2936 .2465601 .1544918 .1397587 .2306547 .3263777
CFO_DER1 2936 .8959188 1.074301 .200712 .5623794 1.174755
CFO_DER2 2936 .3044959 .460272 0 0 1
CFO_DER3 2936 1.983098 1.594473 .7608252 1.673511 2.790132
CFO inside debt (tsd) 2936 2400.844 4078.974 227.6245 840.99 2668.849 CFO debt/equity ratio 2936 .4233435 1.845415 .0486668 .1525582 .3799712 Total assets (mlns) 2936 15691.64 49375.47 1524.35 3748.3 11020.16 Current compensation (tsd) 2936 561.9195 377.3638 388.6655 491.2405 615 VEGA 2936 75.33491 96.57756 14.36681 40.20557 94.90934 DELTA 2936 88.7733 112.7242 17.78102 48.0043 113.127 SIZE 2936 8.367935 1.450098 7.329323 8.229058 9.307481 MTB 2936 3.058496 4.205902 1.599998 2.402001 3.672971 SALESGROWTH 2936 .0540072 .1546715 -.0157173 .0559005 .1179594 CASH_SURPLUS 2936 .0919905 .0713833 .0478628 .0841473 .1284525 TOT_CUR_COMP 2936 6.219948 .4018211 5.962719 6.196934 6.421622
31 | P a g e
Panel A of Table 5 gives an overview of the Pearson correlations between my variables of interest, dependent and control variables for the CEO sample. All the correlations are assessed for potential multicollinearity problems in the regressions, by using variance inflation factors (VIF). I find that the highest value is 2.30,14 therefore I conclude that multicollinearity is not a concern. The correlation coefficients show that CAPEX and RND are negatively and
significantly related, indicating that they are substitutes, which is consistent with prior
literature (Coles et al., 2006). That is, when CEOs increase (decrease) risk-taking, one way to do this is by reallocating investment dollars from capital expenditures (R&D) towards R&D (capital expenditures). In addition, the correlation between the measures for CEO inside debt range from 0.60 to 0.78, indicating there is little variation in CEO_DER1, CEO_DER2 and CEO_DER3.
VEGA is negatively (positively) and significantly related to CAPEX (RND), which is consistent with the argumentation that stronger incentives resulting from stock-return volatility leads to more risk-taking behavior. However, VEGA is also negatively related to HHI for some reasons. Contrary to the expectation, DELTA moves in the same way as VEGA. Only the negative and significant correlation between DELTA and HHI is in line with the stream of research that argues that DELTA decreases risk-taking behavior. The finding that DELTA is negatively (positively) and significantly related to CAPEX (RND) could be explained by the fact that higher delta could also encourage risk-taking because the value of the CEO’s equity portfolio will increase from an increase in the stock price. However, in this case the equity’s sensitivity for a one percent change in stock price is probably too low to decrease risk-taking behavior.
With regard to the correlations between the relative CEO debt/equity ratio measures and dependent variables, I find that all CEO_DER measures are positively and significantly related to RND, which is not consistent with the hypothesis. This finding would indicate that higher CEO debt/equity ratios are associated with more risk-taking (RND as risk-taking measure). CEO_DER2 and CEO_DER3 are negatively related with HHI which is consistent with the view that larger inside debt holdings leads to more diversification. However, this correlation is not significant. Lastly, CEO_DER2 is positively, but not significantly, related to CAPEX, which is in line with prior literature.
14
A rule of thumb is that VIF values exceeding 10 are indicative of multicollinearity problem in the model (Neter et al., 1996).
32 | P a g e
Table 5: Pearson Correlation Matrix
Table 5 - Panel A 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1.CAPEX 1 2.HHI 0,136* 1 0,000 3.RND -0,136* 0,121* 1 0,000 0,000 4.CEO_DER1 -0,014 0,016 0,098* 1 0,462 0,398 0,000 5.CEO_DER2 0,028 -0,004 0,065* 0,710* 1 0,132 0,818 0,001 0,000 6.CEO_DER3 -0,019 -0,002 0,237* 0,779* 0,600* 1 0,312 0,937 0,000 0,000 0,000 7.TOT_CUR_COMP -0,046* -0,152* -0,021 -0,050* -0,012 0,151* 1 0,015 0,000 0,273 0,009 0,512 0,000 8.VEGA -0,083* -0,123* 0,180* 0,002 0,029 0,375* 0,462* 1 0,000 0,000 0,000 0,917 0,119 0,000 0,000 9.DELTA -0,045* -0,044* 0,196* -0,019 -0,001 0,375* 0,388* 0,823* 1 0,017 0,021 0,000 0,321 0,946 0,000 0,000 0,000 10.SIZE -0,104* -0,209* 0,067* -0,129* -0,047* 0,113* 0,615* 0,626* 0,510* 1 0,000 0,000 0,000 0,000 0,013 0,000 0,000 0,000 0,000 11.MTB 0,032* 0,024 0,074* 0,033* 0,058* 0,097* 0,023 0,111* 0,143* 0,050* 1 0,096 0,207 0,000 0,077 0,002 0,000 0,227 0,000 0,000 0,009 12.SALESGROWTH 0,013 0,039** 0,068* 0,010 0,008 0,046 -0,003 0,013 0,108 0,008 0,047* 1 0,484 0,039 0,000 0,610 0,691 0,015 0,880 0,495 0,000* 0,685 0,014 13.CASH_SURPLUS 0,011 0,119*** 0,518* 0,162* 0,170* 0,321* 0,046* 0,239* 0,292* 0,067* 0,164* 0,103* 1 0,577 0,000 0,000 0,000 0,000 0,000 0,016 0,000 0,000 0,000 0,000 0,000 14.MKT_LEV -0,019 0,023 -0,095* -0,164* -0,146* -0,198* 0,033* -0,117* -0,136* 0,039* -0,089* -0,066* -0,280* 1 0,326 0,234 0,000 0,000 0,000 0,000 0,085 0,000 0,000 0,039 0,000 0,001 0,000
Bold coefficients = significant at 10% level
Panel B of Table 5 gives an overview of the correlations between the variables that are used for the test between CFO inside debt and the level of discretionary accruals. CFO_DER2 is negatively and significantly related to DA, which is in line with my second hypothesis. The three measures of CFO inside debt (CFO_DER1-3) are significantly related to each other and range from 0.677 to 0.859, indicating that, although there is some variation in the measures, each measure captures to some extent unique details (Cassell et al., 2006).
INCENTIVE_RATIO is negatively and significantly related to DA, suggesting that CFOs with more equity incentives use less discretionary accruals to manipulate earnings. The positive and significant relation between INCENTIVE_RATIO and GROWTH suggest that firms with higher equity incentives are fast-growing, which is consistent with prior literature (Lambert and Larcker, 1987). Besides, LEV is negatively and significantly related to all the CFO_DER measures. This is logical because when a firm takes on more debt (higher LEV), the CFO to firm debt/equity ratio will decrease since the level of firm’s debt is in the denominator of the measure.
