CEO career horizon, inside debt holdings and R&D expenditure
Name: Dominique Blom Student number: 10631380 Thesis supervisor: Máté Széles Date: June 25, 2018
Word count: 19,253
MSc Accountancy & Control, specialization Control
Statement of Originality
This document is written by Dominique Blom 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
Substantial academic literature has paid attention to the horizon problem. Proponents argue that CEOs become more risk averse and short-term oriented as their career horizons shorten, which expresses itself in the opportunistic curtailment of investments with relatively long and uncertain payback periods like R&D. However, till date, there is no consensus in the literature about the underlying mechanisms or even the existence of the horizon problem. I argue that the inconclusive results with respect to the horizon problem can be explained by the substantial amounts of inside debt that CEOs near retirement hold. Specifically, I suggest that CEOs near retirement opportunistically reduce R&D spending in order to safeguard the value of their inside debt holdings. Drawing upon a sample of 3,481 firm-year observations of U.S. firms from 2007-2017, I investigate whether CEOs near retirement are more likely to reduce R&D spending than CEO who are not when they have greater inside debt compensation. Contrary to my expectations, I do not find evidence that CEOs near retirement reduce R&D spending to safeguard the value of their inside debt compensation, indicating that the horizon problem cannot be explained by debt-like compensation.
Contents
1. Introduction ... 5
2. Literature review and hypothesis development ... 8
2.1 Executive compensation structure ... 9
2.2 Horizon problem ... 13
2.3 Horizon problem and CEO inside debt holdings ... 15
3. Research Design ... 17
3.1 Data and sample selection ... 17
3.2 Variable measurement ... 19 3.2.1 Dependent variables ... 19 3.2.2 Independent variables ... 20 3.2.3 Control variables ... 22 3.3 Empirical model ... 24 4. Results ... 25 4.1 Descriptive statistics ... 25 4.2 Correlations ... 28 4.3 Regression analysis ... 31 4.3.1 Main results ... 31 4.3.1 Robustness checks ... 35
5. Conclusion and Discussion ... 37
References ... 41
Appendices ... 47
Appendix A: Estimating option values ... 47
Appendix B: Estimating vega and delta... 49
Appendix C: Robustness check 1 ... 50
Appendix D: Robustness check 2 ... 51
1. Introduction
Prior academic literature reports that research and development (R&D) spending stimulates firm growth and provides a significant source of competitive advantage (Cheng, 2004). However, CEOs with short career horizons might be motivated to reduce R&D spending as suggested by the proponents of the horizon problem phenomenon. The horizon problem was first introduced by Smith and Watts (1982) and proponents argue that the priorities and incentives of top executives change as they near retirement, with older executives becoming more risk averse and short-term oriented (Barker and Mueller, 2002; Brickley et al., 1999). Particularly, scholars suggest that this results in the curtailment of investments with relatively long and uncertain payback periods, such as R&D (Dechow and Sloan, 1991; Kalyta, 2009). This phenomenon is expected to be most prominent when the CEO is planning to retire rather than to leave the firm and stay on the job market (Gibbons and Murphy, 1992).
However, empirical research examining the behavior of CEOs near retirement finds mixed results. While several studies find that CEOs who are approaching retirement are incentivized to curtail R&D expenditure (Antia et al., 2010; Davidson et al, 2007; Dechow and Sloan, 1991), others have failed to provide evidence in favor of the horizon problem (Conyon and Florou, 2006; Serfling, 2014). Scholars argue that these mixed findings might be due to the compensation incentives provided to executives with short horizons (Dechow and Sloan, 1991; Gibbons and Murphy, 1992b). Agency theory suggests that executive compensation should be designed in a way that it incentivizes executives to maximize shareholder value (Jensen and Murphy, 1990). In line with this suggestion, prior studies argue that compensation committees adapt executive pay in order to mitigate CEOs with relatively short career horizons to reduce investment in long-term risky projects (Gibbons and Murphy, 1992). More specifically, scholars suggest that linking CEOs’ wealth to stock price volatility by means of stock options mitigates the incentives of CEOs near retirement age to opportunistically reduce R&D spending (Cheng, 2004; Dechow and Sloan, 1991). Stock options tie the executive’s wealth to stock volatility but limit potential downside risk since stock options do not have to be exercised in case of a drop in share price. These characteristics encourage executives to invest in long-term projects with high outcome uncertainty. Nevertheless, to date there is no consensus in the literature whether a CEO’s career horizon affects the decision to reduce long-term risky investments and what role executive compensation plays in this relationship.
One reason for these inconsistent findings might be that prior literature has largely disregarded the fact that executive pay also contains long-term debt compensation. Debt-based compensation comprises deferred compensation and pension benefits and is referred to as inside debt. Inside debt represent an obligation for the firm to pay a fixed sum of cash in the future to corporate insiders as long as the firm is in solvent condition (Sundaram and Yermack, 2007; Wei and Yermack, 2011). However, prior to the comprehensive Securities and Exchange Commission (SEC) disclosure reform in 2006, only limited information was available concerning executive compensation in the form of debt. This reform greatly enhanced the transparency of pension and deferred compensation, providing the opportunity to sufficiently examine the implications of inside debt compensation. Drawing upon this newly available information, scholars show that CEOs with relatively large inside debt holdings are more risk averse and propose that this is due to two reasons (Cassell et al, 2002; Phan, 2014; Tung and Wang, 2003). First, when CEO inside debt holdings are large relative to CEO equity holdings, the incentive effect of stock-based compensation (i.e. stock options) will be reduced. Second, because the value of inside debt holdings is dependent upon firm solvency, CEOs that hold high values of inside debt are reluctant to engage in actions that increase firm risk. In line with this logic, Cassell et al. (2012) argue that CEOs with relatively large inside debt holdings are discouraged to invest in R&D. Since inside debt holdings represent a fixed debt claim against the firm, extra benefits will not be obtained from successful R&D investments. Contrary, beneficiaries may suffer severely if the investment turns out to be unsuccessful due to increased firm default risk (e.g. increased probability that the full promised compensation cannot be paid).
In this thesis, I argue that the incentive effect of inside debt compensation is especially present for CEOs near retirement. Sundaram and Yermack (2007) and Cen (2011) indicate that the compensation packages of CEOs who are approaching retirement are disproportionately biased towards inside debt. Due to these built-up amounts of inside debt holdings, inside debt becomes relatively more important than other types of compensation (i.e. salary, bonus, shares and stock options) for CEOs near retirement (Sundaram and Yermack, 2007; Kabir et al., 2018). Given the incentive effects of these holdings and the increased likelihood of CEOs near retirement to take actions that maximize their personal wealth, the combined effect of both features might encourage CEOs near retirement to reduce R&D spending even further. Specifically, I suggest that CEOs near retirement are inclined to opportunistically reduce R&D spending in order to safeguard the value of their inside debt holdings and therefore expect the career horizon problem to be most severe when the CEO is
near retirement and holds substantial amounts of inside debt. The following research question logically follows from these propositions:
Are CEOs who are approaching retirement more likely to reduce R&D expenditure than CEOs who are not when they have higher values of inside debt compensation?
Drawing upon a sample of 3,481 firm-year observations of U.S. firms from 2007-2017, I run OLS regressions to provide an answer to my research question. Specifically, I regress R&D spending on the interaction of the CEO career horizon with CEO inside debt holdings. Following Kalyta (2002) and Heyden et al. (2017), I use a proxy for the career horizon that equals one if a CEO is at least 60 years old and zero otherwise. Furthermore, following prior empirical research (Cassell et al., 2012; Phan, 2014; Wei and Yermack, 2011), I adopt three alternative proxies to measure CEO inside debt. First, I use the relative CEO inside debt ratio, defined as the CEO’s equity ratio relative to the firm’s debt-to-equity ratio. Second I use an indicator variable that equals one if the value of the first measure exceeds one, and zero otherwise. For my third proxy of inside debt, I use the relative CEO incentive ratio, which measures the marginal change of CEO wealth for a unit change in overall firm value. I expect the horizon problem to be most severe if the CEO is at least 60 years old and has greater values of inside debt holdings.
My empirical results may be summarized as follows. First, I find that in general CEOs are more likely to reduce R&D spending when they hold inside debt. Second, I find weak evidence that CEOs who are approaching retirement are more likely to curtail R&D expenditure than CEOs who are not and thereby provide weak evidence in favor of the horizon problem. However, when considering the combined effects of both features, I do not find any evidence indicating that CEOs who are approaching retirement are more likely to reduce R&D spending than CEOs who are not when they hold relatively larger amounts of inside debt. Contrary, I find weak evidence that CEOs are more likely to increase R&D spending when they are near retirement and hold larger amounts of inside debt compensation. In sum, I do not find evidence supporting the view that CEOs near retirement opportunistically reduce R&D spending in order to safeguard the value of their inside debt holdings and therefore conclude that the career horizon problem cannot be explained by inside debt compensation.
My research contributes to the academic literature in several ways. First, by focusing on the effect of CEO inside debt holdings, a relatively unexplored dimension of the relationship between CEO career horizon and opportunistic reductions in R&D expenditure is examined. Although some studies have examined the effect of debt-like compensation on the
horizon problem (e.g. Kabir et al., 2018; Kalyta, 2009), this has been restricted to the effect of pension payments. However, ignoring the effects of deferred compensation results in incomplete conclusions with regards to the implications of debt-like compensation (Bebchuk and Jackson, 2005; Cen, 2010; Edmans and Liu, 2011). Taking both pensions and deferred compensation into consideration, my results contradict the findings of Kabir et al. (2018) and Kalyta (2009), who conclude that CEOs near retirement opportunistically reduce R&D expenditure due to their substantial debt-based compensation. Second, I add to the debate in the academic literature surrounding the existence of the horizon problem (e.g. Barker and Mueller, 2002; Conyon and Florou, 2006; Dechow and Sloan, 1991; Murphy and Zimmerman, 1993). I provide weak evidence that CEOs near retirement are more likely to reduce R&D expenditure than CEOs who are not, suggesting that the horizon problem is present in my sample of U.S. CEOs. Furthermore, I respond to the calls of Edmans and Liu (2011) for more research regarding the implications of inside debt holdings. Existing research shows that executive compensation structures have a significant effect on managerial behavior (e.g. Bizjak et al. 1993; Coles et al., 2006; Guay, 1999; Ryan and Wiggins, 2002; Watts and Zimmerman, 1986). However, the focus has predominantly been on the effects of equity-based compensation. The effects of long-term debt compensation remain largely unexplored even though they represent a substantial part of the executive compensation package in the United States (Milidonis et al., 2017; Sundaram and Yermack, 2007; Wei and Yermack, 2011). My findings support the claims of Cassell et al. (2012) that the sensitivity of inside debt holdings on firm solvency incentivizes CEOs to reduce investment in risk-increasing projects.
The rest of the paper is structured as follows: in section 2 I review prior literature that examines the effects of executive pay and the effects of the career horizon on managerial opportunism and develop my hypothesis. I discuss my research methodology in section 3 and my results in section 4. Lastly, section 5 provides my discussion and conclusion.
2. Literature review and hypothesis development
This thesis draws upon two streams of literature to examine the joint effect of the career horizon and inside debt on research and development curtailment. The first stream of literature examines the effect of the executive compensation structure on managerial opportunism and the second investigates the implications of the career horizon problem.
2.1 Executive compensation structure
Several studies have analyzed the relationship between executive compensation and opportunistic behavior. Watts and Zimmerman (1986) introduced the bonus plan hypothesis, arguing that accounting-based bonus plans incentivize managers to select accounting procedures that maximize their bonus awards. In addition, Jensen (1986) argues that short- term compensation in the form of cash bonuses will incentivize executives to focus on projects that pay back quickly and to reduce investment in long-term projects like R&D because of their long and uncertain payback period. He et al. (2003) test this proposition and find that as the cash portion of their compensation increases, CEOs cut back on R&D spending. These findings are supported by Birzjak et al. (1993) who find a negative relationship between CEO cash-based compensation and R&D expenditure.
Other compensation arrangements affect opportunistic behavior as well. Agency theory predicts that the interests of managers are not aligned with the interests of shareholders due to separation of ownership and control. Jensen and Meckling (1976) argue that in an agency relationship, the principal delegates decision-rights to the agent such that the agent can perform a task on behalf of the principal. However, there are reasons to believe that the agent will not always act in the best interest of the principal when both the individuals in the relationship are utility maximizers and have different attitudes towards risk (Eisenhardt, 1989). Jensen and Murphy (1990a) claim that the conflict of interest between CEO and shareholder is a classic example of the principal-agent problem. Shareholders do not have complete information concerning CEO activities and firm’s investment opportunities and therefore cannot specify complete contracts that can enforce desired actions. This provides top management the opportunity to make decisions that are beneficial for their personal wealth and risk preferences but are not in the best interest of the organization (Fama, 1980; Jensen and Meckling, 1976). CEOs will most likely grab this opportunity since they bear the full costs of their efforts generating value for shareholders but capture only a small fraction of the generated value themselves. Additionally, CEOs are generally more risk averse than shareholders since their portfolio of wealth is usually less diversified (Fama, 1980; Jensen and Meckling, 1976). This is alarming since it encourages top management to reduce firm risk by rejecting valuable risk-increasing projects at the expense of shareholder value (Coles et al, 2006).
However, by providing the appropriate incentives, the interest of top management can be aligned with those of the organization (Jensen and Meckling, 1976). More specifically,
stock-based compensation is seen as the solution for the agency problem as it links the CEOs wealth to stock price performance (Jensen and Meckling, 1976; Jensen and Murphy, 1990a; Yermack, 1995). Linking wealth to stock price performance encourages CEOs to make decisions that increase stock price and thus shareholder value. In addition, stock-based compensation has the possibility to counteract potential risk aversion of executives. Reducing risk aversion will lead to increased investment in profitable risk-increasing projects that might otherwise be rejected (Coles et al., 2006; Guay, 1999). However, stock-based compensation contains different components that have different effects on managerial behavior. Stock options are expected to have a positive relationship with the investment in long-term risky projects. Stock options tie the executive’s wealth to stock volatility but limit potential downside risk since stock options do not have to be exercised in case of a drop in stock price. Contrary, compensation in the form of restricted stock might incentivize more conservative behavior. Restricted stock cannot be sold before a minimum time with the firm passes or until a performance goal is reached, and this exposes the CEO to downside risk. Ryan and Wiggins (2002) examine the relationship between R&D investment decisions and compensation policy. When they treat all stock-based compensation as equal, R&D spending is positively related to equity incentives. However, when they split the types of stock-based compensation into stock options and restricted stock, they observe a positive relationship between R&D expenditure and stock options and a negative relationship for equity incentives in the form of restricted stock. Thus, compensation committees should carefully structure compensation schemes, balancing the effects of both types of equity incentives dependent upon the amount of risk they want to impose on the manager (Ryan and Wiggins, 2002).
Nevertheless, solving the agency problem between CEO and shareholder with equity-based compensation might lead to another type of agency conflict, the one between shareholder and outside creditor (Jensen and Meckling, 1976; John and John, 1993; Yermack, 1995). Encouraging executives to engage in excessive risk-taking behavior by providing stock-based compensation (i.e. stock options) may result in risk-shifting from shareholders to debtholders (Jensen and Meckling, 1976). The potential value gains of increased firm risk accrue to equity holders, while the downside risk is shifted towards the firm’s debtholders since they are the ones that suffer when the risk-increasing investment increases the default probability of the firm (Kabir and Veld Merkoulova, 2013). Risk-shifting and the resulting increases in default risk are referred to as the agency cost of debt (Jensen and Meckling, 1976; Sundaram and Yermack, 2006). John and John (1993) argue that outside creditors rationally anticipate risk-shifting, and therefore increase the organization’s borrowing costs
or impose strict debt covenants to protect their interests. This means that, ultimately, the shareholders bear the agency cost of debt. Thus, firms need to commit to compensation policies that encourage behavior that not only minimizes the agency cost of equity but also the agency cost of debt (Edmans and Liu, 2011; John and John, 1993). Jensen and Meckling (1976) argue that long-term debt-based compensation might be able to (partly) eliminate the agency cost of debt. Similarly, Edmans and Liu (2011) hypothesize that debt-based compensation is a superior way to minimize the agency cost of debt since it tilts CEO incentives towards debtholders and therefore balances the competing interests of shareholders and debtholders. Several studies support these propositions and show that when CEO compensation includes debt instruments, the cost of debt financing and the number of restrictive covenants decreases (Anantharaman et al., 2013; Chen et al., 2011). For example, Anantharaman et al. (2013) examine the effect of CEO compensation in the form of debt on loan contracting terms. They find that higher relative CEO debt-based compensation (compared to stock-based compensation) is associated with lower cost of debt and fewer covenants and claim that this indicates incentive alignment between executive and outside debtholder.
However, the majority of empirical studies within the executive compensation literature focuses on the implications of equity-based compensation (Milidonis et al., 2017; Sundaram and Yermack, 2007). Although recently, emerging literature has shifted its attention towards the implications of debt-like instruments (Edmans and Liu, 2011; Milidonis et al., 2017). Debt-based compensation comprises deferred compensation and pension benefits and is referred to as inside debt. Inside debt represent an obligation for the firm to pay a fixed sum of cash in the future to corporate insiders as long as the firm is in solvent condition (Wei and Yermack, 2011). Inside debt holdings represent a substantial part of CEO compensation. Drawing upon a sample of 299 U.S. S&P 1500 firms, Wei and Yermack (2011) find that 84% of the CEOs in their sample hold some form of inside debt with average inside debt holdings of approximately US$10 million. However, limited disclosure requirements prior to 2006 have constrained scholars’ ability to examine the implications of long-term debt-based compensation. In 2006, the comprehensive Securities and Exchange Commission (SEC) disclosure reform enhanced the transparency of pension and deferred compensation, providing the opportunity to sufficiently examine the role of inside debt compensation. Based on this renewed available data, scholars examine the effect of inside debt holdings on managerial decision-making.
Jensen and Meckling (1976) were one of the first to claim that CEOs become more risk averse when their compensation contains inside debt. Nonetheless, Sundaram and Yermack (2007) took the first step to empirically examine the implications of debt-based compensation. They find that CEOs with large inside debt holdings tend to manage firms more conservatively, which expresses itself in fewer investments in risky projects, extending the average maturity of debt outstanding, unlevering the capital structure or reducing payouts to equity holders. However, because disclosure for deferred compensation was extremely limited at the time, the study restricts its analysis to inside debt in the form of pension payments only. Sundaram and Yermack (2007) state that the omission of deferred compensation is unlikely to cause any serious problems since its value represents only a small fraction of the value of pension payments. Yet, Cen (2011) finds that the value of deferred compensation is similar to the value of pension payments. They conclude that incorporating deferred compensation in the measure of inside debt has a significant influence on the implications of inside debt with respect to managerial behavior. Based on the disclosure reform initiated by the SEC, Cassell et al. (2012) incorporate both deferred compensation and pension payments in their measure of inside debt. They investigate the effect of CEO inside debt holdings on firm investment and financial policies. The authors find that large CEO inside debt holdings incentivize a reduction in the riskiness of firm operations which expresses itself in reduced R&D spending and reduced firm leverage. Cassell et al. (2012) claim that this finding is due to two reasons. First, when CEO inside debt is large relative to CEO equity, the incentive effect of stock-based compensation (especially in the form of stock options) will be reduced. Second, because inside debt holdings are responsive to the possibility of bankruptcy and the liquidation value of the firm in case of bankruptcy, CEOs with large inside debt holdings will become more risk averse in order preserve firm value and reduce the likelihood of bankruptcy. Phan (2014) finds a negative relationship between CEO inside debt holdings and risk-taking as well. Specifically, their study shows that higher relative CEO debt-based compensation is negatively related with the propensity to pursue mergers and acquisitions. Mergers and acquisitions tend to increase the default risk of the acquiring firm and thus represent a risk-increasing action for the CEO. In order to safeguard the value of their debt-based compensation, CEOs are inclined to decrease investment in mergers and acquisitions. Tung and Wang (2013) examine the effect of CEO compensation in the form of inside debt on risk-taking activities of the banking sector during the financial crisis. They conclude that higher bank CEO inside debt-to-equity ratio is negatively related to risk-taking behavior but positively related to bank performance during the financial crisis.
Altogether, these studies provide evidence that CEO inside debt holdings in the form of deferred compensation and pension payments discourages CEOs to engage in behavior that increases firm risk.
2.2 Horizon problem
Substantial academic literature has paid attention to the horizon problem. The career horizon problem argues that top executives behave more opportunistically when they have shorter career horizons (Heyden et al., 2017; Krause and Semadeni, 2013). More specifically, proponents of the career horizon problem argue that the priorities and incentives of top executives change as they near retirement, with older executives becoming more risk averse and short-term oriented (Barker and Mueller, 2002; Brickley et al., 1999).
Fama (1980) suggests that the external labor market assesses managers’ abilities since these abilities will be reflected either by pay in the current period or in the subsequent period. Thus, the labor market works as a disciplinary mechanism and lessens the conflict of interest between executives and shareholders. Furthermore, scholars argue that the function of the labor market as a disciplinary mechanism is more effective in the early years of executives’ working lives (Antia et al., 2010; Davidson et al., 2007). Younger executives at the beginning of their careers are more concerned about their value in the external labor market and are thus incentivized to act in the shareholder’s best interest. However, as executives approach retirement, career concerns become less relevant in guiding their behavior and their focus shifts towards maximizing their own utilities (Gibbons and Murphy, 1992). In other words, as executives approach retirement they are expected to be less concerned about reputational costs and ex-post settling up which causes them to prioritize their own personal wealth. This proposition is supported by the upper echelons theory, from which the theory behind the career horizon problem developed. The upper echelons theory was first introduced by Hambrick and Mason (1984) and argues that managerial background characteristics determine strategic choices which in turn determine firm performance. Regarding age, the theory predicts that older CEOs behave more conservative. Aging executives engage in less risk-taking behavior not only because they have greater psychological commitment to the organizational status quo or less physical and mental stamina status, but also because aging CEOs have reached a point in their lives at which financial security has become highly important (Hambrick and Mason, 1984). The horizon problem theory draws upon the last
rationale in examining the effect of career horizon on CEO behavior (Krause and Semadeni, 2014).
Several studies show that CEOs with shorter career horizons are less likely to engage in long-term investment activities with high outcome uncertainty. Dechow and Sloan (1991) were the first to examine the career horizon phenomenon empirically. They provide evidence showing that CEOs become more risk averse and short-term oriented as they approach retirement. Specifically, they find that CEOs near retirement reduce R&D spending to temporarily boost their current performance in order to maximize their short-term bonus compensation. R&D spending represents a typical risky investment decision that might not benefit CEOs with relatively short career horizons due to its highly uncertain and long payback period. They take away scarce cash resources that could have been used in ways that benefit older CEOs personal and financial interest (Kabir et al., 2018). Furthermore, U.S. GAAP requires R&D expenditures to be expensed in the period in which they are incurred. Contrary, the expected future payoffs can only be recognized in the period in which they are realized. As a result, the year’s R&D expenses lower annual accounting earnings. This U.S. GAAP requirement incentivizes CEOs who want to boost current accounting numbers in order to maximize their current compensation to reduce R&D expenditure (Dechow and Sloan, 1991). Barker and Mueller (2002) draw upon this line of reasoning as well and examine whether R&D spending varies with CEO age. They find that older CEOs tend to opportunistically reduce investment in R&D and claim that this is due to the tendency of CEOs close to retirement to undertake actions that personally benefit them. Specifically, they argue that R&D expenditure does not benefit retiring CEOs since these expenditures decrease short-term salary and bonuses. Oh et al. (2016) provide evidence that shows a positive relation between CEOs nearing retirement and opportunistic behavior expressing itself in reduced investments in corporate social responsibility (CSR). Investment in CSR do not benefit short-term oriented and risk averse CEOs because these investments have high outcome uncertainty and possible returns will only be realized once the CEO has already exited the firm. However, several studies find contradicting results regarding the horizon problem. Canyon and Florou (2006) examine whether CEOs on the verge of retiring change their investment pattern but do not find any supporting results. Additionally, Wells (2002) does not find any evidence indicating that CEOs approaching retirement tend to behave opportunistically either. Furthermore, Murphy and Zimmerman (1993) claim that the curtailment of R&D near CEO retirement cannot be attributed to the managerial horizon problem but is rather caused by poor performance of the firm.
Cheng (2004) draws upon the executive compensation literature to explain the inconclusive results. He argues that whether or not scholars observe the horizon problem phenomenon is dependent upon compensation committees altering CEO compensation in order to control for this problem. The results of his research indicate that potential opportunistic reductions in R&D spending are eliminated by providing CEOs with relatively short career horizons larger values of annual option grants since stock options give executives incentives to invest in long-term projects with high outcome uncertainty. Although Dechow and Sloan (1991) find evidence that stock-based compensation mitigates the horizon problem as well, other studies do not find evidence indicating that stock-based incentives prevent CEOs with short career horizons to engage in opportunistic behavior (Matta and Beamish, 2008; Xu and Yan, 2014). Furthermore, several scholars examining the horizon problem have questioned its underlying mechanism. Serfling (2013), for example, provides evidence that CEOs near retirement decrease R&D spending in order to reduce firm risk rather than to increase short-term accounting figures. Furthermore, Davidson et al. (2007) state that if CEOs near retirement are inclined to behave opportunistically in order to increase short-term compensation, this behavior should especially be present when cash-based compensation represents a larger portion of total compensation. Although Davidson et al. (2007) do find that CEOs approaching retirement behave opportunistically, they do not find evidence that this behavior is affected by the level of salary and bonuses.
Overall, there is still a lot of debate about the underlying mechanisms and even the existence of the horizon problem. The following discussion suggests that CEO inside debt might play an important role in the career horizon problem and might possibly explain its underlying mechanism.
2.3 Horizon problem and CEO inside debt holdings
Drawing upon the executive compensation literature, Sundaram and Yermack (2007) propose an alternative explanation for the horizon problem. They note that CEOs near retirement have substantial amounts of inside debt holdings that will not be received in case the firm goes bankrupt. Specifically, Sundaram and Yermack (2007) document that the importance of inside debt holdings as part of overall compensation increases monotonically up to age 65 for their sample of U.S. CEOs. In addition, the ratio of inside debt holdings to equity holdings increases monotonically as well. Thus, the value of inside debt holdings tends to rise more rapidly than the value of equity holdings as CEOs approach retirement. Furthermore, the
value of CEO inside debt holdings is approximately 40% higher than the value of CEO cash-based compensation for their sample. Based on these observations, Sundaram and Yermack (2007) suggest that CEOs near retirement behave more opportunistically because they want to safeguard the value of their inside debt rather than to maximize the value of their short-term bonus compensation. They argue that ways to accomplish such safeguarding requires less involvement in projects that increase firm risk, such as R&D. However, they do not test these propositions empirically.
As explained, the analysis of Sundaram and Yermack (2007) is restricted to inside debt in the form of pension payments only and while some studies have examined the effect of debt-like compensation on the career horizon problem, this has been restricted to the effect of pension payments as well. Kabir et al. (2018), for example, try to explain the inconclusive results with regards to the horizon problem by examining the joint effect of the CEO career horizon and pension compensation on R&D curtailment for a sample of UK CEOs. Their results suggest that their sample CEOs do not curtail R&D when their career horizons shorten in general. However, when they take pension compensation into consideration, the average CEO does reduce investment in R&D when they approach retirement. Based on these observations they conclude that the career horizon problem can, at least partially, be explained by pension compensation. These findings are consistent with those of Kalyta (2009) who takes an incentive approach towards the horizon problem as well. He proposes that some CEOs near retirement have stronger incentives to behave opportunistically dependent upon their retirement plans. To test this proposition, Kalyta (2008) regresses R&D spending on the interaction of the career horizon problem with the contingency of CEO pension benefits on firm performance. His results suggest that CEOs opportunistically curtail R&D expenditure as they approach retirement in order to maximize the value of their pension payments. However, ignoring the effects of deferred compensation can result in incomplete conclusions with regards to the implications of debt-like compensation on managerial behavior (Bebchuk and Jackson, 2005; Cen, 2010; Edmans and Liu, 2011). Consequently, any study that examines the implications of debt-like compensation that does not include deferred compensation should acknowledge that it ignores a substantial part of executive pay. Drawing upon a sample of 1,947 S&P 1500 firms, Cen (2010) shows that deferred compensation represents a significant part of the overall compensation of CEOs and that its value is about the same size as the value of pension obligations. Furthermore, extending the observations of Sundaram and Yermack (2007), Cen (2010) documents that not only the value of pensions but also the value of total inside debt (including both pension benefits and
deferred compensation) tends to rise more rapidly than the value of equity holdings as CEOs approach retirement.
Drawing upon the suggestion of Sundaram and Yermack (2007) and the observations of Cen (2011), I extend existing literature that examines the implications of debt-like compensation on the CEO career horizon by focusing on the implications of CEO inside debt holdings containing both pension and deferred compensation. Specifically, I suggest that the career horizon problem is most severe when CEOs are near retirement and hold substantial amounts of inside debt holdings. The executive compensation literature predicts that inside debt compensation incentivizes CEOs to reduce spending on long-term investment activities with high outcome uncertainty, such as R&D. Since inside debt holdings represent a fixed debt-claim against the firm, extra benefits will not be obtained from successful R&D investments. Contrary, beneficiaries may suffer severely if the investment turns out to be unsuccessful due to increased firm default risk (e.g. increased probability that the full promised compensation cannot be paid). I argue that the incentive effect of inside debt is especially present for CEOs near retirement. Due to their increasing amounts of built-up inside debt holdings, inside debt becomes relatively more important than other types of compensation (i.e. salary, bonus, shares and stock options) for CEOs near retirement. Given the incentive effects of these holdings and the increased tendency of CEOs near retirement to take actions that maximize their personal wealth, I argue that the combined effect of both features will encourage CEOs near retirement to reduce R&D spending even further. In other words, I propose that CEOs near retirement opportunistically reduce R&D spending in order to safeguard the value of their inside debt holdings. In line with this logic, I expect that if CEOs near retirement are inclined to reduce R&D spending in order to safeguard the value of their inside debt holdings, this behavior should especially be present when they have greater inside debt compensation. I therefore hypothesize:
H1: CEOs who are approaching retirement are more likely to reduce R&D expenditure than CEOs who are not when they have higher values of inside debt compensation.
3. Research Design
3.1 Data and sample selection
This research draws upon the Compustat Executive Compensation (ExecuComp) database in order to examine the joint effect of the CEO career horizon and CEO inside debt holdings.
The ExecuComp database includes over 80 different compensation items on more than 12,500 executives in companies listed on the Standard & Poor’s (S&P) 1500 index. The S&P 1500 index combines all stocks included in the S&P 500, S&P MidCap 400 and S&P SmallCap 600. By using multiple S&P indexes, potential biases caused by firm size are limited. Although data is available from 1992 onwards, data for pensions and deferred compensation was not included in the ExecuComp database until 2007. As explained, the Securities and Exchange Commission (SEC) introduced a disclosure reform in 2006 that greatly enhanced the transparency of pension and deferred compensation. Hence, data is collected from 2007 onwards. Furthermore, firm-specific data and data on investment characteristics is obtained from the Center for Research in Security Prices (CRSP)/Compustat merged database. This database links relationships over time between CRPS data, which contains historical descriptive information and market data on more than 27,000 stocks, and Compustat data, which includes more than thousands of annual income statements, balance sheets, cash flows, pensions, supplemental and descriptive data items. Execucomp and the CRSP/Compustat merged databases are obtained from the Wharton Research Data Services (WRDS) system.
Based on the identified databases, I collect a sample of CEOs in the period 2007-2017. By performing a multi-year analysis, the impact of random fluctuations is mitigated. Following prior literature, R&D expenditure is set equal to zero when it is missing from the database (Coles et al., 2006; Ryan and Wiggins, 2002). In addition, firms for which data is missing with respect to CEO, firm or investment characteristics are dropped from the sample. Lastly, all variables are winsorized at the 1% level in both tails in order to account for the influence of outliers (Cassell et al., 2012; Coles et al., 2006; Kalyta, 2009). Overall, the total sample collected in the years 2007-2017 consists of 3,481 firm-year observations for whom complete data is available for the purpose of my research. The sample reconciliation is specified in Panel A of Table 1. The sample distribution across years is outlined in Panel B of Table 1 and indicates that my sample firms are evenly distributed from 2007 to 2017. Panel C of Table 1 shows the distribution of my sample firms across the 2-digit SIC code industry classification. The majority of the sample firms come from manufacturing industries, which includes industries that are indicated by 2-digit SIC codes between 20 and 39.
3.2 Variable measurement
3.2.1 Dependent variables
The dependent variable of interest is R&D expenditure. R&D curtailment is considered to be an outcome of managerial opportunism and is covered widely in the management research literature (Cazier, 2011; Kalyta, 2009). However, prior literature does not provide a definitive measure for R&D expenditure (Lee and Chang, 2004). Therefore, I adopt three proxies for R&D spending that are commonly used in the literature. First, following Cassell et al. (2012) and Lundstrum (2012), the level of R&D spending (R&D exp/sales ratio) is measured by scaling R&D expenditures by total sales. Second, following Bizjak et al. (1993), Coles et al.
Table 1: Sample Selection and Distribution
Panel A: Sample reconciliation
Number of observations in ExecuComp (2007 – 2017) 22,718
Less: Observations with missing data on inside debt compensation, current compensation and CEO age (2,956) Less: Unmatched data from merging with CRSP/Compustat Merged (by fyear & gvkey) (1,106) Less: Observations with missing data on dependent, independent and control variables (15,175)
Final Sample (2007-2017) 3,481
Panel B: Sample distribution by year
Year Frequency Percent
2007 240 6.89 2008 416 11.95 2009 427 12.27 2010 414 11.89 2011 384 11.03 2012 366 10.51 2013 348 10.00 2014 311 8.93 2015 269 7.73 2016 248 7.12 2017 58 1.67 Total 3,481 100
Panel C: Sample distribution by industry
Two-digit SIC code Frequency Percent
20: Food and kindred products 129 3.71
28: Chemicals and allied products 577 16.58
34: Fabricated metal products 106 3.05
35: Industrial and commercial machinery and equipment 464 13.33
36: Electronic & other electrical equipment 532 15.28
37: Transportation equipment 164 4.71
38: Measuring, photographic, medical & optical Goods 588 16.89
39: Miscellaneous manufacturing industries 75 2.15
73: Business services 391 11.23
Industries with < 2% representation 455 13.07
Total 3,481 100.00
Note. This table presents the sample reconciliation (Panel A) and the sample distribution by year (Panel B) and by industry (Panel
C). Data is collected from 2007-2017 for firms with complete data to construct the dependent, independent and control variables. This results in a sample consisting of 3,481 firm-year observations.
(2006), and Ryan and Wiggins (2002), I divide R&D expenditures by total assets (R&D exp/assets ratio). For the third measure of R&D spending, I follow Dechow and Sloan (1991) and Murphy and Zimmerman (1993) to measure the growth rate of R&D spending (R&D growth rate), defined as the natural logarithm of R&D expenditure in year t minus the natural logarithm of R&D expenditure in year t - 1.
3.2.2 Independent variables
Following Kalyta (2009) and Heyden et al. (2017), a proxy for the CEO career horizon problem (CEO at least 60 years old) is created which equals one if the CEO is at least 60 years old, and zero otherwise. Although it is likely that some CEOs remain on the job market after reaching the age of 60 while others will retire before turning 60, imposing this restriction reduces the number of observations with hypothesized R&D curtailment in which the CEO career horizon problem is less likely to be significant (Kalyta, 2009).
Jensen and Meckling (1976) argue that long-term debt-based compensation might be able to minimize the agency cost of debt since it discourages executives to engage in actions that increase firm risk and thus reduces default risk. Specifically they argue that when the CEO debt-to-equity ratio reflects the debt-to-equity ratio of the firm, executives have no longer an incentive to favor debtholder interest over shareholder interest, or vice versa. Contrary, when the CEO debt-to-equity ratio exceeds the firm’s respective ratio, executive incentives are more aligned with the interest of debtholders, which results in more conservative investment choices (Jensen and Meckling, 1976). Following these theoretical predictions and recent empirical applications (e.g., Cassell et al., 2012; Phan, 2014; Wei and Yermack, 2011), my variable of interest is the CEO debt-to-equity ratio relative to the firm’s respective ratio. Specifically, following the studies of Cassell et al. (2012), Phan (2014) and Wei and Yermack (2011), I construct three alternative proxies for the relative CEO debt-to-equity ratio. The first proxy, relative CEO inside debt ratio, is measured by scaling the CEO debt-to-equity ratio by the firm’s debt-to-equity ratio:
Relative CEO inside debt ratio =
The present value of accumulated pension benefits and the total value of deferred compensation comprises CEO inside debt holdings (CEOIDH). The value of CEO equity
holdings (CEOEH) includes stock1 and stock options. To calculate the value of stock held by
the CEO, the number of shares held by the CEO is multiplied by the stock price at the firm’s
fiscal year-end. The Black-Scholes model (1973) is used to determine the total value of CEO stock options, which is explained in detail in Appendix A. Furthermore, the firm’s debt (FirmDebt) and the firm’s equity (FirmEquity) are determined in order to calculate the firm’s
debt-to-equity ratio. The firm’s debt is measured by adding total long-term debt to current debt and the firm’s equity is measured as the market value of equity, which is calculated by multiplying the total number of common shares outstanding by the stock price at firm’s fiscal year-end. For the second proxy of inside debt, relative CEO inside debt ratio > 1, the relative CEO inside debt ratio as calculated above is set equal to one if the ratio is bigger than one and set equal to zero otherwise. This follows from the above-explained propositions of Jensen and Meckling (1976), who theorize that the incentive effects of CEO inside debt holdings are particularly present when the CEOs equity ratio is bigger than the firm’s debt-to-equity ratio.
However, Wei and Yermack (2011) state that a potential limitation of the first two measures is that the relative CEO inside debt ratio captures levels of the values of CEO debt and equity but not changes in these values. This might encounter problems when CEO and firm equity have differing convexity and/or when the CEOs inside debt has different duration than firm debt (Wei and Yermack, 2011). In order to avoid this potential limitation, I follow prior literature and establish a third proxy (Cassell et al., 2012; Phan, 2014; Wei and Yermack, 2011). The third proxy, relative CEO incentive ratio, estimates the marginal change in the value of CEO inside debt to the marginal change in the value of CEO equity given a unit change in overall firm value, all scaled by the marginal change in the value of firm debt to the marginal change in value of firm equity, given the same unit change in overall firm value:
Relative CEO incentive ratio =
Considering that the CEO’s equity consists of both stock and stock options, the value of ΔCEOEH is estimated as:
ΔCEOEH =
where S represent the ‘total share delta’ that equals the number of shares held by the CEO times delta assumed to be equal to 1. Furthermore, is the ‘total option delta’, where N represents the number of stock options held by the CEO and ΔN the option delta as determined by the Black Scholes model (1973)2. The subscript i represents option tranches
(i.e. current option grant, previously granted unexercisable options and previously granted exercisable options), which have different exercise prices and times to maturity. The value of ΔFirmDebt is calculated in a similar manner. However, complete data for all outstanding option tranches issued by the firm is missing from Compustat and therefore the total options outstanding at year-end, their exercise price and an assumed time to maturity of four years are used as inputs to the Black-Scholes formula (Cassell et al., 2012; Phan, 2014; Wei and Yermack, 2011). Finally, following Cassell et al. (2012), Phan (2014) and Wei and Yermack (2011), I make the following simplifying assumption:
ΔCEOIDH/ΔFirmDebt = CEOIDH/FirmDebt
Wei and Yermack (2011) provide two arguments to support this simplifying assumption. First, they argue that ΔCEOIDH and ΔFirmDebt are likely to have small and difficult to estimate values considering that the large majority of firms in the S&P 1500 index are not financially distressed. Second, since U.S. companies are not required to disclose maturity details of debt with a remaining life that exceeds five years, information about the firm’s maturity structure of debt is incomplete (Wei and Yermack, 2011).
3.2.3 Control variables
Several control variables that are associated with R&D expenditure are included in order to increase internal validity (Lee and Chang, 2004). First, I include two control variables to account for the effect of CEO equity-based compensation. Prior empirical research shows that equity-based compensation has a strong causal relationship with investment policy (Coles et al., 2006; Guay, 1999). Specifically, prior research provides evidence that shows a positive relationship between CEO stock options and investments in risk-increasing projects, while a negative relationship is observed between CEO stock holdings and investment in risk-increasing projects (Coles et al., 2006; Guay, 1999). As explained, stock options tie the executive’s wealth to stock return volatility but limit potential downside risk since stock options do not have to be exercised in case of a drop in share price (Coles et al., 2006; Ryan and Wiggins, 2002). Contrary, since stock holdings tie the CEOs wealth to stock-price, it exposes the CEO to both upside potential and downside risk, which results in more conservative investment strategies (Coles et al., 2006; Guay, 1999, Ryan and Wiggins, 2002). Based on these empirical findings, I control for the change in CEO wealth for a one percentage point change in stock price (CEO delta) and for the change in CEO wealth for a 0.01 change in the annualized stock return volatility (CEO vega) (Coles et al., 2006; Guay,
1999). A higher sensitivity to stock price is expected to lead to more conservative investment strategies whereas a higher sensitivity to stock return volatility is expected to lead to more investment in risk-increasing projects. Following Coles et al. (2006), Guay (1999) and Ryan and Wiggins (2002), I estimate the values of delta and vega with the Black-Scholes (1973) option valuation model, modified by Merton (1973) to account for dividend payouts. Details on the calculations of delta and vega are provided in Appendix B.
Second, a variable is added to control for the level of cash compensation by calculating the logarithm of CEOs total current compensation (cash compensation), defined as the sum of CEO salary and bonus. Prior literature (Bernartzi and Thaler, 1999; Guay, 1999) shows that the level of CEO wealth influences their investment strategy. Specifically, as the level of cash compensation increases, the CEO can invest more money outside the firm and thus becomes more diversified which results in increased investment in long-term projects with high outcome uncertainty. Furthermore, I include the market-to-book ratio (market-to-book ratio) and sales growth (sales growth) to control for growth opportunities since high-growth firms are more likely to exploit R&D opportunities (Bhagat and Welch, 1995). Contrary, high-growth firms might also be constrained in undertaking long-term and uncertain investments since they are often young firms who might struggle accessing the needed capital (Hymer and Pashigian, 1962). The market-to-book ratio is estimated by scaling the market value of the firm’s equity by the book value of total assets and sales growth is determined by subtracting the natural logarithm of sales in year t -1 from the natural logarithm of sales in year t (Cassell et al., 2012; Coles et al., 2006). Fourth, the logarithm of total assets is calculated to control for firm size (firm size). Prior literature finds contradicting results regarding the effect of firm size on R&D expenditure. Several scholars find a positive relationship between firm size and R&D spending, arguing that larger firms have more resources available to establish sustained R&D programs (Baysinger et al., 1991; Schumpeter, 1942). Contrary, Graves (1988) and Hansen and Hill (1991) argue that larger firms are accompanied with greater market power which incentivizes executives to reduce investment in innovations that may threaten the status quo. To control for available funds to invest in new projects, cash surplus (cash surplus) is calculated by subtracting depreciation and amortization expenses and adding R&D expenses to the net cash flow from operations, all scaled by lagged total assets (Cassell et al., 2012; Coles et al., 2006; Richardson, 2002). In addition, I control for past accounting performance (return on assets), defined as income before extraordinary items scaled by lagged total assets. Prior empirical literature shows that unprofitable firms are more likely to reduce investment in R&D, indicating that firm
profitability gives executives the confidence needed to invest in risky long-term projects (Barker and Mueller, 2002; Cazier, 2011; Hundley et al., 1996). Furthermore, I include firm leverage (book leverage) since leverage discourages executives to invest in projects like R&D as current cash flows are needed for debt services (Barker and Mueller, 2002). Following Barker and Mueller (2002), Coles et al. (2006) and Cassell et al. (2012), I define firm leverage as the ratio of firm’s total debt to total assets. Lastly, industry (industry) and year (year) dummies are included to control for industry and year fixed effects. Year dummies are included to control for general year-to-year fluctuations in R&D spending caused by changes in macroeconomic conditions, shifts in policy and other variables that potentially influence the general level of R&D spending (Kabir et al., 2018; Richardson, 2002). Controlling for industry is especially important, as significant cross-sectional variation might exist across industries regarding CEO and firm characteristics (Kabir et al., 2018).
3.3 Empirical model
I follow Kalyta (2009) and Heyden et al. (2017) to examine my hypothesis and regress R&D spending on the interaction of the CEO career horizon with CEO inside debt holdings. This leads to the following empirical model:
+ 6 + ,
As common in literature (e.g. Cassell et al, 2012; Kabir et al., 2018; Kalyta, 2009), I perform the analysis using an OLS estimation that includes year and industry specific effects. Furthermore, t-statistics and p-values are based on robust standard errors that are adjusted for heteroskedasticity (White, 1980) and clustered by firm to control for serial dependence (Cassell et al, 2012). My variable of interest is the interaction term between the career horizon and the relative CEO debt-to-equity ratio, Career horizoni,t * Relative CEO debt/equity ratioi,t. According to my hypothesis, the β1 coefficient should be negative and
significant, indicating that the career horizon problem is especially pronounced if CEOs are approaching retirement and hold substantial amounts of inside debt holdings.
4. Results
4.1 Descriptive statistics
Table 2 presents the descriptive statistics for the full sample of CEOs, including the mean, median, standard deviation and interquartile ranges for the variables used in my analysis. Additionally, I provide descriptive statistics for my sample split by the CEO at least 60 years old dummy in Table 3. This results in a subsample including CEOs who are younger than 60 years old and a subsample including CEOs who are at least 60 years old.
Starting with the descriptive statistics for the full sample, I observe that the average firm in my sample spends 7.39% of sales on R&D, which is in line with the findings of Cassell et al. (2012) and Lundstrum (2012). Similar to the findings of Ryan and Wiggins (2002) and Coles et al. (2006), the mean (median) of the second proxy for R&D spending is positive and equals 0.0521 (0.0359), suggesting that the average sample firm spends 5.21% of total assets on R&D. Furthermore, the average R&D growth rate equals 6.21% from year t-1 to year t for the firms in my sample, which is similar to the findings of prior literature as well (Dechow and Sloan, 1991).
Consistent with prior literature, the average CEO in my sample is about 55 years old (Antia et al., 2010; Phan, 2014). Moreover, 26.03% of the CEOs in my sample are at least 60 years old. Regarding CEO compensation variables, the mean CEO holds on average US$ 0.87 million in current compensation (i.e. salary and bonuses). Furthermore, the mean (median) of CEO vega and CEO delta are US$ 0.35 million (US$ 0.22 million) and US$ 0.39 million (US$ 0.19 million) respectively. The higher value of CEO delta implies that CEO wealth is more sensitive to a one percentage point change in stock price than to a 0.01 change in stock return volatility (Cassell et al, 2012; Coles et al., 2006; Guay, 1999). The mean (median) CEO debt-to-equity ratio is 0.2170 (0.0349) indicating that, for the majority of firms included in my sample, CEO equity holdings have greater values than CEO inside debt holdings. Nevertheless, the average CEO holds approximately US$ 4.6 million in inside debt, suggesting that my sample CEOs hold substantial amounts of inside debt (Cen, 2011; Wei and Yermack, 2011). Furthermore, the relative CEO inside debt ratio is 0.5936, which is slightly lower than the ones found in prior literature (Cassell et al., 2012; Wei and Yermack, 2011). However, this might be due to my longer sample period (Kubick, 2014). The mean of the second proxy for the relative CEO debt-to-equity ratio, which equals one if the CEO’s debt-to-equity ratio exceeds the firm’s debt-to-equity ratio, is 0.1910. Although Jensen and
Meckling (1976) argue that the CEO debt-to-equity ratio should mirror the debt-to-equity ratio of the firm, Edmans and Liu (2011) claim that this is inefficient. Contrary, they suggest that an equity bias is required in order to induce CEO effort, indicating that the CEO’s debt-to-equity ratio should be smaller than the firm’s debt-debt-to-equity ratio. The observation that only 19.10% of the CEOs in my sample have a to-equity ratio exceeding the firm’s to-equity ratio is consistent with these statements. The third proxy for the relative CEO debt-to-equity ratio, relative CEO incentive ratio, has a mean (median) of 0.6845 (0.1119) which is in line with the observation of Wei and Yermack (2011). I note, however, that Table 2 shows large variations for the first and third proxy of the relative CEO debt-to-equity ratio. While the lower-quartile values of the two proxies are zero (i.e. no inside debt), CEOs in the third quartile hold equity ratios representing more than 70% of total firm debt-to-equity ratios. Following Anantharaman et al. (2013) and Cassell et al. (2012), I take the natural logarithm of these relative CEO debt-to-equity measures in my regression analyses to ensure that the skewness does not affect my inferences.
With regards to firm-specific variables, the average firm in my sample has good growth prospects. This is suggested by the positive means (medians) for the market-to-book ratio and sales growth, which equal 3.1284 (2.5024) and 1.0632 (1.0600) respectively. The positive Table 2: Descriptive Statistics
N Mean Std. dev. Q1 Median Q3
R&D exp/sales ratio 3,481 0.0739 0.0730 0.0075 0.0456 0.1118 R&D exp/total assets ratio 3,481 0.0521 0.0471 0.0154 0.0359 0.0774
R&D growth rate 3,481 0.0621 0.1711 -0.0264 0.0543 0.1513
CEO age 3,481 55.3036 6.7261 51 55 60
CEO at least 60 years old 3,481 0.2603 0.4388 0 0 1
CEO inside debt holdings ($thsd) 3,481 4,624.5300 8,784.2030 0 495.7510 4800.6440 CEO equity holdings ($thsd) 3,481 28,883.2000 34,427.7000 6,730.9030 16,575.3600 36,875.6800
CEO debt/equity ratio 3,481 0.2170 0.5941 0 0.0349 0.2510
Relative CEO inside debt ratio 3,481 0.5936 1.0937 0 0.1103 0.7093 Relative CEO inside debt ratio > 1 3,481 0.1910 0.3931 0 0 1 Relative CEO incentive ratio 3,481 0.6845 1.4303 0 0.1119 0.7065 Total current compensation ($ thsd) 3,481 848.0044 392.915 568.3310 800.000 1015.000 CEO vega ($thsd) 3,481 349.9908 369.3655 86.4773 217.4679 498.8925 CEO delta ($thsd) 3,481 385.9104 833.6147 69.2672 194.1952 510.9603 Firm size 3,481 7.4565 1.6101 6.2389 7.3874 8.4927 Total assets ($mln) 3,481 6,467.0770 14,350.2600 512.2960 1,1615.5000 4,7879.2000 Market-to-book ratio 3,481 3.1284 2.5730 1.6407 2.5024 3.7997 Sales growth 3,481 1.0632 0.1678 0.9737 1.0600 1.1417 Return on assets 3,481 0.0612 0.0803 0.0268 0.0631 0.0990 Book leverage 3,481 0.3620 0.1712 0.2266 0.3682 0.4790 Cash surplus 3,481 0.1164 0.0807 0.0643 0.1030 0.1647
Notes. This table presents the descriptive statistics for the full sample of CEOs, including the number of observations, mean, standard deviation and
mean (median) of 0.1164 (0.1030) for cash surplus suggests that the average firm in my sample has funds available to finance new projects (Kabir et al., 2018). Furthermore, sample firms have on average US$ 6.5 billion in assets with a median asset size of US$ 1.6 billion, suggesting that firms are diverse with regards to firm size (Cassell et al., 2012). Moreover, the average return on assets for the sample firms is 6.12%, which indicates that the average firm is profitable (Kabir, et al. 2018). Lastly, the amount of book leverage for the average firm in my sample equals 36.20% of total assets.
To provide better insights regarding the differences between CEOs who are approaching retirement and CEOs who are not, descriptive statistics for the sample split by the CEO at least 60 years old dummy are provided in Table 3. Table 3 shows the mean, standard deviation, median, and statistical difference in mean values based on the t-test for the variables used in my analysis. The results suggest that CEOs who are at least 60 years old spend on average less on R&D than their younger counterparts, as indicated by the lower means (medians) for the three proxies of R&D spending. This observation provides initial evidence that the horizon problem is present in my sample. However, the mean is significantly different for the R&D exp/sales ratio and the R&D exp/assets ratio only.
Regarding the CEO characteristics, I observe that CEOs near retirement hold less current compensation, have higher values of CEO delta and similar values of CEO vega
Table 3: Descriptive statistics for split sample
CEOs younger than 60 years (n =2,575) CEOs at least 60 years old (n = 903)
Mean Std. dev. Median Mean Std. dev. Median Difference t-statistic
R&D exp/sales ratio 0.0770 0.0742 0.0496 0.0654 0.0689 0.0386 0.0116 4.1242***
R&D exp/total assets ratio 0.0544 0.0482 0.0379 0.0457 0.0431 0.0306 0.0087 4.8090***
R&D growth rate 0.0648 0.1685 0.0574 0.0545 0.1781 0.0508 0.0104 1.5703
CEO age 52.3076 4.6773 53 63.8290 3.6894 63 -11.5114 -67.0961***
CEO inside debt holdings ($thsd) 3,788.8110 7,683.3470 324.615 6,999.7790 11,007.8600 1,520.864 -3,210.9680 -9.5859*** CEO equity holdings ($thsd) 26,256.0300 31,609.0400 15,355.8600 36,350.04 40,503.4200 21,626.2700 -10,094.0200 -7.6528***
CEO debt/equity ratio 0.1722 0.3649 0.0244 0.3444 0.9782 0.0893 -0.1722 -7.5656***
Relative CEO inside debt ratio 0.4982 0.9351 0.0660 0.8645 1.4190 0.2569 -0.3663 -8.7629*** Relative CEO inside debt ratio > 1 0.1635 0.3699 0 0.2693 0.4438 0 -0.1058 -7.0157*** Relative CEO incentive ratio 0.5777 1.2460 0.0719 0.9882 1.8236 0.2265 -0.4106 -7.4904*** Total current compensation ($ thsd) 824.7082 379.4175 779.4210 914.2158 422.2193 869.4530 -89.5076 -5.8949*** CEO vega ($thsd) 345.6620 366.5153 213.9544 362.2939 377.2825 233.9979 -16.6319 -1.1658 CEO delta ($thsd) 313.0926 376.7769 177.4353 420.1799 459.5541 239.0679 -107.0873 -6.9314*** Firm size 7.4364 1.6243 7.3306 7.5136 1.5683 7.5564 -0.0772 -1.2417 Total assets ($mln) 6,6446.0880 14,325.3600 1,526.3550 6,561.3110 14,428.3900 1,913.0250 -115.2228 -0.2078 Market-to-book ratio 3.1941 2.6287 2.5350 2.9418 2.3993 2.4340 0.2533 2.5397** Sales growth 1.0678 0.1704 1.0632 1.0502 0.1595 1.0480 0.0175 2.7072*** Return on assets 0.0609 0.0819 0.0624 0.0623 0.0758 0.0646 -0.0014 -0.4453 Book leverage 0.3633 0.1698 0.3693 0.3583 0.1752 0.3633 0.0050 0.7558 Cash surplus 0.1183 0.0817 0.1061 0.1111 0.0773 0.0978 0.0072 2.3096**
Notes. This table presents the descriptive statistics for the sample split by CEO at least 60 years old dummy and includes the mean, median, standard deviation and
statistical difference in mean values for my dependent, independent and control variables. Difference shows the difference in mean values between the two samples calculated as the mean value for the CEOs younger than 60 years old sample minus the mean value for the CEOs at least 60 years old sample. T-statistic represents the
