Option Incentives, Leverage, and Risk-Taking

Transcription

1 Option Incentives, Leverage, and Risk-Taking Kyonghee Kim 1 Trulaske College of Business University of Missouri - Columbia kimkyo@missouri.edu Sukesh Patro College of Business Northern Illinois University spatro@niu.edu Raynolde Pereira Trulaske College of Business University of Missouri - Columbia pereirar@missouri.edu Abstract While extensive research examines the relation between equity-linked incentives in executive compensation and risk-taking by managers, the impact of capital structure on this relationship has received little empirical attention. Prior work suggests that heightened managerial career concerns arising from financial risk and monitoring by debtholders will result in leverage having a dampening effect on the relation between managerial risk-taking and equity incentives. We empirically evaluate this contention and find leverage significantly weakens the positive relation between option-based incentives in flow compensation and managerial risk-taking. These results hold after accounting for the endogeneity of both firm leverage and incentive compensation decisions. The attenuating effect holds for both short-term and long-term components of debt but is stronger for the short-term component. Overall, the results highlight the influence of capital structure on the relationship between option incentives and managerial risk-taking. Keywords: Option incentives, risk-taking, leverage, debtholder monitoring, career concerns JEL Classification Codes: G30, G32 1 Corresponding author: Kyonghee Kim, School of Accountancy, 327 Cornell Hall, University of Missouri-Columbia, Columbia MO 65211, Tel.: , kimkyo@missouri.edu

2 Option Incentives, Leverage, and Risk-Taking 1. Introduction The view that incentives linked to equity, particularly stock options, could encourage managers to undertake risk-increasing activities is widely shared by investors and policy makers, and is supported by findings in academic research (e.g., Coles, Daniel, and Naveen 2006; Low 2009; Dong, Wang, and Xie 2010; Athanasakou, Ferreira, and Goh 2013). However, there is relatively less evidence on how firm leverage affects the relationship between such incentives and managerial risk-taking. Existing theoretical and empirical research suggests that firm leverage is likely to dampen the intended impact of risk-increasing incentives provided to managers. This paper empirically evaluates the validity of this hypothesis. Our main interest in undertaking this inquiry is to better understand the relation between equity incentives provided via incentive contracts to encourage risk-taking and the extent to which the incentives induce managerial risktaking (henceforth incentives-risk-taking ). In addition to addressing the gap in academic research, such an understanding is of relevance to renewed interest by regulators on how executive compensation is related to firm risk management. 2 We argue that the moderating impact of firm leverage on the incentives-risk-taking relation will arise via the impact of firm leverage on managers career concerns and as a result of increased intensity of monitoring by debtholders. Higher leverage is accompanied by increases in the 2 The SEC in an attempt to improve transparency around the issue of corporate risk management, has adopted a new disclosure rule (effective February 28, 2010) requiring firms to provide information about how their compensation policies and practices are related to risk management. The new rule, aimed at helping investors determine whether the company has incentivized excessive or inappropriate risk-taking by employees, also outlines required disclosures about board leadership structure and the board s role in risk oversight. See 1

3 likelihood of financial distress including events such as violation of financial covenants, debt defaults, and bankruptcy. In addition to diminishing the risk-taking capacity of the firm directly, these events can also impact managers appetite for risk-taking because such events increase the likelihood of management turnover and diminish their ensuing career prospects. Gilson (1989) examines financially distressed firms and finds a higher than 50% likelihood that senior management experiences turnover if firms are either in default on their debt, bankrupt, or privately restructuring their debt to avoid bankruptcy. Further, he finds these managers are not hired by another exchange-listed firm for at least three years. Eckbo and Thorburn (2003) find CEOs of firms that file bankruptcy suffer large declines in income relative to those of matched non-filing firms. Similarly, Nini, Smith, and Sufi (2012) find that there is a significant increase in the likelihood of CEO turnover following a financial covenant violation. These findings suggest that, all else equal, the risk-increasing impact of equity compensation will decrease in firm leverage due to a shift in the manager s risk-taking appetite related to leverage. Related to the above, prior research also suggests that firm leverage is likely to increase the cost to risk-averse, under-diversified managers of making risky investments. For example, Chava and Roberts (2008) find that covenant violations, when associated with creditors accelerating the loan, can lead to a decrease in investments in capital. Tan (2013) finds that managers are more conservative in financial reporting after a covenant violation and Vashishtha (2014) finds that firms reduce disclosure after a covenant violation. This evidence, consistent with managers increased prudence in response to more stringent debtholder monitoring after a covenant violation, identifies channels via which financing impacts managers investing and reporting choices. We argue that in addition to these impacts, increased monitoring by debtholders and the consequences of covenant violation are also likely to increase the personal cost to managers of undertaking risk- 2

4 increasing activities. This would imply that incentive compensation that is intended to induce managerial risk-taking would be less effective, all else equal. To summarize, to the extent that riskaverse managers perceive higher leverage to be associated with greater career concerns and/or more stringent monitoring by debtholders, it will be more costly for firms with higher leverage to induce managerial risk taking via incentive contracts. We test this hypothesis using a general cross-section of U.S. firms covering the years Following prior research, we assess risk-taking in terms of the firm s volatility of stock returns (e.g. Low 2009, Bargeron, Lehn, and Zutter 2010). Option value increases in risk. In addition, option incentives introduce the convexity between firm performance and managers wealth, thereby motivating managers to take on risky but value-enhancing projects. Consistent with this view, Guay (1999) finds that stock options, not common stockholdings, significantly increase the sensitivity of CEO wealth to stock return volatility. Since we are interested in examining the impact of incentive contracts, as part of firms ongoing compensation policy, on managerial risk taking, we focus on option-based incentives in CEO annual incentive contracts. Specifically, we use three alternative measures of option-based incentives in the CEO s annual flow compensation: the proportion of total CEO compensation in the form of stock options (Option-Pct), the sensitivity of newly granted CEO option-based compensation to stock price (Option Delta Ratio), and the sensitivity of newly granted CEO option-based compensation to stock volatility (Option Vega Ratio). We relate these measures of option incentives to future return volatility over one-, two-, and three-year periods subsequent to the year in which firm leverage and option incentives are measured. Prior research often focuses on managers existing ownership of options in addition to option incentives in flow compensation. Therefore, we also estimate risktaking incentives from the CEO s option portfolio (i.e., CEO ownership of vested or unvested 3

5 options at the beginning of the fiscal year and options granted during the fiscal year) and use them to check the robustness of our results. We note a twofold challenge to testing the impact of leverage on the incentives-risk-taking relation. First, both firm leverage and risk-increasing incentives are endogenously determined. Second, these variables themselves are closely related. Extensive prior work suggests that firm leverage impacts the determination of optimal compensation contracts (see for example, John and John 1993). Prior work also finds that compensation contracts can affect the leverage choice (Coles, Daniel, and Naveen 2006; Ortiz-Molina 2007) and debt maturity (Brockman, Martin, and Unlu 2010; Chava and Purnanandam 2010). Similarly, compensation contracts can also have an impact on the cost of debt (Ortiz-Molina 2006). To overcome these challenges, we adopt an identification strategy that involves panel regressions with firm fixed effects, an instrument variable approach, and the use of financial crisis as an exogenous shock to the treatment effect. Additionally, exploiting the fact that a firm s credit rating reflects the risk arising from its debt burdens but is unlikely to be as tightly associated with CEO compensation as leverage, we use credit ratings as an alternative to leverage to capture the CEO s career concerns when examining the effect of option incentives on managerial risk taking. Our findings are as follows. In panel regressions that treat firm leverage and CEO compensation as exogenous, we find that while future stock return volatility increases in optionbased incentives as documented in prior studies, the positive association is significantly attenuated by firm leverage. The moderating effect of firm leverage gets weaker as the horizon over which risk-taking is measured is extended but persists significantly for volatility measured in up to a three-year horizon subsequent to the measurement of leverage and incentives. A one-percent increase in total long-term leverage causes more than a three-percent decrease in the positive 4

6 impact of option incentives on risk-taking. We expect that impending cash outflows for the payment of short-term debt impose greater restrictions on managerial risk-taking induced by option incentives. Consistent with this conjecture, we find that the short-term portion of leverage (debt maturing within the next year) is a significantly stronger contributor to the attenuating effect than is long-term leverage. While the corresponding effect of the long-term component of leverage is weaker, it is also significant. These findings are robust to controlling for CEOs total firmspecific wealth, firm-fixed effects and secular variation in firm-risk. We next use an instrumental variable (IV) approach to estimate these relationships. Lemmon, Roberts, and Zender (2008) show that the leverage ratios of firms remain significantly stable over long periods of time and vary significantly with their leverage at the initial public offering (IPO) and with industry leverage. Therefore, we instrument firm leverage with the firm s initial leverage and industry median leverage. For the option-based incentive measures, we use the median option compensation of the firm s industry, indicators for high-technology industries (Business Service, Computers, and Electronic Equipment), and the firm s cash holdings as instruments. Because we estimate the moderating impact of firm leverage, we also instrument the interaction term of firm leverage and the option measures (Woolridge 2002). The IV approach yields a pattern of results that closely resembles the results using the panel estimation method. Here also, the impact of a one-percent increase in firm leverage is large, suggesting more than a five-percent reduction in the positive impact of option incentives on risk-taking for a one percent increase in leverage. To further address the endogeneity issue, we first re-examine the effect of leverage on the incentives-risk-taking relationship using the financial crisis during as an exogenous shock. To the extent that the financial crisis exacerbated (perceived) bankruptcy risks associated with leverage, we expect the dampening effect of leverage to be stronger during the financial crisis 5

7 period. Expectedly, we find that the attenuating impact of leverage strengthens during the years of the financial crisis. However, the impact in the non-crisis sample years persists and is strong. Next, we replace firm leverage with the firm s credit rating as an alternative measure of bankruptcy or default risk and test whether credit ratings similarly moderate the effect of option incentives on managerial risk taking. We find that our inferences remain unchanged when credit ratings are used to test our hypothesis. To further examine the sensitivity of our results, we re-estimate our measures of risk-taking incentives using the CEO s option portfolio and rerun our panel regressions. Similar to our findings with Option Delta (Vega) Ratio we find that the corresponding ratios yield very similar results. We also re-examine our hypothesis using alternative measures of CEO risk-taking. These include the volatility of the firm s future earnings and its future investments in research and development (R&D). The findings show that our inference is robust to using these alternative measures of risktaking. We contribute to several strands of extant research. The impact of compensation design on managerial behavior has long been an important area of interest in finance and accounting research (Indjejikian 1999; Lambert 2001; Rajgopal and Shevlin 2002; Core, Guay, and Larcker 2003). Our findings add to this body of evidence. Further, despite the intuitive appeal of the argument that incentive compensation can induce managerial risk-taking, the empirical evidence on the effect of equity-based incentives on managerial risk-taking behavior is inconclusive (Low 2009, see also Hayes, Lemmon, and Qiu 2012). One explanation offered for this weak evidence is that estimation of the equity incentives-risk-taking relation is complicated by endogeneity issues 6

8 wherein managerial compensation itself is assumed to be impacted by firm risk (Coles et al. 2006). 3 Although the endogeneity issue is an important explanation, relatively limited attention has been paid to other firm-specific policies that impose incentive effects and that may counteract or strengthen the influence of equity-based incentives on risk-taking. Accounting for the counterbalancing impact of firm leverage helps reveal the strength of the positive relationship between equity-based incentives and risk-taking throughout our estimation we find that the measures of option incentives show a strong positive association with measures of risk-taking. To place our results in perspective, it is useful to relate our findings to those of Brockman, Martin, and Unlu (2010). The authors assume that option incentives, due to the sensitivity of option value to firm risk, will induce managerial risk-taking and argue that, as a result, such incentives will also induce managers to opt for short-term debt to attenuate the effect of compensation risk on the cost of debt. By contrast, our evidence is supportive of the contention that leverage moderates the incentives-risk-taking relation itself. This finding has relevance to the debate over the effect of option compensation on excessive managerial risk-taking and increasing disclosure requirements by the Securities and Exchange Commission (SEC) on executive compensation. 4 Our finding suggests that the effect of option compensation on managerial risk-taking needs to be evaluated in the context of various counter-balancing factors, such as firm leverage. Our results also contribute to the literature that examines the value relevance of capital structure. Prior work suggests that leverage can impact managerial behavior by reducing the buildup of larger cash balances (Jensen 1986), by deterring overinvestment (Stulz 1990), and by 3 Taking into account this simultaneous relation, Coles, Daniel, and Naveen (2006) document evidence supportive of a positive relation between equity-based incentives and risk-taking. 4 SEC final rule A (Executive Compensation and Related Person Disclosure) effective November 7, 2006, and SEC final rule (Proxy Disclosure Enhancements) effective February 28,

9 better aligning managers interests with those of shareholders (Grossman and Hart 1982; Harris and Raviv 1988). The moderating effect of firm leverage on the risk-inducing impact of option incentives suggests an indirect effect of leverage that has not been sufficiently recognized in this literature. Our contributions are subject to the following caveats. As noted above, prior research finds that managerial incentives can impact firm leverage and vice versa. While we attempt to address this endogeneity by using an IV approach, it is unlikely that such an approach is sufficient to properly reveal the full extent of the relation between managerial incentives and leverage. In this regard, it is reassuring that our results hold across different specifications and measures of risktaking and leverage. A further issue is that firm risk itself can impact compensation contracts and ultimately even firm leverage. We attempt to address this issue by considering the volatility of future firm returns and panel regressions with firm fixed effects. The rest of the paper is organized as follows. Section 2 discusses measurement of key variables, outlines the sample formation process, and presents descriptive statistics. Sections 3 and 4 present the results and robustness tests, and Section 5 concludes the paper. 2. Variable Measurement and Sample Description 2.1. Measurement of Dependent and Key Independent Variables Option-Based Incentive Measures: We use three measures of the extent of option-based incentives in the CEO's flow compensation as reported in the Execucomp database. Our first measure, Option-Pct, is the ratio of Black-Scholes value of newly granted stock options (Current Option Grants) to annual flow compensation (Total Comp). To maintain consistency across the change in reporting requirements due to the passage of Statement of Financial Accounting 8

10 Standards (SFAS) 123R, we estimate the Black-Scholes value of stock options in the post-sfas 123R period in line with the methodology used by Execucomp. Option Pct = Current Option Grants (Total Comp) The second measure is based on the delta of newly granted stock options. Managers can improve firm value by investing in risky but positive net present value (NPV) projects. Option delta is expected to be positively associated with risk-taking because of the incentive effects created by correlating CEO pay to firm value. For example, Datta, Datta, and Raman (2001) and Agarwal and Mandelker (1987) find that CEOs make variance-increasing acquisitions when their pay has a larger equity-based component. We use Core and Guay's (2002) algorithm to calculate option delta for new option grants made to the CEO in the fiscal year. In our analysis, following Bergstresser and Philippon (2006), we use the ratio of the current option delta to the sum of salary, bonus and current option delta (Option Delta Ratio). This measure captures the fractional contribution the CEO s current option grants make to CEO compensation due to a one percent appreciation in the share price. Because this ratio captures option incentives in a scale invariant way, it avoids issues arising from having to model the quantity of incentives. 5 Our use of scaled incentives is also similar in sprit to Edmans, Gabaix, and Landier (2009) who scale the sensitivity of total CEO wealth to firm value by the annual flow compensation. They demonstrate that incentive compensation scaled by annual pay is independent of firm size, and hence a desirable empirical measure of incentives. Option Delta Ratio = Current Option Delta (Current Option Delta + Salary + Bonus) 5 This measure corresponds to INCENTIVE_RATIO CG in Bergstresser and Philippon (2006) but does not include shares owned by the CEO in the numerator. 9

11 The more recent literature that examines managers risk-taking incentives focuses on option vega the sensitivity of option value to a change in firm risk (e.g. Coles, Daniel, and Naveen 2006; Chava and Purnanandam 2010). Thus, our next measure is based on current option vega (Guay 1999). 6 To be consistent with our use of option delta, we also use current option vega scaled by the sum of current option vega, salary and bonus (Option Vega Ratio). Option Vega Ratio = Current Option Vega (Currnet Option Vega + Salary + Bonus) Managerial Risk Taking: Our main dependent variable is risk-taking by managers measured as the standard deviation of monthly stock returns (RetStd). RetStd is calculated in the 12-, 24- or 36- month period immediately following the fiscal period in which executive compensation and firm leverage are measured. As a robustness check, we use two other measures of risk-taking. These are the earnings volatility and the R&D expenditures of the firm, each measured over the subsequent 3-year period. Leverage: We use the ratio of book value of total long-term debt to total assets (total leverage, TLev) to capture default risk and debtholder monitoring. We later split TLev into the portion that matures within the next year (short-term portion of long-term leverage, SLev) and the remaining (long-term leverage, LLev). As an alternative measure of the firm s default risk, we use the firm s long-term credit rating (CR) as provided by Standard & Poor s (discussed further in Section 3.3 below). Similar to firm leverage, a larger value of CR reflects greater default risk. 6 While vega is theoretically a more direct measure of risk-taking, evidence suggests that just the use of options is associated with greater risk-taking. For example, DeFusco, Johnson and Zorn (1990) find that firms that approved stock option plans had increases in stock return volatility. Similarly, Agrawal and Mandelker (1987) find that CEOs with higher levels of stock and option ownership take on more variance-increasing acquisitions. 10

12 2.2. Sample We identify all firms in the Execucomp database with sufficient data to calculate option delta and vega for the CEO of the firm. This limits our sample to fiscal periods ending during 1995 through and yields an initial sample of 25,629 firm years. We further limit our sample to firms with required financial data in the Compustat database. This step reduces the sample to 21,663 observations. Requirement of return data in the CRSP monthly database to calculate stock return volatility (RetStd) further reduces the sample to 21,269 firm years. From this, we remove firms whose primary business is in the finance or utility industries. This yields our base sample of 17,923 firm years. 8 We winsorize the variables used in the study at the top and bottom 1% level when required Descriptive Statistics Table 1 reports the descriptive statistics of the key variables. The mean values of current option delta and current option vega for the sample are $34,390 and $28,750, respectively, for the period and are comparable to the level reported in prior studies (e.g., Hayes, et al. 2012). The mean value of Option Delta Ratio (Option Vega Ratio) is 2.30% (1.80%). The sample firms are generally mature and large average firm age is 23 years and the mean (median) book value of assets and sales are $4.941 ($1.153) billion and $5.140 ($1.168) billion, respectively. Total longterm leverage has a mean value of 20.9%, 18.0% of which is from debt with maturity greater than one year. Average credit ratings (CR) is reflecting an S&P rating for the average firm of between BBB (9) and BBB- (10). Standard deviation of monthly returns (RetStd) measured over the 12-, 24-, and 36-month period is in the rate of 12~13%. Earnings volatility measured as the 7 We stop at 2011 because we need the leading three years to measure firm risk. 8 The sample size is smaller in certain analyses due to additional data requirements. 11

13 standard deviation of the ratio of income before extraordinary items to assets has an average (median) value of (0.024). Detailed variable descriptions are reported in Appendix 1. Insert Table 1 Here Correlations reported in Table 2 show that our measures of option incentives are highly correlated with each other the Pearson correlation between Option Delta Ratio and Option Vega Ratio is 90%, and the correlation between Option Delta Ratio (Option Vega Ratio) and Option- Pct is 73% (69%). These high correlations suggest that Option-Pct, Option Delta Ratio, and Option Vega Ratio capture the same construct, i.e., incentives to take on risky, value-increasing projects. Consistent with the notion that firms with higher growth options use executive stock options more intensively, the market-to-book ratio of the firm (MTB) is positively correlated (p-value < 1%) with each of the option incentive measures. R&D expenditure (R&D) and capital expenditure (CAPEX) are similarly positively correlated (p-value < 1%) with the option incentive measures. Asset tangibility (PPE Ratio), generally used as an inverse measure of growth option, is negatively correlated to the option incentive measures. Interestingly, the credit rating (CR) is significantly (pvalue < 1%) negatively correlated to each of the option incentive measures, suggesting that CEOs receive less option-based incentives when default risk is higher. Insert Table 2 Here 3. Results The discussion in this section is divided into three parts. Section 3.1 presents univariate and panel estimation tests of our main hypothesis on the effect of firm leverage on the incentives-risktaking relationship. Section 3.2 outlines the methodology and presents results of tests in which the firm s leverage and incentive choices are endogenous. Section 3.3 examines how the impact of leverage varied during the recent financial crisis ( ). 12

14 3.1. Effect of Leverage on the Relation between Option Incentives and Risk-taking Univariate Tests Table 3 reports Pearson correlations between option-based incentives and firm risk across quintiles of firm leverage. Firm risk is measured by the standard deviation of monthly returns (RetStd) over the 1-, 2-, or 3-year period after the fiscal year end in which incentives and leverage are measured. The three measures of option incentives are Option-Pct, Option Delta Ratio, and Option Vega Ratio (as described in Section 2). The results show that the correlations between RetStd and the three measures of option-based incentives in the lowest leverage quintile are positive and significant at the 1% level, ranging from a high of 17.30% for the Option Delta Ratio (1-year RetStd) to a low of 8.70% for the Option Vega Ratio (3-year RetStd). These correlations uniformly and monotonically decrease as we move across quintiles of increasing leverage for each combination of incentive measure and return volatility horizon. In the highest leverage quintile, the correlations are all negative and significant at the 1% level, ranging from a low magnitude of 5.70% for the Delta Ratio (2-year RetStd) to a high magnitude of 13.30% for the Vega Ratio (3-year RetStd). Fisher's Z-tests indicate the changes in correlation from the lowest to highest quintile of leverage are all significant at the 1% level. These results constitute the first piece of evidence supporting our main prediction that leverage is associated with an attenuation of the positive relationship between option-based incentive compensation and managerial risk-taking. Insert Table 3 Here Panel Estimation Results 13

15 We next estimate the following regression model to examine the relationship between optionbased incentive compensation and managerial risk taking conditional on firm leverage (TLev). RetStd i,t+1,t+3 = α0 + α1option Ratioit + α2(option Ratioit TLevit) + α3tlevit + α 4 log(sales) it + α 5 PPE Ratio it + α 6 CAPEX it + α 7 MTB it + α 8 R&D it + α 9 Total Wealth i,t 1 + Firm fixed effects + Year fixed effects + ε i,t+1,t+3 The dependent variable (RetStd) for firm i is the standard deviation of monthly returns measured over the 12-month (year t+1), 24-month (years t+1 through t+2), or 36-month (years t+1 through t+3) horizon as described before. Option Ratioit is option-based incentive compensation for firm i in year t and is measured as Option-Pct, Option Delta Ratio or Option Vega Ratio. The interaction between each of the option incentive variables and firm leverage (Option Ratio*TLev) captures the moderating effect of firm leverage on the relationship between option-based incentives and risk-taking. We include several control variables that are known in the literature to influence managerial risk taking and incentive contracts. These include firm size (measured by the log transform of sales, log(sales)) and asset tangibility (measured by the proportion of total assets in the form of property, plant and equipment, PPE Ratio). We also include the market-to-book ratio of equity (MTB), the ratio of capital expenditure to assets (CAPEX), and the ratio of R&D expense to assets (R&D) as measures of growth opportunities (Coles, Daniel, and Naveen 2006). Existing equity ownership of CEOs can influence both their compensation contracts and risktaking. Therefore, we include in the model total firm-specific wealth of the CEO (Total Wealth), i.e., the value of the CEO s stock and option holdings at the beginning of the fiscal period. To account for unobserved across-firm heterogeneity in risk-taking we include firm-fixed effects in 14

16 the model. Finally, year-fixed effects are included in the model to account for secular variation in firm risk. Insert Table 4 Here The results reported in Table 4 support the hypothesis that leverage attenuates the impact of option-based incentive pay on managerial risk-taking. The coefficient on the interaction between CEO option incentives and firm leverage (Option Ratio*TLev) is negative and significant at the 1% level in each of the nine regression models (i.e., all combinations of the three measures of option-based incentives and standard deviation of returns measured over three different windows). The t-statistics range from a low magnitude of 4.08 (for 3-year RetStd using the Option Vega Ratio) to a high magnitude of 6.78 (for 1-year RetStd using the Option-Pct). 9 The magnitude of the coefficient on this interaction term and its significance decrease as the horizon over which the RetStd is measured increases the drop is over 20% for the delta- and vega-based incentive ratios, and a little over 10% for Option-Pct. Consistent with prior studies (Guay 1999; Rajgopal and Shevlin 2002; Coles, Daniel, and Naveen 2006; Armstrong and Vashishtha 2012), the main effect of option incentives on return volatility is positive and significant with t-statistics ranging from 4.44 to Leverage (TLev) also shows a strong positive correlation with RetStd t-statistics are in the range from 6.35 to Consistent with prior evidence, RetStd is highly negatively (pvalue < 1%) related to firm size and positively correlated with measures of firm growth including the ratios of capital expenditure and R&D expenses to assets. The association of RetStd with lagged value total firm-specific wealth of CEOs (Total Wealth) is positive but its significance is low. 9 T-statistics are based on standard errors that are robust to heteroscedasticity and clustering of observations at the firm level. 15

17 The moderating effect of leverage documented in Table 4 is economically significant. For example, a one-percent increase in firm leverage reduces the positive impact of incentives on risktaking by over 4% 10 when incentives are measured by Option-Pct. For Option Vega Ratio and Option Delta Ratio, the impact of a similar increase in leverage suggests a decrease of over 3%. Thus, an increase in leverage of one standard deviation (0.17) causes a reduction in the impact of risk-taking option incentives in the range of 55% (for 1-year RetStd using Option Delta Ratio) to 94% (for 3-year RetStd using Option-Pct) Effect of Debt Maturity From a theoretical perspective (Douglas 2006), we would expect that the moderating effect of leverage on the managerial incentives and risk-taking relationship derives mainly from long-term debt. This is consistent with the assumption in Douglas (2006) that it is relatively more difficult for the firm to renegotiate its capital structure than it is to readjust managerial incentives. However, it is also possible that short-term debt places more immediate constraints on managers risk taking due to its impending cash flow demand. To examine whether and how this distinction in debt maturity affects our results, we split long-term debt into two parts, the portion of debt maturing within one year (SLev) and the portion with a maturity greater than one year (LLev), each measured as a fraction of total assets. Insert Table 5 Here Table 5 shows that the moderating effect of debt on the relationship between option incentives and return volatility arises both from the long-term and short-term components of leverage. For 10 Based on Table 4, for a 1% increase in leverage, the impact of Option-Pct on RetStd decreases by 0.01* As a proportion of the direct impact of options, this translates to 0.01*(0.094/0.022) = 4.27% decline. 16

18 Option Delta Ratio, the t-statistics on the interaction term with LLev range between and and with SLev from to Corresponding ranges of t-statistics for Option Vega Ratio are from to (LLev) and from to (SLev); and for Option-Pct they are from to (LLev) and from to (SLev). Thus, the moderating impact of both components of leverage is significant at the 1% level across all the incentive measures and risk estimation horizons. However, the test results reported at the bottom of Table 5 show that the impact of the short-term component is generally stronger than that of long-term component of leverage. For 1- and 2-year RetStd, the difference is significant at or below the 1% level for all three measures of option incentives. For 3-year RetStd, the difference is significant at the 10% level for Option Vega Ratio and Option Delta Ratio. In terms of economic significance, the magnitudes of the interaction terms are comparable to those reported in Table 4. Also, the main effects for the long-term and short-term components of leverage show that they are each highly positively (p-value < 1%) correlated with RetStd SLS Estimation The above tests treat the CEO s incentives and firm leverage as being predetermined. However, the firm s compensation policy and capital structure are endogenously determined. A manager may determine the firm s capital structure based on firm-specific and manager-specific factors, including his compensation contract (Coles, Daniel, and Naveen 2006; Ortiz-Molina 2007). Alternatively, the board may design the manager s compensation contracts given the firm s leverage (John and John 1993; Douglas 2006). Such endogeneity can lead to bias in OLS parameter estimates. We implement an instrumental variable approach to address this issue. Specifically, we re-estimate the effect of leverage on incentives-risk-taking relation using two-stage least squares 17

19 regressions (2SLS), where firm leverage, the option-incentive measures, and the interaction of the option incentive measures and firm leverage are endogenous regressors. Lemmon et al. (2008) show that leverage ratios of U.S. firms display remarkable time-series stability. In particular, they show that the firm s initial leverage is the most significant of a set of known determinants of the firm s current leverage ratio. Based on this evidence of the time-series dependence of firm leverage, we use the firm s initial leverage ratio as an instrumental variable (IV) for its current leverage ratio. In addition, we also use the median leverage ratios of 2-digit SIC Compustat peers of the firm the second most significant determinant in the model of Lemmon et al. (2008). Instrumental variables (IVs) for the option incentives measures are the respective industry median incentive ratios (based on all 2-digit SIC firms in the Execucomp database), indicators for high-tech industries (Business Service, Computers, and Electronic Equipment), and the ratio of cash to book value of assets. Because our main variable of interest is the interaction of option-based incentives and leverage, we follow the approach proposed by Wooldridge (2002, pages ) for estimating the coefficient on the interaction effect in 2SLS we use the interaction term as an additional endogenous regressor and use all excluded exogenous variables in the system, i.e., IVs for leverage and option incentive measures, as IVs of the interaction term. Insert Table 6 Here Table 6 reports the 2SLS regression results. Because results for the different horizons of risk measurement are quite similar, we report the results for risk measured over the 36-month horizon. Further, for reasons of brevity, we only report the second stage results for the main equation of 18

20 interest, i.e., where firm risk is the dependent variable. 11 Adjusted R 2 of the first stage regressions with only excluded instrumental variables are in the range of 5-17% (not reported) and the F- statistics for IVs are highly significant (p < 0.00). Hansen s J statistics reported at the bottom of Table 6 show that the model does not reject the overidentification restrictions of the IVs these diagnostic test results suggest that the 2SLS estimates reliable. The 2SLS result show that our main inference remains largely unaltered the impact of leverage on the incentives-risk-taking relationship is still significantly negative. 12 As before, the interaction term is significant at the one percent level in each of the models and the impact is quite strong a one percent increase in leverage is associated with a 4.50% to 5.00% reduction in the impact of option-based incentives on risk-taking. The main effects of incentives and leverage also remain positive and significant (pvalue < 1%). The impact of the other control variables remains similar firm risk is negatively associated with Sales and the PPE Ratio, and positively associated with capital expenditure (CAPEX) and R&D. In addition, the positive relationship between firm risk and total firm-specific CEO wealth (Total Wealth) is now significant at the 5 percent level Effect of Leverage Exogenous Shock from the Financial Crisis of The financial crisis resulted in the worst economic recession since the Great Depression in 1930s and caused a significant liquidity shock in the global economy. Declines in corporate profits and the availability of external credit along with an uncertain macroeconomic outlook significantly increased the riskiness of firms existing debt. 13 In this section, we use the 11 First stage regressions are reported in Appendix The results are robust to using 3SLS full information estimation where leverage and options are simultaneously determined. 13 The S&P 500 index fell from a peak of 1,565 in Oct 2007 to 676 in March 2009, a decline of 56.8% (Federal Reserve Bank of St. Louis). U.S. corporate profits decreased from $1,541.7 billion in 2007 to $ billion by the first quarter of 2009 according to the Bureau of Economic Analysis (News Release, 3 rd quarter, 2009). 19

21 financial crisis as an exogenous shock to the risk-related impact of firm leverage. We expect that the moderating effect of leverage on the incentive-risk-taking relation strengthened as the risk associated with firm leverage increased during the financial crisis period. To test this hypothesis and also to check whether our results are solely driven by effects of the financial crisis, we re-estimate the regression model with the interaction term of option incentives and leverage conditioned on whether the observation falls within the crisis period of (Option Ratio*TLev Financial Crisis) or not (Option Ratio*TLev Non-Financial Crisis). Regression results reported in Table 7 show that the impact of leverage during both periods was significant (p-value < 1%) t-statistics range from a low magnitude of 3.68 to a high magnitude of Further, test result at the bottom of the table shows that the moderating impact of leverage, as expected, is generally stronger during the years of the financial crisis. 14 When incentives are measured by Option Delta Ratio and Option Vega Ratio, the impact of leverage during the financial crisis is stronger at the one percent level. For Option-Pct, however, there is no significant difference across the two regimes. The other effects in the model remain largely unchanged. Insert Table 7 Here 4. Robustness Tests 4.1. Effect of Credit Ratings on the Relation between Option Compensation and Risk-taking In this section, we use firms credit ratings as an alternative to leverage and re-examine our hypothesis. Graham and Harvey (2001) provide survey evidence that credit ratings follow financial flexibility as the second most important concern for Chief Financial Officers (CFOs) when 14 We also estimate a regression model that includes the main effect of the financial crisis, i.e., RetStd = f(option Ratio, Option Ratio*TLev, Option Ratio*TLev* Financial Crisis indicator, TLev, Financial Crisis indicator, control variables, and firm fixed effects). The results are similar and our inference remains the same. 20

22 determining capital structure. Prior studies also document that firms existing debt obligations are also an important determinant of credit ratings (e.g., Adams, Burton, and Hardwick 2003). These prior findings suggest that credit ratings are systematically associated with firm leverage. Further, similar to firm leverage, credit ratings are often used as a measure of borrowers repayment or credit risk (e.g., Murfin 2012; Acharya, Davydenko, and Strebulaev 2012). However, credit ratings are unlikely to vary with incentive compensation as strongly as firm leverage. Therefore, we use credit ratings as an alternative measure of firm leverage that is less subject to the endogeneity problem. Insert Table 8 Here Table 8 reports results of tests where firm leverage is replaced by the credit rating of the firm (ranging from 1 for AAA to 22 for D). Similar to our main results, the interaction of the option incentive measures and the firm s credit rating (Option Ratio*CR) is negative and significant for each incentive measure and risk horizon combination. For Option-Pct, the t-statistics range from to and for Option Delta Ratio from to For Option Vega Ratio, the results are a bit weaker with t-statistics of -1.99, and for the 12-, 24-, and 36-month horizon, respectively. Also, similar to firm leverage, credit ratings are significantly positively correlated with firm volatility with t-statistics ranging from a low of 7.09 to a high of The main effects of option incentives (Option-Pct, Option Delta Ratio, and Option Vega Ratio) also remain positive and highly significant. Overall, tests of the main hypothesis using credit ratings yield results that are quite similar to the main results Alternative Measures of Option-Based Incentives (Using the CEO s Option Portfolio) 21

23 In this sub-section, we re-examine our panel regression results of Table 4 after estimating our measures of option-based incentives from the CEO s option portfolio. Option portfolio includes vested and unvested options held by the CEO from prior grants and the grants made during the fiscal year. We measure option delta and vega of the CEO s option portfolio, and similar to current option delta and current option vega, scale them by the sum of salary, bonus, and delta (or vega) of option portfolio. Option Portfolio in Table 9 denotes either Option Delta Ratio or Option Vega Ratio of CEOs option portfolio. Insert Table 9 Here The results based on risk taking incentives from CEOs option portfolio are generally consistent with the hypothesis that firm leverage moderates option-incentives-risk-taking relation. Table 9 shows that similar to the results based on option-incentives in CEO flow compensation, the coefficient on interaction between Option Portfolio and TLev is negative and significant at the p < 0.05 level across all three return volatility measures, whether Option Portfolio is measured as Option Delta Ratio or Option Vega Ratio. We also find that Option Delta Ratio is positive and significantly associated with return volatility. The coefficient on Option Vega Ratio is positive but insignificant Alternative Measures of Risk-taking In this sub-section, we test our hypothesis using the firm s earnings volatility and the ratio of R&D expenditures to assets as alternative measures of risk-taking by managers. Panel A of Table 10 reports results of our tests using earnings volatility measured over the three years following the year in which option incentives and leverage are measured. Firm leverage has the similar negative moderating impact (p-value < 1%) on managerial risk-taking induced by option-based incentives. 22

24 As observed earlier, the magnitude of coefficient on the interaction term, Option Ratio * TLev, suggests that the moderating effect of leverage is both statistically and economically significant a one percent change in leverage is associated with a reduction in earnings volatility ranging from 2.84% (Option Vega Ratio) to 5.77% (Option-Pct). The main effects of option-based incentives and leverage on risk-taking are positive and significant. However, only the market-to-book ratio impacts earnings volatility all the other control variables do not have a significant impact. Insert Table 10 Panels A and B Here R&D investments involve higher risk because the investment outcomes are more uncertain, particularly in comparison to investments in capital expenditures and other low innovation assets. Consistent with several prior studies, we use R&D expenditures as an alternative measure of risktaking by managers (Coles, Daniel, and Naveen 2006; Bargeron, Lehn, and Zutter 2010). Specifically, we rank firms as being a high-r&d firm if the average ratio of R&D expenditure to assets measured over the leading three years is in the top quartile of the sample. We then estimate a logit model for the likelihood of high-r&d investment. Consistent with prior work the model has the following control variables firm size measured by the log transform of assets, asset tangibility (PPE Ratio) and market-to-book ratio (MTB), firm age, the Altman Z-score, and an indicator variable for a dividend cut in the prior fiscal period. As before, the model includes a measure of option incentives, firm leverage, the interaction of the two, total firm-specific CEO wealth, and firm and year fixed effects. Results reported in Panel B of Table 10 show that leverage has a similar negative impact on the relation between option incentives and risk-taking as before the t-statistics on the interaction term are between and Overall, the results suggests that the results are robust to using the alternative measures of risk taking. 5. Conclusion. 23

25 This paper presents evidence that option-based incentives in flow compensation are less effective in inducing risk-taking by managers as firm leverage increases. Our examination is motivated by the gap in the literature that examines the impact of incentive compensation and leverage on risk-taking. The testable hypothesis pursued in the paper stems from the argument that firm leverage imposes labor risk on managers in the form of consequences that could arise from financial covenant violation, financial distress, or bankruptcy. This suggests that in the presence of risky debt, risk inducement mechanisms such as stock options will be less effective in motivating under-diversified managers to take on more risk. A similar prediction stems from the argument that monitoring by debtholders can increase the cost to risk-averse and under-diversified managers of undertaking risk-increasing activities in the firm s project portfolio. Our documentation of a significant moderating impact of firm leverage on the incentives-risktaking relation adds to the evidence on the impacts of financing on investment. More recently, research has focused on the channels through which financing impacts investment. Our finding suggests that the impact on the manager s risk-taking is one such channel. Our finding also contributes to the evidence on the impact of incentive compensation on managerial risk-taking. Despite persuasive arguments that incentive contracts can be effectively used to induce managerial risk-taking, the empirical evidence is mixed. The finding that firm leverage counteracts some of the incentive effects provides a better understanding of the relationship between managerial risktaking and the incentives deployed to induce it. Thus, our findings are also relevant to the broader literature on the resolution of agency problems via incentive contracts and the firm s capital structure. 24

26 REFERENCES: Acharya, V., Davydenko, S., and Strebulaev, I., Cash holdings and credit risk, Review of Financial Studies 25(12), Adams, M., Burton, B., and Hardwick, P., The determinants of credit ratings in the United Kingdom insurance industry, Journal of Business Finance & Accounting 30(3&4), Agrawal, A., Mandelker, G., Managerial incentives and corporate investment and financing decisions. Journal of Finance 42, Armstrong, C., and Vashishtha, R Executive stock options, differential risk-taking incentives, and firm value, Journal of Financial Economics 104, Athanasakou, V., Ferreira, D., and Goh, L., Corporate investment and changes in CEO stock option grants. ECGI Finance Working Paper No Bargeron, L. L., Lehn, K. M., and Zutter, C. J., Sarbanes-Oxley and corporate risk-taking. Journal of Accounting and Economics 49, Bergstresser, D. B., and Philippon, T., CEO incentives and earnings management. Journal of Financial Economics 80, Brockman, P., Martin, X. and Unlu, E., Executive compensation and the maturity structure of corporate debt. Journal of Finance 65, Chava, S. and Purnanandam, A., CEOs and CFOs: Incentives and corporate policies. Journal of Financial Economics 97 (2), Chava, S., and Roberts, M., How does financing impact investment? The role of debt covenants, Journal of Finance 63(5), Coles, J. L., Daniel, N. D., and Naveen, L., Managerial incentives and risk-taking. Journal of Financial Economics 79, Coles, J., Daniel, N., and Naveen, L Calculation of compensation incentives and firmrelated wealth using Execucomp: Data, Program, and Explanation, Working Paper Core, J., and Guay, W Estimating the value of employee stock option portfolios and their sensitivities to price and volatility. Journal of Accounting Research 40, Core, J., Guay, W., and Larcker, D Executive equity compensation and incentives: A survey. Economic Policy Review 9(1), Datta, S., Datta, M.I., and Raman, K., Executive compensation and corporate acquisition decisions, Journal of Finance 56(6), DeFusco, R., Johnson, R. and Zorn, T., The effect of executive stock option plans on stockholders and bondholders. Journal of Finance 45, Dong, Z., Wang, C., and Xie, F., Do executive stock options induce excessive risk taking? Journal of Business & Finance 34(10), Douglas, A. V. S., Capital structure, compensation and incentives. Review of Financial Studies 19 (2), Eckbo, B. E., and Thorburn, K. S., Control benefits and CEO discipline in automatic bankruptcy auctions. Journal of Financial Economics 69 (1), Edmans, A., Gabaix, X., and Landier, A A multiplicative model of optimal CEO incentives in market equilibrium, Review of Financial Studies 22, Gilson, S.C., Management turnover and financial distress. Journal of Financial Economics 25 (2),