Are debt and incentive compensation substitutes in controlling the free cash flow agency problem?

This paper investigates the governance implications of a firm's capital structure and managerial incentive compensation in controlling the free cash flow agency problem. The results suggest: debt and executive stock options act as substitutes in attenuating a firm 's free cash flow problem; failure to incorporate the substitutability and endogeneity leads to underestimates of the magnitude and economic implication of the disciplinary role of both mechanisms; firm characteristics differ across the prevalence of debt usage versus option usage, suggesting the heterogeneity in the costs and benefits of the monitoring devices; and all the above effects are more pronounced in firms that tend to have more severe agency problem.

**********

Agency theory predicts that self-serving managers may make nonvalue-maximizing decisions with internal free cash flow by investing in negative net present value projects for private benefits (Jensen, 1986). This behavior is known as overinvestment associated with the free cash flow agency problem, which reflects the conflicts between shareholders and managers. It is important to analyze effective mechanisms to mitigate this agency problem. These mechanisms can either restrict the available resources for overinvestment or align the interests between shareholders and managers. In this paper, I focus on debt and incentive compensation that represent the two types of devices, respectively. Debt directly reduces the free cash flow due to the precommitment of interest payment. On the other hand, incentive compensation prevents managers from investment overaggressiveness through interest alignment. In addition, these two are influential mechanisms considering the importance of capital structure and compensation structure in a firm's policy decisions. Imposing discipline by setting managerial incentives or adjusting capital structure is relatively easy as compared with other mechanisms. (1)

Previous empirical studies have looked at the effects of debt and incentive compensation separately. For example, Harvey, Lins, and Roper (2004) argue that debt mitigates the free cash flow problem in emerging markets, where overinvestment agency costs are potentially extreme. Hadlock (1998) and Broussard, Buchenroth, and Pilotte (2004) examine the effects of incentives on firm's investment cash flow sensitivity. While Hadlock (1998) indicates that overinvestment is not a dominant problem for firms in the 1970s, Broussard, Buchenroth, and Pilotte provide evidence that chief executive officer (CEO) incentives alleviate the overinvestment problem for firms over the sample period 1993-1997. There is only one theoretical paper by Garvey (1997) that jointly examines debt and incentive compensation in the free cash flow agency framework. Garvey models an explicit comparison of using capital structure versus compensation structure to resolve the free cash flow problem, suggesting that a manager's private information, cash flow risk, and the possibility of undoing the incentive alignment may explain a firm's optimal policy choice. However, there is no empirical analysis that addresses both the governance capacity of the two mechanisms and their interaction. My paper seeks to fill the gap in the literature. First, I look at the interaction of debt and incentive compensation when dealing with the free cash flow problem. They can be substitutes considering their overlapping governance capacity; they can also be complements as they alleviate the agency problem in different ways and may be employed in concert to achieve an optimal effect. Second, I investigate whether the effects of debt monitoring and incentive alignment as documented in prior studies will change when incorporating their interaction. For a given firm, the relation between the two mechanisms should depend on their benefits and costs. I address this issue by comparing some firm characteristics across the prevalence of debt usage and incentive compensation usage. The interaction of the two devices appears more interesting given the evidence that some firms predominantly grant stock options, but issue too little debt even though they have excess debt capacity. Therefore, an investigation of this issue has important corporate governance implications on a firm's optimal choice of capital structure and compensation structure.

To explore the free cash flow governance implications of debt and incentive compensation, I first investigate the direct effects of debt and incentive compensation on a firm's investment policy. In this analysis, I recognize that debt and incentives attenuate the free cash flow problem in different ways. Debt, which bas a direct claim on a firm's earnings, restricts overinvestment by reducing available cash flow. A negative relationship between investment rate as measured by capital expenditures scaled by beginning-of-year capital and leverage is documented to provide preliminary support for the constraining effect of debt. However, the negative relationship may be driven by a firm switching from capital expenditures to research and development (R&D). To explore this possibility, I analyze the effect of leverage on total investment defined as capital expenditures plus R&D. I also calculate a residual investment measure, obtained by regressing investment rate on growth potential. This represents the abnormal investment that is not explained by the firm's growth opportunity. Both total and residual investments decrease with leverage, consistent with the hypothesis that debt servicing obligations help discourage overinvestment of free cash flow by self-serving managers.

Incentive compensation exerts a constraining effect through a manager/shareholder interest alignment mechanism. High alignment reduces the tendency of managers to squander internal free cash flows. Therefore, incentive compensation mitigates the overinvestment problem by reducing the sensitivity of investment to available cash flows. Previous studies either focused on managerial share ownership (Hadlock, 1998) or CEO pay-for-performance sensitivity from the total shareholdings (Broussard, Buchenroth, and Pilotte, 2004). However, I demonstrate that direct stockholdings tend to make managers entrenched and exacerbate the overinvestment problem. Therefore, I use the value of CEO option holdings scaled by his or her total equity wealth to investigate the governance of incentive compensation on free cash flow.

In addition to estimating the impact of debt and incentive compensation for the average firm, I explicitly consider a firm's potential for over or underinvestment. Doing so enables me to isolate the hypothesized effects of agency costs from those of alternative hypotheses. The agency-related overinvestment problem is more serious in mature firms with low growth perspectives (Jensen, 1986). The low-growth firms surfer from a shortage of positive net present value projects; therefore, it is more likely that the availability of additional cash flow may be associated with excess investment spending in those firms. On the other hand, an underinvestment problem is expected to be more pronounced for high-growth firms. Those firms have more profitable projects and/or more severe informational asymmetry when a large proportion of the firm's value is attributed to growth opportunities of which the quality is, to a large extent, unverifiable ex ante (Myers, 1977).

The study begins with a single-equation regression. I document a firm's investment (capital expenditure, total investment, and abnormal investment) decreases with leverage on average, an effect that is more pronounced in low-growth firms. In addition, the CEO's option holding reduces investment cash flow sensitivity. This effect is centered in low-growth firms. The single-equation results provide preliminary support that debt and options attenuate the free cash flow problem.

I then use a simultaneous equations technique to test the substitutability of debt and executive options. Within the simultaneous equations framework, the substitution hypothesis predicts the constraining effects of debt on investment and options on investment cash flow sensitivity, as well as a negative interaction between the two mechanisms. At the same time, this approach allows me to address the potential interaction of investment, capital structure, and managerial incentives. The endogeneity issue has not been explored in the previous investment cash flow literature, partly because those studies do not examine both capital structure and managerial incentives.

Controlling for endogeneity generates three implications. First, both debt and executive options mitigate the free cash flow agency problem, consistent with the single-equation results. Second, I find a negative relationship between leverage and CEOs' option holdings, especially for firms with more severe free cash flow problems, suggesting the two mechanisms are substitutes in controlling for this problem. These results provide a partial answer to the puzzle as to why some firms grant abundant stock options, but issue little debt, despite their excess debt capacity. Therefore, explicitly accounting for the substitution and interaction of the two mechanisms is important. Third, the single-equation model underestimates the magnitude and economic significance of the disciplinary role of both mechanisms. This downward bias can be attributed to the substitution effect bias and/or omitted variable bias.

The above tests focus on debt level and option holdings. An alternative perspective on the disciplinary role of capital structure and compensation structure would suggest that firms with greater costs of free cash flow tend to use more debt or executive options. Thus, I extend the analysis by investigating the change of capital structure and flow compensation. I find that a firm's debt issuance increases with free cash flow, especially in low-growth firms; and the likelihood of low-growth firms granting CEO stock options increases with free cash flow. However, no effect of cash flow on the likelihood of restricted stock grants is identified. In addition, I document a negative relationship between the net debt issuance and CEO stock option grants, indicating that firms trade-off their capital structure adjustment and stock option grants in response to internal free cash flows.

Given the substitutability of debt and incentive compensation in dealing with the free cash flow problem, I expect some firm heterogeneity may account for a firm's policy choice as the benefits and costs of each mechanism differ across firms. To examine these variations, I focus on firms with high leverage and low option compensation versus firms with low leverage and high option compensation. I find large differences in the characteristics between the two groups, suggesting the cost-benefit considerations may determine a firm's policy choice.

The remainder of the paper is constructed as follows. I present background literature and summarize the hypotheses in Section I and describe the sample and variable construction in Section II. Section III tests the effects of debt and incentive compensation on the free cash flow problem in a single-equation framework. Section IV examines the substitutability of the two mechanisms and the bias of the single-equation method. Section V tests whether firms trade off the adjustment of their capital structure and compensation structure to address the free cash flow problem. Section VI considers firms' relative benefits and costs of using the two mechanisms, and Section VII presents my conclusions.

I. Background Literature and Hypotheses

A. Debt Effect

There are a number of factors that make a firm's investment policy dependent on its financial position. In the agency setting, debt servicing obligations help discourage overinvestment of free cash flow by self-serving managers because of the precommitted payment of interest as documented by Jensen (1986), Stulz (1990), Hart and Moore (1995), and Zweibel (1996). Flannery (1986) and Ross (1977) demonstrate that debt may signal managerial willingness to pay out cash flow or to be monitored by lenders.

However, debt itself can generate agency costs. If debt is given exogenously, owner-managers of a levered firm tend to overinvest and choose risky and often negative net present value projects due to their limited liability. This leads to asset substitution (Jensen and Meckling, 1976). Alternatively, Myers (1977) argues that risky debt may lead to underinvestment due to the wealth transfer from shareholders to creditors that would occur upon investment. Therefore, the following hypotheses address the effect of debt on investment.

The free cash flow hypothesis predicts that leverage is negatively related to investment. In addition, I expect a positive relationship between debt financing and free cash flow level, suggesting firms adjust their capital structure to restrict the free cash flow agency problem. The relationships are expected to be stronger in low-growth firms as the agency-related overinvestment problem is more serious in those firms.

Under the asset substitution hypothesis, investment is expected to increase with leverage, as levered firms tend to overinvest and choose risky projects due to their limited liability.

The underinvestment hypothesis also predicts a negative relationship between investment and leverage, but the relationship is expected to be more pronounced in firms with high growth prospects as those firms are more likely to have underinvestment issues.

B. Incentive Compensation Effect

The literature provides three explanations to account for a positive relationship between investment and free cash flow. First, the positive relationship is a manifestation of financial constraints. A firm's investment is sensitive to internally generated funds when financial markets have less information about the true net present value of a project and, thus, demand a risk premium on financing. Consequently, a shortage of internally generated funds will lead to underinvestment. Supportive evidence is provided by Fazzari, Hubbard, and Petersen (1988), followed by a series of empirical work, but is challenged by Kaplan and Zingales (1997) and Cleary (1999).

Second, investment cash flow sensitivity is a statistical artifact. Specifically, Erickson and Whited (2000) argue after controlling for the measurement error of Q, there is no positive relationship between investment and free cash flow. However, their results should be interpreted with caution given that they fail to find the estimates on Q consistent with Q theory.

Third, agency theory argues that the observed positive investment cash flow sensitivity can be attributed to managers engaging in self-serving expenditures. Specifically, managers may place a discount on excess internal funds and overinvest by undertaking negative net present value projects as they derive more private benefits from "empire building" activities (Conyon and Murphy, 2000; Freund, Prezas, and Vasudevan, 2003; Dyck and Zingales, 2004).

The compensation literature suggests that equity-based compensation serves to align the interests between shareholders and managers (Murphy, 1999). The effects of CEOs' equity holdings on investment cash flow sensitivity have been employed to test the financial constraints versus agency story. Hadlock (1998) finds a positive relationship between managerial stock ownership and investment cash flow sensitivity, consistent with the financial constraints argument. However, Broussard, Buchenroth, and Pilotte (2004) document a negative relationship between CEOs' pay-for-performance sensitivity (delta) and investment cash flow sensitivity, supporting the agency story of overinvestment. My paper differentiates from the aforementioned papers by focusing on the substitution of capital structure and compensation structure on mitigating the free cash flow agency problem and by separating the different effects of stock options and direct stock holdings. Specifically, I test the following hypotheses noted below.

The free cash flow hypothesis ofoverinvestment agency costs predicts that investment cash flow sensitivity decreases with incentive compensation and the relationship is stronger for low-growth firms than for high-growth firms. In addition, I expect firms would grant more equity-based compensation to alleviate the free cash flow problem, predicting a positive relationship between equity grant and a firm's free cash flow level. The free cash flow hypothesis implies that incentive compensation serves as a mechanism to mitigate the agency costs associated with overinvestment.

However, the overinvestment problem can be exacerbated if managers are entrenched, as such managers may expropriate the rights of minority shareholders and pursue an overly aggressive investment policy. Therefore, the entrenchment hypothesis predicts a positive relationship between investment cash flow sensitivity and incentive compensation. Again, the agency problem is more pronounced in low-growth firms.

Under the financial constraints hypothesis, investment cash flow sensitivity is expected to increase with managerial incentives. When shareholder-manager interests are less aligned, managers may accept the excessive risk premium of the financial markets and invest in negative net present value projects. With liquidity constraints present, the underinvestment problem is expected to be more pronounced for high-growth firms. The informational asymmetry for these firms is more severe when a large portion of the firm's value is attributed to growth opportunities whose quality is, to a large extent, unverifiable ex ante.

C. Substitutability of Debt and Incentive Compensation

If debt and managerial incentives are substitutes as governance mechanisms to attenuate the free cash flow problem, I expect that in a simultaneous framework, leverage and managerial incentives are negatively interacted. They both demonstrate their effects on investment as predicted by the free cash flow hypothesis. These relationships should be more pronounced in firms with severe overinvestment problems. In addition, firms will trade off the use of capital structure adjustment and compensation grants to control the free cash flow problem.

Alternatively, the two mechanisms can be complements as they mitigate the free cash flow problem in different ways. The complement hypothesis predicts a positive relationship between debt and incentive compensation.

II. Data, Variable Selection, and Sample Characteristics

I use Standard & Poor's ExecuComp 2004 version to construct the variables characterizing CEO incentives. I measure the incentive variables for firms with positive CEO total current compensation from 1993 to 2004. Firm characteristics data are collected from the Compustat Industrial Annual database. I exclude financial firms (Standard Industrial Classifications [SIC] codes 6000-6999) and utilities (SIC codes 4900-4999) considering the marked differences in capital structure and corporate governance between those industries and other sectors of the economy. Each firm's stock return volatility is calculated from the Center for Research in Security Prices (CRSP) daily stock return file. Firms with less than 60 trading days for that fiscal year are excluded. The sample is restricted to those firm-years in which investment, cash flow, leverage, and incentive measures can be constructed. For the purpose of fixed effect analysis, firms are required to have at least four time-series observations.

The final sample consists of 12,738 CEO-years on 1,557 firms. Missing data for some variables generate a sample of 12,231 CEO-years for regressions using lagged inventive variables and 11,073 CEO-years for the capital structure change and flow compensation analysis. The proxies used for the dependent and explanatory variable are described below. I describe the use of each variable briefly and then explain how I calculate each proxy.

A. Investment and Free Cash Flow

I use three variables to measure a firm's investment. First, I use capital expenditures on fixed assets scaled by the firm's beginning-of-year capital, which represents a firm's investment rate on fixed assets. Second, I employ the total investment measure defined as capital expenditures plus R&D to explore the possibility that an observed decrease (increase) in capital expenditures is simply an investment switching between capital expenditures and R&D. This measure is also scaled by the beginning-of-year capital. Third, I use residual investment that is obtained by regressing the investment rate on growth potential proxied by Tobin's Q. Thus, the residual investment represents the part that is not explained by the firm's growth opportunity.

I follow the existing literature to define the free cash flow variable as operating income before extraordinary items plus the depreciation net of common and preferred dividends. This approximately represents the discretionary internal funds after interest, tax, and dividend payments. The cash flow measure is deflated by the firm's beginning-of-year capital.

B. Capital Structure

The leverage measure used in the analysis is defined as the ratio of total debt over total assets. The book leverage and market leverage are obtained depending on whether the book value or the market value of common equity (and total assets) is used. The main results are robust to both measures. For brevity, results are only presented for regressions using book leverage.

In addition, I use the net change in leverage ([DELTA]Lev) to analyze a firm's capital structure adjustment to address the free cash flow cost. Here, [DELTA]Lev is defined as a firm's net debt issuance (NetDIss) minus the net equity issuance (NetEIss), where NetDIss = (long-terre debt issuance long-term debt reduction + change in current debt) / (total assets) and NetEIss = (equity issued -equity repurchased) / (total assets).

C. Incentive Compensation

I argue that the CEO's direct stock holdings and options demonstrate different effects regarding the free cash flow problem. To measure the CEO's personal wealth tied up in options, I construct the measure OPTSHR defined as the Black-Scholes value of the CEO's option holdings scaled by his or her total equity wealth. (2) ExecuComp provides sufficient information to apply the Black-Scholes model to the current option grants. For previously granted options, I use Core and Guay's (2002) method to compute the option value and incentives.

The variables of flow compensation, which reflect a firm's compensation structure adjustment, are defined as the fraction of restricted stock grant or stock option grant over the CEO's total current compensation for that year.

Measures of stock ownership and the manager-shareholder interest alignment incentives are tested first to motivate the construction of the OPTSHR variable. The CEO's stock ownership (Alpha) is defined as the CEO's total shareholdings over the firm's total shares outstanding, which is the sum of the CEO's direct stock holdings (Alpha_stk) and option holdings (Alpha_opt). I also test on DELTA and VEGA, which measure the sensitivity of the stock and option value with respect to a 1% change in stock price and a 0.01 change in the annualized standard deviation of stock returns, respectively. DELTA is decomposed into the incentives from share holdings (DELTA_stk) and from option holdings (DELTA_opt).

Panel A of Table I reports correlations of the incentive, leverage, size, and growth variables. Note that stock-related measures (Alpha_stk and DELTA_stk) are negatively related to option-related measures (OPTSHR, Alpha_opt, DELTA_opt, and VEGA). The univariate relationship between leverage and incentives is generally negative. The correlation between incentives and size is negative, consistent with Schaefer (1998). The correlation between incentives and growth is positive, consistent with Mehran (1995) and Core and Guay (1999).

D. Control Variables

Control variables for the investment equation are motivated by standard investment models that focus on the determinants of corporate investment. Abel and Blanchard (1989) assume that investment grows in conjunction with a firm's sales. I use net sales scaled by capital to capture the implication of the sales accelerator model, and expect a positive relation between sales and investment.

The Q-theory predicts that a firm's investment rate should depend on its future optimal level of capital stock; the effect is captured by Tobin's Q. I use the approximate average Q = (market value of equity + liquidating value of preferred stock + total debt) / (total assets). (3) A positive coefficient for Q is hypothesized.

Further, I use cash and marketable securities to capture the impact of available financial slack on investment, which controls for what was present in the firm at the beginning of the fiscal year. A positive association between investment and cash stock is expected.

The logarithm of a firm's market capitalization interacted with cash flow is also included to control for the effect of firm size on the investment sensitivity to cash flow. Large firms are subject to closer scrutiny from the analyst and investment community and also disclose more detailed information. This suggests that firm size will proxy for a variety of governance mechanisms that are expected to mitigate both over- and underinvestment agency costs.

Control variables for the leverage equation include modified Altman's Z-score and profitability. Controls for the compensation equation are volatility and CEO's cash compensation. More will be discussed in Section IV.

Panel B of Table I reports descriptive statistics on the major variables for the full sample. All variables are winsorized at the 1st and 99th percentiles to reduce the influence of outliers. Statistics indicated that the variables exhibit large cross-sectional differences, which may explain firms' differing policies regarding investment, financing structure, and incentive package.

III. Effects of Debt and Incentive Compensation in a Single-Equation Model

As starting point, I estimate panel regressions of the single-investment equation. The single-equation model serves as a benchmark to test the governance capacity of the two mechanisms and to compare with the results of simultaneous regressions in later sections.

[(Investment).sub.it] = [[alpha].sub.0] + [[alpha].sub.1] x [Leverage.sub.it-1] + [[[alpha].sub.2] + [[[alpha].sub.3] x [Incentive.sub.it-1] + [[[alpha].sub.Size] x [Size.sub.it-1]] x [(FXF).sub.it] + [beta] x [Controls.sub.it-1] + [[tau].sub.t] + [[lambda].sub.i] [[epsilon].sub.it]. (1)

In Equation (1), the coefficient [[alpha].sub.1] measures the effect of leverage on investment. The free cash flow hypothesis predicts higher leverage leads to a lower investment rate (i.e., [[alpha].sub.1] < 0). The sensitivity of investment to free cash flow, which measures the impact of an additional dollar of free cash flow on the investment rate, is then estimated by the sum in the square brackets. The free cash flow hypothesis predicts [[alpha].sub.2] > 0 and [[alpha].sub.3] < 0, while alternative hypotheses predict [[alpha].sub.2] > 0 and [[alpha].sub.3] > 0.

I use lagged values for leverage and the investment cash flow sensitivity determinants, thus testing the effects of capital structure and managerial incentives on the subsequent year's investment. By using lagged explanatory variables, I hope to mitigate the possibility of a reverse causation from incentive and capital structure to investment and/or a spurious correlation.

The parameters [[tau].sub.t] and [[lambda].sub.i] represent year effects and firm fixed effects, respectively. The year effects control for any changes in the macroeconomic environment that affect the capital investment. The firm fixed effect estimates adjust for the possibility that unobservable firm-specific factors that are not captured by the independent variables influence the level of capital spending across firms.

A. Different Effects of Executive Stock Options and Direct Stock Holdings

Table II reports the regression results using capital expenditures scaled by the beginning-of-year capital as the dependent variable. Results are robust to total investment and residual investment measures. The CEO incentive variables are DELTA and VEGA in Column 1 and Alpha in Column 2. The variables are further decomposed into stock- and option-related measures in Columns 3 and 4. To offer a direct comparison of high- versus low-incentive firms, I use dummies for incentive variables equal to one for above median values and zero otherwise. The results are robust to specifications using level variables.

The coefficients on the control variables generally have signs in line with accepted theories. Therefore, I focus the analysis on the leverage and incentive variables. A negative relationship between investment and leverage across all model specifications is consistent with the interpretation that debt restricts ex post investment scale by reducing free cash flows. The investment cash flow sensitivity increases in firms with high DELTA or high Alpha, but the effects are not statistically different from zero. When decomposing DELTA and Alpha into incentives from stocks and from options, the stock incentives do not show a constraining effect while the option incentives do. The investment cash flow sensitivity is actually higher in firms with high DELTA_stk (or Alpha_stk) though the difference is not significant. On the other hand, the investment cash flow sensitivity decreases with DELTA_opt or Alpha_opt. For example, using the estimates in Column 3, the sensitivity decreases by 17% (from 0.088 to 0.073) in firms with high DELTA_opt and low DELTA_stk; and the results are statistically significant at 5% level. (4)

Obviously, options and direct stock holdings have different free cash flow governance implications. To further investigate the difference, Columns 5 and 6 report regressions of subsamples based on growth potential. I only present the results of the decomposed incentive variables DELTA_stk and DELTA_opt. Other specifications give similar results. Both stock and option effects are isolated in low-growth firms. Moreover they show opposite effects: the investment cash flow sensitivity decreases with option-related variables, while increases with direct stock-related variables. Though it is beyond the scope of this paper to explain the differences, it is interesting to note that the difference cannot be explained by the pay for performance (as measured by DELTA and sometimes Alpha) and risk incentives (as measured by VEGA) that have been generally documented in the literature. I propose that options are more closely associated with cash flow rights that increase the managerial cost of overinvestment. On the other hand, direct stock holdings have more voting rights implications as CEOs with high stock ownership are entrenched and stock holdings are likely to exacerbate the overinvestment problem. Consistent with the argument, research (Ofek and Yermack, 2000; Huddart and Lang, 2003) indicates that a typical manager sells most shares acquired through option exercise and most exercises occur long before expiration. Therefore, to analyze the effect of incentive compensation on restricting free cash flow, I use the variable OPTSHR, defined as the Black-Scholes value of the CEO's option holdings scaled by his or her total equity wealth, which effectively measures the CEO's personal wealth tied up in options.

B. Single-Equation Investment Regressions

Table III presents the results of Equation (1) using OPTSHR as the incentive variable. Panel A reports the regressions on the full sample. In Columns 1 to 4, the dependent variable is capital expenditures scaled by the beginning-of-year capital. Columns 3 and 4 also include the FCF-HLEV interaction variable to explore the possible financing constraint of debt, where HLEV is a dummy variable equal to one if the leverage is above median and zero otherwise. The financing constraint of debt predicts that higher-levered firms have higher investment cash flow sensitivity. In Columns 5 and 6, the dependent variable is total and residual investment, respectively. The coefficient on leverage is significantly negative and investment cash flow sensitivity is decreasing in options across all specifications. In addition, the FCF-HLEV variable, though positive, is insignificant. (5) The results suggest that both leverage and executive options serve to curtail a firm's free cash flow problem. However, the negative relationship between investment and leverage also supports the underinvestment hypothesis of debt.

To further investigate the hypotheses, I split the sample into firm-years with above and below median levels of firm growth potential as represented by Tobin's Q. Panel B reports the results, as well as t-statistics, testing the difference in slope coefficients between the subsamples. (6) Columns 1 to 3 correspond to capital expenditures, total investment, and residual investment as the dependent variable.

The free cash flow problem is more severe in firms with low growth prospects while underinvestment (financial constraints) is more likely to occur in firms with high growth prospects. If there is an overinvestment problem, we expect to see a negative impact of leverage on investment and a negative impact of options on investment cash flow sensitivity that will be more pronounced in low-growth firms. The underinvestment hypothesis predicts a larger negative coefficient on leverage and a larger positive coefficient on option-cash flow interaction variables in high-growth firms.

The results further confirm the free cash flow hypothesis. Leverage restricts investment for both groups across all specifications, but the effect is more pronounced in low-growth firms. Moreover, the negative relationship between options and investment cash flow sensitivity is isolated in low-growth firms. For high-growth firms, the coefficient on the free cash flow option interaction variable is insignificant. The differences in coefficients for high- and low-growth firms are statistically significant. I also find a significantly negative coefficient on leverage in high-growth firms. Given that high-growth firms are more likely to have underinvestment problems, debt may lead to financial constraint in those firms. Therefore, the effects of the governance mechanisms are not homogenous across firms. The free cash flow constraining implications are more valuable and applicable to firms with limited investment opportunities.

To test the interaction of debt and executive options in the single-equation framework, I ran the regression adding leverage and OPTSHR interaction variables. The results are reported in Table IV. Though the positive coefficients on the interaction terms (Leverage x OPTSHR and FCF x OPTSHR x Leverage) are consistent with the substitution hypothesis, it should be interpreted with caution especially considering the different functional channels of debt and incentives on the free cash flow problem. For example, the positive and significant Leverage x OPTSHR in high-growth firms may suggest a combined effect of risk incentives provided by options and asset substitution of leverage leads to higher investment. The same argument can be applied to low-growth firms, but the effect is not significant. Therefore, in the following sections, I use a simultaneous equations system to directly test the relationship between leverage and incentives while controlling for the endogeneity of the policy variables.

IV. Substitutability of Debt and Executive Stock Options

Results from the single-equation regressions indicate that both debt and executive stock options attenuate the free cash flow problem. The key question is whether debt and options are substitutes. The substitution hypothesis implies a negative relationship between leverage and executive options. To test the hypothesis, I estimate a simultaneous equation system with investment, leverage, and executive options as endogenous variables, allowing me to jointly test the governance capacity of debt monitoring and incentive alignment, as well as their interaction.

In addition, there are econometric and economic reasons for adopting a simultaneous equations approach. First, it is possible that investment and some explanatory variables are simultaneously influenced by the same omitted variables. A fixed effects model can control for the unobserved characteristics as long as the unobservables are relatively invariant over time. However, some of the unobserved determinants are likely to change significantly over time, generating a correlation between the explanatory variables and the error term in the investment equation. An instrumental variable technique can be used to control for the omitted variable bias.

Second, there is theoretical and empirical evidence that each of the variables of primary interest (investment, leverage, and incentives) is a function of the other two. The documented interactions are as follows. First, the firm's capital structure is a determining factor of compensation policy. Under this argument, higher-leveraged firms will design lower incentive alignment contracts either because leverage substitutes for incentive pay to discipline managers or incentive contracts are designed in anticipation of their effect on shareholder-debtholder conflicts. Additionally, managerial incentives influence a firm's financing and investment policies. For instance, Berger, Ofek, and Yermack (1997) argue that entrenched managers have influence over financial policy and tend to avoid debt. Coles, Daniel, and Naveen (2006) demonstrate that leverage and risky investment increase with VEGA but decrease with DELTA. Finally, capital structure, compensation contracts, and investment policy are determined simultaneously (Agrawal and Knoeber, 1996; Ryan and Wiggins, 2002).

In designing the simultaneous equations framework, I treat investment, leverage, and OPTSHR (CEO's option value scaled by his or her total equity wealth) as jointly determined. A nonlinear two-stage estimation procedure is adopted as the endogenous variables involve a cash flow-OPTSHR interaction term. To conform to the underlying reasoning for simultaneous equations, the endogenous variables take contemporaneous values. Specifically, I run the following regression system. The free cash flow and substitution hypotheses predict that investment decreases with leverage ([[alpha].sub.1] < 0), OPTSHR reduces investment cash flow sensitivity ([[alpha].sub.2] > 0 and [[alpha].sub.3] < 0), and leverage and OPTSHR are negatively correlated ([b.sub.1] < 0 and [c.sub.l] < 0).

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]. (2)

Table V reports the two-stage least square (2SLS) results of the full sample. Capital expenditures, total investment, and residual investment are used as dependent variables in the investment equation in Systems 1 to 3, respectively. Table VI reports results on subsamples based on growth opportunities with capital expenditures as the dependent variable in the investment equation. Specifications with total investment and residual investment indicate similar results and are not reported.

Each equation has its unique instrument variables, which are drawn from the prior literature, to satisfy the identifying restrictions and Hansen J statistics indicate the models are not overidentified. The instruments for the investment equation, including sales and cash stock divided by the beginning-of-period capital stock, are discussed above. The proposed instruments for leverage are financial distress (proxied by modified Altman's Z-score) and profitability (proxied by ROA). The static trade-off capital structure theory predicts a negative relationship between leverage and Z-score and a positive relationship between leverage and profitability. The pecking order theory predicts opposite relationships. Based on the existing literature, I hypothesize that OPTSHR is likely to be influenced by the following exogenous variables: 1) the uncertainty in the firm's environment (represented by Stock Risk) and 2) the CEO's cash compensation. The endogenous variables include the nonlinear cash flow-OPTSHR interaction terre; therefore, instruments also include interactions and squared terms to approximate the reduced form for the (nonlinear) endogenous variables. (7) Finally, I use firm size and growth opportunities as control variables and add firm and year fixed effects in all three equations. (8)

The results lead to three important conclusions. First, they support the free cash flow hypothesis of both debt and executive options. The investment equation indicates that high leverage leads to lower investment, and the negative effect of leverage on investment is only significant in low-growth firms. High OPTSHR leads to lower investment cash flow sensitivity. The negative effect of options on investment cash flow sensitivity is more pronounced in low-growth firms. Therefore, the result that debt and options attenuate the free cash flow problem is robust to the single-equation method (which is less sensitive to model specification) and the two-stage simultaneous equations method (which controls for endogeneity). Again, the constraining effects are not homogenous across all firms and are especially valuable in low-growth firms.

Second, debt and options are substitutes in attenuating the free cash flow problem. As shown in Table V, the leverage equation predicts that high OPTSHR leads to lower leverage. In the OPTSHR equation, the coefficient on leverage is negative though insignificant. The negative relationship supports the hypothesis that options and leverage are substitutes to disciplined managers. I acknowledge that the explanation to account for the negative relationship should be multidimensional. For example, an alternative hypothesis as proposed by John and John (1993) suggests that option's risk-taking incentive may lower a firm's debt capacity. Zhang (2007) indicates that this argument is true for firms with severe asset substitution problems, particularly all equity firms with high growth potential. Table VI confirms that the negative interaction between leverage and OPTSHR is isolated in low-growth firms, suggesting the substitution effect is more pronounced for firms with severe free cash flow problems. The results are consistent with the argument that the constraining effects of debt and options are more valuable for those firms. Therefore, the substitutability provides a partial answer to the puzzle as to why some firms issue little debt despite excess debt capacity; those firms tend to prevalently use executive options.

Third, the magnitude of the coefficient estimates on leverage and cash flow-OPTSHR is much higher than that in the single-equation model for the full sample and low-growth subsample. The strengthening occurs because the single-equation model does not account for the fact that firms tend to predominantly use just one mechanism to mitigate the free cash flow problem. Hence, failure to incorporate the substitution effect of debt and options underestimates the effectiveness of both mechanisms. In addition, the downward bias of the single-equation model may be due to an omitted variable bias if the omitted variable is positively (or negatively) related to both investment and leverage or to both investment cash flow sensitivity and executive stock options. (9)

Finally, the results confirm that investment, leverage, and incentives are jointly determined on average. Consistent with the literature, investment negatively influences the leverage level and positively influences option compensation. However, the effects from investment to leverage and option compensation are not manifested in the low-growth firms. Although this weakens the "joint determination" argument, it suggests that the causal relationship in those firms is mainly from the debt and compensation to investment and reinforces the governance implications of the two mechanisms.

A. Economic Significance

To gauge economic significance, I calculate the change in investment rate and investment cash flow sensitivity as leverage, OPTSHR, and size increase by one standard deviation. Table VII presents the economic significance of these variables based on the single-equation and 2SLS coefficient estimates. The effects of leverage and incentives are economically significant for the full sample and, more important, for low-Q firms. The economic significance based on 2SLS is dramatically stronger than that based on single-equation regressions. Aone standard deviation increase in the leverage ratio decreases investment by 31% in the full sample as compared with 9% in the single-equation model. Aone standard deviation increase in leverage decreases investment by 44% in low-growth firms and by only 2% in high-growth firms as compared with 16% and 3.5% in the single-equation model. The effect of one standard deviation increase in OPTSHR reduces the investment cash flow sensitivity by about 170% of the mean sensitivity as compared to 24% in the single-equation model. The effect is 227% for low-Q firms and 55% for high-Q firms as compared to 34% and 11% in the single-equation model. The effect of size on investment cash flow sensitivity based on 2SLS estimates is not significant across all regressions, both statistically and economically.

B. Robustness

As robustness checks, the system is reestimated on the following subsamples: 1) the sample for single-equation regressions of 12,231 firm-years, 2) a subsample of 10,728 firm-years that eliminates firm-years with negative free cash flow to exclude the effect of financial distress, and 3) a sample of 10,713 observations that delete firm-years in which firms change their CEOs to exclude the effect of CEO turnover.

Next, I conduct the split sample analysis using the variable assets in place to proxy for the potential severity of the overinvestment problem. It is expected that firms with high levels of assets in place are more likely to have overinvestment problems. I further compare the firm-years of low-Q-high assets in place and high-Q-low assets in place. The results are consistent.

The empirical work of Morck, Shleifer, and Vishny (1988) suggests that the relationship between insider holdings and the alignment of interests is not monotonic. I take this into consideration by adding squares of incentive variables to the regression models. Results indicate that taking nonlinearity into account does not change the basic effects.

It is well known that a simultaneous equation system is sensitive to model specifications. For robustness, I first adopt a generalized method of moments (GMM) estimation technique instead of 2SLS. Billett, King, and Mauer (2007) use GMM to estimate a system of equations, documenting that GMM controls for the heteroskedasticity, autocorrelation, and the nonlinearity of endogenous variables. Then, I rely on the industry instrumental variables approach to specify the leverage equation using the industry-level capital intensity (defined as fixed assets to total assets) and profitability in place of firm-level variables. The industry averages are appealing instrumental variables as they are unlikely to be affected by firm-specific shocks. For industry variables, I use all Compustat firms that are in the same four-digit SIC code. I require a minimum of four observations to compute an industry average. If there are less than four observations in the four-digit SIC code, I use the three-digit SIC code (and two-digit SIC code). The industry average is constructed by weighting each firm's capital intensity or profitability by its share on total industry assets. The main results hold for different estimation methods and different instruments.

V. Adjustment of Capital Structure and/or Compensation Structure

The above analyses using level variables demonstrate that debt and executive stock options act as substitutes to reduce a firm's free cash flow agency problem. An alternative perspective on the disciplinary role of capital structure and compensation structure suggests that firms with greater potential of free cash flow tend to use more debt and/or executive options. If debt and stock options have the governance capacity, it is optimal for firms to adopt the two mechanisms to control the free cash flow problem. And, because they are substitutes, firms may choose to use one predominantly. Therefore, in this section, I investigate the relationship between a firm's financing and/or compensation structure adjustment in response to the free cash flow level (which represents the potential of overinvestment due to free cash flow).

I first examine how capital structure changes with the level of free cash flows. The test is designed by regressing the capital structure adjustment on the free cash flow level.

[(Capital Structure Adj.).sub.it] = [[alpha].sub.0] + [[alpha].sub.1] x [FCF.sub.it] + [[alpha].sub.2] x [Leverage.sub.it-1] + [[alpha].sub.3] x [Size.sub.it-1] + [[alpha].sub.4] x [MTB.sub.it-1] + [[alpha].sub.5] x [Z-score.sub.it-1] + [[alpha].sub.6] x [tenure.sub.it-1] + [[tau].sub.t] + [[lambda].sub.i] + [[epsilon].sub.it]. (3)

Table VIII reports the firms' fixed effects results on the full sample (Panel A) and subsamples based on growth opportunities (Panel B). The free cash flow hypothesis predicts a positive relationship between a firm's issuance of debt versus equity and the free cash flow level, and the effect is expected to be greater for low-growth firms. This is supported by the positive and significant coefficient on the free cash flow variable in the leverage change regressions. In addition, the positive relationship is more pronounced for low-growth firms.

However, it is possible that the positive relationship between leverage change and free cash flow is simply driven by a reduction in equity. Therefore, I ran regressions using net debt issuance and net equity issuance as dependent variables separately. The results indicate that firms with high levels of free cash flows will have more net debt issuance; this positive relationship appears isolated in low-growth firms. Though a detailed analysis on repurchase is beyond the scope of this paper, I do find evidence that firms use internal cash flow to repurchase shares. Repurchase may serve as a mechanism to mitigate the free cash flow problem. However, this effect is not manifested in low-growth firms, confirming the importance of debt governance in firms with severe free cash flow problems. I also decompose net debt issuance into debt issuance and debt reduction to explore the possibility that firms use free cash flow to buy back debt instead of issuing new debt. No effect is documented in the debt reduction regressions. A positive relationship between debt issuance and free cash flow in low-growth firms provides further evidence consistent with the free cash flow hypothesis. The above the results are robust to various regression methods (e.g., industry fixed effects, Fama-MacBeth (Fama and MacBeth, 1973), and firm fixed effects).

Next, I examine the connection of the CEO's equity compensation grant with the firm's free cash flow level. Looking into flow compensation is useful as it represents the compensation policy under board control and because it is through equity grants that CEOs build their portfolios. Specifically, I estimate probit models of the probability that a CEO is granted stock options and restricted stock, respectively. Table IX reports the results on the full sample (Panel A) and subsamples based on growth opportunities (Panel B).

[OG.sub.it](or [RS.sub.it]) = [[alplha].sub.0] + [[alplha].sub.1] x [FCF.sub.it] + [[alpha].sub.2] x [Leverage.sub.it- 1] + [[alpha].sub.3] x [Size.sub.it-1] + [[alpha].sub.4] x [MTB.sub.it-1] + [[alpha].sub.5] x [Riskit.sub.it-1] + [[alpha].sub.6] x [CashComp.sub.it] + [[alpha].sub.7] x [Tenure.sub.it] + [[tau].sub.t] + [[lambda].sub.i] + [[epsilon].sub.it]. (4)

For these models, the dependent variable OG (RS) equals one if firms grant stock options (restricted stock) to their CEOs. Model 1 reports the pooled probit with two-digit SIC and year dummies and Model 2 reports the random effects probit results, l0 The free cash flow hypothesis predicts [[alpha].sub.1] > 0 (i.e., firms with high free cash flows tend to grant more options or restricted stocks).

No significant relationship between the likelihood of option grants and free cash flow is identified in the full sample (Panel A). However, Panel B confirms that CEOs in low-growth firms with higher cash flow are more likely to receive stock option grants. The effect is significant at 10% (pooled probit) or 5% (random effects probit) level. The results suggest that shareholders grant more options to CEOs in firms with potentially high agency costs of free cash flow. In addition, CEOs in more levered firms are less likely to receive stock option grants. This is consistent with the literature and suggests that shareholders use less option grants in highly levered firms.

I fail to identify any effect of free cash flow on the probability of restricted stock grants. The effects do not appear different for low- and high-growth firms, so the results are not reported. This is consistent with prior results that options, but not stocks, have a constraining effect on the free cash flow problem.

The final question is whether firms trade off the adjustment of capital structure and compensation structure to control the free cash flow problem. To address this issue and also to control for the potential endogeneity of a firm's capital raising and compensation decisions, I estimate the simultaneous equations system treating net leverage change (or net debt issuance) and the fraction of option grant to total current compensation as endogenous variables: (11)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (5)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]. (6)

Table X reports results on the full sample (Panel A) and on subsamples based on growth opportunity (Panel B). Two implications can be drawn from the results. First, firms that grant more options use less debt issuance and vice versa, supporting the trade-off argument. Second, firms with high free cash flow tend to issue more debt and the positive relationship between the net debt issuance and free cash flow is only found in low-growth firms. In addition, firms with large free cash flows use more option grants, and the positive relationship between fractional option grants and free cash flow levels is persistent only in low-growth firms. Overall, the evidence supports the hypothesis that firms with greater agency costs of free cash flow issue more debt or grant more options. That is, they trade off the adjustment of capital structure and compensation structure to control the overinvestment associated with free cash flows.

VI. Firm Characteristics and Monitoring Mechanisms

The evidence I have presented supports the hypothesis that debt and incentive compensation are substitutes when dealing with the free cash flow agency problem. Given the fact that both mechanisms have costs, the issue is whether there are firm characteristics that would suggest the determinants of a firm's choice between debt and executive stock options as free cash flow governance mechanisms.

To investigate this issue, I focus on the low-growth firms that have more critical free cash flow concerns. I divide the firms into four groups based on the medians of leverage and OPTSHR. Table XI reports the characteristics of high-leverage/low-OPTSHR firms versus low-leverage/high-OPTSHR firms, along with t-statistics and Wilcoxon sum-rank p-values for tests of differences between the two groups.

The mean free cash flow, market capitalization, and total assets are not different for the two groups though the median differences are significant. There are some significant differences. In general, firms in the high-leverage/low-option group tend to have lower Q, lower stock returns, and lower volatility than their counterparts in the low-leverage/high-option firms. The results suggest that it may be costly for firms with high growth and high risk to borrow. Cash compensation and total current compensation are higher, while tenure is lower in low-leverage/high-option firms. If cash compensation proxies for the pay to a manager's personal skill and talent, the result is consistent with Garvey (1997) in that the incentive approach is more prevalent if a manager's private information is sufficiently valuable. However, if high tenure means managers are more entrenched, it may be optimal for those firms to adopt more external monitoring.

VII. Conclusion

Agency theory reveals that when managers' objectives differ from those of shareholders and when a firm also lacks efficient monitoring mechanisms, high managerial discretion may lead managers to overinvest. Both debt and incentive compensation have the potential to attenuate the agency problem. In this paper, the two mechanisms are explored jointly and their interactions are investigated.

I find that a firm's debt servicing obligations mitigate the free cash flow problem by directly restricting the firm's investment level. Managerial incentive compensation restricts overinvesting internal cash flows by reducing investment cash flow sensitivity. It is important to point out that managers' direct stock holding and options have different agency implications, and only options appear to help alleviate the free cash flow problem.

To investigate the substitutability of debt and options in attenuating the free cash flow problem and, at the same time, address the endogeneity among policy variables, I estimate a nonlinear simultaneous equation system treating investment, leverage, and options as endogenous variables. The results reinforce the single-equation model and demonstrate that the single-equation model understates the effects of the two mechanisms. In addition, debt and options negatively interact with each other, indicating they act as substitutes in controlling overinvestment.

I provide evidence that firms adjust capital structure and option grants to control the free cash flow problem. Specifically, firms with potentially severe free cash flow problems tend to issue more debt and grant more options. Net debt issuance and option grants are negatively associated, suggesting that firms trade off the adjustment of capital structure and compensation structure.

All the above effects concerning the governance mechanisms are not homogenous across firms. The free cash flow constraining implications are more valuable and applicable to firms with limited investment opportunities.

Finally, I document different firm characteristics across the prevalence of debt usage versus option usage. While these results are suggestive of heterogeneity in the benefits and costs of the two monitoring devices, more research is needed to provide a better understanding of the determinants of the firm's optimal choice. In addition, factors driving the different effects of restricted stock and stock options merit refined theory and further empirical investigation.

I am grateful to Matt Billett, Jon Garfinkel, David Mauer, Yiming Qian. Jay Sa-Aadu. Anand Vijh, and an anonymous referee for many helpful comments and suggestions. I thank seminar participants at the University of Iowa, the University of Texas at San Antonio. Kansas State University, Loyola University Chicago, the University, of Toronto, and the University of North Dakota. I also thank Wendy Jennings and Michael O'Doherty, for careful editing and proofreading. All errors are my responsibility.

References

Abel, A. and O. Blanchard, 1989, "Investment and Sales: Some Empirical Evidence," NBER Working Paper 2050.

Agrawal, A. and C.R. Knoeber, 1996, "Firm Performance and Mechanisms to Control Agency Problems between Managers and Shareholders," Journal of Financial and Quantitative Analysis 31, 377-397.

Berger, G., E. Ofek, and D.L. Yermack, 1997, "Managerial Entrenchment and Capital Structure," Journal of Finance 52, 1411-1438.

Billett, M.T., T.D. King, and D.C. Mauer, 2007, "Growth Opportunities and the Choice of Leverage, Debt Maturity and Covenants," Journal of Finance 62, 697-730.

Broussard, J.P., S.A. Buchenroth, and E.A. Pilotte, 2004, "CEO Incentives, Cash Flow, and Investment," Financial Management 33, 51-70.

Cleary, S., 1999, "The Relationship between Firm Investment and Financial Status," Journal of Finance 54, 673-692.

Coles, J.L., N.D. Daniel, and L. Naveen, 2006, "Executive Compensation and Managerial Risk-Taking," Journal of Financial Economics 79, 431-468.

Conyon, M. and K.J. Murphy, 2000, "The Prince and the Pauper? CEO Pay in the U.S. and the UK," Economic Journal 110, 640-671.

Core, J.E. and W. Guay, 1999, "The Use of Equity Grants to Manage Optimal Equity Incentive Levels," Journal of Accounting and Economics 28, 151-184.

Core, J.E. and W. Guay, 2002, "Estimating the Value of Stock Option Portfolios and Their Sensitivities to Price and Volatility," Journal of Accounting Research 40, 613-630.

Dyck, A. and L. Zingales, 2004, "Private Benefits of Control: An International Comparison," Journal of Finance 59, 537-600.

Erickson, T. and T.M. Whited, 2000, "Measurement Error and the Relationship between Investment and q," Journal of Political Economy 108, 1027-1057.

Fama, E. and J. MacBeth, 1973, "Risk, Return, and Equilibrium: Empirical Tests," Journal of Political Economy 81, 607-636.

Fazzari, S.M., R.G. Hubbard, and B.C. Petersen, 1988, "Financing Constraints and Corporate Investment," Brookings Paper on Economic Activity 1, 141-195.

Flannery, M., 1986, "Asymmetric Information and Risky Debt Maturity Choice," Journal of Finance 41, 19-37.

Freund, S., A.P. Prezas, and G.K. Vasudevan, 2003, "Operating Performance and Free Cash Flow of Asset Buyers," Financial Management 32, 87-97.

Garvey, G.T., 1997, "Marketable Incentive Contracts and Capital Structure Relevance," Journal of Finance 52, 353-378.

Hadlock, C.J., 1998, "Ownership, Liquidity and Investment," RAND Journal of Economics 29, 487-508.

Hart, O. and J. Moore, 1995, "Debt and Seniority: An Analysis of the Role of Hard Claims in Constraining Management," American Economic Review 85, 567-585.

Harvey, C.R., K.V. Lins, and A.H. Roper, 2004, "The Effect of Capital Structure When Expected Agency Costs Are Extreme," Journal of Financial Economics 74, 3-30.

Huddart, S. and M. Lang, 2003, "Employee Stock Option Exercises: An Empirical Analysis," Journal of Accounting and Economics 21, 5-43.

Ingersoll, J.E., 2006, "The Subjective and Objective Evaluation of Incentive Stock Options," Journal of Business 79, 453-487.

Jensen, M.C., 1986, "Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers," American Economic Review 76, 323-339.

Jensen, M.C. and W. Meckling, 1976, "Theory of the Firm: Managerial Behavior, Agency Costs, and Ownership Structure," Journal of Financial Economics 3, 305-360.

John, T.A. and K. John, 1993, "Top Management Compensation and Capital Structure," Journal of Finance 48, 949-974.

Kaplan, S.N. and L. Zingales, 1997, "Do Investment-Cash Flow Sensitivities Provide Useful Measures of Financing Constraints?" Quarterly Journal of Economics 169-215.

Kelejian, H.H., 1971, "Two-Stage Least Squares and Econometric Systems Linear in Parameters but Nonlinear in the Endogenous Variables," Journal of the American Statistical Association 66, 373-374.

Mehran, H., 1995, "Executive Compensation Structure, Ownership, and Firm Performance," Journal of Financial Economics 38, 163-184.

Morck, R., A. Shleifer, and R. Vishny, 1988, "Management Ownership and Market Valuation--An Empirical Analysis," Journal of Financial Economics 20, 293-315.

Murphy, K.J., 1999, "Executive Compensation," in O. Ashenfelter and D. Card, Eds., Handbook of Labor Economics, Vol. 3, Amsterdam, North Holland.

Myers, S., 1977, "Determinants of Corporate Borrowing," Journal of Financial Economics 5, 147-175.

Ofek, E. and D. Yermack, 2000, "Taking Stock: Equity-Based Compensation and the Evolution of Managerial Ownership," Journal of Finance 55, 129-149.

Perfect, S. and K. Wiles, 1994, "Alternative Constructions of Tobin's q: An Empirical Comparison," Journal of Empirical Finance 1, 313-341.

Ross, S.A., 1977, "The Determination of Financial Structure: The Incentive-Signaling Approach," Bell Journal of Economics 8, 23-40.

Ryan, H.E. and R.A. Wiggins, 2002, "The Interaction between R&D Investment Decisions and Compensation Policy," Financial Management 31, 5-29.

Schaefer, S., 1998, "The Dependence of the Pay Performance Sensitivity on the Size of the Firm," Review of Economics and Statistics 80, 436-443.

Stulz, R.M., 1990, "Managerial Discretion and Optimal Financing Policies," Journal of Financial Economics 26, 3-27.

Zhang, Y., 2007, "Effects of the Agency Cost of Debt on Capital Structure: What Can We Learn from All-Equity Firms?" University of Iowa Working Paper.

Zweibel, J., 1996, "Dynamic Capital Structure under Managerial Entrenchment," American Economic Review 86, 1197-1215.

(1) For example, mechanisms such as takeovers and board intervention require substantial changes. Moreover, the literature does not find a significant relationship between firm value and board structure. Dividend payout and share repurchase have the same effect as debt on the free cash flow problem. The effect of share repurchase is partially addressed in this paper by analyzing the capital structure adjustment.

(2) Numerous authors note that the Black-Scholes model overvalues executive stock options since it does not account for the fact that such options are nontradable and are held by undiversified, risk-averse executives. On the other hand, senior executives are the insiders and know how they will respond to the incentives provided, making the value of the options to them higher than the Black-Scholes value. In any case, my results are robust when I calculate the executive option value and managerial incentives using the marginal value framework of Ingersoll (2006).

(3) Perfect and Wiles (1994) demonstrate that the improvements obtained from the more involved computation of Q are fairly limited, particularly when regressions are estimated with firm fixed effects.

(4) The investment cash flow sensitivity is calculated as [partial derivative]Investment/[partial derivative]FCF = [[alpha].sub.FCF] + [[alpha].sub.FCFxDELTA_stk] x DELTA_stk + [[alpha].sub.FCFxDELTA_opt] x DELTA_opt, where [alpha] represents the coefficient estimate.

(5) This interaction variable is also insignificant in all split sample regressions (not reported). I argue that debt and executive incentives mitigate the free cash flow problem through different mechanisms, as estimated by investment debt sensitivity and investment cash flow sensitivity, respectively. Therefore, in order not to complicate the structure, I do not include this interaction variable in the simultaneous regressions.

(6) The tests of difference in slope coefficients between subsamples are based on a standard dummy variable technique. The dummy variable Dq = 1 for high-Q firm-years and 0 for low-Q firm-years is interacted with all right-hand-side variables in Equation (1) and the interaction variables are added as explanatory variables. The coefficients on Dq interaction variables estimate the difference in coefficients across the different subsamples. The corresponding t-statistics test the null that the coefficients for high-Q firms do not differ significantly from the corresponding coefficients for low-Q firms.

(7) Kelejian (1971 ) demonstrates that if the functional forms of the reduced-form equations are not known and are, therefore, approximated by polynomials, the polynomials must be of the same degree (using the linear terms, higher powers, and cross products) if the 2SLS estimates are to be consistent.

(8) All instrument and control variables enter the system lagged one period to ensure exogeneity. As most of the variables are highly serially correlated, this procedure does hOt alter significantly the explanatory power of the regressions.

(9) If he true model is Y = [alpha] + [[beta].sub.1] x [X.sub.1] + [[beta].sub.2] x [X.sub.2] + u, we erroneously believe the model is Y = a + [b.sub.1] x [X.sub.1] + v. Then, the estimate of [b.sub.1] [approximately equal to] [[beta].sub.1] + [[beta].sub.2] Cov([X.sub.1], [X.sub.2]) / Var([X.sub.1]).

(10) A fixed effects probit model does not produce consistent parameter estimates. Therefore, I estimate a random effects probit model to control for the omitted variables. Compared with the fixed effects model, the random effects model greatly reduces the number of parameters to be estimated, but requires no correlation between the regressors and the unobserved individual effects.

(11) I also estimate a simultaneous equations system with net leverage change (or net debt issuance), option grant, and restricted stock grant as endogenous variables. The relationship between restricted stock and net debt issuance and between restricted stock and free cash flow is insignificant. The results are not reported to conserve space.

Yilei Zhang, Yilei Zhang is an Assistant Professor in the College of Business and Public Administration at the University of North Dakota, Grand Forks, ND.

Table I. Descriptive Statistics

This table presents the descriptive statistics for the sample of
12,738 firm-year observations on 1,557 firms over the 1993-2004
time period. Financial and restricted utility firms are excluded.
The sample is restricted to those firm-years in which investment,
cash flow, leverage, and incentive measures can be constructed.
Firms are required to have at least four time-series
observations. Panel A reports the Pearson correlation coefficient
of variables. Panel B reports the statistics of major dependent
and independent variables. Here, I is the capital expenditures
and Total 1 is the capital expenditure plus R&D; FCF is the free
cash flow defined as net income after tax plus depreciation less
common and preferred dividends; K is the beginning-of-year
capital stock; Bklev and Mktlev are the book leverage and market
leverage, respectively; [DELTA]Lev is the net debt issuance (NetDlss)
subtracting net stock issuance (NetElss), all scaled by total
assets; OPTSHR is the Black-Scholes value of the CEO's option
holdings scaled by the total value of his or her holdings of
stock and stock options; Alpha is the CEO's fractional stock and
options holding, decomposing into fractional stock holding
(Alpha_stk) and options holding (Alpha_opt); DELTA and VEGA
measure the CEO's equity value sensitivity with respect to a one
percentage change in stock price and a 0.01 change in standard
deviation, respectively; DELTA_stk and DELTA-opt are delta from
CEO's shareholdings and options holdings, respectively; StkG is
the restricted stock grant value as a fraction of total current
compensation; OptG is the Black-Scholes value of current option
grants as a fraction of total current compensation; Size is the
logarithm of a firm's market capitalization; Q is the Tobin's Q
estimate; Sales is the net sales scaled by K; Cash is the cash
and marketable securities scaled by K; Z-score is the modified
Altman's Z-score = [3.3 x (operating income after depreciation) +
sales + 1.4 x (retained earnings) + 1.2 x (current assets current
liability)]-total assets; ROA is operating income before
depreciation divided by assets; Stock Risk is the logarithm of
percentage standard deviation of stock returns; StkRet is the
annual stock returns; CashComp is the logarithm of CEO's current
cash compensation; and Tenure is CEO's tenure. All firm
characteristic controls are measured at the prior fiscal
year-end. Compensation data are from the ExecuComp database 2004,
firm characteristic data are from Compustat, and stock returns are
from CRSP.

Panel A. Correlation between Incentive and Leverage Variables

Variable           OPTSHR      Alpha       Alpha_stk   Alpha_opt

  OPTSHR            1.00
  Alpha            -0.48 ***    1.00
  Alpha_stk        -0.56 ***    0.98 ***    1.00
  Alpha_opt         0.34 ***    0.15 ***   -0.05 ***    1.00
  DELTA            -0.24 ***    0.40 ***    0.41 ***   -0.04 ***
  DELTA_stk        -0.33 ***    0.46 ***    0.48 ***   -0.09 ***
  DELTA_opt         0.23 ***   -0.06 ***   -0.10 ***    0.18 ***
  VEGA              0.25 ***   -0.10 ***   -0.13 ***    0.13 ***
  Bklev            -0.05 ***   -0.14 ***   -0.04 ***   -0.13 ***
  Mktlev           -0.03 ***   -0.10 ***    0.01       -0.11 ***
  Size              0.05 ***   -0.10 ***   -0.07 ***   -0.14 ***
  Q                 0.01        0.07 ***    0.06 ***    0.03 ***

Variable           DELTA       DELTA_stk   DELTA_opt   VEGA

  OPTSHR
  Alpha
  Alpha_stk
  Alpha_opt
  DELTA             1.00
  DELTA_stk         0.96 ***    1.00
  DELTA_opt         0.46 ***    0.23 ***    1.00
  VEGA              0.33 ***    0.14 ***    0.81 ***    1.00
  Bklev            -0.09 ***   -0.01 ***   -0.10 ***   -0.07 ***
  Mktlev           -0.15 ***   -0.12 ***   -0.14 ***   -0.02 ***
  Size             -0.09 ***   -0.07 ***   -0.14 ***   -0.14 ***
  Q                 0.07 ***    0.06 ***    0.06 ***   -0.11 ***

Variable           Bklev       Mktlev      Size        Q

  OPTSHR
  Alpha
  Alpha_stk
  Alpha_opt
  DELTA
  DELTA_stk
  DELTA_opt
  VEGA
  Bklev             1.00
  Mktlev            0.87 ***    1.00
  Size             -0.00       -0.08 ***    1.00
  Q                -0.30 ***   -0.41 ***    0.17 ***       1.00

Panel B. Summary Statistics for Dependent and Independent Variables

Variables                          N          Mean      Std. Dev.

Investment, Cash Flow, and Capital Structure

  I/K                           12,738        0.32        0.29
  Total I/K                     12,738        0.65        0.94
  FCF/K                         12,738        0.50        1.14
  Bklev                         12,738        0.22        0.17
  Mktlev                        12,738        0.15        0.14
  [DELTA]Lev (a)                11,073        0.01        0.15
  NetDlss (a)                   11,073        0.01        0.11
  NetElss (a)                   11,073        0.00        0.09

Managerial Incentives

  OPTSHR                        12,738        0.51        0.33
  Alpha (b)                     12,231        0.05        0.07
  Alpha_stk (b)                 12,231        0.03        0.07
  Alpha_opt (b)                 12,231        0.01        0.01
  DELTA ($million) (b)          12,231        0.74        1.91
  DELTA_stk ($million) (b)      12,231        0.51        1.71
  DELTA_opt ($million) (b)      12,231        0.21        0.42
  VEGA ($million) (b)           12,231        0.10        0.18
  StkG (a)                      11,073        0.05        0.12
  OptG (a)                      11,073        0.36        0.29

Firm and CEO Characteristic Control Variables

Size                            12,738        7.01        1.59
  Q                             12,738        2.14        1.70
  Sales/K                       12,738        7.28       10.21
  Cash/K                        12,738        0.97        1.65
  Z-score                       12,738        2.13        1.28
  ROA                           12,738        0.15        0.10
  Stock Risk                    12,738        0.97        0.44
  CashComp                      12,738        6.76        0.74
  StkRet (a)                    11,073        0.21        0.73
  Tenure (a)                    11,073        8.07        7.68

                                           Percentile

Variables                         Min        25th        50th

Investment, Cash Flow, and Capital Structure

  I/K                             0.00        0.14        0.23
  Total I/K                       0.00        0.18        0.32
  FCF/K                          -4.42        0.16        0.33
  Bklev                           0.00        0.06        0.21
  Mktlev                          0.00        0.03        0.12
  [DELTA]Lev (a)                 -1.60       -0.03        0.00
  NetDlss (a)                    -1.60       -0.02        0.00
  NetElss (a)                    -2.36       -0.01        0.00

Managerial Incentives

  OPTSHR                          0.00        0.19        0.55
  Alpha (b)                       0.00        0.01        0.02
  Alpha_stk (b)                   0.00        0.00        0.00
  Alpha_opt (b)                   0.00        0.00        0.01
  DELTA ($million) (b)            0.00        0.07        0.19
  DELTA_stk ($million) (b)        0.00        0.02        0.06
  DELTA_opt ($million) (b)        0.00        0.02        0.07
  VEGA ($million) (b)             0.00        0.01        0.03
  StkG (a)                        0.00        0.00        0.00
  OptG (a)                        0.00        0.03        0.34

Firm and CEO Characteristic Control Variables

Size                              3.81        5.86        6.82
  Q                               0.41        1.11        1.60
  Sales/K                         0.32        2.34        4.39
  Cash/K                          0.01        0.05        0.22
  Z-score                        -2.78        1.41        2.15
  ROA                            -0.25        0.10        0.15
  Stock Risk                      0.05        0.65        0.96
  CashComp                        4.97        6.23        6.73
  StkRet (a)                     -0.95       -0.15        0.10
  Tenure (a)                      0.03        2.67        5.69

                                    Percentile

Variables                        75th         Max

Investment, Cash Flow, and Capital Structure

  I/K                             0.38        1.75
  Total I/K                       0.66        5.79
  FCF/K                           0.67        6.09
  Bklev                           0.33        0.73
  Mktlev                          0.23        0.86
  [DELTA]Lev (a)                  0.06        5.50
  NetDlss (a)                     0.04        5.70
  NetElss (a)                     0.01        1.25

Managerial Incentives

  OPTSHR                          0.80        1.00
  Alpha (b)                       0.05        0.39
  Alpha_stk (b)                   0.02        0.38
  Alpha_opt (b)                   0.02        0.08
  DELTA ($million) (b)            0.54       14.61
  DELTA_stk ($million) (b)        0.24       13.54
  DELTA_opt ($million) (b)        0.20        2.90
  VEGA ($million) (b)             0.09        1.16
  StkG (a)                        0.00        1.00
  OptG (a)                        0.59        1.00

Firm and CEO Characteristic Control Variables

Size                              8.00       11.41
  Q                               2.48       10.14
  Sales/K                         7.83       70.11
  Cash/K                          0.99        6.28
  Z-score                         2.87        5.35
  ROA                             0.20        0.41
  Stock Risk                      1.29        2.05
  CashComp                        7.25        8.66
  StkRet (a)                      0.38       17.73
  Tenure (a)                     10.92       54.95

*** Significant at the 0.01 level.

(a) Variables for capital and compensation structure adjustment,
for a sample of 11,073 firm-years on 1,397 firms. Sample size
decreases due to the availability of data.

(b) Variables used for baseline single-equation model on a sample
of 12,231 firm-years on 1,510 firms. Sample size decreases due to
the availability of data.

Table II. Different Effects of Executive Stock
Options and Direct Stock Holdings

This table reports firm fixed effects regression results for the
single-equation model using different incentive variables. The
dependent variable is investment (capital expenditure) scaled by
the beginning-of-period capital stock K. All incentive variables
(DELTA, VEGA, Alpha, Alpha_stk, Alpha_opt, DELTA_stk, DELTA_opt)
are dummy variables equal to 1 for above-median values and 0
otherwise. Variables are defined as in Table 1. All regressions
include year dummies. The t-statistics are based on robust
standard errors and are reported in parentheses below the
parameter estimates.

                            Dependent Variable: I/K

                                Full Sample

                     (1)          (2)          (3)

FCF                 0.080 ***    0.070 ***    0.088 ***
                   (3.53)       (3.21)       (4.02)

Leverage           -0.165 ***   -0.165 ***   -0.165 ***
                  (-6.74)      (-6.74)      (-6.77)

FCF x DELTA         0.010
                   (1.20)

FCF x VEGA         (0.01)

FCF x Alpha                      0.010
                                (1.28)

FCF x DELTA_stk                               0.011
                                             (1.37)

FCF x DELTA_opt                              -0.015 **
                                            (-1.98)

FCF x Alpha_stk

FCF x Alpha_opt

FCF x Size         -0.007 **    -0.006 **    -0.007 **
                  (-2.31)      (-2.08)      (-2.40)

Sales/K             0.011 ***    0.011 ***    0.010 ***
                   (8.81)       (8.88)       (8.75)

Q                   0.057 ***    0.057 ***    0.057 ***
                  (17.20)      (17.57)      (17.60)

Cash/K              0.043 ***    0.043 ***    0.043 ***
                   (9.38)       (9.34)       (9.39)

Within [R.sub.2]    0.30         0.30         0.30

Observations       12,231       12,231       12,231

                         Dependent Variable: I/K

                                Low Q        High Q

                     (4)          (5)          (6)

FCF                 0.087 ***    0.079 ***    0.109 ***
                   (3.98)       (2.87)       (4.04)

Leverage           -0.165 ***   -0.219 ***   -0.077 *
                  (-6.75)      (-7.65)       (1.90)

FCF x DELTA

FCF x VEGA

FCF x Alpha

FCF x DELTA_stk                  0.015 **     0.004
                                (2.58)       (0.47)

FCF x DELTA_opt                 (0.02)       (0.01)
                               (-1.57)      (-0.56)

FCF x Alpha_stk     0.009
                   (1.16)

FCF x Alpha_opt    -0.015 **
                  (-2.41)

FCF x Size         -0.007 **    (0.00)       -0.010 **
                  (-2.40)      (-1.31)      (-2.80)

Sales/K             0.010 ***    0.009 ***    0.014 ***
                   (8.74)       (6.36)       (6.56)

Q                   0.057 ***    0.127 ***    0.048 ***
                  (17.55)      (12.69)      (12.95)

Cash/K              0.043 ***    0.010 ***    0.052
                   (9.39)       (1.07)       (8.92)

Within [R.sub.2]    0.30         0.21         0.29

Observations       12,231        6,115        6,116

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

* Significant at the 0.10 level.

Table III. Effects of Leverage and Option
Compensation: Single-Equation Model

This table reports firm fixed effects regressions using CEO's
option value scaled by his or her total equity wealth (OPTSHR) as
the incentive variable. Here, HLEV is a dummy variable that
equals 1 if leverage is above median and 0 otherwise. The
dependent variables are investment (capital expenditure) scaled
by the beginning-of-period capital stock (I/K), total investment
(capital expenditure plus R&D) scaled by K (Total UK), and
residual investment scaled by K (Resid UK). Other variables are
defined as in Table 1. All regressions include year dummies (not
reported). The t-statistics are based on robust standard errors
and are reported in parentheses below the parameter estimates.
Panel B reports firm fixed effects regressions on subsamples
based on Q. All regressions include year dummies (not reported).
The t-statistics testing the null that the coefficients for a
subsample do not differ significantly from zero are given in
parentheses below the parameter estimates. The t-statistics
testing the null that the coefficients for high-Q firms do not
differ significantly from the corresponding coefficients for low-Q
firms are given in square brackets.

Panel A. Full Sample Regressions

Independent             Dependent Variables
Variables
                             (1) I/K

Variables              (1)            (2)

FCF                   0.097 ***      0.094 ***
                     (5.09)         (4.85)
Leverage             -0.166 ***     -0.172 ***
                    (-6.80)        (-6.85)
FCF x HLEV

FCF x OPTSHR         -0.030 ***     -0.028 **
                    (-2.73)        (-2.42)
FCF x Size           -0.006 **      -0.006 **
                    (-2.28)        (-2.15)
Sales/K               0.010 ***      0.010 ***
                     (8.62)         (8.57)
Q                     0.057 ***      0.060 ***
                   (-17.53)       (-16.25)
Cash/K                0.043 ***      0.044 ***
                    (-9.40)        (-9.44)
OPTSHR                              -0.005
                                   (-0.44)
Size                                -0.009 *
                                   (-1.65)
Within [R.sup.2]      0.30           0.30
Observations         12,231         12,231

Independent            Dependent Variables
Variables
                         (2) Total I/K

Variables               (3)            (4)

FCF                   0.095 ***      0.091 ***
                     (4.91)         (4.64)
Leverage             -0.175 ***     -0.182 ***
                    (-6.82)        (-6.84)
FCF x HLEV            0.011          0.012
                     -1.44          -1.51
FCF x OPTSHR         -0.030 ***     -0.028 **
                    (-2.75)        (-2.43)
FCF x Size           -0.006 **      -0.006 **
                    (-2.29)        (-2.15)
Sales/K               0.010 ***      0.010 ***
                     (8.49)         (8.42)
Q                     0.058 ***      0.060 ***
                   (-17.58)       (-16.23)
Cash/K                0.044 ***      0.044 ***
                    (-9.43)        (-9.48)
OPTSHR                              -0.005
                                   (-0.48)
Size                                -0.009 *
                                   (-1.74)
Within [R.sup.2]     0.30            0.30
Observations         12,231         12,231

Independent            Dependent Variables
Variables
                     Total I/K      Resid I/K

Variables               (5)            (6)

FCF                   0.093 ***      0.098 ***
                     (4.91)         (5.11)
Leverage             -0.222 ***     -0.162 ***
                    (-5.58)        (-6.45)
FCF x HLEV

FCF x OPTSHR         -0.030 ***     -0.026 **
                    (-2.71)        (-2.26)
FCF x Size           -0.006 **      -0.007 ***
                     (0.00)         (0.00)
Sales/K               0.023 ***      0.010 ***
                     (9.67)         (8.50)
Q                     0.128 ***
                   (-14.23)
Cash/K                0.162 ***      0.042 ***
                   (-14.49)        (-9.14)
OPTSHR                0.012         -0.004
                     (0.49)        (-0.40)
Size                 -0.083 ***     -0.022 ***
                    (-6.97)        (-4.59)
Within [R.sup.2]      0.28           0.12
Observations         12,231         12,231

Panel B. Effects of Leverage and Option Compensation
on Subsamples Based on Growth Opportunity

                          Independent
                           Variables

                            (1) I/K

                       Low Q          High Q

FCF                  0.084 ***       0.090 ***
                    (3.40)          (3.49)
                                    [0.13]

Leverage            -0.221 ***      -0.090 **
                   (-7.43)         (-2.19)
                                    [5.34] ***

FCF x OPTSHR        -0.045 ***      -0.01
                   (-5.43)         (-0.71)
                                    [2.95] ***

FCF x Size          -0.003          -0.008 **
                   (-1.23)         (-2.23)
                                   [-1.71] *

Sales/K              0.009 ***       0.014 ***
                    (6.14)          (6.31)
                                    [1.26]

Q                    0.124 ***       0.056 ***
                   (10.37)         (12.54)
                                   [-4.90] ***

Cash/K               0.010           0.055 ***
                    (1.06)          (9.17)
                                    [5.51] ***

OPTSHR               0.009          -0.018
                    (0.81)         (-0.91)
                                   [-1.51]

Size                 0.002          -0.036 ***
                    (0.37)         (-3.30)
                                   [-2.81] ***

Within [R.sub.2]     0.21           0.28
Observations         6,115          6,116

                           Dependent
                           Variables

                        (2) Total I/K

                     Low Q          High Q

FCF                  0.053 ***       0.026
                    (2.81)          (0.80)
                                   [-0.08]

Leverage            -0.236 ***      -0.230 ***
                   (-7.21)         (-3.24)
                                    [3.30] ***

FCF x OPTSHR        -0.056 ***      -0.005
                   (-4.67)         (-0.25)
                                    [2.16] **

FCF x Size           0.004          -0.001
                    (0.39)         (-0.07)
                                   [-1.14]

Sales/K              0.013 ***       0.035 ***
                    (4.66)          (7.62)
                                    [1.28]

Q                    0.145 ***       0.140 ***
                    (7.57)         (12.41)
                                   [-5.26] ***

Cash/K               0.056 ***       0.191 ***
                    (2.94)         (12.93)
                                    [6.29] ***

OPTSHR               0.028           0.001
                    (1.43)          (0.02)
                                    [0.18]

Size                -0.012          -0.202 ***
                   (-1.16)         (-7.14)
                                   [-2.64] ***

Within [R.sub.2]     0.20            0.30
Observations         6,115           6,116

                            Dependent
                            Variables

                          (3) Resid I/K

                     Low Q          High Q

FCF                  0.015           0.092 ***
                    (0.42)          (3.62)
                                    [2.03] **

Leverage             -0.224 ***     -0.075 *
                    (-7.50)        (-1.83)
                                    [5.09] ***

FCF x OPTSHR         -0.046 **      -0.006
                    (-1.96)        (-0.45)
                                    [1.98] **

FCF x Size           0.007          -0.009 **
                    (1.27)         (-2.64)
                                   [-2.90] ***

Sales/K              0.009 ***      0.013 ***
                    (6.07)         (6.04)
                                   [0.46]

Q

Cash/K               0.012          0.055 ***
                    (1.32)         (9.29)
                                   [4.28] ***

OPTSHR               0.011         -0.019
                    (0.94) ***    (-0.94)
                                  [-2.14] **

Size                 0.014        -0.062 ***
                    (2.68)       (-6.80)
                                  [4.61] ***

Within [R.sub.2]     0.16          0.14
Observations         6,115         6,116

* Significant at the 0.01 level.

** Significant at the 0.05 level.

*** Significant at the 0.10 level.

Table IV. Single-Equation Model Including Leverage and Incentive

Interaction Variables

This table reports the fixed effects single-equation model
including leverage and incentive interaction variables. All
regressions include year dummies (not reported). The t-statistics
are based on robust standard errors. The t-statistics
testing the null that the coefficients do not differ
significantly from zero are given in parentheses below the
parameter estimates and the t-statistics testing the null that
the coefficients for high-Q firms do not differ significantly
from the corresponding coefficients for low-Q firms are given in
square brackets (to conserve space t-statistics on control
variables are not reported).

                         Dependent Variable
Independent
Variables                   Full Sample

FCF                   0.092 ***       0.094
                     (4.76)          (4.88) ***

Leverage             -0.280 ***      -0.175
                    (-8.17)         (-6.86) ***

Leverage              0.208 ***
x OPTSHR             (4.24)

FCF x OPTSHR         -0.025 **       -0.032
                    (-2.18)         (-2.62) **

FCF x OPTSHR                          0.036
x Leverage                           (1.29)

FCF x Size           -0.006 **       -0.006
                    (-2.14)         (-2.17) **

Sales/K               0.010 ***       0.010
                      8.56           (8.51) ***

Q                     0.059 ***       0.060
                    (16.13)         (16.25) ***

Cash/K                0.044 ***       0.044
                     (9.50)          (9.47) ***
OPTSHR               -0.051 ***      -0.006
                    (-2.97)         (-0.57)
Size                 -0.008          -0.009
                    (-1.55)         (-1.71) *
Within [R.sup.2]      0.30            0.30
Observations         12,231          12,231

                       Dependent Variable
Independent
Variables            Low Q           High Q

FCF                   0.085           0.089 ***
                     (3.42) ***      (3.45)
                                     [0.16]

Leverage             -0.256          -0.203 ***
                    (-6.43) ***     (-3.61)
                                     [4.10] ***

Leverage              0.070           0.208 **
x OPTSHR             (1.24)          (2.54)
                                     [0.85]

FCF x OPTSHR         -0.045          -0.008
                    (-4.93) ***     (-0.57)
                                     [2.62] **

FCF x OPTSHR
x Leverage

FCF x Size           -0.003          -0.008 **
                    (-1.23)         (-2.23)

Sales/K               0.009           0.014 ***
                     (6.14) ***      (6.31)

Q                     0.124           0.055 ***
                    (10.36) ***     (12.41)

Cash/K                0.010           0.055 ***
                     (1.07)          (9.18)
OPTSHR               -0.01           -0.051 **
                    (-0.48)         (-2.03)
Size                  0.002          -0.035 ***
                    (-0.33)         (-3.18)
Within [R.sup.2]      0.21            0.29
Observations         6,115           6,116

                       Dependent Variable

Independent
Variables            Low Q           High Q

FCF                   0.084 ***       0.090-
                     (3.48)          (3.50)
                                     [0.15]

Leverage             -0.240 ***      -0.090 **
                     (7.82)          (2.17)
                                     [5.67] ***

Leverage
x OPTSHR

FCF x OPTSHR         -0.085 **       -0.01
                    (-2.56)         (-0.68)
                                     [2.58] **

FCF x OPTSHR          0.140 **       -0.003
x Leverage           (2.53)         (-0.08)
                                    [-2.72] **

FCF x Size           -0.003          -0.008 **
                    (-1.23)         (-2.22)

Sales/K               0.009 ***       0.014 ***
                     (6.14)           6.31

Q                     0.126 ***       0.056-
                    (10.64)         (12.44)

Cash/K                0.009           0.055 ***
                     (0.97)          (9.16)
OPTSHR                0.008          -0.018
                     (0.70)         (-0.89)
Size                  0.001          -0.036 ***
                    (-0.09)         (-3.29)
Within [R.sup.2]      0.22            0.29
Observations         6,115           6,116

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

* Significant at the 0.10 level.

Table V. Substitutability of Leverage and Option Compensation:
Nonlinear Two-Stage Least Squares Estimates The table reports
firm fixed effects regression results for the simultaneous
equation model of CEO's option value scaled by his or her total
equity wealth (OPTSHR), (book) Leverage, and Investment
variables, estimated by nonlinear two-stage least squares.
Variables are defined as in Table 1. All regressions include
year effects (not reported). Coefficients on cross-products
are not reported. The t-statistics are based on robust standard
errors and are reported in parentheses below the parameter
estimates.

                            System (1)

Independent            Dependent Variables

Variables         I/K        Leverage      OPTSHR

FCF             0.202 ***
               (5.69)

Leverage       -0.569 ***                -0.077
              (-3.00)                   (-0.42)

OPTSHR                      -0.262 ***
                           (-4.57)

FCF x OPTSHR   -0.285 ***
              (-4.77)

I/K (Total                  -0.143 ***    0.080 *
I/K, or                    (-3.99)       (1.68)
Resid I/K)

FCF x Size     -0.0003
              (-0.07)

Sales/K         0.010 ***
               (6.43)

Cash/K          0.040 ***
               (7.59)

Z-score                     -0.035 ***
                           (-7.47)

ROA                          0.125 ***
                            (2.72)
Dependent Variables

Stock Risk                               -0.003
                                        (-0.04)

Cash Comp                                 0.272 ***
                                         (4.46)

Size           -0.009       -0.016 ***    0.002
              (-1.21)      (-5.34)       (0.36)

Q               0.058 ***    0.003       -0.014 ***
              (12.53)       (1.13)      (-3.92)

Observations   12,738       12,738       12,738

                            System (2)

Independent             Dependent Variables

Variables      Total I/K      Leverage      OPTSHR

FCF             0.204 ***
               (2.93)

Leverage       -0.593 ***                   -0.17
              (-2.71)                      (-1.54)

OPTSHR                       -0.320 ***
                            (-5.94)

FCF x OPTSHR   -0.319 ***
              (-2.88)

I/K (Total                   -0.017 **      0.035 **
I/K, or                     (-2.21)        (2.35)
Resid I/K)

FCF x Size     -0.002
              (-0.31)

Sales/K         0.022 ***
               (7.84)

Cash/K          0.164 ***
              (14.61)

Z-score                      -0.045 ***
                            (-9.42)

ROA                           0.245 ***
                             (5.42)

Stock Risk                                  -0.035
                                           (-0.45)

CashComp                                     0.259 ***
                                            (4.30)

Size           -0.088 ***    -0.014 ***      0.005
              (-6.43)       (-4.97)         (1.15)

Q               0.131 ***    -0.002         -0.011 ***
              (13.26)       (-0.78)        (-4.00)

Observations   12,738        12,738         12,738

                            System (3)

Independent             Dependent Variables

Variables      Resid 11K     Leverage     OPTSHR

FCF             0.162 ***
               (5.40)

Leverage       -0.201 **                  -0.129
              (-2.02)                    (-1.17)

OPTSHR                      -0.334 ***
                           (-6.11)

FCF x OPTSHR   -0.185 ***
              (-3.87)

I/K (Total                  -0.084 ***     0.069
I/K, or                    (-3.60)        (1.46)
Resid I/K)

FCF x Size     -0.004
              (-1.06)

Sales/K         0.010 ***
               (7.43)

Cash/K          0.042 ***
               (8.89)

Z-score                     -0.043 ***
                           (-8.89)

ROA                          0.267 ***
                            (5.69)

Stock Risk                                -0.034
                                         (-0.42)

CashComp                                   0.261
                                          (4.31)

Size           -0.023 ***   -0.014 ***     0.004
              (-4.38)      (-5.11)        (0.52)

Q                           -0.005 ***    -0.006 **
                           (-3.07)       (-2.47)

Observations   12,738       12,738        12,738

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

* Significant at the 0.10 level.

Table VI. Substitutability of Leverage and Option Compensation:
Subsample Analysis

This table reports firm fixed effects regression results for the
simultaneous equation model of CEO's option value scaled by his
or her total equity wealth (OPTSHR), (book) Leverage, and
Investment (1-K), estimated by nonlinear two-stage least squares
on subsamples stratified based on growth opportunity. Variables
are defined as in Table 1. All regressions include year effects
(not reported). Coefficients on cross-products are not reported.
The t-statistics are based on robust standard errors and are
reported in parentheses below the parameter estimates.

Independent                 Dependent Variables
Variables
                                  Low Q

                 I/K            Leverage       OPTSHR

FCF               0.209 ***
                 (3.40)

Leverage         -0.554 ***                    -0.628 ***
                (-2.81)                       (-2.87)

OPTSHR                           -0.308 ***
                                (-4.31)

FCF x OPTSHR     -0.320 **
                (-2.57)

I/K                               0.007        -0.042
                                 (0.09)       (-0.47)

FCF x Size       -0.0001
                (-0.06)

Sales/K           0.009 ***
                 (3.77)

Cash/K            0.003
                 (0.22)

Z-score                          -0.045 ***
                                (-6.97)

ROA                               0.079
                                 (0.78)

Stock Risk                                      0.087
                                               (0.67)
Cash Comp                                       0.338 ***
                                               (3.38)

Size             -0.0003         -0.013 **     -0.013
                (-0.03)         (-2.56)       (-1.55)

Q                 0.131 ***      -0.001         0.009
                 (5.81)         (-0.10)        (0.43)

Observations     6,369           6,369          6,369

Independent                Dependent Variables
Variables
                                 High Q

                 I/K            Leverage       OPTSHR

FCF               0.118 ***
                 (2.89)

Leverage         -0.04                         0.718 **
                (-0.14)                       (2.10)

OPTSHR                           -0.094
                                (-1.30)

FCF x OPTSHR     -0.099 *
                (-1.70)

I/K                              -0.051        0.149 ***
                                (-1.58)       (2.91)

FCF x Size       -0.0009
                (-1.32)

Sales/K           0.015 ***
                 (6.92)

Cash/K            0.056 ***
                 (9.43)

Z-score                          -0.036 ***
                                (-5.01)

ROA                               0.060
                                 (1.10)

Stock Risk                                    -0.246 *
                                             (-1.73)

Cash Comp                                      0.143
                                              (1.49)

Size             -0.042 ***      -0.017 ***    0.018 *
                (-3.29)         (-3.07)       (1.68)

Q                 0.062 ***      -0.0002      -0.017 ***
                (11.53)         (-0.06)      (-3.87)

Observations     6,369           6,369         6,369

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

* Significant at the 0.10 level.

Table VII. Economic Significance of the Estimates

This table displays the change in the investment rate as leverage
increases by one standard deviation, calculated as the regression
coefficient of leverage times the standard deviation of leverage,
and the change in the investment cash flow sensitivity as OPTSHR
or Size increases by one standard deviation, calculated as the
regression coefficient on cash flow interacted with OPTSHR or
Size variable times the standard deviation of OPTSHR or Size. The
economic significance of the single/equation model is calculated
using regression coefficients from Model 2 of Table III Panel A
(full sample) and Model 1 of Panel B (subsamples). The economic
significance of the 2SLS model is calculated from coefficient
estimates in System 1 of Table V (full sample) and Table VI
(subsamples). The average sensitivity is calculated as the sum of
regression coefficient of the cross/term of a variable (including
Constant 1) and cash flows, times the mean value of the variable,
that is, the mean value of ([[alpha].sub.1] x 1 + [[alpha].sub.3]
x OPTSHR + [[alpha].sub.Size] x Size). The numbers in parentheses
measure the effect of leverage or OPTSHR on AI or A[partial derivative]
I/[partial derivative]FCF relative to the mean investment or investment
cash flow sensitivity.

[DELTA]/ or
[partial
derivatives]      Change in Investments Rate ([DELTA]/
[DELTA]           and Investment Cash Flow Sensitivity
[parallel]        ([partial derivatives][DELTA][parallel]
[partial          [DELTA][partial derivatives]FCF
derivatives]
FCF                      Single Equation
Resulting from
One Std Dev       Full
Increase in:      Sample    Low Q       High Q

Leverage          -0.029    -0.034      -0.014
                  (-9%)     (-16%)      (-4%)
OPTSHR            -0.009    -0.014      -0.003
                  (-24%)    (-34%)      (-11%)
Size              -0.01     -0.005      -0.013
                  (-26%)    (-12%)      (-48%)

Average Investment Rate and Investment
Cash Flow Sensitivity

Avg I/K           0.306     0.217       0.395
Avg sensitivity   0.038     0.041       0.027

[DELTA]/ or
[partial
derivatives]
[DELTA]
[parallel]
[partial
derivatives]
FCF                      2SLS Estimates
Resulting from
One Std Dev       Full
Increase in:      Sample    Low Q       High Q

Leverage          -0.097    -0.097      -0.007
                  (-31%)    (-44%)      (-2%)
OPTSHR            -0.095    -0.102      -0.034
                  (-170%)   (227%)      (-55%)
Size              -0.0005   -0.0002     -0.001
                  (-0.9%)   (0.4%)      (2%)

Average Investment Rate and Investment
Cash Flow Sensitivity

Avg I/K           0.316     0.221       0.410
Avg sensitivity   0.056     0.045       0.062

Table VIII. Regressions of the Capital Structure Changes

This table reports firm fixed effects of the capital structure
change regressions. The dependent variables are change in
leverage [DELTA]Lev, net debt issuance NetDlss, and net equity
issuance NetElSS, respectively. Here, NetDlss equals long-term
debt issuance (Debtlss) minus long-term debt reduction (DebtBack)
plus change in current liabilities; Debtlss and DebtBack are also
used as dependent variable; FCF is the free cash flow, defined as
net income after tax plus depreciation less common and preferred
dividends; Control variables correspond to the prior fiscal
year-end; Leverage is the book leverage; LogTA is the logarithm of
total assets; MTB is the market-to-book ratio; Z-score is the
modified Altman's Z-score = [3.3 x (operating income after
depreciation) + sales + 1.4 x (retained earnings) + 1.2 x (current
assets--current liability)]-total assets; and Stock return is the
annual stock returns. All regressions include year dummies (not
reported). The t-statistics are based on robust standard errors
and are reported in parentheses below the parameter estimates.
Panel B reports firm fixed effects of capital structure change
regressions on subsamples based on growth opportunity. The
t-statistics testing the null that the coefficients for a subsample
do not differ significantly from zero are given in parentheses
below the parameter estimates and the t-statistics testing the
null that the coefficients for high-Q firms do not differ
significantly from the corresponding coefficients for low-Q firms
are given in square brackets.

Panel A. Regressions on Full Sample

                   Dependent Variable

Independent
Variables          [DELTA] Lev          NetDlss       Debtlss

FCF                 0.075 ***            0.048         0.046
                   (1.92)               (1.39)        (1.30)

Leverage           -0.509 ***           -0.364 ***    -0.013
                  (18.05)             (-18.21)       (-0.47)

LogTA               0.032 ***           -0.017 ***    -0.020 ***
                   (5.36)              (-3.80)       (-3.28)

MTB                -0.002                0.003 **      0.001
                  (-1.07)               (2.23)        (1.15)

Z-score            -0.007               -0.015        -0.016
                  (-0.54)              (-1.26)       (-1.38)

Stock Return       -0.007 **            -0.0001        0.003
                  (-2.46)              (-0.06)        (0.90)

Within [R.sub.2]    0.11                 0.11          0.02

Observations       11,073               11,073        11,073

                   Dependent Variable

Independent
Variables          DebtBack             NetElss

FCF                -0.002               -0.026 **
                  (-0.12)              (-1.98)

Leverage            0.304 ***            0.145 ***
                 (-13.14)              (-9.12)

LogTA              -0.004               -0.049 ***
                  (-0.82)             (-14.88)

MTB                -0.001                0.004 ***
                  (-1.47)               (3.57)

Z-score            -0.001               -0.008 ***
                  (-0.39)              (-3.14)

Stock Return        0.003                0.007 ***
                   (0.93)               (3.05)

Within [R.sub.2]    0.03                 0.12

Observations       11,073               11,073

Panel B. Capital Structure Change Regressions on Subsamples
Divided Based on Growth Opportunity

Independent        Dependent Variable
Variables

                   [DELTA] Lev
                   Low Q                High Q

FCF                 0.031 **              0.018 *
                   (2.32)               (-1.92)
                                        [-2.00] **

Leverage           -0.491 ***            -0.504 ***
                 (-17.42)               (-8.65)
                                         [2.58] **

LogTA               0.002                 0.055 ***
                  (-0.30)               (-5.09)
                                         [5.96] ***

MTB                 0.042 ***            -0.001
                   (3.99)               (-0.84)
                                        [-4.43] ***

Z-score             0.009 **             -0.016
                   (2.15)               (-0.85)
                                        [-1.22]

Stkret             -0.012 **             -0.009 **
                  (-2.49)               (-2.15)
                                        [-0.19]

Within [R.sub.2]    0.20                  0.10

Observations       5,536                 5,537

Independent        Dependent Variable
Variables

                   NetDlss
                   Low Q                High Q

FCF                 0.024 **            -0.008
                  (-2.04)               (0.00)
                                       [-1.86] *

Leverage           -0.400 ***           -0.320 ***
                 (-22.98)              (-7.46)
                                        [4.99] ***

LogTA              -0.015 ***           -0.020 **
                  (-3.80)              (-2.39)
                                        [5.05] ***

MTB                 0.055 ***            0.003 **
                  (-7.59)               (2.08)
                                       [-4.10] ***

Z-score             0.004               -0.023
                  (-1.64)              (-1.25)
                                       [-1.39]

Stkret             -0.007 **            -0.004
                  (-2.05)              (-1.35)
                                       [-0.49]

Within [R.sub.2]    0.18                 0.09

Observations        5,536                5,537

Independent        Dependent Variable
Variables

                   Debtlss
                   Low Q                High Q

FCF                 0.033 *              -0.012
                  (-1.71)               (-0.41)
                                        [-1.24]

Leverage            0.019                 0.024
                   (0.47)                (0.32)
                                         [1.76] *

LogTA              -0.014                -0.027 **
                   (0.00)                (2.08)
                                        [-2.22] **

MTB                 0.079 ***             0.001
                  (-4.09)                (0.55)
                                        [-4.35] ***

Z-score             0.0004               -0.024
                  (-0.07)               (-1.33)
                                        [-1.43]

Stkret             -0.005                 0.002
                   (0.00)                (0.56)
                                         [0.64]

Within [R.sub.2]   0.02                   0.02

Observations       5,536                 5,537

Independent        Dependent Variable
Variables

                   DebtBack
                   Low Q                High Q

FCF                 0.011                -0.004
                   (0.55)               (-0.18)
                                        [-0.40]

Leverage            0.366 ***             0.270 ***
                   (8.45)                (9.42)
                                        [-1.94] *

LogTA              -0.003                -0.004
                  (-0.26)               (-0.71)
                                        [-0.90]

MTB                 0.030                -0.002 *
                   (1.54)               (-1.96)
                                        [-1.67] *

Z-score            -0.003                -0.001
                  (-0.50)               (-0.43)
                                         [0.02]

Stkret             -0.002                 0.004
                  (-0.17)                (1.20)
                                         [0.77]

Within [R.sub.2]    0.03                  0.05

Observations        5,536                5,537

Independent        Dependent Variable
Variables

                   NetElss
                   Low Q                High Q

FCF                -0.007                -0.070 **
                  (-0.06)               (-2.28)
                                        [-2.63] ***

Leverage            0.091 ***             0.184 ***
                   (8.07)                (6.29)
                                         [2.44] **

LogTA              -0.017 ***            -0.075 ***
                  (-6.40)              (-12.68)
                                        [-3.37] ***

MTB                 0.013 ***             0.004 ***
                   (2.97)               (-3.10)
                                         [2.25] **

Z-score            -0.005 *              -0.006 *
                  (-1.84)               (-1.91)
                                        [-0.33]

Stkret              0.005 **              0.005 *
                   (2.69)               (-1.73)
                                        [-0.40]

Within [R.sub.2]    0.08                   0.15

Observations       5,536                  5,537

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

* Significant at the 0.10 level.

Table IX. Probit Models of Stock Options and
Restricted Stock Grants

This table reports the results for the probit model of the stock
options grants and restricted stock grants, which estimate the
probability that a CEO is granted stock option or restricted
stock. The dependent variables are OG and RS that equal 1 if a
CEO is granted stock options or restricted stocks in that year,
respectively. Variables are defined as in Table I. Model 1
reports results of pooled probit model. Model 2 reports results
of random effects probit models. All regressions include year
dummies (not reported). The Z-statistics are in parentheses.
Panel B presents results for the probit model of the stock
options grants and restricted stock grants on subsamples based on
growth opportunity. Model 1 reports results of the pooled probit
model. Model 2 reports results of the random effects probit
models. The Z-statistics testing the null that the coefficients
for a subsample do not differ significantly from zero are in
parentheses, and the Z-statistics testing the null that the
coefficients for high-Q firms do not differ significantly from
the corresponding coefficients for low-Q firms are given in
square brackets.

Panel A. Regressions on Full Sample

                             Dependent Variable

OG RS                         (1)           (2)
                             Pooled         RE
                             Probit       Probit

FCF                         -0.0003       0.088
                           (-0.00)       (0.70)

Leverage                    -0.171 *     -0.525 ***
                           (-1.78)      (-3.76)

LogTA                        0.052 ***    0.086 ***
                            (3.99)       (4.19)

MTB                          0.015 *      0.015
                            (1.78)       (1.29)

Stkret                      -0.025       -0.039
                           (-1.16)      (-1.56)

Risk                        -0.003       -0.163 **
                            (0.00)      (-2.63)

CashComp                     0.158 ***    0.125 ***
                            (7.04)       (5.90)

Tenure                      -0.027 ***   -0.034 ***
                          (-15.75)     (-12.53)

2-digit SIC dummies        Yes          No

Year dummies               Yes          Yes

Goodness-of-fit test

Pseudo [R.sub.2]             0.08

Wald chi-square            930.62       365.4

(p > chi-square)            (0.000)      (0.000)

Observations               11,073       11,073

                               Dependent Variable

OG RS                         (1)           (2)
                             Pooled         RE
                             Probit       Probit

FCF                        -0.024        0.054
                          (-0.23)       (0.31)

Leverage                    0.519 ***    0.640 ***
                           (4.94)       (3.55)

LogTA                       0.069 ***    0.150 ***
                           (4.63)       (5.64)

MTB                        -0.030 **    -0.059 ***
                          (-1.96)      (-2.99)

Stkret                     -0.024        0.008
                          (-0.86)       (0.22)

Risk                       -0.471 ***    0.482 ***
                          (-9.73)      (-6.20)

CashComp                    0.102 ***    0.117 ***
                           (3.08)       (3.85)

Tenure                     -0.022 ***   -0.033 ***
                         (-10.24)      (-8.93)

2-digit SIC dummies        Yes          No

Year dummies               Yes          Yes

Goodness-of-fit test


Pseudo [R.sub.2]           0.11

Wald chi-square            1,168.13     486.4

(p > chi-square)           (0.000)      (0.000)

Observations               11,073       11,073

Panel B. Probit Model of Stock Option Grants on
Subsamples Based on Growth Opportunity

                           Dependent Variable: OG

Independent                  (1) Pooled Probit
Variables

FCF                         0.200 *        -0.088
                           (1.72)         (-0.60)
                                          [-3.84] ***

Leverage                   -0.338 **        0.025
                            (-2.27)        (0.18)
                                           [1.84] *

LogTA                       0.057 ***       0.043 **
                           (2.92)          (2.42)
                                          [-0.711

MTB                         0.322 ***      -0.004
                           (3.90)         (-0.46)
                                          [-2.88] ***

Stkret                     -0.183 ***     -0.004
                          (-3.88)        (-0.16)
                                          [2.21] **

Risk                       -0.085         0.152 **
                          (-1.26)        (-2.21)
                                          [2.19] **

CashComp                    0.262 ***      0.108 ***
                           (7.07)         (4.41)
                                          [-0.14]

Tenure                     -0.022 ***     -0.033 ***
                          (-8.81)        (-13.52)
                                          [-2.86] ***

2-digit SIC dummies        Yes            Yes

Year dummies               Yes            Yes

Goodness-of-fit test

Pseudo [R.sub.2]           0.09           0.10

Wald chi-square          551.10         548.05

(p > chi-square)          (0.0000)       (0.0000)

Observations               5,536          5,537

                             Dependent Varialble: OG

Independent                     (2) RE Probit
Variables

FCF                         0.322 **       -0.087
                           (1.95)         (-0.43)
                                          [-2.31] **

Leverage                   -0.520 ***      -0.331
                          (-2.69)         (-1.63)
                                           [1.24]

LogTA                       0.088 ***       0.074 **
                           (3.22)          (2.44)
                                          [-0.06]

MTB                         0.328 ***      -0.001
                           (3.25)         (-0.62)
                                          [-1.95] *

Stkret                     -0.186 ***      -0.019
                          (-3.55)         (-0.62)
                                           [1.89] *

Risk                        0.051          -0.034
                          (-0.77)          (0.00)
                                          [-1.86] *

CashComp                    0.255 ***       0.089-
                           (5.68)          (3.33)
                                          [-0.63]

Tenure                     -0.028 ***      -0.039 ***
                          (-7.98)         (-9.49)
                                          [-1.44]

2-digit SIC dummies        No              No

Year dummies               Yes             Yes

Goodness-of-fit test

Pseudo [R.sub.2]

Wald chi-square            164.60        206.50

(p > chi-square)           (0.0000)      (0.0000)

Observations               5,536         5,537

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

* Significant at the 0.10 level.

Table X. 2SLS Estimation on the Adjustment of Capital
Structure and Compensation Structure

This table presents firm fixed effects regression results for the
simultaneous equations model of the change in leverage ([DELTA]Lev) or
net debt issuance (NetDlss), and Black-Scholes value of option
grants to total current compensation (OptG), estimated by the
two-stage least squares. All instrument variables are lagged one
period. All regressions include year dummies (not reported). The
t-statistics are based on robust standard errors and are reported
in parentheses below the parameter estimates.

Panel A. Estimates on Full Sample

                          Dependent Variables
Independent
Variables                 [DELTA]Lev      OptG

[DELTA]Lev (NetDlss)                     -10.010 **
                                         (-1.94)

OptG                      -0.175 ***
                         (-5.85)

FCF                        0.138 ***       1.275 **
                         (13.77)          (1.98)

Leverage                  -0.137 ***      -1.294 **
                        (-13.21)         (-2.12)

Z-score                   -0.008
                         (-0.70)

Stock Risk                                -0.262
                                         (-1.29)

CashComp                                  -0.040
                                         (-1.46)

Tenure                                    -0.003
                                         (-1.54)

LogTA                      0.017 ***       0.138 ***
                         (13.23)         (-2.60)

MTB                       -0.001          -0.014
                         (-0.50)         (-0.77)

Adjusted [R.sup.2]         0.06            0.00

Observations              11,073          11,073

                          Dependent Variables
Independent
Variables                 NetDlss         OptG

[DELTA]Lev (NetDlss)                      -9.195 ***
                                         (-2.77)

OptG                      -0.045 **
                         (-2.12)

FCF                       -0.010          -0.056
                         (-1.38)         (-0.73)

Leverage                  -0.075 ***      -0.747 ***
                        (-10.08)         (-3.16)

Z-score                    0.012
                         (-1.42)

Stock Risk                                 0.066 *
                                         (-1.74)

CashComp                                  -0.006
                                         (-0.48)

Tenure                                    -0.003 *
                                         (-1.67)

LogTA                     -0.0001          0.023 **
                         (-0.06)         (-2.31)

MTB                        0.004 ***       0.004 ***
                         (-5.56)         (-4.18)

Adjusted [R.sup.2]         0.03            0.01

Observations              11,073          11,073

Panel B. 2SLS on Estimation on the Adjustment of Capital
Structure and Compensation Structure on Subsamples
Based on Growth Opportunity

Independent               Dependent Variables
Variables
                          Low Q

                          [DELTA]
                          Lev             OptG

[DELTA]Lev (NetDlss)                      -4.218 *
                                         (-1.85)
OptG                      -0.195 ***
                         (-4.76)
FCF                        0.053 ***       0.219 *
                          (4.81)          (1.89)
Leverage                  -0.190 ***      -0.816 **
                        (-15.85)         (-2.17)
Z-score                    0.006
                         (-0.60)
Stock Risk                                -0.030
                                         (-0.45)
CashComp                                  -0.031 **
                                         (-2.06)
Tenure                                    -0.001
                                         (-0.67)
LogTA                      0.012 ***       0.063 ***
                         (-6.40)          (5.52)
MTB                        0.053 ***       0.240 ***
                         (-7.89)          (3.08)
Adjusted [R.sup.2]         0.08            0.03
Observations              5,536           5,536

Independent               Dependent Variables
Variables
                          High Q

                          [DELTA]
                          Lev               OptG

[DELTA]Lev (NetDlss)                       -19.092
                                           (-0.75)
OptG                       -0.079 **
                          (-2.18)
FCF                         0.189 ***        3.480
                          (11.57)          (-0.76)
Leverage                   -0.061 ***       -0.120 ***
                          (-3.72)          (-0.75)
Z-score                    -0.0003
                          (-0.02)
Stock Risk                                 -0.393
                                          (-0.49)
CashComp                                   -0.092
                                          (-0.68)
Tenure                                     -0.010
                                          (-1.19)
LogTA                       0.020 ***       0.354
                          (10.43)          (0.84)
MTB                        -0.005 ***      -0.075
                          (-3.79)         (-0.66)
Adjusted [R.sup.2]          0.08            0.00
Observations               5,537            5,537

Independent               Dependent Variables
Variables
                          Low Q

                          NetDlss         OptG

[DELTA]Lev (NetDlss)                      -3.113 **
                                         (-2.46)
OptG                      -0.078 **
                         (-2.35)
FCF                        0.024 ***       0.104 **
                         (-2.68)         (-2.10)
Leverage                  -0.130 ***      -0.500 ***
                        (-13.35)         (-3.18)
Z-score                    0.020 **
                          (2.32)
Stock Risk                                 0.065 ***
                                         (-3.59)
CashComp                                  -0.019 **
                                         (-2.29)
Tenure                                    -0.002 **
                                         (-2.13)
LogTA                      0.002           0.045 ***
                          (1.58)          (9.64)
MTB                        0.043 ***       0.219 ***
                          (7.92)          (4.31)
Adjusted [R.sup.2]         0.07            0.05
Observations              5,536           5,536

Independent               Dependent Variables
Variables
                          High Q

                          NetDlss         OptG

[DELTA]Lev (NetDlss)                      -7.149 ***
                                         (-2.75)
OptG                       0.030
                         (-1.21)
FCF                       -0.041 ***      -0.200
                         (-3.62)         (-1.64)
Leverage                  -0.015          -0.107
                         (-1.35)         (-1.19)
Z-score                    0.039 ***
                          (2.81)
Stock Risk                                 0.253 ***
                                         (-5.29)
CashComp                                  -0.010
                                         (-0.66)
Tenure                                    -0.005 ***
                                          (0.00)
LogTA                     -0.003 **        0.030 ***
                         (-2.51)          (2.75)
MTB                        0.001 *         0.018 ***
                         (-1.83)          (3.00)
Adjusted [R.sup.2]         0.01            0.01
Observations               5,537           5,537

*** Significant at the 0.01 level.

** Significant at the 0.05 level.

* Significant at the 0.10 level.

Table XI. Characteristics of Firms Predominantly Using Debt versus
Firms Predominantly Using Executive Stock Options

This table presents summary statistics for a group of 1,618 firms
with high (low) debt and low (high) OPTSHR based on the medians
of book leverage and OPTSHR in low-growth firms. Here, Stock
return is the annual stock return over the fiscal year, Monthly
volatility is the variance of 60 months of stock returns, and
Daily volatility is the variance of daily stock returns over the
fiscal year. Other variables are defined as in Table 1.

                               High Leverage/    Low Leverage/
                                Low OPTSHR        High OPTSHR

                             Mean    Median      Mean    Median

Sorting
Characteristics
 Leverage                     0.40     0.38       0.15     0.17
 OPTSHR                       0.25     0.24       0.80     0.80
Benefits-costs
determinants
 Free Cash Flow               0.33     0.19       0.28     0.26
 Market cap ($ million)      4,007      664      4,097      781
 Total assets ($ million)    6,188    1,197      5,067    1,092
 Q                            1.08     1.09       1.18     1.15
 Stock Return (%)             9.17     3.99      26.23    14.78
 Volatility (monthly %)       0.37     0.34       0.42     0.38
 Volatility (daily %)         0.08    0.055       0.10    0.061
 Cash Comp ($ thousand)      1,229      860      1,294      956
 Current Comp ($thousand)    2,956    1,571      4,126    2,254
 Tenure                       9.14     6.92       5.01     3.59

                                      Test for
                                    Differences

                             t-statistic      p-value

Sorting
Characteristics
 Leverage                    74.73 ***        < 0.0001
 OPTSHR                     -97.38 ***        < 0.0001
Benefits-costs
determinants
 Free Cash Flow               1.51            < 0.0001
 Market cap ($ million)      -0.13              0.0017
 Total assets ($ million)     1.24              0.0030
 Q                           -4.01 ***        < 0.0001
 Stock Return (%)            -6.27 ***        < 0.0001
 Volatility (monthly %)      -7.09 ***        < 0.0001
 Volatility (daily %)        -3.48 ***          0.0019
 Cash Comp ($ thousand)      -1.26            < 0.0001
 Current Comp ($thousand)    -5.07 ***        < 0.0001
 Tenure                      15.99 ***        < 0.0001

*** Significant at the 0.01 level.